Molecular Signatures of Cancer

Esther Edlundh-Rose

Royal Institute of Technology

School of Biotechnology

Stockholm 2005

© Esther Edlundh-Rose

School of Biotechnology

Department of Gene Technology

Royal Institute of Technology

AlbaNova University Centre

SE-106 91 Stockholm

Sweden

Printed at Universitetsservice US-AB

Box 700 14

SE-100 44 Stockholm

Sweden

ISBN 91-7178-348-2

Esther Edlundh-Rose (2006). Molecular Signatures of Cancer. School of Biotechnology, Royal Institute of Technology, Stockholm, Sweden. ISBN 91-7178-348-2

ABSTRACT

Cancer is an important public health concern in the western world, responsible for around 25% of all deaths. Although improvements have been made in the diagnosis of cancer, treatment of disseminated disease is inefficient, highlighting the need for new and improved methods of diagnosis and therapy. Tumours arise when the balance between proliferation and differentiation is perturbed and result from genetic and epigenetic alterations. Due to the heterogeneity of cancer, analysis of the disease is difficult and a wide range of methods is required. In this thesis, a number of techniques are demonstrated for the analysis of genetic, epigenetic and transcriptional alterations involved in cancer, with the purpose of identifying a number of molecular signatures. Pyrosequencing proved to be a valuable tool for the analysis of both point mutations and CpG methylation. Using this method, we showed that oncogenes BRAF and NRAS, members of the Ras-Raf-MAPK pathway, were mutated in 82% of melanoma tumours and were mutually exclusive. Furthermore, tumours with BRAF mutations were more often associated with infiltrating lymphocytes, suggesting a possible target for immunotherapy. In addition, methylation of the promoter region of the DNA repair gene MGMT was studied to find a possible correlation to clinical response to chemotherapy. Results showed a higher frequency of promoter methylation in non-responders as compared to responders, providing a possible predictive role and a potential basis for individually tailored chemotherapy. Microarray technology was used for transcriptional analysis of epithelial cells, with the purpose of characterization of molecular pathways of anti-tumourigenic agents and to identify possible target genes. Normal keratinocytes and colon cancer cells were treated with the antioxidant N-acetyl L-cysteine (NAC) in a time series and gene expression profiling revealed that inhibition of proliferation and stimulation of differentiation was induced upon treatment. ID-1, a secreted protein, was proposed as a possible early mediator of NAC action. In a similar study, colon cancer cells were treated with the naturally occurring bile acid ursodeoxycholic acid (UDCA) in a time series and analysed by microarray and FACS analysis. Results suggest a chemopreventive role of UDCA by G1 arrest and inhibition of cell proliferation, possibly through the secreted protein GDF15. These investigations give further evidence as to the diversity of cancer and its underlying mechanisms. Through the application of several molecular methods, we have found a number of potential targets for cancer therapy. Follow up studies are already in progress and may hopefully lead to novel methods of treatment. Key words: BRAF, NRAS, MGMT, methylation, pyrosequencing, microarray technology, gene expression analysis

I feel so extraordinary

Something’s got a hold on me

I get this feeling I’m in motion

A sudden sense of liberty

True Faith, New Order

List of publications

This thesis is based on the following publications:

1. Edlundh-Rose E, Kupershmidt I, Gustafsson AC, Parasassi T, Serafino A, Bracci-

Laudiero L, Greco G, Krasnowska EK, Romano MC, Lundeberg T, Nilsson P and Lundeberg

J. Gene expression analysis of human epidermal keratinocytes after N-acetyl L-cysteine

treatment demonstrates cell cycle arrest and increased differentiation. Pathobiology.

2005;72(4):203-12.

2. Gustafsson AC, Kupershmidt I, Edlundh-Rose E, Greco G, Serafino A, Krasnowska EK,

Lundeberg T, Bracci-Laudiero L, Romano MC, Parassassi T and Lundeberg J. Global gene

expression analysis in time series following N-acetyl L-cysteine induced epithelial

differentiation of human normal and cancer cells in vitro. BMC Cancer. 2005 Jul 7;5:75.

3. Bresso F, Edlundh-Rose E, D’Amato M, Are A, Grecius G, Lidén A, Sjöqvist U, Löfberg

R, Arulampalam V, Lundeberg J and Pettersson S. Investigating the chemopreventive role of

ursodeoxycholic acid in colorectal cancer cells. Manuscript.

4. Edlundh-Rose E, Egyházi S, Omholt K, Månsson-Brahme E, Hansson J, and Lundeberg

J. Lower age at diagnosis in melanomas with BRAF as compared to NRAS mutations.

Submitted.

5. Edlundh-Rose E*, Ma S*, Hansson J, Lundeberg J, Egyházi S. Hypermethylation status of

the MGMT promoter in melanoma tumours in relation to clinical response to chemotherapy.

Manuscript. *These authors contributed equally to this work

Contents

INTRODUCTION .............................................................................. 5

1. From peas to -omics ............................................................................5

2. Cancer .................................................................................................7

2.1 Carcinogenesis ................................................................................. 7

2.2 Epigenetics of cancer ...................................................................... 10

2.3 DNA repair and drug resistance ........................................................ 13

3. Mutation analysis ..............................................................................14

3.1 Sanger sequencing ......................................................................... 14

3.2 Single-strand conformation polymorphism, SSCP ................................ 15

3.3 Pyrosequencing .............................................................................. 15

4. Methylation analysis..........................................................................18

4.1 Methylation-specific PCR (MSP)......................................................... 18

4.2 Pyrosequencing .............................................................................. 19

5. Microarray Technology ......................................................................20

5.1 Microarray platforms ....................................................................... 21 5.1.1 GeneChip® technology.......................................................................... 21 5.1.2 Spotted arrays .................................................................................... 21

6 Gene expression analysis ...................................................................24

6.1 Experimental design........................................................................ 24

6.2 Data analysis ................................................................................. 25

6.3 Validation of array data ................................................................... 28

PRESENT INVESTIGATION ...............................................................31

7. Gene expression analysis of epithelial cell differentiation .................31

7.1 Gene expression profiling of N-acetyl L-cysteine (NAC) induced

differentiation of epithelial cells (Papers 1 and 2)...................................... 31

7.2 Investigating the chemopreventive role of ursodeoxycholic acid in

colorectal cencer cells (Paper 3) ............................................................. 32

8. Cutaneous melanoma ........................................................................34

8.1 NRAS and BRAF mutations in melanoma tumours in relation to clinical

characteristics – a study based on mutation screening by pyrosequencing

(Paper 4) ............................................................................................ 36 8.1.2 RAS-RAF-MEK-ERK pathway .................................................................. 36

8.2 Hypermethylation status of the MGMT promoter in melanoma tumours in

relation to clinical response to chemotherapy (Paper 5) ............................. 39 8.2.1 O6-methylguanine DNA-methyltransferase, MGMT .................................... 39

9. Concluding Remarks..........................................................................43

Acknowledgements ...............................................................................45

References ............................................................................................49

5

INTRODUCTION

1. From peas to -omics

Human genetics dates back to Darwin’s theory of evolution in 1856 and Mendel’s study of

garden peas in 1865, where it was observed that certain traits are inherited. It was Avery,

however, who in 1944 discovered that DNA is the material behind which genes and

chromosomes are made [1]. This paved the way for Watson and Crick’s work, which

described the double-helical structure of the DNA molecule [2], for which they won the Nobel

Prize in 1962. Francis Crick also postulated the “central dogma”, which describes how the

genetic information found in the DNA molecule is replicated as cells divide and translated

into functional proteins via transcription to messenger RNA (figure 1) [3]. These

revolutionizing discoveries have created a platform on which modern molecular and genetic

science is based.

Figure 1. The Central Dogma was postulated by Crick in 1958.

6

2001 saw the completion of the first draft of the human genome [4, 5], with the number of

genes estimated today at around 20000-25000 (http://www.ensembl.org/). With this came the

concept of “genomics” and an increased potential to identify and gain an understanding of

genes involved in human physiology and disease. We have now entered the “post-genomic

era”, from which has sprung a plethora of new terms: proteome, transcriptome, epigenome,

metabolome etc., along with a barrage of new techniques to study these. Microarray

technology has made it possible to analyze thousands of samples in parallel and has a variety

of applications [6]; the human proteome atlas is well under way and will eventually produce

antibodies toward all human proteins [7] (http://www.proteinatlas.org/) and antisense

technology and micro RNAs have given us a method for studying gene function [8].

There is much still to learn before we can characterize and cure all human disease, however

the means for doing so are growing and the amount of knowledge gained is increasing

rapidly. We may no longer need only to dream of what the future holds in store.

7

2. Cancer

Cancer has been recognized for thousands of years, the first description being found on

Egyptian papyri written between 3000-1500 BC. Hippocrates was the first to differentiate

between benign and malignant tumours and it was he who gave origin to the medical terms

that are still used today: oncos, meaning swelling in greek and carcinos, which means crab

(the cross section of a malignant tumour somewhat resembling a crab). Today cancer is

considered one of the major human diseases, responsible for around 25% of deaths in the

western world [9]. Thus, cancer prevention and control are major health issues and one of the

largest fields of scientific research.

2.1 Carcinogenesis

Carcinogenesis means literally the creation of cancer and describes the process of

transformation from normal to cancer cells. Cancer arises when the homeostatic balance of a

cell is disrupted, leading to an increase in cell proliferation. There is much evidence

suggesting that carcinogenesis is a multistep process, where several events lead to the growth

of a malignant tumour. These steps reflect genetic or epigenetic alterations, where normal

cells go through several pre-malignant states before resulting in invasive cancer. Tumour

development can be compared to Darwinian evolution, where a mutation, or series of

mutations, gives a cell a selective advantage over surrounding cells [10, 11]. Several types of

genetic alterations affecting cell growth have been identified, including point mutations,

aneuploidy, chromosome translocations and gene amplification. These alterations can produce

oncogenes with a dominant gain of function, or tumour suppressor genes with a recessive loss

of function [12]. Epigenetic alterations can also lead to inhibition of transcription and gene

silencing (reviewed below). Cancer is a very heterogenous disease, making it difficult to study

and get a good understanding of. The determinants of cancer are many and varied including

8

genetic predisposition, environmental influences, infectious agents, nutritional factors and

radiation, to name a few [13]. There are some typical molecular, biochemical and cellular

characteristics that are common in most human cancers. One of the main theories of

carcinogenesis is that the following 6 alterations are essential for the onset of cancer [14]:

Self sufficiency in growth signals

Normal cells require mitogenic growth signals in order to progress from quiescence to

proliferation. These signals are transmitted into the cell by transmembrane receptors that bind

distinct classes of signalling molecules: diffusible growth factors, extracellular matrix

components and cell-cell adhesion/interaction molecules. Tumour cells are self sufficient and

are not dependent on exogenous growth signals. Oncogenes are generally constitutively

expressed and are most often found in pathways leading to activation of transcription, thus the

tumour cell creates its own growth signal [15]. Cancer cells can also switch expression of

extracellular matrix receptors (integrins) to favour those that transmit pro-growth signals [16,

17] . Both these types of growth signal can lead to activation of the Ras-Raf-MAPK pathway

(see below) [17, 18].

Insensitivity to growth-inhibitory signals

Antigrowth signals maintain cellular quiescence and tissue homeostasis either by forcing cells

out of the cell cycle into the G0 phase, or inducing cells to enter a post mitotic differentiation

state. Differentiation is characterized by expression of integrins and adhesion molecules and is

often switched off in tumour cells [19].

The tumour suppressor gene p53 plays a central role in tumorigenesis and is often mutated in

cancer cells [20]. Inactivation of p53 renders the cell insensitive to anti-growth signals, thus

stimulating proliferation.

Evasion of apoptosis

Apoptosis is an organized, genetically programmed cell death, by which multicellular

organisms specifically destroy, dismantle and dispose of cells [21]. It is essential in both

embryonic and adult tissues to eliminate unwanted or potentially harmful cells and maintain

cellular homeostasis. Apoptosis can be triggered by a variety of factors such as DNA damage,

cellular stress, hypoxia or cytotoxic drugs. The Bcl-2 protein family as well as p53 and pRb

are key regulators of this process and mutations in these have been shown to play a

fundamental role in the ability of cancer cells to evade apoptosis [20, 22, 23].

9

Limitless replicative potential

Telomeres cap the ends of chromosomes and are essential for maintaining genomic integrity

and stability [24]. Telomeres are shortened with every replication, although this process is

delayed by telomerase, a DNA polymerase that has the ability to elongate them [25].

Normally a cell has a limited copy number, as telomere length decreases from replication to

replication. Tumour cells, however, have been seen to have an upregulation of telomerase,

resulting in good telomere maintenance and a limitless replicative potential [26].

Sustained angiogenesis

Crucial to cell function and survival is the supply of oxygen and nutrients, thus requiring

vicinity to capillary blood vessels. For tumors to develop in size and metastatic potential they

must make an "angiogenic switch" through perturbing the local balance of proangiogenic and

antiangiogenic factors [27]. Here integrins and cell adhesion molecules mediating cell-matrix

and cell-cell association are again implicated. Vascular epithelial growth factor (VEGF) is an

example of one of these and is thought to play a central role in the angiogenic switch from

vascular quiescence [28, 29].

Tissue evasion and metastasis

A tumour will eventually begin to outgrow its surroundings and will send off a group of cells

to invade different sites and grow new colonies, or metastases. Metastases cause 90% of

human cancer deaths [30]. Yet again cell-cell adhesion molecules, calcium-dependent

cadherins and integrins are involved in this process by linking cells to the extracellular matrix

and allowing them to be transported to new locations [18]. E-cadherin is expressed in

epithelial cells and acts as a suppressor of invasion and metastasis and its functional

elimination results in an acquisition of metastatic ability [31].

The acquisition of these 6 capabilities depends on gene mutation, which is an inefficient

process as the cell has its own set of “caretakers”, or repair enzymes, which mend damaged

DNA, making it highly unlikely for a tumour genome to occur during a human life span [32,

33]. However, cancer does occur frequently and therefore there must be some additional

factors involved. Alterations of genes that encode DNA damage response proteins, such as

p53 or repair enzymes, can result in genomic instability causing a heightened predisposition

to cancer [34]. Inactivation of DNA mismatch repair genes (MMR), such as MSH2 or MLH1

can give rise to alterations at short DNA repeat sequences throughout the genome, resulting in

10

microsatellite instability (MIN) [35, 36]. Tumours with such profiles are referred to as

displaying a mutator phenotype [37]. In most cases genomic instability arises from larger

chromosomal changes, such as amplifications and translocations, which lead to chromosomal

instability (CIN) [33]. These observations suggest that genetic instability plays an important

role in tumourigenesis, although Darwinian selection is more likely to be the initiating and

main driving force for the on-set and progression of cancer [38].

2.2 Epigenetics of cancer

Epigenetics can be defined as alterations in gene expression that are not related to changes in

genotype. Epigenetic regulation of gene expression plays a critical role in development and

differentiation, X inactivation, genomic imprinting and several human diseases including

cancer [39].

DNA methylation is one of the most common and best studied epigenetic events taking place

in the mammalian genome and because it is reversible, it makes it a useful therapeutic target.

DNA methylation is a covalent chemical modification, resulting in the addition of a methyl

group (CH3) at the carbon 5 position of the cytosine ring and occurs predominantly at CpG

dinucleotides [40-43]. The human genome, however, is not uniformly methylated, but

contains regions of unmethylated DNA scattered with methylated regions [44]. Regions

containing short runs of cystosine-guanine repeats are known as CpG islands and are often

found in the promoter region of genes. In cancer cells, gene silencing has been associated with

aberrant hypermethylation at CpG islands in the promoter region, which are normally

unmethylated [45]. In mammals there are three biologically DNA active methyl transferases

(DNMTs): DNMT1, working mainly on maintenance of methylation and DNMT3a and

DNMT3b, whose main activity is de novo methylation [46-48].

The mechanisms of transcriptional repression by epigenetic regulation are at present not fully

understood, although several models have been proposed. Previously it was believed that

silencing through DNA methylation was caused by physical hindrance of binding of

transcription factors to binding sites in the promoter region [49]. It seems, however, that DNA

methylation in itself does not result in gene silencing; instead it serves as a target for

recruitment of other proteins, which are required for the formation of heterochromatin to

silence genes. Binding of methyl binding domain proteins (MBDs) to methylated CpG islands

recruits a corepressor complex containing histone deacetylases (HDACs), resulting in a

change in chromatin structure (figure 2) [50]. It is well established that there is a link between

11

DNA methylation, histone deacetylation and methylation at lysine 9 of H3, although the

precise string of events is not yet fully understood. One theory proposes that H3K9

methylation creates a foundation where an adaptor molecule e.g. HP1 can be recruited, which

in turn binds a DNMT, methylating local CpG dinucleotides. MBD can then bind to me-CpG,

recruiting HDAC complexes, which are required for H3K9 methylation [52]. H3K9

methyltransferases (HMTs) can then methylate lysine 9 of H3 and thus epigenetic information

can flow back and forth from histone to DNA [53].

DNA methylation and cancer has recently become focus of investigation. Tumour cells show

major disruptions in DNA methylation patterns as compared to normal cells and several genes

have been identified as being frequently hypermethylated in cancer. Groups of genes

particularly susceptible are cell cycle genes (p16INK4a, p15INK4a, Rb, p14ARF), genes associated

with DNA repair (MGMT, BRCA1), apoptosis, drug resistance, detoxification, differentiation,

angiogenesis and metastasis [54]. Many tumours show hypermethylation in several genes, an

example of this is a study on lung cancer, where more than 40 genes were found to have

alterations in DNA methylation patterns [55]. Methylation profiling may be a potential

clinical application in cancer diagnosis, prognosis or therapeutics. Development in

methylation detection techniques has led to an array of tools, the most common being

bisulphite sequencing, where bisulphite causes the deamination of unmethylated cytosines to

uridine, thus enabling discrimination between methylated and unmethylated cytosine residues

through sequence analysis. Methylation profiling may be useful as a prognostic factor for

chemotherapeutic response (see paper 5 in Present Investigation).

Epigenetic changes are reversible, which makes them an attractive target for novel therapeutic

agents. Clinical trials are presently in progress using HDAC inihibitors and DNA

demethylating drugs [56-59]. Perhaps the future will see an approach where agents inducing

reversal of epigenetic silencing in tumour cells are combined with conventional chemotherapy

and tailored to suit every individual tumour and patient.

12

Figure 2. Gene silencing by DNA methylation (figure modified from Molecular Biology of

the Cell 4th edition [51]).

13

2.3 DNA repair and drug resistance

Chemotherapy and radiation are currently the two main manners of cancer treatment.

However, resistance to these poses a considerable problem in clinical oncology, leading to the

death of a large number of patients. It is therefore of great importance to find tools to predict

response to chemotherapy and to find new therapeutic targets. Cytotoxic agents work by

inducing DNA damage in cells, activating several pathways starting with recognition of

damage and resulting in programmed cell death [60]. The human cell has developed an

efficient system for repair of DNA-lesions, which is crucial in the maintenance of genomic

integrity and cell survival. DNA repair is therefore an important mechanism for therapeutic

resistance and an understanding of these pathways may reveal new targets for treatment.

DNA repair mechanisms

The simplest mechanism of DNA-repair is direct reversal, which involves a single enzyme

reaction for the removal of certain types of damage directly from the DNA. Removal of alkyl

adducts by MGMT is the best described example of this (reviewed in more detail below) [61].

Base excision repair (BER) is a DNA-repair pathway for single-base abnormalities [61] and is

the main pathway dealing with damage induced by cellular metabolites [62].

Nucleotide excision repair (NER) is a DNA repair complex consisting of recognition, removal

and synthetic proteins that repair bulky adducts and kinks, e.g. as induced by cisplatin, in a

transcription-coupled manner [60, 61].

Mismatch repair (MMR) is a repair complex, containing mismatch recognition proteins that

detects and repairs incorrectly paired nucleotides that have been introduced during replication

[61]. Cisplatin adducts are recognized by MMR complexes and loss of MMR has been linked

to acquired tumour resistance due to failure to enter apoptosis in the presence of DNA damage

[63].

Double-strand break (DSB) repair functions through homologous or non-homologous

recombination [60]. Double-strand breaks are among the most fatal DNA lesions and if

unrepaired can result in chromosome rearrangement and nucleotide loss, which commonly

occur in cancer [64].

14

3. Mutation analysis

As discussed above, cancer is often a result of genetic alterations in a cell and can be large

chromosomal aberrations, such as amplifications, translocations or deletions, or subtle

changes in nucleotide sequence such as single base substitutions, insertions or deletions.

Although Sanger sequencing has been regarded as the “gold standard” for many years and has

been considerably improved, as the need for high-throughput, enhanced speed and low cost

increases, this technique faces many limitations. Several other methods have been developed

for the analysis of point mutations and can be classed into two categories: scanning methods,

which aim at finding unknown mutations in candidate genes; and screening methods, which

aim at finding known mutations, preferably with high-throughput. Scanning technologies

include: direct DNA sequencing, conformation-based techniques, cleavage-based techniques

and enzyme-based methods [65]. Screening technologies include: ligation-based methods,

hybridisation-based techniques and techniques based on polymerase extension [66]. Recently,

novel methods such as biosensors and array-based techniques have also been described for

these purposes [65]. A number of these methods are reviewed by Ahmadian and Syvanän

[67-69].

Here is a short description of some of the most popular methods of mutation analysis:

3.1 Sanger sequencing

The first DNA sequencing techniques were introduced in 1977: the Maxam and Gilbert

method of chain degradation [70] and the Sanger method of chain termination [71]. In Sanger

sequencing four primer extension reactions are performed in the presence of one of the four

dNTPs and a low concentration of dideoxy-dNTPs (ddNTP). ddNTPs have a missing OH-

group at the 3’-end preventing addition of another nucleotide and incorporation results in

15

chain termination. This produces a number of fragments of differing lengths, which are then

separated by electrophoresis and the sequence determined by reading the band pattern.

Originally, radioactively labelled nucleotides or primers were used for detection. Modern

methods, however, use dye-terminators, where each of the dNTPs is labelled with a different

fluorescent dye [72]. The major advantage of this is that the complete sequencing set can be

performed in a single reaction making it easily automated. This, together with the introduction

of capillary electrophoresis, has increased throughput considerably as well as reducing cost.

3.2 Single-strand conformation polymorphism, SSCP

SSCP is a simple and fast method for scanning many samples for DNA polymorphisms.

Single-stranded DNA molecules can assume varying secondary and tertiary structures,

depending on their nucleotide sequences. SSCP analysis detects point mutations based on

electrophoretic mobility differences that result from these conformational changes [73]. PCR

amplified DNA, covering the region of interest, is denatured using a denaturing agent, such as

formamide, and heat. The single-stranded molecules are then separated on a non-denaturing

polyacrylamide gel and compared to wild-type or known mutant DNA. Results are typically

visualized by autoradiography[74], silver staining or capillary electrophoresis based methods

[75]. Mutant samples are usually confirmed by DNA sequencing.

3.3 Pyrosequencing

A fairly novel method for performing sequence-based DNA analysis is the pyrosequencing

technique [76]. Due to the ability of automation, this system can easily be used for high-

throughput screening. Pyrosequencing is based on the “sequencing by synthesis” principle

[77], where nucleotides are sequentially added to a primed single-stranded DNA template of

which the sequence can be determined by the order of the incorporated nucleotides. Four

enzymes are involved: the Klenow fragment of E. coli DNA polymerase I, ATP sulphurylase,

luciferase and apyrase. If the added nucleotide is incorporated into the growing DNA strand,

pyrophosphate, PPi, will be released by the Klenow DNA polymerase. PPi serves as a

substrate for ATP sulphurylase, which produces ATP. Luciferase then converts ATP into

visible light, which can be detected by a CCD camera. Nucleotides are added to the reaction

one at a time and excess nucleotides are degraded by apyrase in between additions. If the

added nucleotide does not form a base-pair with the template DNA strand, it will not be

16

incorporated and no light will be detected. The overall reaction takes place within 3-4 seconds

at room temperature (figure 3) [76].

One of the major drawbacks of pyrosequencing has been its limited read-length ability, which

originally allowed read lengths of up to only 20-30 bp. However, many improvements have

been made to the technique since the earliest version. It was discovered that false signals were

obtained when dATP was added to the pyrosequencing reaction, due to dATP being a

substrate of luciferase. This problem was solved by the substitution to dATPαS, which is inert

to luciferase, in the polymerization reaction [76]. Secondary structured in the single-stranded

DNA template, as well as unspecific primer binding, can interfere with the pyrosequencing

reaction, also reducing read-length. The addition of ss-DNA binding protein (SSB) can reduce

this problem and has recently been introduced to commercial kits [78] (www.biotage.com).

Pyrosequencing has many other applications including: SNP genotyping [79], CpG

methylation analysis (see below) [80, 81], resequencing [82] and has recently been developed

into an ultra high-throughput method for whole genome sequencing by 454 Life Sciences

[83].

17

Figure 3. Schematic diagram of the priniciples of pyrosequencing.

18

4. Methylation analysis

In the recent years, it has become apparent that cancer is as much a disease of misdirected

epigenetics as it is a disease of genetic mutations [42]. Hypermethylation of promoter regions

is associated with gene silencing, through chromatin conformational changes and is a

common event in carcinogenesis (see section 2.3). Detection of these events is therefore of

great interest in both clinical diagnostics and therapeutics. Most methods for analysis of DNA

methylation patterns are based on bisulphite genomic sequencing of amplified products [84].

Sodium bisulphite converts unmethylated cytosines to uracil, leaving methylated cytosines

unchanged [85]. PCR amplification of treated template results in a C/T polymorphism, which

can be detected by various sequencing methods. Novel techniques include pyrosequencing

[80, 81] and array-based methods [86, 87]. The method of choice depends on the desired

application, whether it is to quantify overall methylation in genomic DNA, to map methylated

cytosines in specific DNA sequences or to find new methylation hot spots (several methods

are reviewed by Fraga and Esteller 2002 [88]).

Two methods are briefly described below:

4.1 Methylation-specific PCR (MSP)

Methylation-specific PCR is the most widely used technique for studying the methylation

profile of CpG islands. Primers are designed to anneal specifically to methylated or

unmethylated bisulphite-converted sequence. Because the strands are no longer

complementary, the primer-design must be customized for each chain [88]. MSP eliminates

the false positive results inherent to previous PCR-based approaches which relied on

differential restriction enzyme cleavage to distinguish methylated from unmethylated DNA

19

[89]. This method has had a huge impact on the field of cancer epigenetics by making DNA

methylation analysis accessible to a wide number of laboratories [84].

4.2 Pyrosequencing

Pyrosequencing has recently been adapted for methylation analysis (figure 4) [80, 81, 90].

The main advantage of the pyrosequencing technology, compared to MSP, is the

quantification of the methylation status of each individual CpG in the sequence. This is of

particular interest as the degree of methylation can differ significantly at various CpGs within

a CpG island [90].

c/mcgcagactgcctcaggcccg

u/mcguagautguutuagguuug

bisulphite treatment

t/cgtagattgttttaggtttg

PCR amplification

1200

1300

1400

1500

E S A T C A G T C T A G A T G T T A G T C A G T C A G T C A G T C T G

T:52,6%C:47,4%

T:100,0%C:0,0%

T:74,0%C:26,0%

T:70,7%C:29,3%

T:70,4%C:29,6%

T:69,7%C:30,3%

5 10 15 20 25 30

Pyrosequencing analysis

c/mcgcagactgcctcaggcccg

u/mcguagautguutuagguuug

bisulphite treatment

u/mcguagautguutuagguuug

bisulphite treatment

t/cgtagattgttttaggtttg

PCR amplification

t/cgtagattgttttaggtttg

PCR amplification

1200

1300

1400

1500

E S A T C A G T C T A G A T G T T A G T C A G T C A G T C A G T C T G

T:52,6%C:47,4%

T:100,0%C:0,0%

T:74,0%C:26,0%

T:70,7%C:29,3%

T:70,4%C:29,6%

T:69,7%C:30,3%

5 10 15 20 25 30

Pyrosequencing analysis

1200

1300

1400

1500

E S A T C A G T C T A G A T G T T A G T C A G T C A G T C A G T C T G

T:52,6%C:47,4%

T:100,0%C:0,0%

T:74,0%C:26,0%

T:70,7%C:29,3%

T:70,4%C:29,6%

T:69,7%C:30,3%

5 10 15 20 25 30

Pyrosequencing analysis

Figure 4. CpG methylation analysis by pyrosequencing.

20

5. Microarray Technology

With the completion of the human genome project, comes the ability to identify genes of

interest and find a link to human disease. Along with this comes the need for an efficient high

throughput method of screening large numbers of genes at once. Previously, methods such as

Northern Blot Analysis and RT-PCR have been used, which allow expression analysis of a

small number of genes. Serial Analysis of Gene Expression (SAGE) was one of the first

methods of global gene expression analysis [91] and is a powerful method, although it has

several limitations. Microarray technology, however, allows for parallel and global analysis of

several thousands of genes and is one of the most widely used and versatile methods. The

earliest recorded study using this technology was published by Augenlicht et al. in 1987,

where 4000 cDNA clones were spotted on a nylon membrane to examine gene expression in

colon cancer [92]. During the past few years development of microarrays has bloomed and

progressed from obscurity to becoming routinely used in biological and medical research for

gene expression studies. Furthermore, the microarray platform can be used in a variety of

other applications and several new methods are emerging, such as protein interaction assays

e.g. CHiP-on-chip (on chip immunoprecipitation) [93], genome-wide epigenetic analysis [86,

87] and resequencing [94].

21

5.1 Microarray platforms

5.1.1 GeneChip® technology

Array manufacture

The first commercially available microarrays were high-density oligonucleotide chips,

reported by Affymetrix in 1995 [95, 96]. Oligonucleotide probes are chemically synthesized

on the surface of a chip, using lithographic masks in a similar technique to that used in the

production of computer chips [97, 98]. For each gene represented on the chip, 11-20 probes,

preferably chosen at unique regions, are used. The probes are approximately 25 nt in length

and are generated in pairs with a perfect match (PM) and a mismatch (MM), where one

nucleotide in the complementary sequence has been changed. The MM probes are used later

to correct for non-specific binding and background signals [99].

Sample preparation, labelling and hybridization

cDNA is synthesized from isolated mRNA using reverse transcriptase and a poly-T primer.

Samples are amplified using T7 RNA polymerase in the presence of biotin-labelled UTP and

CTP, yielding 50-100 copies of biotin-labelled cRNA. The cRNA is then fragmented into 25-

300 nt long fragments, to avoid formation of secondary structures, and hybridized to the chip.

After washing, biotin-labelled target is stained using an antibody amplification procedure

involving binding to streptavidin-phycoerythrin. Finally the chip is scanned in a confocal laser

scanner [100].

5.1.2 Spotted arrays

Spotted arrays are considerably cheaper to use than GeneChip® and have the advantage that

they can be manufactured in-house.

Array manufacture

The first step in the construction of a microarray is to identify which transcripts are to be

spotted. Spotted arrays can be designed to cover any combination of transcripts focusing on

the field of interest and the user is not limited to pre-designed arrays, which may be of a more

general nature. There are several publically available databases where genes of interest can be

identified e.g. UniGene (http://www.ncbi.nlm.nih.gov/UniGene/), where gene nucleotide

22

sequences can be found in a set of non-redundant clusters [101] and Ensembl

(http://www.ensembl.org/, [102]). Another public source of cDNA clones is the Integrated

Molecular Analysis of Genomics and their expression (IMAGE) Consortium [101]. Once a

suitable collection of genes and ESTs has been chosen, bacterial clones containing plasmids

with cDNAs of interest are grown. After extraction, the cDNAs are amplified by PCR using

vector sequences as priming sites. PCR-products are then purified and sequence-verified

before being spotted onto aminosilane or polysilane coated glass slides by a robotic

microarrayer. An alternative to cDNA clones is to spot conventionally synthesized long-mer

oligonucleotides [103], which have gained increased interest, due to higher specificity in

hybridization and simplicity in array manufacturing.

Sample preparation, labelling and hybridization

Total or mRNA is isolated from cells or tissues of interest, quantitated and checked for purity.

Labelling with fluorescent dyes, typically Cy3 and Cy5, is then performed either in a direct or

indirect reaction. In a direct labelling reaction dye-labelled nucleotides are incorporated

during cDNA synthesis and in an indirect reaction chemically modified nucleotides are

incorporated to which fluorescent dyes are subsequently coupled. The most common method

involves incorporation of amino-allyl modified nucleotides, which are coupled to Cy3- or

Cy5-esters [104, 105]. Although indirect labelling involves an extra step, it is often preferred

as it avoids bias due to uneven incorporation of the different fluorophores, often gives higher

signals and is usually cheaper.

Before hybridization, the microarray slide is subjected to a prehybridization bath, where

reactive groups on the slide surface are blocked by BSA, to avoid high background signals.

Labelled samples are mixed with hybridization buffer, generally containing denaturing agents

such as sodium dodecyl sulphate (SDS) and formamide, salts and Cot-1 and poly-A DNA to

block repetitive sequences. Hybridization is performed manually, by applying the mixture to

the slide and placing it in a hybridization chamber, or automatically in a hybridization station

for 16-24h. Slides are then washed and scanned in a confocal laser scanner (figure 5) [100].

23

Figure 5. Schematic diagram of the principles of cDNA microarray analysis.

24

6 Gene expression analysis

Gene expression analysis or transcript profiling, where simultaneous analysis of gene

expression levels in various tissues is performed, can give us novel information on genes and

pathways involved in various diseases or states. The process involves several steps and can be

performed with both GeneChip® and spotted arrays.

6.1 Experimental design

Before the experiment is performed, it is important to plan well. A poorly planned experiment

may result in a load of useless data, where time, money and perhaps valuable samples will

have gone to waste. The first thing to consider is the aim of the experiment, whether it is to

identify differentially expressed genes, to search for specific gene expression patterns or to

classify tumour subtypes. Once this has been established, one has to think about which

samples one wants to look at. The most common analyses are where two samples are

compared to each other e.g. normal and diseased tissues, or treated and untreated cell samples.

Here one has to take into account aspects such as, what is considered normal and the

heterogeneity of tissue samples, especially tumours. Time is also an aspect, where a disease or

a treatment may have very different effects on gene expression over time; in this case it might

be suitable to compare several samples taken over a period of time. Biological fluctuation is

one of the major sources of variation in a microarray experiment and it is therefore extremely

important to have several biological replicates, the minimum recommendation being five

[106, 107]. This, however, is not always accomplished, often due to lack of biological

material. Technical replicates, which allow only effects of measurement variability to be

reduced, are not often required, although they can be of use in quality control experiments or

cases where the number of samples is small [108].

25

In the case of spotted arrays, there are three main design choices: direct comparison, where

the two samples of interest are hybridized against each other on one slide; reference design,

where all samples are hybridized on separate slides against a common reference; or loop

design, where samples are hybridized to one another in a chain [109, 110]. Direct comparison

is the simplest choice and reduces technical variance, although dye swap experiments should

be done to avoid bias from either of the fluorophores. This method is, however, limited in its

uses as only the two samples hybridized together can be compared. If there is a desire to make

several comparisons with many different samples a reference design might be preferred. This

has become the most common design method as it allows any comparison to be made between

samples with equal efficiencies and it also allows additional samples to be added to the

experiment at a later date, providing availability of the reference sample [110]. An alternative

to reference design is the loop design, which involves fewer slides [111]. Efficiency may,

however, be lost if loops are large and many comparisons are made, or if an array falls out

[110].

6.2 Data analysis

Data preprocessing and normalisation

Microarray experiments produce an overwhelming amount of data and the handling of this is

a whole science of its own. Raw data is obtained by scanning the arrays with a confocal laser

scanner after which image processing done and foreground and background intensities of the

spots measured. In the case of spotted arrays, where samples are hybridized in two channels, a

ratio of the spot intensities is used for analysis. In order to present up- and downregulated

genes equally, these ratios are log-transformed, most commonly using a base-2 logarithm

[100]. The array fabrication and hybridization process, especially for spotted arrays,

introduces a lot of noise, which may interfere with results of downstream analyses. It is,

therefore, common to process the data through a series of filtering steps before proceeding

with the analysis. Examples of some filtering steps are removal of spots with intensities close

to background, differing mean-to-median pixel ratios, saturated intensities or deviating size

[100]. Filtering is, however, considered unnecessary by some, who suggest that aberrant spots

will display a high variance between slides and will therefore be removed in downstream

stastistical analysis.

26

Normalization is performed, to adjust for variability across a slide or between experiments.

There are several algorithms “on the market” and to date no consensus on which is the best.

Local weighted linear regression, or lowess, analysis is widely used and takes into account

intensity dependent effects. A local, or print-tip, lowess normalisation corrects for both

intensity and spatial variation between spotting tips [112]. Other normalisation methods

include total intensity normalization, which relies on the assumption that mRNA quantities

are equivalent for both samples hybridized and that equal amounts of genes are up- and

downregulated [113]. Housekeeping genes can be used for normalization, however, this

assumes that all housekeeping genes have constant levels of expression, which is rarely true.

The use of spike-in controls for normalization purposes is attractive, although it involves

more work in array design etc. [114].

Identifying differentially expressed genes

Once normalized expression ratios have been obtained, it is usually desirable to identify

which genes are most likely to be differentially expressed. Early microarray gene expression

analyses often used simply a fold change cut-off of 2 to define genes as differentially

expressed, which is now considered as an inadequate method [115, 116] as it does not take

into account variance and offers no associated level of confidence [115, 117]. Subtle,

although potentially important changes, may also be missed when using the fc cut-off method.

More recent methods involve ranking the genes in order of evidence for differential

expression and setting a suitable value of significance [118]. One choice of ranking genes is

to use the t statistic, where variability between replicates of each particular gene is taken into

account. However, some genes, especially those with low intensities, will have small sample

variance and result in large t-values even though they are not in fact differentially expressed

[118]. An alternative choice is to use a parametric empirical Bayes approach, where the B-

statistic is an estimate of the posterior log-odds that each gene is differentially expressed

[119]. It is not always obvious where to choose the cut-off value as both type I and type II

error needs to be kept low. It is generally preferable to allow a certain amount of false

positives in order not to miss the true positives. Other popular statistical methods for

identifying differentially expressed genes are Significance Analysis of Microarrays (SAM)

[120] and ANOVA-based approaches [111, 121].

As many thousands of genes are measured simultaneously, correction must be done for

multiple testing. Bonferroni correction is a family wise error rate control procedure and limits

the probability of making a type I error due to multiple testing [122]. However, as it is often

27

deemed acceptable to include a number of false positives, Bonferroni correction is usually

considered too stringent. Benjamini and Hochberg came up with a procedure called false

discovery rate (FDR) which is often considered more appropriate [123].

Data mining

Microarray analysis often gives rise to long lists of potentially differentially expressed genes

and poses the problem of how to interpret these from a biological point of view. A number of

tools have been developed to deal with this and the choice of which to use depends on the

questions being asked.

Genes involved in a common pathway, or that respond to a particular environmental factor are

expected to be co-regulated. It is, therefore, of interest to study genes which share similar

patterns of expression and this is done by clustering analysis. Clustering is based on distance

metrics, where a “distance” is calculated between each gene expression vector. Various

clustering algorithms are then used to group genes of similar expression pattern. Unsupervised

methods, which require no prior knowledge, such as hierarchical clustering [124], k-means

clustering [125], self-organizing maps (SOM) [126] and PCA [127] are commonly used.

Supervised methods are used for classification of e.g. cancer sub-types and require previous

information about which genes will cluster together. The data is trained on the classes of

genes already known, so as to distinguish between members and non-members, and is then

applied to the test data [128]. Examples of supervised clustering methods are: linear

discriminant analysis, nearest neighbour classifiers [129], and neural networks such as support

vector machine (SVM) [130].

It may also be of interest to find out more about the function of the genes and how they are

related to other biological pathways and processes. For this purpose, the Gene Ontology (GO)

Consortium has constructed a vocabulary to describe gene and gene product attributes to

biological processes, cellular components and molecular functions [131]. A number of open-

access tools are available, which can calculate overrepresentation statistics for all GO terms

with respect to a given data set, so called gene class testing (GCT) [108], e.g. EASE [132],

MAPPfinder [133] and Onto-Express [134].

Identification of metabolic pathways is also an important part of gene expression profiling, to

acquire a mechanistic understanding of disease or drug treatment etc. One of the biggest

efforts in this field is the Kyoto Encyclopedia of Genes and Genomes (KEGG) consisting of

28

graphical representations of most known biochemical pathways [135]. There are various tools

available, which allow the user to link pathway information to gene expression data or to

build new pathways, such as the open-source GenMAPP [136] or commercially available

PathwayStudio [137] (Ariadne Genomics, MD, USA) and GeneSpring (Agilent Technologies,

USA).

The management and analysis of microarray data has become more and more demanding,

making it quite a challenge. The requirement for comprehensive and flexible tools has led to

the development of a number of both commercial and open-source tools. The R open-source

language and environment for data analysis and visualization provides implementations for a

broad range of statistical and graphical techniques [138](http://www.r-project.org/). Together

with the Bioconductor (http://www.bioconductor.org/) open-source software package, R can

be used to cover most of the above mentioned data analysis steps and is applicable to all array

platforms.

In 2001, the Minimum Information About a Microarray Expermiment (MIAME) was

introduced by the MGED society [139] in an effort to standardize the presentation of

microarray data. These standards have been adopted by the microarray community and are the

basis of several array databases.

6.3 Validation of array data

Although microarray technology and data analysis is becoming more and more standardized

and thus more reliable, artifacts can still be introduced at any time during an experiment and it

is therefore necessary to validate the results. Once a number of target genes have been

identified by the microarray analysis, it is desirable to confirm RNA expression levels by an

independent method. The most common method is quantitative RT-PCR [140] because of its

speed and inexpensiveness. Northern blot has also been reported to give consistent results as

compared to microarray [141].

In addition to validating transcript levels, it is important to measure levels of the

corresponding proteins, as we know that they are not always correlated. This can be done by

various antibody based methods e.g.Western blot, immunohistochemistry or IHC via tissue

microarrays [142-144]. These methods are dependent on antibody availability, which can pose

a problem in the case of unknown transcripts. However, with the Human Proteome Atlas

(http://www.proteinatlas.org/) well on the way, this should soon be solved.

29

A key strategy in target validation is to determine the phenotypic effects when the activity of

a gene is blocked. RNA interference (RNAi) technology is a gene knockdown strategy, which

can be used for this purpose [145]. This method coupled with large scale cell based screening

can also be performed, perhaps leading to identification of novel therapeutic targets [146].

30

31

PRESENT INVESTIGATION

7. Gene expression analysis of epithelial cell

differentiation

Cancer arises when the balance between proliferation, differentiation and apoptosis is in some

way disturbed. Human cancers often arise in epithelial cells, highlighting the importance of

gaining knowledge about the underlying molecular mechanisms involved in differentiation

and proliferation in these cell types.

7.1 Gene expression profiling of N-acetyl L-cysteine (NAC) induced

differentiation of epithelial cells (Papers 1 and 2)

N-acetyl L-cysteine, a sulfhydryl reductant, has been shown to induce a 10-fold more rapid

differentiation in normal human epidermal keratinocytes (NHEK) and a reversion from

neoplastic proliferation to apical-basolateral differentiation in the colon carcinoma cell line

Caco-2. Decreased cell proliferation in these cell types occurred without compromising cell

viability and without induction of apoptosis [147]. Rearrangement of the cytoskeleton was

observed, demonstrated by an increase in cellular junctions and basal localization of actin, all

characteristics of differentiation.

To study the effects of NAC on gene expression, microarray studies were performed using

both the Affymetrix and cDNA platforms. The Affymetrix GeneChip Human U95Av2

containing 12000 transcripts was used to compare NAC treated to untreated NHEK and Caco-

2 in a time series. The KTH HUM30k cDNAarray containing 29760 transcripts was used to

compare NAC treated to untreated NHEK in a time series. Quantitative RT-PCR was

performed to verify the array results.

32

In the Affymetrix study NAC induced differentiation in Caco-2 and NHEK gave 253 and 414

definitely expressed targets respectively (according to our criteria) across the whole time

series. This corresponds to approximately 200 genes in both cell types after consideration to

multiple appearances.

In the cDNA microarray experiment 1007 genes were found to be differentially expressed in

NAC treated NHEK (according to our criteria).

Both experiments showed a limited early transient response at 30 min-3h, with an increasing

late response at 12-48h. A significant number of genes were classified as being involved in

differentiation/proliferation processes and reorganization of the cytoskeleton, which was in

concordance with previous morphological studies. Although both NHEK and Caco-2

displayed differential expression of genes involved in similar processes, the actual responses

seemed to be lineage specific, with little overlap between gene lists.

Interestingly, inhibitor of differentiation 1 (ID-1), which has previously been shown to be

required in G1 cell cycle progression, was found to be downregulated at 1h/30 min in both

NHEK and Caco-2, suggesting a common mechanism for NAC induced differentiation and

inhibition of proliferation.

7.2 Investigating the chemopreventive role of ursodeoxycholic acid in

colorectal cencer cells (Paper 3)

Colorectal cancer (CRC) is one of the leading causes of cancer death in the western world,

with numbers increasing every year [148]. CRC is believed to arise through a multistep

process characterized by histopathological precursor lesions and molecular genetic alterations

[149]. By interfering with these events, chemoprevention could inhibit or reverse the

development of adenomas or the progression from adenoma to cancer. Nonsteroidal anti-

inflammatory drugs (NSAIDs) are the best studied chemopreventive agents against CRC and

have proven efficiency, though they do have certain side-effects. Ursodeoxycholic acid

(UDCA) is a naturally occurring bile acid and is widely used in cholestatic liver disease with

minimal side effects [150]. Preclinical studies have demonstrated colon cancer

chemopreventive effects of UDCA [151-153]. Apoptotic and anti-proliferative properties of

UDCA derivatives have been described in colon cancer cells, with cell cycle arrest at the G1

phase [151, 154]. Hence, UDCA may be of use in the chemoprevention of colon cancer,

although the molecular mechanisms are yet to be elucidated.

33

Global gene expression analysis was performed using cDNA microarrays, in order to

characterize the molecular mechanisms of action and find marker genes for the

chemopreventive effect of UDCA.

Colonic epithelial SW480 adenocarcinoma cells treated with 500 µM UDCA were compared

to ethanol treated control cells in a time series including time points 2, 6, 10, 24 and 48h. A

list of 62 statistically significant differentially expressed genes was extracted from time point

10h and 17 of these were verified by quantitative RT-PCR. Many genes had fold changes of 2

or more at time points 24 and 48h, although not statistically significant. QRT-PCR performed

at time points 2h and 24h reflected previous morphological observations and microarray

results, which suggested increasing transcriptional regulation over time.

Among the genes found to be upregulated at 10h, was growth differentiation factor 15

(GDF15), a divergent member of the TGF-β superfamily, which has been seen to regulate

apoptosis [155] and to be involved in the differentiation of epithelium [156]. GDF15 is a

secreted protein [156], suggesting a role as a regulator of late/secondary transcriptional

response and making it an interesting candidate for further analysis.

Interestingly, a comparison of gene expression levels in SW480, by QRT-PCR, revealed that

UDCA also has a gene regulatory effect on hepatocarcinoma cell line HEPG2 and breast

cancer cell line MCF-7, suggesting a potential chemopreventive use of UDCA in other cancer

types.

Cell cycle analysis by FACS showed a G1 arrest in UDCA treated SW480 cells as compared

to control cells treated with ethanol or deoxycholic acid (DCA). These results support the

microarray data, where both GDF15 and growth arrest- and DNA damage- inducible gene

(GADD45), were upregulated.

In conclusion, our results demonstrate a clear antiproliferative effect of UDCA on colon

cancer cells. Together with previous studies, our results support UDCA as a potential

chemothpreventive agent in patients at risk for colon cancer and may even have its uses in

other cancer types. Although the molecular mechanisms are not entirely clear, a number of

target genes have been identified and follow up studies on these are in the pipeline.

34

8. Cutaneous melanoma

Malignant melanoma is a growing disease increasing at a rate of 2.5-4% every year [157].

Primary tumours are easily detected and can be cured in 95% of cases by surgery. However,

patients with disseminated melanoma are generally resistant to all current therapies and have a

poor prognosis with a one year survival rate of approximately 50% [158].

Clark suggested a model for the progression of melanoma in 1984, which is based on a series

of histopathological and clinical properties. The model is based on 5 stages of melanoma

progression (figure 6): 1. common nevus; 2. dysplastic nevus; 3. radial growth phase (RGP);

4. vertical growth phase (VGP); 5. metastatic melanoma [159]. Common nevi, or moles, may

be present at birth (congenital nevi) or may appear later. They are benign, but are considered a

potential precursor to melanoma. A nevus is defined as dysplastic when it begins to differ

from the normal, i.e. it becomes atypical, and is considered a high risk marker of melanoma.

At the RGP stage, the tumour has begun to spread horizontally, but is still confined to the

epidermis. These tumours are also known as cancer in situ and do not have the ability to

metastasize. In VGP, the tumour has become invasive and has spread to the papillary and

reticular dermis. At this point the tumour has acquired the ability to metastasize and can

continue to develop into metastatic melanoma [159].

Figure 6. Clark’s model for the progression of melanoma.

35

There are several histological subtypes of cutaneous melanoma: superficial spreading

melanoma (SSM) which is the most common type; nodular melanoma (NM) are ball shaped

and lack RGP, making them highly aggressive; lentigo maligna melanoma (LMM) are often

located on the face and acral lentiginous melanoma (ALM), which are commonly found on

the palms of the hand, soles of the feet and subungally and are sometimes difficult to diagnose

as they often look benign.

Melanocytes are epidermal cells derived from the neural crest [157]. Melanocyte precursors

migrate to the basal layer of the epidermis where they differentiate and acquire the ability to

synthesise melanin (figure 7). These cells do not divide rapidly, but are stimulated by

environmental factors such as UV-radiation, which induce division and production of

cytoprotective pigments [157].

Figure 7. The different layers of the skin (figure modified from www.infomedica.se).

Melanoma is a disease of homeostatic imbalance in the skin, where a number of components

influence tumour development: epidermal keratinocytes, dermal fibroblasts, endothelial and

inflammatory cells. In normal skin there is a fine balance of these, where keratincocytes keep

melanocytes under control from constant proliferation and a well-defined basement

membrane complex keeps melanocytes out of the dermis [157].

E –cadherin is normally expressed in melanocytes, allowing them to adhere and thus

preventing invasion [160]. In melanoma progression, the expression of E-cadherin is

36

progressively lost and completely absent in melanoma cells [157]. Instead they acquire the

expression of N-cadherin, which promotes interactions with fibroblasts and endothelial cells

allowing migration and invasion of the tumour [161].

There are a number of gene mutations associated with malignant melanoma and the risk of

developing melanoma is considerably higher if there is a family history [162, 163]. CDKN2A

and CDK4 are high penetrance genes, which give a high risk of developing melanoma if

mutated [164, 165]. The CDKN2A gene locus encodes the tumour suppressors p16INK4A and

p14ARF. p16INK4A controls proliferation by binding pRb and thus blocking E2F activated gene

transcription, leading to cell cycle arrest in the G1 phase. p14ARF binds mdm2, blocking

degradation of p53, which plays an important roll in apoptosis [166, 167]. Thus, a mutation in

this gene locus can lead to increased proliferation and/or dysregulation of apoptosis.

There is strong evidence that the Ras-Raf-MAPK signalling pathway is involved in cutaneous

melanoma [168, 169]. Studies have shown mutations in NRAS or BRAF in approximately 80%

of tumours. The most common BRAF mutation is a switch from valine to glutamic acid at

codon 600 in exon 15 leading to hyperactivation of BRAF and increased transcription through

MAPK activation [170] (see below for more details).

8.1 NRAS and BRAF mutations in melanoma tumours in relation to clinical

characteristics – a study based on mutation screening by pyrosequencing

(Paper 4)

8.1.2 RAS-RAF-MEK-ERK pathway

The MAPK pathway plays a central role in the control of proliferation, differentiation and cell

survival. Growth factors activate RAS through a G-protein coupled receptor, which in turn

recruits RAF to the cell membrane where it is activated by phosphorylation [171, 172]. RAF

is a serine/threonine kinase, which activates the protein kinase MEK by phosphorylation,

which in turn phosphorylates and activates ERK, also a protein kinase [173]. ERK

phosphorylates and activates cytosolic and membrane-localized cytoskeletal proteins,

regulating cell shape and migration [174]. ERK can also translocate to the nucleus and

activate transcription factors, regulating gene expression [173]. There are three mammalian

RAF proteins: Raf-1, ARAF and BRAF [175, 176]. Activation of Raf-1 and ARAF requires

signalling from both RAS and Src, whereas BRAF has a higher basal kinase activity and is

activated by RAS alone [175, 177]. Consequently, BRAF requires fewer steps to become

37

activated, which suggests that it will be activated under a greater variety of conditions and

may be the isoform that is primarily responsible for signalling between RAS and MEK in the

majority of cells [173]. This may also provide an explanation as to why BRAF is more

susceptible to mutation in cancer.

Activating mutations in NRAS are found in up to 30% of human melanomas [178-180] and

BRAF is mutated in 40-67% of cases [170, 181, 182].

This study

Screening and identification of gene mutations in tumours is generally performed by DNA

sequencing, which is high throughput and reliable, but very costly, or single strand

conformational polymorphism (SSCP), a cheap and simple pre-screening method, where

sensitivity and specificity has been questioned.

Pyrosequencing is a real time, sequence by synthesis method, which is high throughput, very

specific and reliable. Sivertsson et al. (2002) showed that mutations in NRAS could be

screened using pyrosequencing [183]. Here we expanded the study to include BRAF

mutations. Samples were amplified in an outer multiplex reaction and 2 separate inner PCR

reactions were done for each of the genes. Pyrosequencing was then performed on a PSQ HS

96A Pyrosequencer.

In our study we screened 219 tumours from 176 patients for mutations in NRAS codon 61 and

BRAF codon 600, using pyrosequencing. In addition, 75 tumours from 43 patients with

previously reported NRAS/BRAF mutation status [182] were added to the analysis. Thus,

data from 294 tumours from 219 patients were analysed. Mutations in NRAS were found in 86

(29%) of the tumours and 156 (53%) in BRAF. In total, 82% of the tumours analysed

displayed mutations in either BRAF or NRAS, and were mutually exclusive in all cases but 2,

which contained mutations in both genes. Multiple tumours were analysed in 57 of the 294

cases and NRAS/BRAF mutations were found in 47 of them, where all but 3 had the same

genotype, indicating that these mutations occur early, before metastasis takes place.

Significant differences with respect to association of a pre-existing nevus and to lymphocyte

infiltration were seen between the three genotype groups: NRAS mutants, BRAF mutants and

wt NRAS and BRAF. BRAF mutants were more commonly associated with a pre-existing

nevus and a higher infiltration of lymphocytes.

38

The two groups of patients defined with NRAS and BRAF mutations, respectively, were

compared to each other only, in a clinical analysis including parameters such as overall

survival. BRAF mutations were associated with an earlier onset than NRAS, with mean ages at

diagnosis of 58 and 64 years respectively. In addition, NRAS mutations were associated with a

higher Clark level of invasion than BRAF mutations. Ulceration, however, seemed to be more

common in tumours with mutations in BRAF, although the difference was not statistically

significant.

In our study, BRAF mutations were more commonly found in tumours on the trunk and lower

extremities, whereas NRAS mutations were found more frequently on the upper extremities,

with a more continual exposure to sun. This observation is not surprising, as the nature of the

mutations in NRAS and BRAF differ from each other. NRAS codon 61 mutations are classical

UVB related modifications with a formation of pyrimidine dimers, caused by UV-radiation,

resulting in a CC → TT or C → T switch, wheras the most common BRAF mutation is a T →

A at nucleotide 1799, where the mechanism is still unknown.

In conclusion, our results show that NRAS and BRAF mutations arise early and are preserved

from primary tumours to metastases. We have also shown that different genotype patterns are

associated with different clinical parameters. Perhaps the most interesting finding is that

tumours with BRAF mutations are more often associated with lymphocyte infiltration,

suggesting a potential target of immunotherapy.

39

8.2 Hypermethylation status of the MGMT promoter in melanoma tumours

in relation to clinical response to chemotherapy (Paper 5)

8.2.1 O6-methylguanine DNA-methyltransferase, MGMT

The gene O6-methylguanine DNA-methyltransferase (MGMT) encodes the repair protein

MGMT, which removes alkylating lesions at the O6 position of guanine. MGMT scans the

double-stranded DNA flipping each base into its active-site pocket, which contains a cysteine

to which the alkyl group is covalently transferred (figure 8) [184, 185]. After this, MGMT is

released from the DNA and subsequently ubiquitinylated and degraded [186]. If the lesion is

not repaired, a GC→AT mutation will occur in 90% of cases during replication due to

incorrect base-pairing by the DNA polymerase [187].

MGMT is interesting from several points of view: first of all it is important in the protection

of normal cells from exogenous carcinogens and endogenous methylating agents. Several

reports show that overexpression of MGMT suppresses tumour development in mouse models

[188-191] and that MGMT knockout mice are more sensitive to methylating agents than wt

[192].

Secondly it may increase drug sensitivity when used as a target of inhibition in the use of

alkylating chemotherapeutic agents. Alkylating agents are often used in chemotherapy and

work on the basis of their capacity to transfer alkyl groups such as methyl, ethyl or

chloroethyl to nucleophilic sites in DNA. Alkylation at the O6-position is one of the most

cytotoxic lesions and mainly responsible for the effects of chemotherapy [193]. Resistance to

these cytotoxic agents may be a result of efficient repair of O6-AG adducts by MGMT.

MGMT can be upregulated by response signals to DNA-damage. The promoter region

includes several transcription-factor recognition sites [194-196] and hypermethylation of CpG

islands in this region in tumour cells turns off MGMT expression (see below). Loss of

MGMT may promote mutagenic conversion in cells by allowing O6-mG lesions to remain

unrepaired, triggering activation of oncogenes or inhibition of tumour suppressor genes. This

may be a good early marker of carcinogenesis in the detection of tumours [61]. Demethylating

agents induce expression of MGMT by reducing promoter methylation [197] and may be of

potential clinical use.

40

Figure 8. MGMT repairs alkylating adducts at the O6 position of guanine in a suicide reaction.

Regarding expression of MGMT, there are two sides of the coin: tumours expressing low

levels of MGMT probably have a worse prognosis, due to an increased number of oncogenic

mutations; on the other hand these tumours may respond better to chemotherapy, although it

is still unclear whether these effects can be solely attributed to loss of MGMT [61].

This study

The purpose of this study was to evaluate the methylation status of the MGMT promoter

region in melanoma tumours and see if there is a correlation to clinical outcome of

chemotherapeutic response. We used methylation specific PCR (MSP) and pyrosequencing to

analyse 64 tumours from 64 patients with cutaneous melanoma, 29 responders and 45 non-

responders. Good concordance was seen between the two methods, in a comparison of

methylation status, with only 4 out of 64 samples showing contrasting results.

In a comparison to clinical response to DTIC/TMZ chemotherapy, tumours were defined as

having a methylated MGMT promoter if they: a. gave a PCR product in MSP analysis,

regardless of methylation level; or b. methylation index (MtI, the average methylation level

over all CpGs in the analysed region) was > 5% in the pyrosequencing analysis. In the MSP

analysis, methylation of the MGMT promoter was observed in 7 out of 19 (37%) responders

compared to 9 out of 45 (20%) non-responders, receiving DTIC/TMZ-based chemotherapy.

When DTIC or TMZ was administered as a single agent, 4 out of 6 (67%) responders

41

displayed tumours with MGMT promoter methylation compared to 6 out of 27 non-responders

(22%). Results were similar in the pyrosequencing analysis, with tumours from 7 out of 19

(37%) responders to DTIC/TMZ-based chemotherapy defined as methylated at the MGMT

promoter compared to 11 out of 45 (24%) non-responders. In a comparison to single agent

therapy tumours from 4 out of 6 (67%) responders displayed methylation of the MGMT

promoter compared to 7 out of 27 (26%) non-responders. Thus, our results show that tumours

from patients responding to DTIC/TMZ-based chemotherapy tend to have MGMT promoter

methylation more often than non-responders. These results are, however, not statistically

significant and the study needs to be expanded to include more patients.

To conclude this study, we have seen that MGMT promoter methylation may provide a

predictive role in the clinical response to DTIC/TMZ-based chemotherapy in patients with

cutaneous melanoma. Thus, analysis of the MGMT promoter methylation status could

potentially be used as a basis for individually tailored chemotherapy.

42

43

9. Concluding Remarks

Cancer is an important public health concern in the western world, coming second only to

heart disease in causes of death. In the past few years progress has been made in the early

diagnosis and monitoring of cancer and many patients are successfully treated. However,

most current therapies have a limited efficacy in curing late-stage disease, thus underlining

the importance of finding improved methods of diagnosis, treatment and chemoprevention. It

is well established that cancer is not one disease, but a collective term describing a number of

diseases arising from an imbalance in cell growth and differentiation. Heredity, somatic

mutations and epigenetic events all play a role in the initiation and progression of cancer by

altering the pattern of gene and protein expression. Because of the heterogeneity of the

disease and its origin, an assortment of methods is needed to explain these genetic and

molecular changes.

In this thesis, a number of methods are described for the identification of the molecular

signatures of cancer at different genetic or transcriptional levels. Pyrosequencing has proved

to be a valuable, high throughput tool in the analysis of both point mutations and detection of

DNA methylation. Through this method we have managed to characterize several molecular

features of cutaneous melanoma, including the role of activation of proto-oncogenes and a

possible mechanism of drug resistance by the inactivation of a DNA repair gene. Microarray

technology has given us a method of determining the effects on global gene expression of

prospective anti-tumourigenic drugs. Our results suggest possible pathways of action and

offer a few genes that may be of potential therapeutic interest. Further investigations are

already in the pipeline as a follow up to these studies, to hopefully take us even closer to an

understanding of the underlying mechanisms of cancer and to find more efficient and specific

methods of therapy.

44

45

Acknowledgements

Jaha, är du nu jag ska öva inför Oscarsgalan? Jag började här på labbet 2001 (samma år som humana genomet publicerades), så jag har ju varit här ett tag även om själva doktorandtiden varit kort. Det finns många personer som hjälpt mig på vägen och jag vill passa på här att tacka några stycken. Jag vill tacka min handledare Joakim Lundeberg för att du tog emot mig först som examensarbetare och sedan doktorand. Du lyckas alltid hitta något positivt i mina resultat när jag själv tycker att det ser hopplöst ut. Tack också för att du varit så uppmuntrande nu på slutet, jag har behövt det. Mattias Uhlén, en stor inspiration till oss alla. Tack för att du alltid är så glad och positiv och tar dig tid att lära känna oss. Tack alla PIs på DNA corner, ni bidrar alla till den sköna stämning här och jag tycker väldigt mycket om er allihopa. Tack till mina roomies gamla och nya, jag tycker om er allihopa: Cissi, tack för diskussioner om microarray och en massa annat... och för alla konferenser vi varit på tillsammans (särskilt Bergen!). Emma, ett fräscht och mycket välkommet nytillskott. Fortsätt att vara sådär smart o cool så kommer du hur långt som helst. Karin L, tack för en del intressant skvaller och för att du också fryser ibland. Tack också för att du är så himla trevlig. Mats, den sötaste guldmunken på DNA corner. Jag har saknat dina spännande telefonsamtal sen du lämnade oss. Det brukar vara rätt kul på dina fester också! Olle-snut, vårt rum har aldrig riktigt hämtat sig doftmässigt sen du lämnade oss. Tack för att alltid dansar tryckare med mig, I’m never gonna dance again... Ronny, det har varit skönt att ha någon att dela skrivarglädjen med, även om det varit lite kyligt ibland, tack för att du låtit mig höja. Tove, min ”partner in crime”, tack för alla samtal om precis allt möjligt och för att du är världens bästa partybrud och kompis. Vi har inte alltid varit överens om temperaturen, men det var värt alla diskussioner för att få sitta bredvid dig! Puss. Flera på DNA corner: Malin A, tack för att du är en bra kompis och för alla långa samtal om mat och rumpor. Det är härligt att du äntligen fattat var man ska bo nånstans.

46

Torun, Emilie, Mikaela och Carro, tack för allt vi gjort tillsammans och för att ni är mina vänner. Erik F, tack för att vi blev vänner när jag var ny och för alla långa samtal och roliga fester, du är saknad. Maria S, du var bästa ex-jobbshandledaren och du inspirerade mig att börja doktorera. Du är också en av de mest sympatiska människorna jag känner.Vi ses på Dello. Anna G, min mentor och partypolare. Tack för alla samtal om vetenskap och annat och för allt roligt vi haft på konferenser och hemma. Du har verkligen varit en inspiration. Tack hela microarraygruppen. Johan L, min teddy-bä. Tack för all hjälp med data analys och för att du haft sån tålamod med mig. Valtteri och Marcus, tack för hjälp med array-stuff och annat skvaller. Nipe, tack för att du alltid är glad och trevlig. Jag känner att jag kan alltid fråga dig om jag behöver hjälp. Kicki, tack för att du alltid tar dig tid att hjälpa mig med pyro trots att du jämt har fullt upp själv. Det är trevligt att ha någon att skvallra med också! Jacob O, tack för intressanta samtal (även vetenskapliga!) och för att jag kan fråga dig om precis vad som helst och du vet svaret. Anna W, tack för alla microarray hybbar du gjort åt mig och för schysst skvaller, du är cool. Annika Å, tack för hjälp med pyro och för alla bra samtal, både roliga och allvarliga, du är också cool. Tack alla samarbetspartners: Suzanne, Shuhua, Johan, Katarina, Marianne och Doris på CCK tack för ett gott samarbete. Det har varit ett nöje att jobba med er. Tack särskilt Suzanne och Shuhua som jobbat jättehårt för att få ihop manuset till avhandlingen. Francesca, tack för att du är så skön och italiensk! Det har varit kul att jobba ihop med dig, särkilt när jag fick vara hos er o köra FACS. Gesan, tack för att du och Francesca har jobbat så hårt för att få ihop manuset till avhandlingen. Tack även Sven och Mauro för ett trevligt samarbete. Uppsalagänget: Figge, Anna, Helena och Sara m.fl. Tack för roliga middagar och hudcancerkonferensen i Sevilla där alla låg vid poolen o solade. Just det, tack även för intressanta vetenskapliga cancermöten.

47

Thank you Tiziana, Ewa, Eugenia and everyone else at the CNER in Rome for our collaboration and for welcoming me to the lab last October. Tack alla vänner: KMSSK: Jånni, Lenny och Hanni-Banni, utan er hade jag aldrig uthärdat KC i fyra år. Vi hade fantastiskt kul ihop i Lund och den tiden kommer jag alltid att minnas. Sara & David, tack för alla roliga fester med mycket intriger och för att jag alltid får tea and biscuits hos er. Mattias E, tack för att vi alltid har kul ihop och för att ditt missförstånd blev världens godaste drink! Mattias J, tack för hjälpen med bilden och för sköna efterfesther. Kakan, Lisa, Anna o Holmen, tack för alla telefonjourer på helgmorgnar med summering av förra kvällens händelser... Ni är mina bästisar och jag älskar er. Kakan, min lilla kanariefågel, tack för alla samtal om livet, labbet och killar och för att du alltid vill dricka vin. Jag har ju följt i dina fotspår: doktorera, singelliv (med allt vad det innebär), horan, disputera och snart fyller jag även 30. Tack för du visar mig vägen, jag är nyfiken på vad som kommer härnäst. Lisa (aka Big mouth strikes again), va mycket kul vi haft tillsammans i Rom, på Street mm. Tack för att du alltid ställer upp när jag behöver dig, du är fin. Anna Maria, tack för att du alltid (med ytterst få undantag) vill ta en öl med mig, du har varit min räddning många gånger (inte minst i sommar). Tack för ”brandy”, snurra flaskan och allt annat kul vi hittat på! Holmen, tack för att du inte försvann i bodelningen. Se nu till att inte försvinna i parträsket! Toffe, tack för allt du gjort för mig och stått ut med, du är fantastisk. Mummy and Daddy, thank you for all your support and encouragement through my whole life. You both know that I often need a little extra “shove” to get things done and if you hadn’t been there to help me, I probably wouldn’t be where I am today. Carl, my baby brother, you’re the only one who really understands my silly sense of humour and by the way “look out for mice, in Swedish!” I love you all.

48

49

References

1. Avery OT, MacLeod CM, and McCarty M, Studies on the chemical nature of the substance inducing transformation of pneumococcal types. Inductions of transformation by a desoxyribonucleic acid fraction isolated from pneumococcus type III. J Exp Med, 1979. 149(2): p. 297-326.

2. Watson JD and Crick FH, Molecular structure of nucleic acids: a structure for deoxyribose nucleic acid. J.D. Watson and F.H.C. Crick. Published in Nature, number 4356 April 25, 1953. Nature, 1974. 248(5451): p. 765.

3. Crick FH, On protein synthesis. Symp Soc Exp Biol, 1958. 12: p. 138-63. 4. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, et al., Initial

sequencing and analysis of the human genome. Nature, 2001. 409(6822): p. 860-921. 5. Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG, et al., The sequence

of the human genome. Science, 2001. 291(5507): p. 1304-51. 6. Hoheisel JD, Microarray technology: beyond transcript profiling and genotype

analysis. Nat Rev Genet, 2006. 7(3): p. 200-10. 7. Uhlen M, Bjorling E, Agaton C, Szigyarto CA, Amini B, Andersen E, et al., A human

protein atlas for normal and cancer tissues based on antibody proteomics. Mol Cell Proteomics, 2005. 4(12): p. 1920-32.

8. Wagner RW, Gene inhibition using antisense oligodeoxynucleotides. Nature, 1994. 372(6504): p. 333-5.

9. Jemal A, Siegel R, Ward E, Murray T, Xu J, Smigal C, et al., Cancer statistics, 2006. CA Cancer J Clin, 2006. 56(2): p. 106-30.

10. Foulds L, The experimental study of tumor progression: a review. Cancer Res, 1954. 14(5): p. 327-39.

11. Nowell PC, The clonal evolution of tumor cell populations. Science, 1976. 194(4260): p. 23-8.

12. Balmain A, Cancer genetics: from Boveri and Mendel to microarrays. Nat Rev Cancer, 2001. 1(1): p. 77-82.

13. Marahatta SB, Sharma N, Koju R, Makaju RK, Petmitr P, and Petmitr S, Cancer: determinants and progression. Nepal Med Coll J, 2005. 7(1): p. 65-71.

14. Hanahan D and Weinberg RA, The hallmarks of cancer. Cell, 2000. 100(1): p. 57-70. 15. Vats TS and Emami A, Oncogenes: present status. Indian J Pediatr, 1993. 60(2): p.

193-201. 16. Lukashev ME and Werb Z, ECM signalling: orchestrating cell behaviour and

misbehaviour. Trends Cell Biol, 1998. 8(11): p. 437-41. 17. Giancotti FG and Ruoslahti E, Integrin signaling. Science, 1999. 285(5430): p. 1028-

32. 18. Aplin AE, Howe A, Alahari SK, and Juliano RL, Signal transduction and signal

modulation by cell adhesion receptors: the role of integrins, cadherins, immunoglobulin-cell adhesion molecules, and selectins. Pharmacol Rev, 1998. 50(2): p. 197-263.

19. Mizejewski GJ, Role of integrins in cancer: survey of expression patterns. Proc Soc Exp Biol Med, 1999. 222(2): p. 124-38.

20. Harris CC, p53 tumor suppressor gene: from the basic research laboratory to the clinic--an abridged historical perspective. Carcinogenesis, 1996. 17(6): p. 1187-98.

21. Cheung HH, Arora V, and Korneluk RG, Abnormalities of cell structures in tumors: apoptosis in tumors. Exs, 2006(96): p. 201-21.

50

22. Weinberg RA, The retinoblastoma protein and cell cycle control. Cell, 1995. 81(3): p. 323-30.

23. Vaux DL, Cory S, and Adams JM, Bcl-2 gene promotes haemopoietic cell survival and cooperates with c-myc to immortalize pre-B cells. Nature, 1988. 335(6189): p. 440-2.

24. Kim Sh SH, Kaminker P, and Campisi J, Telomeres, aging and cancer: in search of a happy ending. Oncogene, 2002. 21(4): p. 503-11.

25. Collins K and Mitchell JR, Telomerase in the human organism. Oncogene, 2002. 21(4): p. 564-79.

26. Bryan TM and Cech TR, Telomerase and the maintenance of chromosome ends. Curr Opin Cell Biol, 1999. 11(3): p. 318-24.

27. Folkman J, Role of angiogenesis in tumor growth and metastasis. Semin Oncol, 2002. 29(6 Suppl 16): p. 15-8.

28. Veikkola T and Alitalo K, VEGFs, receptors and angiogenesis. Semin Cancer Biol, 1999. 9(3): p. 211-20.

29. Hynes RO and Wagner DD, Genetic manipulation of vascular adhesion molecules in mice. J Clin Invest, 1996. 98(10): p. 2193-5.

30. Sporn MB, The war on cancer. Lancet, 1996. 347(9012): p. 1377-81. 31. Birchmeier W and Behrens J, Cadherin expression in carcinomas: role in the

formation of cell junctions and the prevention of invasiveness. Biochim Biophys Acta, 1994. 1198(1): p. 11-26.

32. Loeb LA, Mutator phenotype may be required for multistage carcinogenesis. Cancer Res, 1991. 51(12): p. 3075-9.

33. Lengauer C, Kinzler KW, and Vogelstein B, Genetic instabilities in human cancers. Nature, 1998. 396(6712): p. 643-9.

34. Markowitz S, DNA repair defects inactivate tumor suppressor genes and induce hereditary and sporadic colon cancers. J Clin Oncol, 2000. 18(21 Suppl): p. 75S-80S.

35. Thibodeau SN, Bren G, and Schaid D, Microsatellite instability in cancer of the proximal colon. Science, 1993. 260(5109): p. 816-9.

36. Parsons R, Li GM, Longley MJ, Fang WH, Papadopoulos N, Jen J, et al., Hypermutability and mismatch repair deficiency in RER+ tumor cells. Cell, 1993. 75(6): p. 1227-36.

37. Raptis S and Bapat B, Genetic instability in human tumors. Exs, 2006(96): p. 303-20. 38. Sieber OM, Heinimann K, and Tomlinson IP, Genomic instability--the engine of

tumorigenesis? Nat Rev Cancer, 2003. 3(9): p. 701-8. 39. Jaenisch R and Bird A, Epigenetic regulation of gene expression: how the genome

integrates intrinsic and environmental signals. Nat Genet, 2003. 33 Suppl: p. 245-54. 40. Robertson KD and Wolffe AP, DNA methylation in health and disease. Nat Rev

Genet, 2000. 1(1): p. 11-9. 41. Bird A, DNA methylation de novo. Science, 1999. 286(5448): p. 2287-8. 42. Jones PA and Takai D, The role of DNA methylation in mammalian epigenetics.

Science, 2001. 293(5532): p. 1068-70. 43. Bestor TH, The DNA methyltransferases of mammals. Hum Mol Genet, 2000. 9(16):

p. 2395-402. 44. Bird AP, CpG-rich islands and the function of DNA methylation. Nature, 1986.

321(6067): p. 209-13. 45. Esteller M, Hamilton SR, Burger PC, Baylin SB, and Herman JG, Inactivation of the

DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation is a common event in primary human neoplasia. Cancer Res, 1999. 59(4): p. 793-7.

51

46. Rhee I, Jair KW, Yen RW, Lengauer C, Herman JG, Kinzler KW, et al., CpG methylation is maintained in human cancer cells lacking DNMT1. Nature, 2000. 404(6781): p. 1003-7.

47. Rhee EG, Mendez S, Shah JA, Wu CY, Kirman JR, Turon TN, et al., Vaccination with heat-killed leishmania antigen or recombinant leishmanial protein and CpG oligodeoxynucleotides induces long-term memory CD4+ and CD8+ T cell responses and protection against leishmania major infection. J Exp Med, 2002. 195(12): p. 1565-73.

48. Costello JF and Plass C, Methylation matters. J Med Genet, 2001. 38(5): p. 285-303. 49. Tate PH and Bird AP, Effects of DNA methylation on DNA-binding proteins and gene

expression. Curr Opin Genet Dev, 1993. 3(2): p. 226-31. 50. Soejima H, Joh K, and Mukai T, Gene silencing in DNA damage repair. Cell Mol Life

Sci, 2004. 61(17): p. 2168-72. 51. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, and P W, Molecular Biology of the

Cell. 4th Edition ed. 2002: Garland Science Publishing. 52. Rea S, Eisenhaber F, O'Carroll D, Strahl BD, Sun ZW, Schmid M, et al., Regulation of

chromatin structure by site-specific histone H3 methyltransferases. Nature, 2000. 406(6796): p. 593-9.

53. Fuks F, DNA methylation and histone modifications: teaming up to silence genes. Curr Opin Genet Dev, 2005. 15(5): p. 490-5.

54. Das PM and Singal R, DNA methylation and cancer. J Clin Oncol, 2004. 22(22): p. 4632-42.

55. Tsou JA, Hagen JA, Carpenter CL, and Laird-Offringa IA, DNA methylation analysis: a powerful new tool for lung cancer diagnosis. Oncogene, 2002. 21(35): p. 5450-61.

56. Plumb JA, Strathdee G, Sludden J, Kaye SB, and Brown R, Reversal of drug resistance in human tumor xenografts by 2'-deoxy-5-azacytidine-induced demethylation of the hMLH1 gene promoter. Cancer Res, 2000. 60(21): p. 6039-44.

57. Lubbert M, DNA methylation inhibitors in the treatment of leukemias, myelodysplastic syndromes and hemoglobinopathies: clinical results and possible mechanisms of action. Curr Top Microbiol Immunol, 2000. 249: p. 135-64.

58. Strathdee G and Brown R, Epigenetic cancer therapies: DNA methyltransferase inhibitors. Expert Opin Investig Drugs, 2002. 11(6): p. 747-54.

59. Thiagalingam S, Cheng KH, Lee HJ, Mineva N, Thiagalingam A, and Ponte JF, Histone deacetylases: unique players in shaping the epigenetic histone code. Ann N Y Acad Sci, 2003. 983: p. 84-100.

60. Bradbury PA and Middleton MR, DNA repair pathways in drug resistance in melanoma. Anticancer Drugs, 2004. 15(5): p. 421-6.

61. Gerson SL, MGMT: its role in cancer aetiology and cancer therapeutics. Nat Rev Cancer, 2004. 4(4): p. 296-307.

62. Lindahl T and Wood RD, Quality control by DNA repair. Science, 1999. 286(5446): p. 1897-905.

63. Fink D, Aebi S, and Howell SB, The role of DNA mismatch repair in drug resistance. Clin Cancer Res, 1998. 4(1): p. 1-6.

64. Khanna KK and Jackson SP, DNA double-strand breaks: signaling, repair and the cancer connection. Nat Genet, 2001. 27(3): p. 247-54.

65. Jiang T, Minunni M, and Mascini M, Towards fast and inexpensive molecular diagnostic: the case of TP53. Clin Chim Acta, 2004. 343(1-2): p. 45-60.

66. Balogh K, Patocs A, Majnik J, Racz K, and Hunyady L, Genetic screening methods for the detection of mutations responsible for multiple endocrine neoplasia type 1. Mol Genet Metab, 2004. 83(1-2): p. 74-81.

52

67. Syvanen AC, Toward genome-wide SNP genotyping. Nat Genet, 2005. 37 Suppl: p. S5-10.

68. Ahmadian A and Lundeberg J, A brief history of genetic variation analysis. Biotechniques, 2002. 32(5): p. 1122-4, 1126, 1128 passim.

69. Syvanen AC and Taylor GR, Approaches for analyzing human mutations and nucleotide sequence variation: a report from the Seventh International Mutation Detection meeting, 2003. Hum Mutat, 2004. 23(5): p. 401-5.

70. Maxam AM and Gilbert W, A new method for sequencing DNA. Proc Natl Acad Sci U S A, 1977. 74(2): p. 560-4.

71. Sanger F, Nicklen S, and Coulson AR, DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci U S A, 1977. 74(12): p. 5463-7.

72. Prober JM, Trainor GL, Dam RJ, Hobbs FW, Robertson CW, Zagursky RJ, et al., A system for rapid DNA sequencing with fluorescent chain-terminating dideoxynucleotides. Science, 1987. 238(4825): p. 336-41.

73. Orita M, Iwahana H, Kanazawa H, Hayashi K, and Sekiya T, Detection of polymorphisms of human DNA by gel electrophoresis as single-strand conformation polymorphisms. Proc Natl Acad Sci U S A, 1989. 86(8): p. 2766-70.

74. Harnevik L, Fjellstedt E, Molbaek A, Tiselius HG, Denneberg T, and Soderkvist P, Identification of 12 novel mutations in the SLC3A1 gene in Swedish cystinuria patients. Hum Mutat, 2001. 18(6): p. 516-25.

75. Dong Y and Zhu H, Single-strand conformational polymorphism analysis: basic principles and routine practice. Methods Mol Med, 2005. 108: p. 149-57.

76. Ronaghi M, Karamohamed S, Pettersson B, Uhlen M, and Nyren P, Real-time DNA sequencing using detection of pyrophosphate release. Anal Biochem, 1996. 242(1): p. 84-9.

77. Melamede RJ, Automatable process for sequencing nucleotide, U.S. Patent, Editor. 1985, New York Medical College, Valhalla, NY.

78. Ehn M, Ahmadian A, Nilsson P, Lundeberg J, and Hober S, Escherichia coli single-stranded DNA-binding protein, a molecular tool for improved sequence quality in pyrosequencing. Electrophoresis, 2002. 23(19): p. 3289-99.

79. Ahmadian A, Gharizadeh B, Gustafsson AC, Sterky F, Nyren P, Uhlen M, et al., Single-nucleotide polymorphism analysis by pyrosequencing. Anal Biochem, 2000. 280(1): p. 103-10.

80. Tost J, Dunker J, and Gut IG, Analysis and quantification of multiple methylation variable positions in CpG islands by Pyrosequencing. Biotechniques, 2003. 35(1): p. 152-6.

81. Colella S, Shen L, Baggerly KA, Issa JP, and Krahe R, Sensitive and quantitative universal Pyrosequencing methylation analysis of CpG sites. Biotechniques, 2003. 35(1): p. 146-50.

82. Ronaghi M, Pyrosequencing sheds light on DNA sequencing. Genome Res, 2001. 11(1): p. 3-11.

83. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, Bemben LA, et al., Genome sequencing in microfabricated high-density picolitre reactors. Nature, 2005. 437(7057): p. 376-80.

84. Laird PW, The power and the promise of DNA methylation markers. Nat Rev Cancer, 2003. 3(4): p. 253-66.

85. Shapiro R, Braverman B, Louis JB, and Servis RE, Nucleic acid reactivity and conformation. II. Reaction of cytosine and uracil with sodium bisulfite. J Biol Chem, 1973. 248(11): p. 4060-4.

53

86. Adorjan P, Distler J, Lipscher E, Model F, Muller J, Pelet C, et al., Tumour class prediction and discovery by microarray-based DNA methylation analysis. Nucleic Acids Res, 2002. 30(5): p. e21.

87. Mund C, Beier V, Bewerunge P, Dahms M, Lyko F, and Hoheisel JD, Array-based analysis of genomic DNA methylation patterns of the tumour suppressor gene p16INK4A promoter in colon carcinoma cell lines. Nucleic Acids Res, 2005. 33(8): p. e73.

88. Fraga MF and Esteller M, DNA methylation: a profile of methods and applications. Biotechniques, 2002. 33(3): p. 632, 634, 636-49.

89. Herman JG, Graff JR, Myohanen S, Nelkin BD, and Baylin SB, Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A, 1996. 93(18): p. 9821-6.

90. Dupont JM, Tost J, Jammes H, and Gut IG, De novo quantitative bisulfite sequencing using the pyrosequencing technology. Anal Biochem, 2004. 333(1): p. 119-27.

91. Velculescu VE, Zhang L, Vogelstein B, and Kinzler KW, Serial analysis of gene expression. Science, 1995. 270(5235): p. 484-7.

92. Augenlicht LH, Wahrman MZ, Halsey H, Anderson L, Taylor J, and Lipkin M, Expression of cloned sequences in biopsies of human colonic tissue and in colonic carcinoma cells induced to differentiate in vitro. Cancer Res, 1987. 47(22): p. 6017-21.

93. Buck MJ and Lieb JD, ChIP-chip: considerations for the design, analysis, and application of genome-wide chromatin immunoprecipitation experiments. Genomics, 2004. 83(3): p. 349-60.

94. Hacia JG, Resequencing and mutational analysis using oligonucleotide microarrays. Nat Genet, 1999. 21(1 Suppl): p. 42-7.

95. Lipshutz RJ, Morris D, Chee M, Hubbell E, Kozal MJ, Shah N, et al., Using oligonucleotide probe arrays to access genetic diversity. Biotechniques, 1995. 19(3): p. 442-7.

96. Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, et al., Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol, 1996. 14(13): p. 1675-80.

97. Fodor SP, Read JL, Pirrung MC, Stryer L, Lu AT, and Solas D, Light-directed, spatially addressable parallel chemical synthesis. Science, 1991. 251(4995): p. 767-73.

98. Pease AC, Solas D, Sullivan EJ, Cronin MT, Holmes CP, and Fodor SP, Light-generated oligonucleotide arrays for rapid DNA sequence analysis. Proc Natl Acad Sci U S A, 1994. 91(11): p. 5022-6.

99. Lipshutz RJ, Fodor SP, Gingeras TR, and Lockhart DJ, High density synthetic oligonucleotide arrays. Nat Genet, 1999. 21(1 Suppl): p. 20-4.

100. Sievertzon M, Nilsson P, and Lundeberg J, Improving reliability and performance of DNA microarrays. Expert Review of Molecular Diagnostics, 2006. In press.

101. Wheeler DL, Church DM, Federhen S, Lash AE, Madden TL, Pontius JU, et al., Database resources of the National Center for Biotechnology. Nucleic Acids Res, 2003. 31(1): p. 28-33.

102. Birney E, Andrews D, Bevan P, Caccamo M, Cameron G, Chen Y, et al., Ensembl 2004. Nucleic Acids Res, 2004. 32(Database issue): p. D468-70.

103. Nakanishi T, Oka T, and Akagi T, Recent advances in DNA microarrays. Acta Med Okayama, 2001. 55(6): p. 319-28.

104. Randolph JB and Waggoner AS, Stability, specificity and fluorescence brightness of multiply-labeled fluorescent DNA probes. Nucleic Acids Res, 1997. 25(14): p. 2923-9.

54

105. Schroeder BG, Peterson LM, and Fleischmann RD, Improved quantitation and reproducibility in Mycobacterium tuberculosis DNA microarrays. J Mol Microbiol Biotechnol, 2002. 4(2): p. 123-6.

106. Pavlidis P, Li Q, and Noble WS, The effect of replication on gene expression microarray experiments. Bioinformatics, 2003. 19(13): p. 1620-7.

107. Tsai CA, Chen YJ, and Chen JJ, Testing for differentially expressed genes with microarray data. Nucleic Acids Res, 2003. 31(9): p. e52.

108. Allison DB, Cui X, Page GP, and Sabripour M, Microarray data analysis: from disarray to consolidation and consensus. Nat Rev Genet, 2006. 7(1): p. 55-65.

109. Yang YH and Speed T, Design issues for cDNA microarray experiments. Nat Rev Genet, 2002. 3(8): p. 579-88.

110. Churchill GA, Fundamentals of experimental design for cDNA microarrays. Nat Genet, 2002. 32 Suppl: p. 490-5.

111. Kerr MK and Churchill GA, Experimental design for gene expression microarrays. Biostatistics, 2001. 2(2): p. 183-201.

112. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, et al., Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res, 2002. 30(4): p. e15.

113. Quackenbush J, Microarray data normalization and transformation. Nat Genet, 2002. 32 Suppl: p. 496-501.

114. van de Peppel J, Kemmeren P, van Bakel H, Radonjic M, van Leenen D, and Holstege FC, Monitoring global messenger RNA changes in externally controlled microarray experiments. EMBO Rep, 2003. 4(4): p. 387-93.

115. Hsiao A, Worrall DS, Olefsky JM, and Subramaniam S, Variance-modeled posterior inference of microarray data: detecting gene-expression changes in 3T3-L1 adipocytes. Bioinformatics, 2004. 20(17): p. 3108-27.

116. Miller RA, Galecki A, and Shmookler-Reis RJ, Interpretation, design, and analysis of gene array expression experiments. J Gerontol A Biol Sci Med Sci, 2001. 56(2): p. B52-7.

117. Budhraja V, Spitznagel E, Schaiff WT, and Sadovsky Y, Incorporation of gene-specific variability improves expression analysis using high-density DNA microarrays. BMC Biol, 2003. 1(1): p. 1.

118. Smyth GK, Yang YH, and Speed T, Statistical issues in cDNA microarray data analysis. Methods Mol Biol, 2003. 224: p. 111-36.

119. Lönnstedt I and Speed T, Replicated microarray data. Statistica Sinica, 2002. 12: p. 31-46.

120. Tusher VG, Tibshirani R, and Chu G, Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A, 2001. 98(9): p. 5116-21.

121. Cui X and Churchill GA, Statistical tests for differential expression in cDNA microarray experiments. Genome Biol, 2003. 4(4): p. 210.

122. Bland JM and Altman DG, Multiple significance tests: the Bonferroni method. Bmj, 1995. 310(6973): p. 170.

123. Benjamini Y and Hochberg Y, Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B, 1995. 57: p. 289-300.

124. Eisen MB, Spellman PT, Brown PO, and Botstein D, Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A, 1998. 95(25): p. 14863-8.

125. Tavazoie S, Hughes JD, Campbell MJ, Cho RJ, and Church GM, Systematic determination of genetic network architecture. Nat Genet, 1999. 22(3): p. 281-5.

55

126. Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E, et al., Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation. Proc Natl Acad Sci U S A, 1999. 96(6): p. 2907-12.

127. Raychaudhuri S, Stuart JM, and Altman RB, Principal components analysis to summarize microarray experiments: application to sporulation time series. Pac Symp Biocomput, 2000: p. 455-66.

128. Quackenbush J, Computational analysis of microarray data. Nat Rev Genet, 2001. 2(6): p. 418-27.

129. Dudoit S, Fridlyand J, and Speed T, Comparison of discrimination methods for the classification of tumors using gene expression data. Journal of the American Statistical Association, 2002. 97: p. 77-87.

130. Brown MP, Grundy WN, Lin D, Cristianini N, Sugnet CW, Furey TS, et al., Knowledge-based analysis of microarray gene expression data by using support vector machines. Proc Natl Acad Sci U S A, 2000. 97(1): p. 262-7.

131. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al., Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet, 2000. 25(1): p. 25-9.

132. Dennis G, Jr., Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al., DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol, 2003. 4(5): p. P3.

133. Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, and Conklin BR, MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol, 2003. 4(1): p. R7.

134. Draghici S, Khatri P, Martins RP, Ostermeier GC, and Krawetz SA, Global functional profiling of gene expression. Genomics, 2003. 81(2): p. 98-104.

135. Ogata H, Goto S, Sato K, Fujibuchi W, Bono H, and Kanehisa M, KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res, 1999. 27(1): p. 29-34.

136. Dahlquist KD, Salomonis N, Vranizan K, Lawlor SC, and Conklin BR, GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways. Nat Genet, 2002. 31(1): p. 19-20.

137. Nikitin A, Egorov S, Daraselia N, and Mazo I, Pathway studio--the analysis and navigation of molecular networks. Bioinformatics, 2003. 19(16): p. 2155-7.

138. Dudoit S, Gentleman RC, and Quackenbush J, Open source software for the analysis of microarray data. Biotechniques, 2003. Suppl: p. 45-51.

139. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, et al., Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet, 2001. 29(4): p. 365-71.

140. Rajeevan MS, Vernon SD, Taysavang N, and Unger ER, Validation of array-based gene expression profiles by real-time (kinetic) RT-PCR. J Mol Diagn, 2001. 3(1): p. 26-31.

141. Taniguchi M, Miura K, Iwao H, and Yamanaka S, Quantitative assessment of DNA microarrays--comparison with Northern blot analyses. Genomics, 2001. 71(1): p. 34-9.

142. Chuaqui RF, Bonner RF, Best CJ, Gillespie JW, Flaig MJ, Hewitt SM, et al., Post-analysis follow-up and validation of microarray experiments. Nat Genet, 2002. 32 Suppl: p. 509-14.

143. Mousses S, Kallioniemi A, Kauraniemi P, Elkahloun A, and Kallioniemi OP, Clinical and functional target validation using tissue and cell microarrays. Curr Opin Chem Biol, 2002. 6(1): p. 97-101.

56

144. Kallioniemi OP, Wagner U, Kononen J, and Sauter G, Tissue microarray technology for high-throughput molecular profiling of cancer. Hum Mol Genet, 2001. 10(7): p. 657-62.

145. Hannon GJ, RNA interference. Nature, 2002. 418(6894): p. 244-51. 146. Lum L, Yao S, Mozer B, Rovescalli A, Von Kessler D, Nirenberg M, et al.,

Identification of Hedgehog pathway components by RNAi in Drosophila cultured cells. Science, 2003. 299(5615): p. 2039-45.

147. Parasassi T, Brunelli R, Bracci-Laudiero L, Greco G, Gustafsson AC, Krasnowska EK, et al., Differentiation of normal and cancer cells induced by sulfhydryl reduction: biochemical and molecular mechanisms. Cell Death Differ, 2005. 12(10): p. 1285-96.

148. Wingo PA, Cardinez CJ, Landis SH, Greenlee RT, Ries LA, Anderson RN, et al., Long-term trends in cancer mortality in the United States, 1930-1998. Cancer, 2003. 97(12 Suppl): p. 3133-275.

149. Vogelstein B and Kinzler KW, The multistep nature of cancer. Trends Genet, 1993. 9(4): p. 138-41.

150. Paumgartner G and Beuers U, Mechanisms of action and therapeutic efficacy of ursodeoxycholic acid in cholestatic liver disease. Clin Liver Dis, 2004. 8(1): p. 67-81, vi.

151. Martinez JD, Stratagoules ED, LaRue JM, Powell AA, Gause PR, Craven MT, et al., Different bile acids exhibit distinct biological effects: the tumor promoter deoxycholic acid induces apoptosis and the chemopreventive agent ursodeoxycholic acid inhibits cell proliferation. Nutr Cancer, 1998. 31(2): p. 111-8.

152. Earnest DL, Holubec H, Wali RK, Jolley CS, Bissonette M, Bhattacharyya AK, et al., Chemoprevention of azoxymethane-induced colonic carcinogenesis by supplemental dietary ursodeoxycholic acid. Cancer Res, 1994. 54(19): p. 5071-4.

153. Ikegami T, Matsuzaki Y, Shoda J, Kano M, Hirabayashi N, and Tanaka N, The chemopreventive role of ursodeoxycholic acid in azoxymethane-treated rats: suppressive effects on enhanced group II phospholipase A2 expression in colonic tissue. Cancer Lett, 1998. 134(2): p. 129-39.

154. Park SE, Choi HJ, Yee SB, Chung HY, Suh H, Choi YH, et al., Synthetic bile acid derivatives inhibit cell proliferation and induce apoptosis in HT-29 human colon cancer cells. Int J Oncol, 2004. 25(1): p. 231-6.

155. Schlittenhardt D, Schober A, Strelau J, Bonaterra GA, Schmiedt W, Unsicker K, et al., Involvement of growth differentiation factor-15/macrophage inhibitory cytokine-1 (GDF-15/MIC-1) in oxLDL-induced apoptosis of human macrophages in vitro and in arteriosclerotic lesions. Cell Tissue Res, 2004. 318(2): p. 325-33.

156. Uchida K, Chaudhary LR, Sugimura Y, Adkisson HD, and Hruska KA, Proprotein convertases regulate activity of prostate epithelial cell differentiation markers and are modulated in human prostate cancer cells. J Cell Biochem, 2003. 88(2): p. 394-9.

157. Satyamoorthy K and Herlyn M, Cellular and molecular biology of human melanoma. Cancer Biol Ther, 2002. 1(1): p. 14-7.

158. Balch CM, Soong SJ, Gershenwald JE, Thompson JF, Reintgen DS, Cascinelli N, et al., Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol, 2001. 19(16): p. 3622-34.

159. Clark WH, Jr., Elder DE, Guerry Dt, Epstein MN, Greene MH, and Van Horn M, A study of tumor progression: the precursor lesions of superficial spreading and nodular melanoma. Hum Pathol, 1984. 15(12): p. 1147-65.

160. Danen EH, de Vries TJ, Morandini R, Ghanem GG, Ruiter DJ, and van Muijen GN, E-cadherin expression in human melanoma. Melanoma Res, 1996. 6(2): p. 127-31.

57

161. Furukawa F, Fujii K, Horiguchi Y, Matsuyoshi N, Fujita M, Toda K, et al., Roles of E- and P-cadherin in the human skin. Microsc Res Tech, 1997. 38(4): p. 343-52.

162. Ford D, Bliss JM, Swerdlow AJ, Armstrong BK, Franceschi S, Green A, et al., Risk of cutaneous melanoma associated with a family history of the disease. The International Melanoma Analysis Group (IMAGE). Int J Cancer, 1995. 62(4): p. 377-81.

163. Goldstein AM and Tucker MA, Genetic epidemiology of familial melanoma. Dermatol Clin, 1995. 13(3): p. 605-12.

164. Goldstein AM and Tucker MA, Genetic epidemiology of cutaneous melanoma: a global perspective. Arch Dermatol, 2001. 137(11): p. 1493-6.

165. Kefford RF and Mann GJ, Is there a role for genetic testing in patients with melanoma? Curr Opin Oncol, 2003. 15(2): p. 157-61.

166. Zhang Y, Xiong Y, and Yarbrough WG, ARF promotes MDM2 degradation and stabilizes p53: ARF-INK4a locus deletion impairs both the Rb and p53 tumor suppression pathways. Cell, 1998. 92(6): p. 725-34.

167. Pomerantz J, Schreiber-Agus N, Liegeois NJ, Silverman A, Alland L, Chin L, et al., The Ink4a tumor suppressor gene product, p19Arf, interacts with MDM2 and neutralizes MDM2's inhibition of p53. Cell, 1998. 92(6): p. 713-23.

168. Dong J, Phelps RG, Qiao R, Yao S, Benard O, Ronai Z, et al., BRAF oncogenic mutations correlate with progression rather than initiation of human melanoma. Cancer Res, 2003. 63(14): p. 3883-5.

169. Pollock PM, Harper UL, Hansen KS, Yudt LM, Stark M, Robbins CM, et al., High frequency of BRAF mutations in nevi. Nat Genet, 2003. 33(1): p. 19-20.

170. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, et al., Mutations of the BRAF gene in human cancer. Nature, 2002. 417(6892): p. 949-54.

171. Kolch W, Meaningful relationships: the regulation of the Ras/Raf/MEK/ERK pathway by protein interactions. Biochem J, 2000. 351 Pt 2: p. 289-305.

172. Moodie SA and Wolfman A, The 3Rs of life: Ras, Raf and growth regulation. Trends Genet, 1994. 10(2): p. 44-8.

173. Gray-Schopfer VC, da Rocha Dias S, and Marais R, The role of B-RAF in melanoma. Cancer Metastasis Rev, 2005. 24(1): p. 165-83.

174. Sahai E, Olson MF, and Marshall CJ, Cross-talk between Ras and Rho signalling pathways in transformation favours proliferation and increased motility. Embo J, 2001. 20(4): p. 755-66.

175. Marais R, Light Y, Paterson HF, Mason CS, and Marshall CJ, Differential regulation of Raf-1, A-Raf, and B-Raf by oncogenic ras and tyrosine kinases. J Biol Chem, 1997. 272(7): p. 4378-83.

176. Hagemann C and Rapp UR, Isotype-specific functions of Raf kinases. Exp Cell Res, 1999. 253(1): p. 34-46.

177. King AJ, Wireman RS, Hamilton M, and Marshall MS, Phosphorylation site specificity of the Pak-mediated regulation of Raf-1 and cooperativity with Src. FEBS Lett, 2001. 497(1): p. 6-14.

178. Albino AP, Nanus DM, Mentle IR, Cordon-Cardo C, McNutt NS, Bressler J, et al., Analysis of ras oncogenes in malignant melanoma and precursor lesions: correlation of point mutations with differentiation phenotype. Oncogene, 1989. 4(11): p. 1363-74.

179. van 't Veer LJ, Burgering BM, Versteeg R, Boot AJ, Ruiter DJ, Osanto S, et al., N-ras mutations in human cutaneous melanoma from sun-exposed body sites. Mol Cell Biol, 1989. 9(7): p. 3114-6.

180. Omholt K, Karsberg S, Platz A, Kanter L, Ringborg U, and Hansson J, Screening of N-ras codon 61 mutations in paired primary and metastatic cutaneous melanomas:

58

mutations occur early and persist throughout tumor progression. Clin Cancer Res, 2002. 8(11): p. 3468-74.

181. Gorden A, Osman I, Gai W, He D, Huang W, Davidson A, et al., Analysis of BRAF and N-RAS mutations in metastatic melanoma tissues. Cancer Res, 2003. 63(14): p. 3955-7.

182. Omholt K, Platz A, Kanter L, Ringborg U, and Hansson J, NRAS and BRAF mutations arise early during melanoma pathogenesis and are preserved throughout tumor progression. Clin Cancer Res, 2003. 9(17): p. 6483-8.

183. Sivertsson A, Platz A, Hansson J, and Lundeberg J, Pyrosequencing as an alternative to single-strand conformation polymorphism analysis for detection of N-ras mutations in human melanoma metastases. Clin Chem, 2002. 48(12): p. 2164-70.

184. Moore MH, Gulbis JM, Dodson EJ, Demple B, and Moody PC, Crystal structure of a suicidal DNA repair protein: the Ada O6-methylguanine-DNA methyltransferase from E. coli. Embo J, 1994. 13(7): p. 1495-501.

185. Wibley JE, Pegg AE, and Moody PC, Crystal structure of the human O(6)-alkylguanine-DNA alkyltransferase. Nucleic Acids Res, 2000. 28(2): p. 393-401.

186. Srivenugopal KS and Ali-Osman F, The DNA repair protein, O(6)-methylguanine-DNA methyltransferase is a proteolytic target for the E6 human papillomavirus oncoprotein. Oncogene, 2002. 21(38): p. 5940-5.

187. Delaney JC and Essigmann JM, Effect of sequence context on O(6)-methylguanine repair and replication in vivo. Biochemistry, 2001. 40(49): p. 14968-75.

188. Dumenco LL, Allay E, Norton K, and Gerson SL, The prevention of thymic lymphomas in transgenic mice by human O6-alkylguanine-DNA alkyltransferase. Science, 1993. 259(5092): p. 219-22.

189. Nakatsuru Y, Matsukuma S, Nemoto N, Sugano H, Sekiguchi M, and Ishikawa T, O6-methylguanine-DNA methyltransferase protects against nitrosamine-induced hepatocarcinogenesis. Proc Natl Acad Sci U S A, 1993. 90(14): p. 6468-72.

190. Becker K, Dosch J, Gregel CM, Martin BA, and Kaina B, Targeted expression of human O(6)-methylguanine-DNA methyltransferase (MGMT) in transgenic mice protects against tumor initiation in two-stage skin carcinogenesis. Cancer Res, 1996. 56(14): p. 3244-9.

191. Liu L, Qin X, and Gerson SL, Reduced lung tumorigenesis in human methylguanine DNA--methyltransferase transgenic mice achieved by expression of transgene within the target cell. Carcinogenesis, 1999. 20(2): p. 279-84.

192. Shiraishi A, Sakumi K, and Sekiguchi M, Increased susceptibility to chemotherapeutic alkylating agents of mice deficient in DNA repair methyltransferase. Carcinogenesis, 2000. 21(10): p. 1879-83.

193. Passagne I, Evrard A, Depeille P, Cuq P, Cupissol D, and Vian L, O(6)-methylguanine DNA-methyltransferase (MGMT) overexpression in melanoma cells induces resistance to nitrosoureas and temozolomide but sensitizes to mitomycin C. Toxicol Appl Pharmacol, 2006. 211(2): p. 97-105.

194. Pegg AE, Repair of O(6)-alkylguanine by alkyltransferases. Mutat Res, 2000. 462(2-3): p. 83-100.

195. Grombacher T, Mitra S, and Kaina B, Induction of the alkyltransferase (MGMT) gene by DNA damaging agents and the glucocorticoid dexamethasone and comparison with the response of base excision repair genes. Carcinogenesis, 1996. 17(11): p. 2329-36.

196. Biswas T, Ramana CV, Srinivasan G, Boldogh I, Hazra TK, Chen Z, et al., Activation of human O6-methylguanine-DNA methyltransferase gene by glucocorticoid hormone. Oncogene, 1999. 18(2): p. 525-32.

59

197. Bhakat KK and Mitra S, Regulation of the human O(6)-methylguanine-DNA methyltransferase gene by transcriptional coactivators cAMP response element-binding protein-binding protein and p300. J Biol Chem, 2000. 275(44): p. 34197-204.