ABERRANT DNA METHYLATION IN HUMAN NON-SMALL CELL LUNG CANCER
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Romulo Martin Brena, B.A.
*****
The Ohio State University
2007
Dissertation Committee: Approved by: Dr. Christoph Plass, Adviser
Dr. Thomas J. Rosol ______Dr. Michael C. Ostrowski Adviser Graduate Program in Molecular Genetics Dr. Gregory A. Otterson
Dr. Laura J. Rush
ABSTRACT
Lung cancer is the leading cause of cancer-related death worldwide. Given its impact on human health, extensive research is being conducted in an effort to reduce the global lung cancer death toll. Specifically, much interest has been placed on the development of biomarkers and the discovery of novel prognostic factors.
Over the past 2 decades it has become evident that the cancer genome is not only affected by genetic abnormalities, such as mutations, deletions and chromosomal rearrangements, but also by epigenetic changes which, together, contribute to the deregulation of transcription profiles. Epigenetic changes are defined as heritable lesions to the DNA affecting gene expression without altering the primary DNA sequence. These lesions typically involve a genomewide reduction in 5-methylcytosine, increased DNA methylation in gene promoter sequences and substitutions in histone tail modifications.
Epigenetic changes have been shown to interact with one another, resulting in genomic instability, silencing of tumor suppressor genes, activation of oncogenes and derepression of transposable elements. As opposed to the irreversible nature of genetic lesions, epigenetic lesions can be reversed. Because of their reversibility, epigenetic alterations have become an attractive target for new therapies, which has resulted in the development of new anticancer compounds, several of which are currently in clinical trials.
ii The field of epigenetics has expanded considerably since its inception.
Integrative research approaches aimed at elucidating the contribution of genetic and epigenetic alterations to the tumorigenic process are being undertaken worldwide, generating promising results. In hopes of furthering the body of knowledge currently afforded by the epigenetics field, the work presented in this thesis is focus primarily on unraveling the role of DNA methylation in the diagnosis, etiology and potential treatment of lung cancer.
Early detection would play a major role in reducing lung cancer-related death.
Since standardized early detection methods for lung cancer are currently lacking,
extensive efforts have been devoted in the lung cancer field to the identification of
molecular markers which might be useful for disease detection or which might afford
improvement in prognostic parameters. In recent months, several studies have reported
important advances in these two areas. Lu and colleagues were able to identify a gene
expression signature that helps predict survival of patients with stage I non-small cell
lung cancer. Following a similar investigative approach, Chen and colleagues reported a
5 gene expression signature that correlates with clinical outcome in non-small cell lung
cancer patients, regardless of stage. Other investigators, such as Guo et al and Raponi
et al have focus primarily on specific lung cancer subtypes and have reported molecular
classifiers that help refine the clinical prognosis of adenocarcinomas and squamous cell
carcinomas, respectively. Other prominent investigators in the lung cancer field have
focused their efforts on the discovery of markers which could be utilized for early
detection. A landmark study on this line of research has recently been published by
Shames and colleagues. These investigators reported a number of molecular markers
that could help detect not only lung cancer, but also other types of common neoplasias,
such as those of the breast and colon. The importance of this study resides in that it is
iii one of the few research endeavors geared towards the discovery of pan-cancer markers. Furthermore, Shames and colleagues decided to focus on aberrant DNA methylation as a marker for neoplasias, an emerging investigative approach of great potential that still remains to be fully explored.
Following this line of investigation, we hypothesize that epigenetic abnormalities,
in particular aberrant DNA methylation, is involved in the etiology of lung cancer.
Furthermore, we hypothesized that different non-small cell lung cancer subtypes can be
distinguished by their aberrant DNA methylation profiles. We tested our hypotheses by
analyzing human primary lung tumors via Restriction Landmark Genomic Scanning
(RLGS) in an effort to identify not only novel DNA methylation targets, but also targets
specific to lung cancer.
In Chapter 1, we introduce DNA methylation as biological process intimately
involved in the regulation of key cellular functions, from early development to adulthood.
We continue to expand on how DNA methylation is involved in human cancer and
particularly discuss the identification of key genes silenced by DNA methylation in lung
cancer. We conclude the chapter by discussing how DNA methylation could be used as
a biomarker and a prognostic marker in lung cancer.
Chapter 2 provides a synopsis of the techniques currently available for the
assessment of DNA methylation, coupled with a discussion on the suitability of each of
these techniques for their application in disease diagnosis and classification. We put
special emphasis on addressing the applicability of these techniques in clinical settings,
in an effort to highlight the current disconnect between bench work achievements and
their translation into novel approaches which could directly impact patient lives.
Chapter 3 describes a novel technique, Bio-COBRA, for the quantitative analysis of DNA methylation. Specifically, we discuss how the determination of DNA methylation
iv levels in a quantitative and reproducible manner could aid in increasing the efficacy of treatments involving DNA demethylating agents.
In Chapter 4, we describe a genomewide DNA methylation scan of primary human lung adenocarcinomas and squamous cell carcinomas. Our scan revealed that these two lung cancer subtypes can in fact be distinguished based on their aberrant
DNA methylation profiles. Also, we report the discovery a novel prognostic factor, oligodendrocyte transcription factor 1 (OLIG1), whose expression at the protein level was strongly correlated with survival in patients suffering from non-small cell lung cancer.
Chapter 5 presents a survey of lung-specific DNA methylation events, as determined by comparing NotI RLGS profiles from human cancers derived from 12
different organs. Our survey showed that lung cancer is the neoplasia with the highest
number of tumor-specific aberrant DNA methylation events. We performed extensive
mRNA expression and DNA methylation analyses in an effort to provide a
comprehensive report of the genes most frequently silenced by lung-specific DNA
methylation.
Finally, we conclude with Chapter 6, where we discuss the future steps that need
to be taken in order to further the current understanding of how aberrant DNA
methylation impacts the etiology of lung cancer.
v
Dedicated to my family and to all those whose constant support has helped me
made a small contribution towards the advancement of science
vi ACKNOWLEDGMENTS
The past 6 years have been a memorable journey. This journey, like many others, seemed, at times, of uncertain destination. However, now that I find myself at the end of its road, I can clearly see the destination had always been there. Knowing this about the journey might enable me to impart some words of wisdom to future travelers, to let them know that although it may seem unattainable at times, they will also get to the journey’s end. Past travelers did impart those same words to me, but they did not resonate as loudly as they would today. Nevertheless, their intention to pass on their experience and encouragement was unmistakable, for which I will always be grateful.
Many people have helped me get to the point where I can finally be writing these words. There are so many in fact that, I will certainly forget to mention some of them, for which I apologize in advance. First, I would like to thank Dr. Christoph Plass, whose incredible patience has made me realize that the best attribute of a great mentor is being able to make his students feel they are always welcome and that no question or concern is too trite for his attention.
Dr. Thomas Rosol and Dr. Laura Rush have been instrumental in my learning to look at scientific problems not just from a molecular standpoint, but as a puzzle that affects an entire being, whether it be human or animal, and that the answer to solving the puzzle most likely entails thinking about that being as a whole. I would also like to thank Dr. Rush for her constant support and for saying the right words at the right time to help me stay on track.
vii A very special thank you goes to Dr. Sandya Liyanarachchi, who not only played a central role in the statistical analysis of the data for almost all my research projects, but who also permanently took time out of her busy schedule to teach me about the theoretical background involved in her work. Dr. Liyanarachchi’s involvement in my graduate career has been fundamental in my training to be able to critically assess the validity of statistical analyses presented in biomedical research publications. Thank you
Dr. Liyanarachchi!
I sincerely thank my friends Kevin Poole, Stephen Lee, Abbey Carter, Kristin
Becknell, Paolo Neviani and Herbert Auer, who have always been there when I needed advice or would skillfully tell the right joke when I simply needed to laugh.
I also would like to thank my entire dissertation committee, Dr Thomas Rosol, Dr,
Laura Rush, Dr. Michael Ostrowski and Dr. Gregory Otterson for their continuous mentorship and support in all my scientific endeavors. Current and former members of the Plass lab have been equally important in my career development, by providing not only personal support, but also valuable scientific discussions which have helped me grow and become a better scientist.
Finally, I will forever treasure the support and encouragement given by my mother, Dr. Josefina Nicolao, who has always had the right words and the incredible wisdom to help me overcome the hurdles needed for me to be writing these words.
viii VITA
October 1, 1977…………………………...... Born – Mar del Plata, Argentina
December 2000…………………………...... B.A. Genetics and Microbiology Ohio Wesleyan University
January 2001-present………………………... Graduate Research Associate Department of Molecular Genetics Division of Human Cancer Genetics The Ohio State University
PUBLICATIONS
Research Publications
1. Dai Z, Lakshmanan RR, Zhu WG, Smiraglia DJ, Rush LJ, Frühwald MC, Brena RM, Li B, Wright FA, Ross P, Otterson GA, Plass C. Global methylation profiling of lung cancer identifies novel methylated genes. Neoplasia 2001 July;3(4):314-323.
2. Dai Z, Zhu WG, Morrison CD, Brena RM, Smiraglia DJ, Rush LJ, Ross P, Molina J, Otterson GA, Plass C. A comprehensive search for DNA amplification in lung cancer identifies inhibitors of apoptosis cIAP2 and cIAP2 as candidate oncogenes. Human Molecular Genetics 2003 Apr;12 (7):791-801.
3. Richard V, Luchin AI, Brena RM, Plass C, Rosol TJ. Quantitative evaluation of alternative promoter usage and 3' splice variants for parathyroid hormone-related protein by real-time reverse transcription- PCR. Clinical Chemistry 2004 Aug;49(8):1398-1402.
4. Sellers RS, Luchin AI, Richard V, Brena RM, Lima D, Rosol TJ. Alternative splicing of parathyroid hormone-related protein mRNA: expression and stability. Journal of Molecular Endocrinology 2004 Aug;33(1):227-241. ix 5. Park J, Brena RM, Gruidl M, Zhou J, Huang T, Plass C, Tockman MS. CpG island hypermethylation profiling of lung cancer using restriction landmark genomic scanning (RLGS) analysis. Cancer Biomarkers 2005 Feb;(1):193-200.
6. Weber F, Fukino K, Sawada T, Williams N, Sweet K, Brena RM, Plass C, Caldes T, Mutter GL, Villalona-Calero MA, Eng C. Variability in organ- specific EGFR mutational spectra in tumour epithelium and stroma may be the biological basis for differential responses to tyrosine kinase inhibitors. British Journal of Cancer 2005 May;92(10):1922-1926.
7. Smith LT, Lin M, Brena RM, Lang JC, Schuller DE, Otterson GA, Morrison CD, Plass C. Epigenetic regulation of the tumor suppressor gene TCF21 on 6q23-q24 in lung and head and neck cancer. Proceedings of the National Academy of Sciences of the United States of America 2006 Jan;103(4):982-987.
8. Brena RM, Auer H, Kornacker K, Hackanson B, Raval A, Byrd JC, Plass C. Accurate quantification of DNA methylation using combined bisulfite restriction analysis coupled with the Agilent 2100 Bioanalyzer platform. Nucleic Acids Research 2006 Feb;34(3):e17.
9. Tada Y, Brena RM, Hackanson B, Morrison C, Otterson GA, Plass C. Epigenetic modulation of tumor suppressor CCAAT/enhancer binding protein α activity in lung cancer. Journal of the National Cancer Institute 2006 Mar;98(6):396-406.
10. Brena RM, Auer H, Kornacker K, Plass C. Quantification of DNA methylation in electrofuidics chips (Bio-COBRA). Nature Protocols 2006 June;1(1):52-58.
11. Brena RM, Huang TH, Plass C. Toward a human epigenome. Nature Genetics 2006 Dec; 38:1359-1360.
12. Brena RM, Plass C, Costello JF. Mining methylation for early detection of common cancers. PLoS Medicine 2006, Dec;3(12)e479.
13. Brena RM, Morrison G, Liyanarachchi S, Jarjoura D, Davuluri RV, Otterson GA, Reisman D, Glaros S, Rush LJ, Plass C. Global DNA methylation profiling of non-small cell lung cancer identifies a differentially methylated gene, OLIG1, as a novel prognostic factor. PLoS Medicine in press
x Invited Reviews
1. Brena RM, Huang TH-M, Plass C. Quantitative assessment of DNA methylation: potential applications for disease diagnosis, classification and prognosis in clinical settings. Journal of Molecular Medicine 2006 May;84(5):365-377.
FIELDS OF STUDY
Major Field: Molecular Genetics
xi TABLE OF CONTENTS
Abstract……………………..………………………..…………………………………….…...... ii
Dedication……………………..………………………..…………………………………...... vi
Acknowledgments……………………………………..………………………………..……...vii
Vita……………………………………………………………………………….………….…....ix
List of Tables…………………………….………………..………………………….….…....xviii
List of Figures……………………………………………..………………………….……...... xx
Chapters:
1. The role of DNA methylation in human lung cancer…………………………………1
1.1 Human lung cancer…………………………………………………………….1
1.2 DNA methylation in cancer...…....…………….…….……………….....….....3
1.3 DNA methylation in human lung cancer...... …………...... …..4
1.4 Biomarkers in human lung cancer…….....………...... ……...5
1.5 Prognostic factors in human lung cancer……………...... ………7
2. Quantitative assessment of DNA methylation: potential applications for disease
diagnosis, classification and prognosis in clinical settings……..……....……..…....8
2.1 Introduction…….....……………………………….…….....………….....….....8
xii 2.2 Defining DNA methylation…...…....…………….…….………………..….....9
2.3 DNA methylation is essential for normal development…………...... …..12
2.4 DNA methylation and environmental exposure…….....………...... ……...13
2.5 DNA methylation in cancer…………….………………………...... ……….14
2.6 Epigenetic crosstalk: DNA methylation and histone modifications…...... 16
2.7 DNA methylation as a biomarker………………………………….....….….17
2.8 Techniques for the sequence-specific analysis of DNA methylation…....18
2.8.1 DNA methylation analysis using the MassARRAY system...... 19
2.8.2 MethyLight...... 20
2.8.3 Quantitative analysis of methylated alleles...... 20
2.8.4 Enzymatic regional methylation assay...... 21
2.8.5 HeavyMethyl: PCR amplification of methylated DNA using
methylation-specific oligonucleotide blockers...... 22
2.8.6 Quantitative bisulfite sequencing using the pyrosequencing
Technology...... 23
2.8.7 Quantification of DNA methylation differences at specific sites
using methylation-sensitive single nucleotide primer
extension………………………………………………………………24
2.8.8 MethylQuant: PCR-based quantification of methylation at specific
cytosines...... 24
2.8.9 Quantitative DNA methylation analysis based on four-dye trace
data from direct sequencing of PCR amplificates...... 25
2.8.10 Oligonucleotide-based microarray for DNA methylation
analysis...... 26
xiii 2.9 Techniques for the genome-wide analysis of methylcytosine content…..27
2.9.1 Reversed-phase high-performance liquid chromatography...... 27
2.9.2 Differential methylation hybridization...... 28
2.9.3 Restriction landmark genomic scanning (RLGS)...... 29
2.9.4 BAC microarrays for the high-resolution genome-wide analysis of
CpG island methylation...... 30
2.10 Concluding remarks...... 32
3. Accurate quantification of DNA methylation using Combined Bisulfite Restriction
Analysis coupled with the Agilent 2100 bioanalyzer platform…...... 34
3.1 Introduction……...... ………………….…….……………….....…...34
3.2 Methods…...….....………...... ….…...... …………….....….....36
3.2.1 Generation of DNA methylation standards and bisulfite DNA
treatment...... 36
3.2.2 PCR amplification and restriction enzyme digestion ...... …..….37
3.2.3 Electrophoresis on the Agilent 2100 bioanalyzer platform...... 38
3.2.4 Data analysis and quantification of DNA methylation...... 38
3.2.5 Quantitative real-time PCR………...... 39
3.3 Results...... 39
3.3.1 Measurement of a gradient of in vitro methylated DNA...... 39
3.3.2 Quantification of DNA methylation percentages...... 43
3.3.3 Determination of sensitivity, reproducibility and accuracy of Bio-
COBRA...... 45
xiv 3.3.4 Quantification of DNA methylation in human lung cancer cell lines
treated with 5-aza-2’deoxycytidine...... 49
3.4 Discussion...... 49
4. Global DNA methylation profiling of non-small cell lung cancer identifies OLIG1
as a novel prognostic factor…...... 52
4.1 Introduction……...... ………………….…….………………...... ….52
4.2 Methods…...….....………...... ….…...... …………….....….....53
4.2.1 Procurement of primary human tissue samples...... 53
4.2.2 Restriction landmark genomic scanning...... 54
4.2.3 Identification of RLGS Fragments...... 55
4.2.4 RNA isolation and quantitative real-time PCR...... 57
4.2.5 Combined Bisulfite Restriction Analysis and Combined Bisulfite
Restriction Analysis coupled with the Agilent 2100 bioanalyzer
platform...... 57
4.2.6 OLIG1 luciferase assay...... 58
4.2.7 5-aza-2’deoxycytidine and TSA treatment of human lung cancer
cell lines...... 61
4.2.8 Assessment of OLIG1 deletions in primary tumors...... 61
4.2.9 Bisulfite DNA sequencing...... 61
4.2.10 Immunohistochemical staining and scoring of primary lung tumor
tissue arrays and a lung cancer cell line array...... 62
4.2.11 Statistical analysis...... 65
xv 4.3 Results...... 66
4.3.1 Genome-wide DNA methylation analysis of human
adenocarcinomas and SCCs of the lung...... 66
4.3.2 Differentially methylated loci in adenocarcinomas and SCCs...... 70
4.3.3 OLIG1 in human lung cancer ...... 74
4.3.4 OLIG1 immunohistochemistry on lung tissue arrays...... 79
4.4 Discussion...... 83
5. Genome-wide and tumor-type specific aberrant DNA methylation is significantly
higher in human lung cancer compared to eleven other primary human
neoplasias...... 86
5.1 Introduction……...... ………………….…….………………...... ….86
5.2 Methods.....……...... ………………….…….………………...... 87
5.2.1 Collection of primary human tumors...... 87
5.2.2 Restriction landmark genomic scanning (RLGS)...... 88
5.2.3 Cloning of RLGS Fragments...... 88
5.2.4 RNA isolation...... 88
5.2.5 Quantitative real-time PCR...... 88
5.2.6 5-aza-2’deoxycytidine and trichostatin A treatment of lung cancer
cell lines……………………………………………………………….91
5.2.7 Combined Bisulfite Restriction Analysis (COBRA)………….……91
5.2.8 Statistical analysis……………………………..…………....…….…93
5.3 Results.....……...... ………………….…….………………...... 93
5.3.1 Aberrant DNA methylation levels in twelve primary human
neoplasias……………………………………………………………..93
xvi 5.3.2 Identification of aberrantly methylated genes in lung cancer…....97
5.3.3 Identification of lung cancer-specific aberrantly methylated
genes………………………………………………………………...101
5.3.4 Chromosomal distribution of aberrant DNA methylation in human
lung cancer…………………………………………………….…….103
5.3.5 Confirmation of DNA methylation in primary human lung cancer
samples via Combined bisulfite restriction analysis (COBRA)...106
5.3.6 Genes identified by RLGS in primary human lung cancers are
epigenetically regulated..……………………………………..……108
5.3.7 mRNA expression in primary lung tumors of genes identified by
RLGS ………………………….…………………………………….110
5.4 Discussion...... ………………….…….………………...... 112
6. Future Directions...... 117
6.1 The impact of aberrant DNA methylation in human lung cancer...... 117
6.2 Closing remarks...... 119
References...... 120
xvii LIST OF TABLES
Table 3.1 Fluorescent signals generated for the 1.6% DNA methylation standard for
SALL3, C/EBPα and TWIST2...... 45
Table 3.2 Percent DNA methylation determined for four different DNA
concentrations...... 46
Table 4.1 Clinical characteristics of the adenocarcinoma samples used for cluster
analysis...... 55
Table 4.2 Clinical characteristics of the squamous cell carcinoma samples used
for cluster analysis...... 56
Table 4.3 Primer sequences and PCR conditions utilized for real-time PCR,
COBRA, Bio-COBRA, luciferase and bisulfite sequencing assays...... 60
Table 4.4: Clinical characteristics of the subset of tumor samples present in tissue
array 1 which met all the quality control criteria to be considered for the
analysis for OLIG1 protein expression...... 64
Table 4.5: Chromosomal location and associated genes for the 33 out of 47 cloned
RLGS loci used to generate the tumor sample clusters...... 71
Table 4.6: Multivariate analysis of TMAs 1 and 2 combined...... 82
Table 5.1: Primer sequences utilized for measuring mRNA gene expression via real- time PCR……………………………………………………………..…...…...90
xviii Table 5.2: Primer sequences utilized for PCR amplification of COBRA
templates...... 92
Table 5.3: Percent DNA methylation, chromosomal location, name, molecular
function and biological process for the 142 genes identified as aberrantly
methylated in human lung cancer…………………………………………...98
Table 5.4: Percent DNA methylation, chromosomal location, name, molecular
function and biological process for the 71 genes identified as aberrantly
methylated exclusively in human lung cancer………………………....…102
Table 5.5: Observed vs. expected aberrant DNA methylation frequency per
chromosome in human lung cancer…………………………………….…106
xix LIST OF FIGURES
Figure 2.1 Common DNA methylation changes observed in cancer...... 11
Figure 3.1 DNA methylation standards for SALL3 and TWIST2...... 42
Figure 3.2 Plots of observed vs. expected methylation values for SALL3, TWIST2,
and C/EBPα methylation standards...... 44
Figure 3.3 Assessment of DNA methylation in clinical CLL samples and a human
lung cancer cell line...... 48
Figure 4.1 Aberrant DNA methylation profile and cluster analysis of
adenocarcinomas and SCCs of the lung...... 69
Figure 4.2: Real-time PCR analysis of three differentially methylated genes in
adenocarcinomas, SCCs and lung cancer cell lines, OLIG1
immunohistochemistry in H1299 cells and OLIG1 deletion analysis...... 73
Figure 4.3 OLIG1 luciferase promoter assay and bisulfite DNA sequencing...... 76
Figure 4.4 OLIG1 mRNA expression in primary tumor samples in relation to OLIG1
DNA methylation levels and deletions at the OLIG1 locus ...... 78
Figure 4.5 OLIG1 immunohistochemistry on a lung tissue array...... 80
Figure 5.1 RLGS analysis of 12 primary human neoplasias...... 96
Figure 5.2 Chromosomal location of aberrant DNA methylation events identified in
lung cancer……………………………………………………….…………..104
Figure 5.3 COBRA analysis of frequently methylated genes. ………………………107
xx Figure 5.4 mRNA expression of the top 30 most frequently methylated genes in
A549 and H719 lung cancer cell lines………………………………….…109
Figure 5.5 mRNA expression of the top 30 most frequently methylated genes in
primary human lung tumors………………………………………………...111
xxi CHAPTER 1
THE ROLE OF DNA METHYLATION IN HUMAN LUNG CANCER
1.1 Human lung cancer
Cancer is the leading cause of death worldwide1. According to the World Health
Organization (WHO), of the total 58 million deaths registered in 2005, 7.6 million or 13% were due to cancer (WHO 2006 Cancer Report). Cancer is an umbrella term utilized to define a group of more than 100 diseases that can affect virtually any part of the body. A hallmark of cancer is the rapid accumulation of abnormal cells which grow beyond their usual boundaries2. Frequently, these cells invade adjoining parts of the body and spread to other organs in a process referred to as metastasis. Metastatic spread is also the most prevalent clinical cause of cancer-relate death.
Of all types of cancer, lung cancer alone accounted for 17% (1.3 million) of all
cancer deaths in 2005, making it the neoplasia with the highest incidence and mortality
rate worldwide. In particular, the mortality due to lung cancer in United States was higher
than that of colorectal, breast and prostate cancers combined, representing 28% of all
cancer deaths (American Cancer Society 2005 report).
An encouraging statistic is that the incidence of lung cancer could be drastically
reduced by refraining from consuming tobacco products. Worldwide, tobacco use is the
single most important risk factor for cancer development. Specifically, tobacco use has
1 been linked to cancer of the lung, larynx, esophagus, stomach, bladder and oral cavity, among others.
It is now well established that environmental tobacco smoke, usually referred to as passive smoking, causes lung cancer. Twelve compounds in tobacco smoke
(benzene, arsenic, ethylene oxide, vinyl chloride, beryllium, chromium, cadmium, polonium-210, nickel compounds, 2-naphthylamine, 4-aminobiphenyl and benzo[a]pyrene) have been catalogued by the International Agency for Research on
Cancer as known human carcinogens3. However, The United States National Toxicology
Program estimates that at least 250 toxic or carcinogenic chemicals are generated by burning tobacco. Thus, extensive efforts have been launched at national and international levels to reduce the sale of tobacco products and to reduce the exposure of individuals to environmental tobacco smoke.
Clinically, lung cancer is divided into 2 main histological categories: small cell lung cancer and non-small cell lung cancer. Non-small cell lung cancer is further classified into 3 subtypes: adenocarcinomas, squamous cell carcinomas and large cell carcinomas. Small cell lung cancers have the poorest prognosis, are inoperable and therefore are generally treated through chemotherapy and radiation therapy4. Non-small
cell lung cancers comprise the majority of cases (<80%), can be surgically resected and
are characterized by better prognosis, which is reflected in longer overall patient
survival.
In light of the extensive body of knowledge currently available concerning the
etiology of lung cancer, it is important to question why this type of neoplasia accounts for
such a disproportionate percentage of cancer-related deaths. One key factor that sets
lung cancer apart from other neoplasias of comparable incidence is the lack of
affordable and effective early detection methods. Thus, a significant number of patients
2 present relatively advanced-stage disease at the time of diagnosis. Late diagnosis is often concomitant with metastatic spread, drastically reducing the patient’s chance of survival. In fact, the 5-year survival for all lung cancer cases combined is approximately
15%, a figure which has seen little improvement over the past two decades.
1.2 DNA methylation in cancer
From the early 1970s, researchers have observed changes in the DNA methylation levels of normal and cancer cells in response to various stimuli5,6. These first
observations primarily reported an overall reduction in the 5-methylcytosine (5meC)
content of cancer genomes7-9. However, studies soon started to emerge where gene-
specific changes in DNA methylation were measured as a result of cell maturation10,11, differentiation12-14 and oncogenic transformation15,16. As these lines of evidence
strengthen and grew in number, the hypothesis that DNA methylation patterns could be
related to the regulation of gene expression emerged17,18.
The DNA methylation field has advanced significantly over the past 2 decades and it is now well-accepted that the establishment and maintenance of DNA methylation patterns is essential for normal development19-22, initiation and preservation of genomic
imprinting23-26, X-chromosome inactivation27-29, overall genomic stability30-32 and regulation of tissue-specific gene expression33,34. Furthermore, scientists now recognize that DNA methylation is commonly altered in neoplastic transformation35-39.
In the realm of oncogenesis, DNA methylation took a central role when it became
clear that a significant part of the alterations observed in the cancer cell transcriptome
could not be explained solely by genetic events40,41. Thus, several techniques were developed that allowed for the assessment of DNA methylation at discrete genomic loci.
These techniques ranged from genomewide approaches, which could interrogate
3 hundreds to thousands of genes in a single assay42-44, to PCR based methods which focused primarily on single genes45,46. Since their inception, these methodologies have
made an enormous contribution towards the understanding of DNA methylation
metabolism in mammalian genomes.
In the year 2000, a landmark study reported that aberrant DNA methylation patterns
are non-random and exhibit tumor-type specificity47. This finding meant that in order to
fully elucidate the impact of DNA methylation in tumorigenesis, future research
endeavors would have to focus on distinct neoplasias and that the contribution of
aberrant DNA methylation in neoplastic transformation might be tissue or organ-type
dependent. Thus, over the past 7 years, a significant portion of studies examining DNA
methylation in cancer have been geared towards understanding the impact of aberrant
DNA methylation in specific cancer types.
1.3 DNA methylation in human lung cancer
There is ample evidence that DNA methylation patterns are profoundly altered in
lung cancer. In 1989, Shiraishi and colleagues reported high levels of DNA methylation
in chromosomes 3p and 13q48. Interestingly, these authors found that high DNA methylation in these chromosomal arms correlated with the retention of heterozygosity, since DNA methylation was essentially absent in tumors which had lost a copy of 3p and/or 13q. This evidence was the first to suggest that DNA methylation could functionally emulate loss of genetic material in lung cancer cells. A study by Makos et al49 reported hypermethylation of chromosome 17p. However, these authors were able
to show that 17p was also often reduced to homozygosity in primary lung tumors. Taken
together, these studies suggested that on specific chromosomal regions, DNA
methylation could inactivate 1 or 2 alleles, irrespective of copy number. The importance
4 of these findings resides in that it became evident that DNA methylation could provide 1 or 2 of the hits necessarily to inactivate tumor suppressor genes, as postulated by
Knudson’s 2 hit hypothesis for oncogenic transformation50.
In the early 1990s, Vertino and colleagues demonstrated for the first time that de novo methylation of CpG islands and demethylation of non-CpG island sequences occur at different stages of immortalization and oncogenic transformation of bronchial epithelial cells51. Since then, over 100 genes, many of them tumor suppressor such as
RASSF1A52, p1653, MLH154, MGMT55, BCL256, DAPK57, TCF2158 and BMP3B59 among
others, have been described as aberrantly methylated and silenced in human lung
cancer55,56,60-65, and the list continues to grow.
1.4 Biomarkers in human lung cancer
One important aspect of lung cancer is its low 5-year survival compared to other
neoplasias of equal or higher incidence. The main reason for the relative poor outcome
of lung cancer patients is that there are currently no standardized early detection
methods. Thus, by the time most patients present clinical symptoms, they also frequently
present advanced-stage disease. In an effort to address this shortcoming, several
research groups investigated the possibility of utilizing DNA methylation as a biomarker
for early detection of lung cancer. This line of research showed great potential, since for
almost 2 decades it has been known that metaplastic cells can be detected in the
sputum of patients with squamous cell carcinoma of the lung66. The outcome of these investigations has been promising; several assays have been developed to detect aberrant DNA methylation at the p16 locus67-69, among others, from bronchial lavage, sputum and serum of patients at risk of developing lung cancer (current or former
5 smokers). Importantly, it has been well-documented that aberrant p16 methylation can be detected in patients several years before the onset of lung cancer53,70,71.
A recent study has examined whether DNA methylation could become a pan-cancer
biomarker72. In their work, Shames and colleagues were able to identify aberrant DNA
methylation signatures common not only to lung, but also to breast, colon and prostate
cancers. This is an exciting finding, since it suggests that despite tissue and organ-
specific aberrant DNA methylation, there seems to be a number of genes which could be
used to accurately detect more than 1 type of neoplasia. Also, this finding can shed light
on the functional aspects these genes, since abrogation of their expression appears to
be important for cancer development and/or progression. However, more work still
needs to be done in order to elucidate the mechanism behind the epigenetic inactivation
of these genes.
Studies such as the one of Shames and colleagues also underscore the importance
of refining and increasing the sensitivity and specificity of DNA methylation detection
methods. In general, the amount of useful biological material that can be extracted via
non-invasive techniques, such as collection of sputum, is limited. Therefore, reliable
assays need to be in place if accurate diagnoses are to be made. In light of this need,
several techniques amenable for quantitative DNA methylation analysis, such as
pyrosequencing73 and matrix-assisted laser desorption/ionization time-to-flight
spectrometry (MALDI-TOF-MS)74, among others, have recently been introduced. These methods have been successfully utilized to assay small amounts of genetic material.
Nevertheless, because both methods require sophisticated and costly equipment, researchers are still interested in developing novel assays that can provide reliable answers to their specific scientific questions in a rapid and cost-effective manner.
6 1.5 Prognostic factors in human lung cancer
It is important to point out that significant progress has been made in the field of lung cancer prognosis. On the one hand, in this past year, numerous studies have reported specific gene expression signatures associated with survival in various subtypes of non- small cell lung cancer75-77. On the other hand, several other investigations focused
mainly on the impact of the mutation status of the epidermal growth factor receptor gene
on the outcome of patients treated with Gefitinib78-81. It should be noted that these
advances, though promising, are yet to be translated into new therapeutic modalities.
This disconnect between the laboratory bench and innovation in bedside treatments is
one of the reasons why the field of lung cancer biomarker discovery still remains wide
open. It should not come as a surprise that several research groups are devoting
extensive efforts in an attempt to identify novel molecular markers whose detection could
be easily integrated into existing protocols routinely performed in clinical settings.
Providing a prognostic marker that could be assessed with existing technology and
requiring only minimal additional training by health professionals and staff should greatly
increase its chances of making an impact on patient treatment. One approach that could
increase the chances of finding such markers is the combination of several non-
overlapping experimental techniques, with the objective of evaluating a given biological
process at different levels and from different angles. For example, the combination of
DNA methylation assays with gene expression arrays and protein detection methods
could prove invaluable if the molecular mechanism of a prognostic marker is to be
elucidated.
Without a doubt, new and exciting advances will continue to be made in the detection
and treatment of human lung cancer and hopefully, these advances will translate in
increase survival and better quality of life for lung cancer patients.
7 CHAPTER 2
QUANTITATIVE ASSESSMENT OF DNA METHYLATION: POTENTIAL
APPLICATIONS FOR DISEASE DIAGNOSIS, CLASSIFICATION AND
PROGNOSIS IN CLINICAL SETTINGS
Published in the Journal of Molecular Medicine 2006 May;84(5):365-377 by Brena RM, Huang T H-M and Plass C
2.1 Introduction
Deregulation of the epigenome is now recognized as a major mechanism involved in the development and progression of human diseases such as cancer. As opposed to the irreversible nature of genetic events, which introduce changes in the primary DNA sequence, epigenetic modifications are reversible and leave the original DNA sequence intact. There is now evidence that the epigenetic landscape in humans undergoes modifications as the result of normal ageing, with older individuals exhibiting higher levels of promoter hypermethylation compared to younger ones. Thus, it has been proposed that the higher incidence of certain disease in older individuals might be, in part, a consequence of an inherent change in the control and regulation of the epigenome. These observations are of remarkable clinical significance, since the
8 aberrant epigenetic changes characteristic of disease provide a unique platform for the development of new therapeutic approaches. In this chapter we address the significance of DNA methylation changes that result or lead to disease, occur with ageing or may be the result of environmental exposure. We provide a detailed description of quantitative techniques currently available for the detection and analysis of DNA methylation and provide a comprehensive framework that may allow for the incorporation of protocols which include DNA methylation as a tool for disease diagnosis and classification, which could lead to the tailoring of therapeutic approaches designed to individual patient needs.
2.2 Defining DNA methylation
Cytosine methylation is the most common base modification in the eukaryotic genome and is defined as the addition of a methyl group to the 5’-carbon of the pyrimidine ring to generate 5-methylcytosine (5meC)82,83. 5meC is preferentially found in the context of 5’-CpG-3’ (CpG) dinucleotides, although cytosine methylation has also been observed in 5’-CpNpG-3’and 5’-CpCpWpGpG-3’ sequences84-87. The methylation
reaction is catalyzed by a family of DNA methyltransferases (DNMTs) which utilize S-
adenosyl methionine (SAM) as a cofactor88. The function of DNA methylation in normal cells is diverse and it includes silencing of transposable elements, inactivation of viral sequences, maintenance of chromosomal integrity, X chromosome inactivation and transcriptional regulation of a large number of genes39,89-95.
Since 5meC has a relatively high propensity to spontaneously deaminate to thymine,
CpG dinucleotides are underrepresented in the human genome90. Interestingly, the
methylation status and distribution of CpG sites in the human genome is not random.
Approximately 80% of all CpG sites are methylated and located primarily in repetitive
9 sequences and the centromeric repeat regions of chromosomes96. The remaining 20% is unmethylated and preferentially found in short sequence stretches which range from 0.5 to 5 kb that occur at average intervals of 100 kb97. These stretches, or CpG islands, are
often methylation-free in somatic tissues and, to a large extent, have been maintained
through evolution. Current estimates indicate that 50% to 60% of human genes are
associated with a CpG island43,98,99.
The functional importance of CpG islands derives from the observation that changes
in their methylation levels results in altered expression of their associated genes (Figure
2.1). In general, genes associated with methylated CpG islands are either silenced or
downregulated100-102. Because of its potential to abrogate gene activity, DNA methylation has been proposed as one of the two hits in Knudson’s two hit hypothesis for oncogenic transformation92.
10
r e of ces y
cancer cells is
rmal tissues the majorit CpG islands as well oth
A) In no . B) The genome of d
and regional hypermethylation of gene enhancers, are methylation free. Repetitive sequen s es, however, are heavily methylate DNA methylation loss of obal l rsed CpG dinucleotid ory sequences. Common DNA methylation changes observed in cancer. d by g e
CpG islands and regulatory elements, such a and intersp characterize gene regulat Figure 2.1:
11 2.3 DNA methylation is essential for normal development
In recent years, the importance of DNA methylation in normal development has become evident. Studies based on knockout mouse models for any of the three DNA methyltransferases (Dnmt1, Dnmt3a and Dnmt3b) have demonstrated that the lack of any of these three enzyme activities in the mouse embryo results in embryonic or perinatal lethality, underscoring the essential role of DNA methylation in normal developmental processes22,103.
In humans, DNA methylation patterns are first established during gametogenesis.
However, the genetic material contributed by each of the gametes undergoes profound
changes after fertilization. A recent report indicates that the paternal genome is actively
demethylated in mitotically active zygotes. This active demethylation phase is followed
by a passive and selective loss of DNA methylation that continues until the morula
stage104. DNA methylation patterns are then reestablished after implantation and maintained through somatic cell divisions105.
A variety of human congenital malignancies are characterized by abnormal DNA
methylation during development. ICF syndrome, a rare disorder typified by
immunodeficiency, chromosomal instability and facial anomalies, has been linked to
mutations in a de novo DNA methyltransferase, DNMT3B, which result in the
hypomethylation of juxtacentrometic regions in chromosomes 1, 9 and 16103. Imprinting
disorders, such as Beckwith-Wiedemann and Prader-Willi/Angelman syndromes, are the
result of defects in the maintenance of the mono-allelic expression of imprinted genes.
Imprinted genes are expressed in a parental-specific manner and their expression is
regulated by DNA methylation of short regulatory domains termed differentially
methylated regions (DMRs). In typical Beckwith-Wiedemann cases, bi-allelic expression
12 of the insulin-like growth factor 2 (IGF2) gene is observed, a gene normally expressed only from the maternal allele106.
2.4 DNA methylation and environmental exposure
Monozygotic (MZ) twins develop when at least two daughter cells from a single embryo undergo independent mitotic divisions107. Given their origin, MZ twins are
considered to be genetically identical. However, it has been observed that phenotypic
discordances between them exist. These discordances include the incidence and/or time
of onset of various pathologies, of which schizophrenia and bipolar disorder have
received particular attention108. Current literature provides little evidence of cases where
a true genetic difference could account for an observed phenotypic discordance between
MZ twins. Furthermore, the cases presented are usually examples of well-established
genetic syndromes107.
Recent reports have highlighted the role of epigenetic mechanisms, especially DNA
methylation, as the potential cause for some of the common discordances and disease
traits observed in MZ twins. Interestingly, one study was able to show that MZ twins are
epigenetically impossible to differentiate at an early age109. However, older monozygotic twins exhibited prominent differences with respect to the distribution and overall content of 5meC. Most remarkable was the finding that those twins who reported having spent less of their lifetime together showed the highest differences with respect to 5meC content109.
Several studies have also focused on the influence of nutrition on DNA methylation.
Of particular interest is the role played by a set of nutrients directly involved in
regenerating or supplying methyl groups. Since methyl groups are intrinsically labile,
chronic deficiency in methyl-supplying nutrients can results in the direct or indirect
13 alteration of SAM to S-adenosylhomocysteine (SAH) ratios, consequently reducing the cellular potential for DNA methylation110. Nutrients that regenerate or supply methyl
groups fall into the category of lipotropes, and include folate, choline, methionine, and
vitamin B12. Riboflavin and vitamin B6 might also contribute to the modulation of DNA methylation processes since both of these nutrients are integral parts in 1-carbon metabolism111.
Studies in which rodents were subjected to diets deficient in different combinations of folate, choline, methionine, and vitamin B12 were able to show a reduction in the SAM to
SAH ratio in those animals. Furthermore, DNA hypomethylation could be detected at the
genomic level not only in specific tissues, but also at specific loci112-115. Taken together, these results suggest that the mechanisms regulating the epigenome can be influenced by environmental factors, such as geographic location, diet and lifestyle. Moreover, the modulation exerted by environmental factors on the epigenome can potentially contribute and/or trigger the development or onset of disease.
2.5 DNA methylation in cancer
Most of the current evidence linking DNA methylation, regulation of gene expression
and disease stems from studies of human cancers. Significant changes in genome-wide
DNA methylation have been observed in cultured cancer cells and primary human
tumors47,116. These changes include global DNA hypomethylation of centromeric repeats
and repetitive sequences and gene-specific hypermethylation of CpG islands. DNA
hypomethylation has been associated with chromosomal instability, resulting in
increased mutation rates and abnormal gene expression32,117,118.
In general, DNA hypermethylation of gene associated CpG islands results in either
downregulation or complete abrogation of gene expression, indicating that aberrant DNA
14 methylation could serve a similar function to genetic abnormalities, such as inactivating mutations or deletions in the disease state38. Numerous studies have indicated that several gene classes, such as adhesion molecules, inhibitors of angiogenesis, DNA repair, cell cycle regulators, and metastasis suppressors, among others, are frequently hypermethylated in human primary tumors70,102,119-123.
As opposed to the irreversible essence of genetic alterations that result in gene
silencing, the importance of understanding the mechanism involved in the epigenetic
abrogation of gene expression lies on the reversible nature of epigenetic processes.
Thus, a number of “epigenetic therapies” geared towards reversing aberrant epigenetic
events in malignant cells have been developed. Most of these therapies rely on the use
of two classic inhibitors of DNA methylation: 5-azacytidine and 5-aza-2’-deoxycytdine,
which were originally synthesized as cytotoxic agents12,124. Both molecules are potent
inhibitors of DNA methylation, and exert their action through a variety of mechanisms.
One of them is their incorporation into the DNA during S-phase, which results in the
trapping of DNMTs through the formation of a covalent bond between the catalytic site of
the enzyme and the pyrimidine ring of the azanucleoside. After the completion of each
cell cycle, concomitant to the depletion of DNMTs from the cellular environment,
heritable DNA demethylation is observed in cells treated with either of these agents 125-
129. Another report has demonstrated that both, 5-azacytidine and 5-aza-2’-deoxycytdine can induce the rapid degradation of DNMT1 by the proteasomal pathway, even in the absence of DNA replication130.
Despite the fact that when used in high concentrations azanucleosides exhibit high cytotoxicity, promising reports have emerged from clinical trials in which low doses of these agents administered in 3 to 10 day courses have been effective in treating some
15 myelodysplastic syndromes and leukemias126,131,132. For a comprehensive review on leukemia clinical trials involving the use of Decitabine please see133.
Recent reports have underscored the commonality of the epigenetic changes observed in cancer with those present in aging cells in normal tissues 38,134,135.
Consequently, a hypothesis has emerged, proposing that age-related methylation may act as a precursor for malignant transformation, thus helping to explain the age- dependent increase in cancer risk136.
2.6 Epigenetic crosstalk: DNA methylation and histone modifications
DNA methylation is not the only regulatory mechanism that comprises the
epigenome. Histone modifications have been the subject of intense investigation for
many years, and have actually been defined as epigenetic modifiers. Histones are the
target of several post-translational modifications, such as methylation, acetylation,
phosphorylation and ubiquitination, among others. Most of these modifications occur at
conserved amino-terminal domains and have been shown to be involved in the
configuration of chromatin structure. Typically, acetylated histones are associated with
relaxed and transcriptionally competent chromatin regions. However, hypoacetylated
histones are generally associated with transcriptionally silent regions, characterized by a
condensed chromatin structure. The term “histone-code” is currently used to describe a
number of histone post-translational modifications and the potential impact different
combinations of these modifications could have on gene expression, among other
cellular processes. However, histone modifications and their interplay with DNA
methylation is not the subject of this chapter. For comprehensive information on this
topic, please see137-141
16 2.7 DNA methylation as a biomarker
Given the role of aberrant DNA methylation in cancer initiation and progression, distinct effort has been put towards the development of strategies which could facilitate early cancer detection. It is now clear that aberrant DNA methylation is an early event in tumor development, as indicated by reports where aberrantly hypermethylated sites could be detected in seemingly normal epithelia from patients years before the overt development of cancer142.Thus, utilizing DNA methylation as a biomarker might prove to be a useful tool not only for early diagnosis, but also for the detection and assessment of high risk individuals. The importance of early detection is evident, since the 5 year survival rate for patients with breast, prostate or colon cancers, for which screening tests are available, is 4-6 times higher than that for lung cancer patients, for which no early detection protocol is currently implemented68.
For a biomarker to be clinically applicable it must be specific, sensitive and
detectable in specimens obtained through minimally invasive procedures. Promising
results have already been obtained, since aberrantly methylated CpG islands have been
detected in DNA samples derived from urine, serum, sputum and stool of cancer
patients143. Of importance, it should be noted that changes in DNA methylation also occur in normal epithelia. Thus, extensive research is currently underway to identify tumor-specific DNA methylation events that afford enough sensitivity and specificity to be utilized as biomarkers. Another major obstacle to overcome is the fact that tumor
DNA is present only in minimal amounts in bodily fluids. Thus, exquisitely sensitive techniques need to be utilized in order to detect and analyze tumor-derived DNA.
A wide array of techniques is currently available to measure DNA methylation genome-wide and at the single gene level. In general, genome-wide techniques for DNA methylation analysis require large amounts of DNA, which makes them unsuitable for
17 the analysis of biomarkers. These techniques, however, have been successfully utilized
to uncover novel tumor suppressor genes and to monitor global changes in DNA
methylation in health and disease47,116,119,144.
2.8 Techniques for the sequence specific analysis of DNA methylation
Over the past decade, a large number of techniques geared towards the analysis of
DNA methylation in short DNA stretches have been developed. Some of these assays,
such as methylation-specific PCR (MS-PCR), bisulfite sequencing, methylation-sensitive
single nucleotide primer extension (MS-SNuPE), and combined bisulfite restriction
analysis (COBRA) are well established in the DNA methylation field46,145,146. Because of
their high impact in previous DNA methylation studies and their current use in large
number of DNA methylation analyses, some well-established techniques will be
described in this chapter. However, we will focus primarily on newly developed assays
and recent technical improvements on well establish methods that have resulted in
either higher specificity or that have provided a quantitative platform for a well-
established technique, thus making them the most attractive candidates for the analysis
of DNA methylation focused towards the discovery and assessment of biomarkers.
Bisulfite treated DNA is the starting material for many DNA methylation techniques,
including most of the ones described in this chapter. Thus, the principle of bisulfite DNA
treatment will be briefly described.
Several DNA methylation assays involve one or more PCR steps. The problem that
stems from the use of PCR on genomic DNA is that the methylation marks found in the
genomic DNA template are not retained in the resulting PCR product. However, treating
genomic DNA with sodium bisulfite provides a solution to this problem. Under the
appropriate conditions, sodium bisulfite induces the deamination of cytosine to uracil
18 while 5meC remains unchanged. The net result of this reaction is DNA in which only methylated cytosines are retained and unmethylated cytosines are converted to uracil.
During PCR, then, a thymine nucleotide is incorporated in the PCR product for every uracil present in the bisulfite treated template. In the same fashion, a cytosine is incorporated in the PCR product for every 5meC found in the bisulfite treated template.
Overall, bisulfite DNA treatment followed by PCR results in the identification of 5meC in a given template by the presence or absence of cytosine residues in the PCR product45.
2.8.1 DNA methylation analysis using the MassARRAY system
This technique uses base-specific cleavage and matrix-assisted laser
desorption/ionization time-to-flight spectrometry (MALDI-TOF MS)74,147,148. After bisulfite treatment of genomic DNA, a T7-promoter tag is introduced through PCR. Next, an in vitro RNA transcription is performed on the reverse strand, followed by an RNaseA base-specific cleavage reaction (U or C). The cleavage products are analyzed in a
MALDI-TOF MS machine, which yields distinct signal patterns for the methylated and unmethylated templates. The MassARRAY system is capable of detecting DNA methylation levels as low as 5%. The main advantage of this technique is its ability to generate quantitative data for multiple CpG sites within a region of interest without the need for cloning of PCR products. Also, its reliance on bisulfite treated DNA makes it suitable for the analysis of samples obtained from various sources, such as paraffin blocks and laser capture microdissected specimens. It should be noted, however, that this technique requires multiple steps and sophisticated equipment which might not be available in all research settings.
19 2.8.2 MethyLight
MethyLight technology provides a tool for the quantitative analysis of methylated
DNA sequences via fluorescence detection in PCR reactions149. MethyLight relies on the bisulfite conversion of genomic DNA followed by a flexible PCR-based analytic platform.
Target sequence discrimination can be achieved at 3 levels: through the design of methylation-specific primers which may or may not overlap with CpG dinucleotides; through the design of the fluorescent probe, which could overlap one or various CpG sites; or both. Typically, primers that amplify both methylated and unmethylated sequences are used, coupled with a fluorescent probe overlapping two or more CpG sites. An attractive feature of MethyLight is that the fluorescent probe design can be used to detect specific DNA methylation patterns, not to simply discriminate methylated from unmethylated sequences. This flexibility could make it an excellent tool for the assessment of specific DNA methylation patterns that have been shown to possess prognostic value. Also, because of its reliance on PCR amplification, this assay is suitable for the analysis of samples where the available DNA amount maybe be small or not of the highest quality. Another attractive feature of this method is that the overall approach might be familiar to most researchers since it is entirely PCR based. However, it should be noted that careful design of primers and fluorescent probes, and the optimization of the PCR reaction itself are key in order to ensure the specific detection of the intended target sequence.
2.8.3 Quantitative analysis of methylated alleles (QAMA)
QAMA150 is a novel quantitative version of MethyLight149, which employs TaqMan
probes based on minor groove binder technology (MGB)151. Because of the improved sequence specificity of the probes, relative quantification of methylated and
20 unmethylated alleles can be achieved in a single reaction. Dual quantification is achieved through the use of different fluorescent dyes (VIC and FAM), to distinguish the signal emitted by the methylated-specific probe from that of the unmethylated-specific probe. The main advantage of QAMA is its simple setup, which makes it suitable for high throughput methylation analysis. Also, the equipment required to perform the assay is available in many research settings, given the frequent use of real-time PCR technology for quantitation of gene expression. It is should be noted that mutations or sequence polymorphisms might affect probe binding, thus yielding measurements not representative of the methylation status of the sequence under study. Finally, because the sequence of the fluorescent probes interrogates more than a single CG dinucleotide, only alleles either completely methylated or completely unmethylated generate a positive reading, excluding partial methylation patterns from the analysis. Thus, QAMA might not be suitable as a discovery tool, since it’s flexibility in the methylation patterns interrogated is limited. However, this method could provide a powerful analytical tool for the assessment of DNA methylation patterns whose clinical relevance has already been determined in large numbers of patient samples.
2.8.4 Enzymatic regional methylation assay (ERMA)
ERMA is a technique designed for the quantification of regional DNA methylation in a given sequence152. Following bisulfite treatment, DNA is amplified using primers specific for bisulfite-converted DNA, tailed with 2 GATC repeats at their 5’ end. The PCR product is subsequently in vitro methylated using 3H-labeled SAM as substrate. The
result of this first methylation reaction is the incorporation of 3H-methyl groups at all CpG sites that were methylated the original DNA template in the region flanked by the PCR primers. A second in vitro methylation reaction is then carried out, using 14C-labeled
21 SAM and dam methyltransferase, an enzyme that methylates cytosines in a GATC sequences. The outcome of this second methylation reaction is the addition 14C-methyl groups to the GATC sequences incorporated by the primers. Since the number of GATC sites is constant for every PCR product, 14C decay is used as an internal control for
normalizing the DNA amount assayed for each sample. Given that the amount of 3H- methyl groups incorporated into the PCR product is directly proportional to the level of
DNA methylation in the PCR amplificate of the original DNA template, the 3H signals can be compared across samples and methylation levels can be determined.
2.8.5 HeavyMethyl: PCR amplification of methylated DNA using methylation- specific oligonucleotide blockers
HeavyMethyl is an innovative real-time variant of the MS-PCR assay46, which because of its unique design, allows for the detection of methylated sequences at remarkably low concentration in a DNA mixture with high specificity153. In this technique, the PCR priming is methylation specific, but the high specificity of the assay stems from the use of non-extendable oligonucleotide blockers. The blockers are designed to bind to the bisulfite-treated DNA template in a methylation-dependent manner and their binding sites are selected so as to overlap with the 3’ primer binding sites. Using primers specific for GSTP1, HeavyMethyl has been successfully used to detect 30pg of in vitro
methylated and bisulfite treated DNA in a background of 50ng unmethylated DNA153.
The high sensitivity of HeavyMethyl makes it suitable for clinical applications, such as the analysis of DNA methylation in serum, where the amount of non-cell bound free- floating DNA in healthy patients is estimated at 10-50ng per milliliter154,155. An interesting feature of HeavyMethyl is that it can be adapted for qualitative as well as quantitative analysis of DNA methylation. It is important to note that HeavyMethyl requires more
22 components and potentially more optimization than conventional MS-PCR, which has been used with high sensitivity and specificity for a large number of genes. Thus,
HeavyMethyl could provide an attractive technical alternative when convention MS-PCR is unsuitable for the goal of a given research endeavor.
2.8.6 Quantitative bisulfite sequencing using the pyrosequencing technology
(QBSUPT)
Pyrosequencing is a sequence-by-synthesis approach that is based on the luminometric detection of pyrophosphate release following nucleotide incorporation156,157.
Depending on the chemistry used, a three to four enzyme cascade converts the released pyrophosphate to ATP, which is immediately hydrolyzed to produce light. Since a single known nucleotide is added sequentially in each step, the sequence of the template can be determined. Reports have indicated that the pyrosequence technology can be used for quantification of DNA methylation at CpG sites on bisulfite treated
DNAs73,158,159. Currently, pyrosequencing allows for the analysis of up to 10 CpG
dinucleotides spanning a 75 nucleotide stretch in a single run 160,161. The main advantage
of QBSUPT over conventional bisulfite sequencing is the fact that quantitative DNA
methylation information can be obtained from whole PCR products, without the need for
cloning and sequencing of a large number of clones in order to obtain statistically
relevant information. However, QBSUPT cannot be used for the analysis of haplotypes-
specific DNA methylation patterns.
23 2.8.7 Quantification of DNA methylation differences at specific sites using methylation-sensitive single nucleotide primer extension (Ms-SNuPE)
Single nucleotide primer extension is a well-established method which has been successfully used for the detection of gene mutations162 and for the quantitation of allele- specific expression163-165. Ms-SNuPE relies on single nucleotide primer extension to
assess DNA methylation at a specific cytosine145. An initial round of PCR is carried out
using bisulfite DNA-specific primers, followed by a second PCR step in which radio-
labeled dCTP and dTTP and an internal primer which terminates precisely 5’ of the
single nucleotide whose methylation status is to be determined are added. The radio-
labeled products are then run on a 15% polyacrylamide gel under denaturing conditions
and by visualized via exposure to an auto radiographic film or a phosphorimage screen.
The intensity of the observed bands can be then quantified to determine the proportion
of C:T at the cytosine of interest. Ms-SNuPE can be carried out in multiplex reactions,
allowing for the quantification of more than a single CpG site per assay. Ms-SNuPE is a
viable alternative when sensitive quantitation of a single or few CpG sites is desired and
small amounts of DNA are available.
2.8.8 MethylQuant: PCR-based quantification of methylation at specific cytosines
MethylQuant can be used to quantify the methylation level of a single cytosine
through the real-time amplification of bisulfite treated DNA166. Quantification is achieved via the comparison of real-time PCR reactions, one of which amplifies the target sequence irrespective of its methylation status (non-discriminative), while the other one only amplifies the methylated target (discriminative). Distinction between methylated and unmethylated sequences is accomplished by the complementary base pairing of the most 3’ end nucleotide in the primer. Through the comparison of the threshold crossing
24 cycle for the non-discriminative and discriminative reactions, a relative ratio between the methylated and unmethylated target can be obtained. One advantage of this method is that quantification can be achieved using SYBR Green I, which eliminates the need for fluorescently labeled probes, thus reducing the overall cost of the assay. Furthermore, given the widespread use of SYBR Green I in conventional real-time PCR assays, this technique could provide a suitable initial approach to DNA methylation analysis for researchers without prior experience in the DNA methylation field. However, as is the case with other PCR-based techniques described, careful primer design and optimization of the PCR reaction are critical in order to ensure the detection of the intended target sequence.
2.8.9 Quantitative DNA methylation analysis based on four-dye trace data from direct sequencing of PCR amplificates
Quantification of DNA methylation via conventional bisulfite sequencing is dependent upon the cloning and sequencing of individual PCR amplicons. This technique has been used extensively in the past with excellent results167-170. The two
main drawbacks of conventional bisulfite sequencing are the need for cloning of PCR
products and the large number of clones that need to be sequenced in order to obtain
statistically meaningful results. These requirements make the technique laborious and
expensive. In a recent report171,172 an algorithm is described that allows for the analysis of four-dye sequencing trace files obtained from direct sequencing of bisulfite PCR products in Applied Biosystems (ABI) machines. This algorithm yields quantitative methylation measurements for each cytosine present in the PCR product without the need for cloning and without the introduction of potential biases due to the cloning step itself. DNA methylation measurements generated from in vitro methylated DNA indicate
25 that this algorithm can yield accurate results for methylation differences of as low as
20%. Although this figure is relatively high compared to the sensitivity of other techniques, the low technical requirements to perform the assay might still make it a suitable choice for quantitation of DNA methylation of several CpG sites in a single run.
It should be noted that the algorithm and software needed to perform the technique just described are currently unavailable for public use.
2.8.10 Oligonucleotide-based microarray for DNA methylation analysis
Traditional PCR-based techniques for detection of DNA methylation are best suited for the analysis of single or a small number of genes. In recent years, however, research studies have focused on the concomitant investigation of DNA methylation in a relatively large number of genes173,174. Oligonucleotide-based microarrays for DNA
methylation analysis consist of pairs of methylated and unmethylated specific probes
that, through hybridization with PCR-amplified bisulfite treated DNA can reveal ratios
between the methylated and unmethylated form of a sequence. Each probe can
interrogate one to several CpG sites175, lending this system remarkable flexibility.
Because variations in the printing amount of oligonucleotide probes between a methylated/unmethylated pair and cross-hybridization between imperfect match probes and targets is likely to occur, a control experiment is required to normalize the system.
The data from the control experiment is typically used to generate a standard curve, so that the DNA methylation for a given locus can be extracted from calculated intensity ratios174. Oligonucleotide arrays have been successfully used to characterize the methylation profile of non-Hodgkin’s lymphomas and breast tumors175,176
26 2.9 Techniques for the genome-wide analysis of methyl-cytosine content
Measurement of the global content of 5meC is a useful parameter for the understanding of not only cellular homeostasis, but also the interplay between genome- wide alterations in DNA methylation and their effect on genomic stability and gene- specific alterations in epigenetic regulation177. Several assays are currently available for the assessment of the global levels of genome-wide methylation in DNAs. If the measurement desired is only the overall content of 5meC in a genome, that is, the ratio between total cytosine and total 5meC in a given sample, a chromatography-based method, such as reversed-phase HPLC can be utilized. On the other hand, if DNA methylation measurements taking place in discrete compartments of the genome, such as CpG islands or repetitive sequences are desired, methods such as RLGS, DMH and
BAC arrays, among others, could be employed.
2.9.1 Reversed-phase high-performance liquid chromatography (HPLC)
For many years reversed-phase HPLC has been the technique of choice for quantitation of global DNA methylation levels. Liquid chromatography-based assays generally rely on the total hydrolysis of genomic DNA by nuclease P1, or snake venom phosphodiesterase, followed by further processing to deoxyribonucleosides by alkaline phosphatase treatment. The free nucleosides, product of the two hydrolysis steps, are then injected into a column containing a silica-hydrocarbon stationary phase, over which a pressurized polar phased is run. The nucleosides are eluted from the column based on their solubility in the mobile polar phase and are detected and quantified through monitoring of ultraviolet (UV) absorbance as they exit the column82,178-181. Positive identification of the separated bases and further specificity has been achieved by combining HPLC technology with mass spectrometry182. It is important to note that the
27 choice of mobile and solid phases can greatly affect the efficiency of separation, as can pH variations in the mobile phase and fluctuations in the temperature at which the assay is carried out. Also, RNA contamination in the DNA preparation can result in overestimation of total 5meC. Reversed-phase HPLC is a good option when an accurate determination of total 5meC in a genome of interest is desired. This technique can be used on DNA extracted from various sources, such as mammalian and plant tissues.
However, relative large amounts of DNA are needed to perform the assay, as well as specialized equipment which may not be available in all research settings.
2.9.2 Differential methylation hybridization
Studies on global changes of DNA methylation at the CpG island level can also be achieved through the use of CpG island arrays. Differential Methylation Hybridization
(DMH) was the first successful attempt to build an array based DNA methylation assay.
The technique has been used to successfully identify epigenetic alterations in breast and ovarian cancers183,184. This technique has been further adapted to a microarray format by
printing 7,776 CpG island clones on a glass slide184 and is currently used on the 12,000
CpG island clone array manufactured by the University Health Network Microarray
Center, Toronto, Canada (http://www.microarrays.ca/). In DMH genomic DNA is digested
with MseI, a methylation insensitive restriction enzyme. Linkers are subsequently ligated
to the digested DNA and the ligation product is then digested with BstUI and HpaII, both
methylation sensitive restriction enzymes. The product of this second round of enzymatic
digestion is amplified by PCR using primers complimentary to the linker sequence. The
net result is the PCR amplification of methylated sequences not digested by BstUI or
HpaII. The PCR products are labeled with fluorescent dyes (Cy3 or Cy5) and then
hybridized to a CpG island array. Variations of this technique have been reported, in
28 which McrBc was used to digest the PCR products (instead of BstUI or HpaII)185,186.
These variations lend the described approach more flexibility and coverage, since the methylation status of different sets of CpG dinucleotides can be attained by simply changing the restriction enzyme combinations used in the experimental procedure. One attractive feature of this technique is that it allows for the potential identification of thousands of CpG islands that are the target of aberrant DNA methylation in a given sample in a single experiment. It should be noted that the specificity of DMH relies on the efficient digestion of genomic DNA by methylation sensitive restriction enzymes.
Thus, incomplete sample digestion could lead to the generation of false positive results.
2.9.3 Restriction landmark genomic scanning (RLGS)
As opposed to chromatography-based techniques, which only provide information on the overall content of 5meC in a genome of interest, RLGS allows for the preferential analysis of DNA methylation in the context of CpG islands. RLGS is a two-dimensional gel electrophoresis approach built upon the use of rare-cutting methylation sensitive restriction enzymes, which provide a platform for the simultaneous assessment of over
2000 loci42,44. The main strength of RLGS resides in the technique’s unbiased approach
towards the analysis of CpG islands irrespective of their association with known genes,
thus providing a unique tool for the discovery of novel hypermethylated sequences
mammalian in genomes. Furthermore, this method can also be applied to any genome
without prior knowledge of the DNA sequence. RLGS has been used for the
identification of novel imprinted genes and genes frequently hypermethylated in several
types of human cancers47,63,89,91,102,169,187-194, as well as regions of genomic hypomethylation195,196. One of the limitations of this approach is that methylation can only
be assessed in CpG islands which contain the sequence for the methylation-sensitive
29 enzyme used in the assay. Also, sequence polymorphisms in any of the enzyme recognition sequences required to perform RLGS or genomic deletions result in the effective loss of signal, which could be erroneously interpreted as DNA methylation.
Thus, other methods should be used in order to confirm RLGS data. Finally, the assay requires relatively large amounts of high molecular weight genomic DNA (greater than
1µg), which makes this approach unsuitable for the analysis of samples where the amount of DNA material recovered is low or highly fragmented.
2.9.4 BAC microarrays for the high-resolution genome-wide analysis of CpG island methylation
One of the difficulties of genome-wide methylation studies focused towards DNA methylation taking place primarily at CpG islands is the identification of the methylated target sequence. Techniques such as RLGS, for example, can provide an accurate overview of DNA methylation at the CpG rich restriction sites NotI and AscI, over 90% of
which occur within CpG islands47,197. However, the process of cloning and identifying the
sequence where the methylation signal is detected can be laborious and time
consuming. One strategy to overcome this difficulty is to work with a platform where
most or all potential target sequences are known a priori. In a recent report, a new
method is described that allows for the interrogation of CpG island methylation using
comparative genomic hybridization on a BAC array platform made of thousands of CpG
island containing BAC clones distributed across the genome198. In this approach, high molecular weight genomic DNA from a test and a reference sample is digested with NotI
and EcoRV. The digested NotI overhangs are then filled with biotin-labeled nucleotides
and purified with streptavidin-coated magnetic beads. The eluted DNA from the test
sample is subsequently labeled with Cy3, while the eluted DNA from the reference
30 sample is labeled with Cy5. Both DNA pools are then hybridized to a BAC array rich in
CpG islands, making it possible to distinguish differential methylation patterns taking place at the methylation-sensitive restriction enzyme site between the two DNA pools.
This approach has been successfully used to identify differences in tissue-specific DNA methylation in humans, as well as evolutionary conservations in tissue-specific DNA methylation patterns across species199. One important feature of this technique is its
flexibility to be adapted for the use of different methylation-sensitive restriction enzymes,
thus providing a platform for the potential assessment of thousands of DNA methylation
events on a single BAC array platform. A limitation of the BAC array, however, as is the
case with other genome-wide approaches based on methylation-sensitive restriction
enzymes, is the presence of sequence polymorphisms, which could result in false
positive or false negative results. Nevertheless, since all target sequences in the BAC
array are known, it is possible to map all known SNPs in the genome of interest so as to
predict which of the tested restriction sites might be affected by sequence
polymorphisms, effectively reducing the error rate of the assay199.
Other techniques using BAC arrays as the platform for analysis of DNA methylation have recently been reported200,201. The main differences among these methods lie in the manner in which methylated DNA is first detected in the genomic DNA pool and the resolution capacity of the array platform based on the type of material hybridized to the
BAC clones. Immunoprecipitation of methylated genomic DNA via the use of antibodies against 5meC followed by hybridization to BAC clones can result in low resolution when it comes to the methylation status of individual CpG islands, since each BAC clone may contain more than a single island in its sequence. However, this approach is suitable if the goal of the experiment is to elucidate average methylation levels in specific genomic regions. Due to the average sequence length of BAC clones and the likelihood that more
31 than one PCR product could hybridize to each clone, the hybridization of labeled PCR products to BAC arrays201 could also results mainly in the assessment of average DNA methylation levels over relatively large genomic regions.
2.10 Concluding remarks
The study of epigenetic alterations in the human genome has taken center stage in
an effort to better understand the molecular basis of human disease beyond the well-
documented realm of genetic events. The analysis of DNA methylation at global and
gene-specific levels has helped shed light on gene function and has also uncovered a
large number of genes whose expression is abolished primarily thought epigenetic
mechanisms in disease. Also, the fact that epigenetic changes are reversible opens a
new spectrum of potential treatment options which may lead to the amelioration or even
elimination of the disease phenotype.
There are currently many different approaches to generate DNA methylation data. A
large number of these are well-established and have been important tools for epigenetic
analysis for many years. However, no single technique provides an unambiguous
approach to DNA methylation data harvesting. Thus, we have tried to provide a
description of the advantages and disadvantages of various techniques, in an attempt to
provide a framework useful when deciding which method to use in order to generate the
most meaningful data.
Finally, we would like to emphasize the critical role of DNA methylation assays as
tools for the assessment of the effectiveness and safety of DNA demethylating agents,
as they potentially develop into standard regiments for cancer therapy. Drugs such as
Decitabine have shown promising results in clinical trials focused on the treatment of
solid and liquid tumors. However, due to the non-specific nature of nucleotide analogs, it
32 is critical to monitor their effect not only on neoplastic cells, but also on normal tissues to ensure no long-term damage is inflicted to unaffected targets.
A large body of evidence now exists indicating that not all possible DNA methylation targets in the human genome are affected equally in the disease state. The biological mechanism behind these observations is currently not fully understood, but could involve selection pressure or an intrinsic difference in sequence susceptibility to aberrant epigenetic changes. Thus the use of sensitive assays to monitor DNA methylation changes will play a key role in the development and implementation of new therapies aimed at modulating the epigenome.
33 CHAPTER 3
ACCURATE QUANTIFICATION OF DNA METHYLATION USING COMBINED
BISULFITE RESTRICTION ANALYSIS COUPLED WITH THE
AGILENT 2100 BIOANALYZER PLATFORM
Published in Nucleic Acids Research 2006 Feb;34(3):e17 by Brena RM, Auer H, Kornacker K, Hackanson B, Raval A, Byrd JC, Plass C
and
Nature Protocols 2006 June;1(1):52-58 by Brena RM, Auer H, Kornacker K, Plass C
3.1 Introduction
Epigenetic modifications, such as DNA methylation, are defined as heritable modifications to the DNA with the potential to alter gene expression while conserving the primary DNA sequence. Over the past decade, it has become evident that aberrant epigenetic alterations are a common feature of human neoplasias and play an important role in their development and progression38,92. DNA methylation occurs primarily in the context of 5’-CpG-3’ dinucleotides84,86,202. In the human genome, almost 90% of all CpG dinucleotides are located in repetitive sequences and are normally methylated. Most of the remaining 10% stay methylation-free, and are found in 0.5-4 kb sequence stretches termed CpG islands43,98. Interestingly, most CpG islands are located in close proximity of
34 genes or actually span gene promoters. The relevance of this observation rests upon the fact that these genes are consistently silenced when their associated CpG island is methylated203. Because of its potential to abrogate gene activity, DNA methylation has
been proposed as one of the two hits in Knudson’s two hit hypothesis for oncogenic
transformation92.
Studies have shown that aberrant DNA methylation can be detected in body fluids and secretions of patients years prior to the clinical diagnosis of cancer, suggesting that aberrant DNA methylation is manifested early in the process of malignant transformation38,39. Thus, much effort is being devoted to further characterize aberrant
DNA methylation patterns in several tumor types in an attempt to uncover specific patterns that might afford clinical diagnostic and prognostic value204-210. However, given the fact that normal DNA methylation patterns can vary among individuals, the specificity of one or several aberrant DNA methylation events might rest not only on which particular CpG dinucleotides are methylated, but also on their methylation frequency109.
This possibility presents an important challenge for the DNA methylation field, since the search for aberrantly methylated loci useful for early disease detection, assessment of disease risk or disease prognosis may involve focusing on subtle changes in DNA methylation. Thus, there is a need for a screening technique that will allow for the rapid and reliable evaluation of DNA methylation in large sample sets, while at the same time providing quantitative information on the level of aberrant DNA methylation and spatial information as to which CpG dinucleotides are preferentially methylated in a genomic region of interest.
Most techniques used to evaluate DNA methylation rely on the bisulfite conversion of
DNA45. One such technique, combined bisulfite restriction analysis (COBRA), involves the PCR amplification of bisulfite converted DNA followed by enzymatic digestion146.
35 COBRA is technically simple, and depending on the region being investigated, information on the methylation status of several CpG sites can be extracted in a single reaction. Because of these reasons, various DNA methylation laboratories use COBRA as a screening method for large sample sets. The main drawback of this assay is that quantitative information cannot be obtained from the visual inspection of restriction patterns.
The Agilent 2100 Bioanalyzer provides a robust platform for the quantification and high resolution of DNA fragments via electrophoresis in microfluidics chips211. This platform has been utilized in various studies, primarily with the goal of replacing or improving existing techniques, such as RFLP212,213, or attaining the visualization of PCR
products that, due to their low concentration, could not be detected in regular agarose
gels214. However, to our knowledge, no study has assessed the full potential of the
Bioanalyzer platform as a quantitative tool for the measurement of DNA methylation.
3.2 Methods
3.2.1 Generation of DNA methylation standards and bisulfite DNA treatment
Genomic DNA was isolated from normal peripheral blood lymphocytes (PBL) as
previously described44. 1µg of sheared DNA was incubated at 37°C for 4hs with 100U of
SssI (New England Biolabs, Beverly MA) and 2µl of 20mM S-adenosyl methionine. The in vitro methylation reaction was carried out twice, to ensure complete methylation. The
DNA was purified using Qiaquick columns (Qiagen, Valencia CA). The methylated and non-methylated DNAs were concentration adjusted to 20ng/µl and mixed in ratios to obtain samples with the following levels of DNA methylation: 1.6%, 3.2%, 6.4%, 12.5%,
25%, 50%, 75%, 87.5%, 93.6%, 96.8% and 100%. 1µg of each DNA mixture was
36 bisulfite treated as previously described46 and diluted to a final volume of 300µl with
ddH2O. 10µl of each mixture was used for PCR amplification.
3.2.2 PCR amplification and restriction enzyme digestion
SALL3, C/EBPα and TWIST2 PCR primers were designed to amplify bisulfite
treated DNA. The sequences of the primers used were: for SALL3, forward 5’-
GTTTGGGTTTGGTTTTTGTT-3’, reverse 5’-ACCCTTTACCAATCTCTTAACTTTC-3’, for
C/EBPα: forward 5’-TTGTTAGGTTTAAGGTTATTG-3’, reverse 5-
TCAACTAAACCCAAATAAAA-3’, for TWIST2: forward 5’-AAGGGGGAGGTAAAATTGAAA-3’, reverse 5’-CTAAACTAAATTACTAAATAATTATC-3’. PCR amplifications were performed as
follows: 95°C x 10', (96°C x 30", annealing x 30", 72°C x 30") for 35 cycles, with a final
step at 72°C for 10 minutes. The annealing temperatures and PCR product sizes were
59°C and 208bp for SALL3, 53°C and 150bp for C/EBPα and 52°C and 141bp for
TWIST2. PCR reactions were carried out in a 50µl volume containing 10X buffer46, 6µl of
each primer (10 pmol), 1µl (10mM) dNTPs, 2 units of Platinum Taq DNA polymerase,
29.25µl ddH2O and 10µl of bisulfite treated DNA. PCR amplifications were performed in a GeneAmp 9700 thermal cycler (Perkin-Elmer, Norwalk CT). PCR products were purified using Qiaquick columns, eluted in 40µl 10mM tris pH 8.0 and concentrated to a final volume of 7µl using a SpeedVac (Eppendorf, Hamburg Germany). Restriction digestions were performed using 10U of BstUI (New England Biolabs) in a total volume of 10µl at 60°C for 4 hours. 5µl of the digestion reaction was electrophoresed in an 8% polyacrylamide gel and visualized by ethidium bromide staining. Complete digestion of the PCR product was assessed by the lack of full length fragments in the 100% in vitro methylated samples.
37 3.2.3 Electrophoresis on the Agilent 2100 bioanalyzer platform
1µl of each of the digestion products was loaded onto a DNA 500 LabChip and assayed using the Bioanalyzer 2100. The chromatograms were visually examined, raw data was exported as CSV-files using the 2100 expert software and subsequently plotted to obtain the fluorescence values for each of the fragments. The sensitivity of the system was examined by determining the lowest percentage of the standard methylation mix that yielded a restriction fragment with fluorescence values above background. The background was defined as the mean plus 3 standard deviations of 10 measurements in front of the fluorescence signal peak.
3.2.4 Data analysis and quantification of DNA methylation
The fluorescence and migration time raw data for each sample were plotted into
Excel graphs. For quantification, the peak height generated by each DNA fragment was utilized. Thus, for each sample, a table was created listing the expected DNA size fragments and the fluorescent signal generated by each of those fragments. The methylation percent value for each sample was calculated using the following formula: fluorescence of methylated products/(fluorescence of methylated products + fluorescence of unmethylated product). The use of this calculation makes it possible to compare methylation percentages across an entire sample set, because the methylation value of each sample is normalized within itself by computing the total fluorescence generated by each sample. The methylation percentages for each of 12 data points of the in vitro methylated standard were plotted and a model was generated for each of the
3 genes tested. The in vitro generated methylation standard was tested at least 3 times
for each of the genes. The r2 values for the models was >0.98 in all cases.
38 3.2.5 Real-Time quantitative PCR
1µg of total RNA extracted from H1299 cells treated with 5-aza-2’-deoxycytidine was incubated with 2U of DNAseI (Invitrogen, Carlsbad CA) for 30 minutes at room temperature. The DNA-free RNA was reverse transcribed using 100U of SuperScript II
(Invitrogen) and 1µg of oligo dT per reaction. Quantitative C/EBPα expression was
measured using SYBR Green I (Bio-Rad, Hercules CA) in an I-Cycler (Bio-Rad).
Calcium/calmodulin-dependent protein kinase kinase 2 (CAMKK2) was used as internal
control. I-Cycler conditions were as follows: 10 min at 95°C; 35 cycles with 30s at 95°C,
followed by 30s at 60°C (for CAMKK2) or 64°C (for C/EBPα) and 30s at 72°C. At the end
of the amplification cycles, a melting curve was generated, yielding a single peak of the
expected melting temperature for the desired products. For the described assay the
following primers were used: C/EBPα forward 5’-TGTATACCCCTGGTGGGAGA-3’ and reverse 5’-TCATAACTCCGGTCCCTCTG-3’; CAMKK2 forward 5’-
CTCTTCCAGTGGGCAAAGAG-3’ and reverse 5’-GTGTCAACAAGGGGCTCAAT-3’. Prior to real-time PCR, a regular PCR was performed on DNAseI incubated but non-RT-treated samples in order to ensure that no DNA contamination was present in the RNA extract, given the fact that C/EBPα is an intron-less gene. The PCR products were run on an 8%
polyacrylamide gel. No product of the expected C/EBPα size was detected in those
reactions, indicating the absence of contaminating genomic DNA in the DNAseI treated
RNA extracts.
3.3 Results
3.3.1 Measurement of a gradient of in vitro methylated DNA
The Agilent 2100 Bioanalyzer provides a platform for the electrophoresis of nucleic
acids on a disposable chip213,215. In this study, chemistry suitable for the resolution of 39 fragments from 25bp to 500bp was utilized. In order to test our method, a methylation gradient was generated by mixing in vitro methylated DNA with PBL DNA. The samples were PCR amplified using 3 pairs of COBRA primer for 3 different genes (SALL3,
C/EBPα and TWIST2) and digested with BstUI (New England Biolabs). Digestion
products were electrophoresed in an 8% polyacrylamide gel and visualized by ethidium
bromide staining. Complete digestion of the PCR products was determined by the lack of
full length PCR fragments (208bp for SALL3 and 141bp for TWIST2) in the 100%
methylated samples (Figure 3.1.A and 3.1.B). 1µl of each digestion product was then
loaded into individual wells on a chip and electrophoresed in the Bioanalyzer. Virtual
gels were visually analyzed and the fluorescence data generated was tabulated and
graphed (Figure 3.1.C and 3.1.D).
40
Figure 3.1: DNA methylation standards for SALL3 (A) and TWIST2 (B). Fragment sizes are indicated to the right of the gels. Methylation percentages for each lane are indicated at the top. The restriction map of the sequence is indicated at the bottom of each gel. BstUI sites are indicated with vertical lines on the restriction map. C) Example
of a SALL3 virtual gel generated by the Bioanalyzer software. D) Fluorescence vs. time data plot for lanes 8 and 9 from Figure C. From right to left, the fluorescence peaks correspond to the following digestion fragments: 208bp, 124bp, 36bp and 26bp. The
22bp fragment overlaps with the front marker. As the methylation percent of the sample increases, there is a decrease in the fluorescence of the 208bp peak and an increase in the fluorescence of the digested peaks (75% vs. 87.5% plots) Plots were used to calculate methylation percentages for all standards and samples tested. RFU: relative fluorescence units.
41 42 3.3.2 Quantification of DNA methylation percentages
The 2100 expert software provides quantification for each DNA fragment.
However, quantification by this software is generated by calculating peak area measurements, which rely heavily on the definition of the start and end points of a peak
216. In order to eliminate the possibility of inaccurate quantification due to poor definition of peak areas, peak height was utilized as the quantification parameter. The peak heights of all digested fragment was added and then divided by the peak height of digested fragments plus undigested fragment for each sample, resulting in the observed methylation value. The observed/expected methylation values were plotted for the 3 genes (Figure 3.2). The equation derived from the model was used to calculate DNA methylation percentages in experimental samples.
43
Figure 3.2: Plots of observed vs. expected methylation values for SALL3, TWIST2, and C/EBPα methylation standards. Trend lines and R2 values are displayed for each
plot.
44 3.3.3 Determination of sensitivity, reproducibility and accuracy of Bio-COBRA
The sensitivity of Bio-COBRA was determined by testing the 1.6% sample of the methylation gradient. Because it is known that PCR amplification efficiency can vary significantly depending on the primer pair utilized and the target sequence, 3 genes were selected so as to provide a variable input for the assay. Fluorescent signals at least 2 fold above background could be generated from the restriction fragments of all 3 genes
(Table 3.1).
Since final DNA concentration after PCR amplification might vary among samples within a sample set, the effect of input DNA concentration on methylation measurements was tested for SALL3 PCR products. Input DNA concentration within the range of
10ng/µl to 65ng/µl showed no influence on methylation measurement. Also no positional effect within the Chip was observed (Table 3.2).
Table 3.1: Fluorescent signals generated for the 1.6% DNA methylation standard for SALL3, C/EBPα and TWIST2. The peaks for all DNA digestions fragments are indicated underneath each gene.
45
Table 3.2: Percent DNA methylation determined for 4 different DNA concentrations. The data generated indicates that DNA concentration in the range of
10-65 ng/µl do not affect DNA methylation measurements in this system.
The accuracy of Bio-COBRA was assessed by comparing TWIST2 methylation results obtained via this method with data generated through Southern blotting of
COBRA digests probed with radioactively labeled primers (for a description of the assay see168). The data generated by both methods was comparable, yielding similar overall methylation percentages for the sample set (Figure 3.3.A). To further validate these analyses, bisulfite DNA sequencing was performed in a subset of the samples168.
The reproducibility of Bio-COBRA was tested by comparing the methylation
percentages generated by the methylation gradient in at least 3 different runs of the
same restriction digest for each of the 3 genes. When different runs of the same gene
were plotted and compared, almost identical equations were derived from each one, all
of them with R2 values >0.98 (data not shown).
46
Figure 3.3: Assessment of DNA methylation in clinical CLL samples and a human
lung cancer cell line. A) Methylation levels of TWIST2 in 19 primary CLL samples
generated by Bio-CoBRA and Southern blot. B) Quantification of SALL3 methylation
levels in A549 cells treated with 5-aza-2’deoxycytidine (5-aza-dC). Three separate
measurements were performed for each sample. Different cultures of this cell line were
incubated with 1µm 5-aza-dC for the times periods indicated on the right. The standard
deviation (SD) and the coefficient of variation (CV) derived from each triplicate run are
indicated. C) Quantification of C/EBPα methylation levels in H1299 cells follow the same
scheme as in panel B. mRNA expression was not detected in the parental cell line by
real-time PCR (ND). Expression was detected after 48hs of treatment with 5-aza-dC.
The expression level measured at this time point was normalized to 1. C/EBPα mRNA
expression increased to 1.4 fold after 72hs of treatment with 5-aza-dC.
47
48 3.3.4 Quantification of DNA methylation in human lung cancer cell lines treated with 5-aza-2’deoxycytidine
DNA methylation levels of SALL3 were examined in A549 cells treated with 1µM
5-aza-dC for 24, 48 and 72hs. As expected, Bio-COBRA results showed a steady
decrease in DNA methylation at this locus with increasing exposure time of the cells to
the DNA demethylating agent (Figure 3.3.B). DNA methylation levels of C/EBPα were
tested in H1299 cells treated with 1µM 5-aza-dC for 48 and 72hs. In an effort to further
validate the results of our technique, Bio-COBRA was performed in 2 different regions of
the C/EBPα promoter (region 1: -1142 to -1121, region 2: -1271 to -1121 relative to the
transcription start site) and the methylation data was correlated with C/EBPα mRNA
expression (see materials and methods). As expected, DNA methylation decreased with
increased exposure of the cells to the DNA demethylating agent, and C/EBPα mRNA
expression increased (Figure 3.3.C).
3.4 Discussion
Epigenetic mechanisms play a major role in the initiation and progression of human
neoplasias. There is currently a large body of evidence that indicates DNA methylation
might be an early event in tumor development, since aberrantly methylated DNA
molecules can be found in secretions and body fluids of individuals years in advance to
the clinical diagnosis of cancer38,134,206. Thus, early detection of aberrant DNA
methylation patterns might provide a gateway for early disease detection and the
assessment of disease risk and disease prognosis. In order to achieve this goal,
however, sensitive, reliable and cost-effective assays for the quantification of DNA
methylation are needed.
49 COBRA is commonly used for screening aberrant DNA methylation in large sample sets. This is because COBRA allows for the interrogation of CpG sites over relatively large sequence stretches, and depending on the sequence being analyzed, the use of several restriction enzymes can increase the number of informative CpG sites examined within that sequence. COBRA is also technically simple and the assay is well- established in most DNA methylation laboratories. However, it is important to note that
COBRA only provides information on the DNA methylation status of those CpG sites which are part of the restriction enzyme’s recognition sequence. Thus, some sequences might not be suitable for COBRA analysis due to their lack of restriction enzyme sites.
The main drawback of COBRA is that quantitative information cannot be readily extracted from the assay. Attempts have been made to quantify COBRA results by blotting and hybridizing the restriction products with radiolabeled primers168. Though successful results were obtained from this approach, the method is laborious and it involves the use of a radioactive isotope. Attempts have also been made at quantifying
DNA fragment intensities through the use of imaging softwares on ethidium bromide stained gels. The problem of this approach is its intrinsically narrow dynamic range, leading to underestimation of strong signals. Furthermore, the fluorescence background of gel images is often variable (from gel to gel or from lane to lane within the same gel), affecting the calculations and thus making it impossible to reliably compare DNA methylation levels across a sample set. Other techniques, such as methylation-sensitive single nucleotide primer extension (MsSNuPE)145 and enzymatic regional methylation
assay (ERMA)152, though sensitive, are time consuming and also require the use of
radioisotopes, making them unsuitable for the high throughput screening needs of a
clinical setting.
50 In this study, we combined a standard COBRA assay with the quantification capability afforded by the Agilent 2100 Bioanalyzer. The main strength of our approach is that it allows for the rapid, accurate and cost-effective determination of DNA methylation percentages on a platform that enables the comparison of these values across large sample sets. As demonstrated, the data generated by this method is highly reproducible and by making use of an in vitro methylated DNA standard, experimental values can be converted to actual methylation values in one step. No DNA methylation standard is needed, however, if the goal of a screen is only a relative comparison of
DNA methylation levels across a sample set. Most importantly, no saturation of the system was observed within the dynamic range tested in our study (10-65ng/µl). The tested dynamic range ensures that virtually any PCR product can be digested without having to adjust the DNA concentration of any sample (this range encompasses PCR reactions performed in a 50µl volume with a total DNA yield ranging from 500ng to
3.25µg).
It should also be noted that the bisulfite DNA conversions performed in this study used 1µg of genomic DNA as substrate. However, bisulfite conversion of DNA has been successfully carried out using much smaller amounts of starting material217. Since the bisulfite converted DNA is later used as PCR template, the sensitivity of the primers and the intrinsic properties of the target sequence are the factors that determine as to how low an amount of DNA is needed as starting material. Based on the presented results,
Bio-COBRA affords an alternative approach to other well established methods, such as pyrosequencing73,160 and quantitative methylation specific PCR (QMSP)218, for quantitative DNA methylation analysis in epigenetic studies.
51 CHAPTER 4
GLOBAL DNA METHYLATION PROFILING OF NON-SMALL CELL LUNG
CANCER IDENTIFIES OLIG1 AS A NOVEL PROGNOSTIC FACTOR
Brena RM, Morrison C, Liyanarachchi S, Jarjoura D, Davuluri RV, Otterson GA, Reisman D, Glaros S, Rush LJ , Plass C. PLoS Medicine 2007 in press
4.1 Introduction
Lung cancer is the leading cause of cancer related death worldwide1. It is estimated that over 1.2 million people are diagnosed with lung cancer annually and 1.1 million die from the disease219. Despite intensive research over the past decades, the 5-year
survival of lung cancer patients remains poor220. Currently, the most accurate prognostic factor for patients with non-small cell lung cancer (NSCLC) is TNM clinico-pathologic staging221. Nevertheless, patients with early-stage lung cancer exhibit a wide spectrum of survival, indicating the need for additional prognostic parameters to better predict the outcome of the disease222. Thus, much effort has been dedicated to identify molecular markers that might improve the classification of NSCLC. Such markers should not only give prognostic information, but could also help identify patients that would benefit from novel therapeutic strategies or alternatively, those for which additional treatment is not
52 needed. A recent example of this is the identification of gene expression profiles that predict high risk of recurrence of localized lung cancer223.
Over the past decade it has become evident that the cancer genome is marked by
epigenetic modifications that contribute to the deregulation of transcription profiles38,39.
Of particular interest is that certain genes demonstrate differential susceptibility to epigenetic deregulation. That is, some genes are targeted for promoter methylation only in some tumor types47,224, while others are common targets for DNA methylation in
several types of neoplasias119. Thus, a genome-wide scan for DNA methylation in
NSCLC could uncover new clinically relevant molecular targets.
We analyzed primary human lung tumor samples via RLGS225 to identify DNA sequences differentially methylated between the two major NSCLC subgroups, adenocarcinomas and squamous cell carcinomas (SCCs). We uncovered promoter methylation patterns characteristic for both NSCLC subtypes and describe a novel marker, OLIG1, whose expression correlates with overall survival in NSCLC patients, as validated by univariate and multivariate analyses.
4.2 Methods
4.2.1 Procurement of primary human tissue samples
Primary lung cancer and adjacent tumor-free tissue samples were procured through the Cooperative Human Tissue Network at The Ohio State University James
Cancer Hospital and The University of Michigan following approved Internal Review
Board protocols. Consent from participants was waived under CFR 46 subpart A. A total of 70 snap-frozen matched tumor-free/adenocarcinomas and 70 snap-frozen matched tumor-free/SCCs were procured. For immunohistochemical studies, 2 lung tissue microarrays (TMAs) were generated. All specimens included in these arrays were cored
53 from formalin-fixed paraffin-embedded tissue blocks. TMA1 was comprised of 67 adenocarcinomas, 82 SCCs and 6 tumor-free lung samples arrayed in quadruplicate.
TMA2 was comprised of 74 adenocarcinomas and 79 SCCs arrayed in triplicate. DNA isolated from peripheral blood mononuclear cells (PBMCs) procured from random cancer-free donors was utilized as a negative control for DNA methylation. In an effort to facilitate the tracking of which tumor specimens were utilized only once in this study vs. those employed in several experimental approaches, each specimen has been assigned a unique identifier. Adenocarcinomas are denoted as Adeno followed by a number, while squamous cell carcinomas are denoted as SCC followed by a number.
4.2.2 Restriction landmark genomic scanning (RLGS)
RLGS was performed as previously described63. To avoid potentially confounding factors, such as age-related DNA methylation135, samples were selected so that gender, race and age range would be comparable between the adenocarcinoma and the SCC tumor subsets (Tables 4.1.A, 4.1.B, 4.2.A and 4.2.B). RLGS profiles of primary tumors and tumor-free lung from the same patient were superimposed and visually inspected for differences in the presence and/or intensity of radiolabeled fragments. The investigator performing the analysis was blinded as to the cancer subtype of each sample. The use of control tissues derived from the same patient as the tumor sample ensured that DNA polymorphisms that might be present at any of the restriction enzymes’ recognition sites would not introduce a bias in the analysis.
54 4.2.3 Identification of RLGS Fragments
RLGS fragments of interest which had not already been identified in our laboratory were cloned with the aid of either a human NotI–EcoRV or a human AscI-EcoRV
plasmid library, as previously described63,194,226 or by a PCR based approach194.
A SAMPLE Tumor Diagnosis Differentiation Gender Race Age ID DNA A 1 Adeno 90% Well F Caucasian 68 A 2 Adeno 90% Poor F Caucasian 80 A 3 Adeno 90% Poor M Caucasian 73 A 4 Adeno 90% Moderate M Caucasian 63 A 5 Adeno 81% Poor M N/A 78 A 6 Adeno 80% Poor F Caucasian 57 A 7 Adeno 90% Well F Caucasian 80 A 8 Adeno 75% Moderate M N/A 74 A 9 Adeno 90% Moderate F Caucasian 68 A 10 Adeno 76% Well F N/A 70 A 11 Adeno 80% Moderate F Caucasian 77 A 12 Adeno 100% Poor M N/A 60 A 13 Adeno 75% Poor F Caucasian 82 A 14 Adeno 100% Moderate M Caucasian 80 A 15 Adeno 77% Moderate F Caucasian 74 A 16 Adeno 70% Poor M Caucasian 71 A 17 Adeno 70% Poor M Black 59 A 18 Adeno 70% Poor F Black 68 A 19 Adeno 71% Well M Black 49
B
Gender Tumor differentiation Race Mean tumor DNA Mean age M 47% Well 21% Caucasian 73% 70 F 53% Moderate 32% Black 16% 82% (49-82) Poor 47% N/A 11%
Table 4.1: A) Clinical characteristics of the adenocarcinoma samples used to generate the clusters in Figure 4.1.A, 4.1.B and 4.1.C. B) Summary of the demographic and clinical features of the samples listed in part a. The age range is indicated in parenthesis. 55
A
SAMPLE Tumor Diagnosis Differentiation Gender Race Age ID DNA S 1 SCC 77% Moderate F N/A 72 S 2 SCC 84% Poor M Caucasian 71 S 3 SCC 90% Poor F Caucasian 66 S 4 SCC 86% Moderate F Caucasian 71 S 5 SCC 75% Well F Caucasian 47 S 6 SCC 88% Moderate N/A N/A N/A S 7 SCC 76% Poor M Caucasian 62 S 8 SCC 85% Poor M Caucasian 74 S 9 SCC 77% Poor F N/A 83 S 10 SCC 86% Moderate F Caucasian 75 S 11 SCC 82% Poor M Caucasian 64 S 12 SCC 70% Poor F Caucasian 70 S 13 SCC 78% Poor F Caucasian 68 S 14 SCC 74% Moderate M Black 65 S 15 SCC 80% Poor M N/A 71 S 16 SCC 100% Well M Black 51 S 17 SCC 80% Poor M Black 69 S 18 SCC 80% Moderate M Caucasian 67 S 19 SCC 75% Moderate M Caucasian 52 S 20 SCC 90% Well F Black 65 S 21 SCC 75% Poor M Caucasian 62
B
Gender Tumor differentiation Race Mean tumor DNA Mean age M 52% Well 15% Caucasian 62% 66 F 43% Moderate 33% Black 19% 81% (47-83) N/A 5% Poor 52% N/A 19%
Table 4.2: A) Clinical characteristics of the squamous cell carcinoma samples used to generate the clusters in Figure 4.1.A, 4.1.B and 4.1.C. B) Summary of the demographic and clinical features of the samples listed in part a. The age range is indicated in parenthesis.
56 4.2.4 RNA isolation and quantitative real-time PCR
Total RNA from primary human samples and human lung cancer cell lines was isolated and purified as previously described227. RNA integrity was assessed with the
Agilent 2100 Bioanalyzer using an RNA 6000 LabChip kit (Agilent Technologies, Palo
Alto CA). Only samples that showed high level of RNA integrity were used for reverse transcription216. For each sample, 1µg of total RNA was reverse transcribed using oligo dT (Invitrogen, Carlsbad CA), as previously described228. Given the fact that OLIG1 is an
intronless gene, regular PCR was performed on DNAseI treated but not reverse
transcribed RNA samples to ensure that no DNA contamination was present in the RNA
extracts. Quantitative OLIG1 expression was measured using SYBR Green I (Bio-Rad,
Hercules CA) in an iCycler (Bio-Rad). Calcium/calmodulin-dependent protein kinase kinase 2 (CAMKK2) was used as internal control228.
4.2.5 Combined bisulfite restriction analysis (COBRA) and combined bisulfite restriction analysis coupled with the Agilent 2100 Bioanalyzer platform (Bio-
COBRA)
COBRA was performed on BAHD1 and DMRTA1 as previously described229.
Briefly, an 181bp and a 218 bp fragments from the BAHD1 and DMRTA1 genes respectively, were amplified by PCR from bisulfite treated DNAs. The PCR products were purified and digested with 10U of BstUI (New England Biolabs, Beverly MA) at
60°C for 4 hs. The digested samples were electrophoresed in an 8% polyacrylamide gel and visualized via ethidium bromide staining.
Bio-COBRA was performed as previously described228,230 on 41 out of the 59
samples utilized to assess deletions at the OLIG1 locus. The reduction in the number of
samples analyzed by Bio-COBRA was due to limitations in the amount of tumor DNA 57 available from some specimens. Briefly, genomic DNA was isolated from human primary lung tumors, which was then mechanically sheared and bisulfite treated45. Bisulfite treated DNAs were PCR amplified with OLIG1 specific primers, purified and digested
with 10U of BstUI (New England Biolabs) at 60°C for 4 hs. 5µl of the digestion reaction was electrophoresed in an 8% polyacrylamide gel and visualized via ethidium bromide staining. 1µl of each digestion products was loaded onto a DNA 500 LabChip and assayed using the Agilent 2100 Bioanalyzer. Chromatograms were visually examined and the raw data generated from the assay was plotted to obtain the fluorescence values for each of the digestion fragments. The methylation percentage for each sample was calculated as follows: fluorescence of methylated products / (fluorescence of methylated products + fluorescence of unmethylated product).
4.2.6 OLIG1 luciferase assay
Four OLIG1 constructs were generated by PCR using primers tagged with NotI or
EcoRV sequence tails. The constructs were directionally cloned into a pGL3-Basic
vector (Promega, Madison WI) modified to contain NotI and EcoRV restriction
sequences in its multiple cloning site. A549 cells were plated at a density of 2 x 104 cells/35 mm well in RPMI-1640 medium (Cellgro, Herndon VA) supplemented with heat- inactivated 10% FBS (Cellgro) the day before transfection. The next day, cells were transfected as previously described227. A promoterless pGL3-Basic vector was used as
the negative control for expression and a pGL3-Basic vector containing the E2F3a
promoter was used as the positive control. Renilla luciferase was used as the
transfection efficiency normalizing factor. Luciferase activity was measured using the
Dual Luciferase assay system (Promega). All measurements were performed in triplicate
and the experiment was repeated 3 times.
58
Table 4.3: A) Primer sequences and PCR conditions used to evaluate mRNA expression in the genes listed. The OLIG1 and CAMKK2 primers were also used to
assess for OLIG1 deletions in primary tumors. B) Primer sequences with their
corresponding annealing temperatures used to amplify the BAHD1 and DMRTA1
sequences for COBRA and Bio-COBRA. C) Primer sequences with their corresponding
annealing temperatures used to amplify the OLIG1 constructs used in the luciferase
assays. D) Primer sequences with their corresponding annealing temperatures used to
amplify the OLIG1 regions selected for bisulfite DNA sequencing. *:The PCR condition
for these reactions was 95°C x 10', [(96°Cx30", 61°Cx30", 72°Cx20") x 35] followed by a
final extension at 72°C for 10' **:The PCR condition for these reactions was 95°C x 10',
[(96°Cx30", ATx30", 72°Cx20") x 35] followed by a final extension at 72°C for 10' AT:
Annealing temperature A: Primer set also used for Bio-COBRA.
59
e z
i
p p p p p p p p p p s t b b b b b b b ze i uc
A
150 150 179 163 b 166 b 191 b 220 247 219 165
s e od p t p r
z b i
b p P
p p p s 1 b t
8 3
218 s s s 0 uc oduc es es e e es es e es es es 2 ze r
l l l l l l l l l l , i 543 b 246 b 863 b
s od c c 1 c c c c c c c c
P 1 A r y y y y y y y y y y n s p p P t c c c c c c c c c c o
ti 5 5 5 g
35 3 3 35 3 35 35 35 35 35
n g 203 b ndi * * i
260 b x x x x x x x n x x x l * * * * * i o ) ) ) ) ) ) ) ) ) )
oduc
*
" " " " " " " " " c " a
r C C al
C C C C 0 0 0 0 0 0 0 0 0 0 ° ° ° n P 0 1 1 3 3 3 ne o i 60° 6 61° 61° 6 6 x 3 x x x x 3 x 2 x 3 x 3 x 3 x 3 n t
nne
a A C C C C C C C C C g A
c °C ° ° i
n f 0 2 2
i
* * i ’ ’ ’ ’ l
62° 6 6 60° 60° 62° 62° 62° 62°
l * * ’ 3 3 3 3
6
- - - - , , , , , , , , p , , 3 a - G G G G " " " " " " " " " " C C m ° ° T T T T 0 0 0 0 0 0 0 0 0 0
A
’ a 6 0 T T T T
3 A 3 3 3 5 5 - C C C C R nne A C x 3 x x x x 3 x 3 x 3 x 3 x 3 x 3 G G G G
C C A G G C C C C C C C C C C ° ° ° T C C P
6 6 6 C C C CCG CCG e
’ 9 9 96° 9 96° 96° 96° 96° 96° 96°
er C C C C ( ( ( ( ( ( ( ( ( ( C 3 ’ m
r CCCC 3 A A , , , - T m ' ' ' A- i e A A -ti C T 3 3', 3 3 3', 3', 3', 3', 3', 3', r l
A A AAA AAA T AA T a
m T T T T C P x x x x x x x i T CCT e r T T T T A e r T C x C x C C x C C C C C C T T CT R e s A ° ° ° C C A A r 5 5 5 C P C C CCT CCT e T m T 9 9 9 95° 95° 95° 95° 95° 95° 95°
C i e C C C C v T C r A AAAAA e s C C C C C A C T
r A A A A R T P
’ C C A T T
e ’
3’ ’ C 3 3’ ’ C 3’ e - C C A
3 - - v - 3’ ’ 3 - A 3 A C C CCT CCT - s A - C - T 3 T ’ A G G T T T T C r - T A T C 3 A CCC A G T A A A A R T A - Re A e A C C A A T T T T T A C T G A A T v A T A A A A A A A C C CC T C A T C C C T G G G G A A A T - - - -
G A C T ’ ’ ’ ’ G T G C C A A Re 5 5 5 5 G C T G C A A A er T A C G T AA T C G A A T GC T C G m C G T C
T A i T T ’
T G G C C r A T C T 3 A
- - - ’ T A CC ’ G ’ T T
G 3 G C - - C 5 G - G P A
’ ’ T ’ A CG
C T T 5 5 G T
T T G e 3 ’ A T - T T C C T 3 s A T C G 5 - ’ G r C G G T G C T
A 3 C C G T ’ ’ e - AA C A G G 3 3 T G A C G
T v C A A - ’ G A CCCA C C G A e 3 C C T T A A - C G G A
G T A A C C C R A C C A TT - A A G
T G G G A T G A r T A T T G G A C T G C G G C T T T
A G A e C A C T A
G G T T T G A G T G A G C T G G G G C T G G C T T er T m C C G
i T G TTT A G T A A G G G T T r C m A A i T A A T G CCG T G T G C C CT e T - - - G A ’ ’ ’ C
C pr T G GGGT A pr 5 5’- 5 5 5’- 5’- 5’- 5’- 5’- 5’- C A m T A d Y i TTTTTG d G T A T r T r T T G TTTTG G A G a
a
G
’ G ’ A ’ ’ ’ T A ’ G GG pr A 3’ ’ w 3 3’ 3 A 3 3 3 C T T 3 w ’ - - - - r A - - - A 3 - d C G G G 3 - A o r T A T - T A G G C GA A A ACCT T A T F C T A a T A A A G T A A G G T For T G G C C G T A T G A G T w TTTTT A A C C CCG CC A TTG G ACC A C G A C A
G G C G G
T A G T A GT A A r T G G - - C C CG CG C ’ ’ T G G A A GT For e C A 5 5 C C G C C A A G T A C G C C m G G T G G i G C C C G T T G TTTTT G G G G G C
T C T G C A GGGT A T C C C C G G A pr C T A A G G G G A C G C - - - - G d G G ’ ’ ’ ’ 31 499 G r A T T C A 5 5 5 5 G G T T A 1 a GG
- G A + - - A C A G ’ ’
G G
5 o 5 A w C A A AG
o CA C G
r A C t e
4 G A C G A T o G C T 7 4 4 2 n A
A F A G 96 t T C G A 6 6 C 8 2 e 9
2 A AAA T G G 8 1 C 31 A 391 t T C 9 - + G 1 -
-2 -5 T A C G A T G G - o o 1 - 1 uc A T C C C C T t T +4 o t
tr G G A C A A CCCT C T T t T I I o
1 1 s A C C G C T CCCC C G G L L 2 2 21 to 21 to t e ------n - - - ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ 6 - O O o n 5 5 5 5 5 5 5 5 5 5 9 1 1 1 1 391 C e - D G G G G I I I I 1
+2 G
L L L L 3 A 1 O O O O T
K
C 7 2 I2 2 R P 1
HD
P A A K
A A ne 1 M 2 1 7 1 e R K D BA Y 1 0 B G G RT P MK I HD R 53B kB C 5 L M F A L P 4 P
C O BA D MA S T N P S A
60 4.2.7 5-aza-2’deoxycytidine and TSA treatment of human lung cancer cell lines
Human NSCLC cell lines A549 and H1299 were cultured for 2 days and then treated with 1µM 5-aza-dC (Sigma-Aldrich, St. Louis MO) for 48 and 72 hs as previously described227. After treatment, total RNA was isolated as previously described227.
4.2.8 Assessment of OLIG1 deletions in primary tumors
DNA was isolated from snap frozen tissues as previously described63. The DNAs were sheared and diluted to a final concentration of 20ng/µl. Real-time PCRs were performed using SYBR Green I (Bio-Rad) in an iCycler (Bio-Rad). Calcium/calmodulin- dependent protein kinase kinase 2 (CAMKK2) was used as internal control. All reactions were performed in triplicate. The OLIG1 threshold crossing (Ct) value for each sample
was normalized to that of its internal control by subtracting the OLIG1 Ct from the
CAMKK2 Ct. The OLIG1 level in the tumor samples was calculated by the ∆Ct method,
setting the normalized OLIG1 values obtained from the matching the tumor-free DNA to
1. A sample was consider to harbor a deletion at the OLIG1 locus if reduction of OLIG1
at the DNA level was assessed to be >25% compared to its matching normal control231.
The overall comparison for the frequency of deletions between the adenocarcinomas and the SCCs was assess by a 1 tail Z-ratio and considered significant if the result of the test was P ≤ 0.050.
4.2.9 Bisulfite DNA sequencing
Bisulfite DNA sequencing was performed on two adenocarcinomas, two SCCs, and
the four tumor-free lung tissues from the same patients, as previously described227. Eight to ten individual clones were sequenced per sample.
61 4.2.10 Immunohistochemical staining and scoring of primary lung tumor tissue arrays and a lung cancer cell line array
Immunohistochemical staining of human primary lung tumor samples was performed on a tissue microarray (TMA1) comprised of formalin-fixed, paraffin- embedded specimens. Each specimen was present 4 times in the array232. The array contained 67 different adenocarcinomas, 82 different SCCs and 6 tumor-free lung samples (Tables 4.3.A and 4.3.B). Brain tissue cores were included as positive controls for OLIG1 staining. Validation of the immunohistochemistry results generated from
TMA1 was performed on an independent sample set (TMA2). This sample set was
comprised of 74 formalin-fixed, paraffin-embedded adenocarcinomas and 79 formalin-
fixed, paraffin-embedded SCCs arrayed in triplicate. A mouse monoclonal anti-OLIG1
antibody (R&D Systems, Minneapolis MN) was used at 1:1,000 dilution for
immunohistochemical detection. Antibody binding was detected by incubating the slides
with a secondary polyclonal anti-mouse IgG antibody (Amersham Biosciences,
Piscataway NJ). Positive staining was visualized by incubating the slides with diaminobenzadine (Sigma-Aldrich).
The slides were examined by an experienced lung pathologist (CM) and reviewed by the primary investigator (RMB). The evaluation of the immunohistochemical results was performed as follows: each tissue core was assigned an “OLIG1 index score”, calculated on two parameters, percent of positive (stained) cells in the tumor epithelium and intensity of staining233. Each parameter was subdivided into 3 categories: for percent of positive cells, 0% to10% was assigned a value of 1; 10% to 50% was assigned a value of 2 and >50% was assigned a value of 3. For the intensity of staining, no staining was assigned a value of 1, weaker than normal lung staining was assigned a value of 2 and staining as strong as normal lung was assigned a value of 3234. The OLIG1 index for 62 each core was then calculated by multiplying the value assigned to each parameter. In order to ensure the accurate assessment of OLIG1 protein expression in each tumor, either 3 or 4 cores of the same sample were placed in the tissue arrays. This designed helped overcome the problem of tumor heterogeneity, which could affect the results depending on what area of tumor is cored. The final OLIG1 index score for each sample was determined by taking the average of the indexes given to each individual core. The goal of the experiment was to be able to classify the samples into 3 categories: OLIG1 positive, OLIG1 negative and weak expression of OLIG1. An average index of 1-3 was
considered OLIG1 negative, an index of 4-5 was considered weak expression and an
index of 6-9 was considered OLIG1 positive. To further ensure the correct OLIG1 index
score was assigned to each sample, only the samples in which all cores were
individually scored within the same category (OLIG1 negative, weak expression or
OLIG1 positive) were counted and tabulated in the final report.
OLIG1 protein levels were also assessed via immunohistochemistry in H1299 cells treated with 1 µM 5-aza-dC for 48 and 72 hs. After treatment, the cells were collected,
embedded in agar pellets and fixed in formalin as previously described 227. After fixation,
each pellet was cored twice and placed on a single slide to create a cell line array.
OLIG1 protein detection was performed following the same protocol utilized on the
human primary tissue arrays, as previously described.
63
A
Total cases Gender T stage N stage M stage Mean age 59 Males 58% T 1 36% N 0 78% M 0 93% 61 Females 42% T 2 57% N 1 22% M 1 7% (32-83) T 3 7% N 2 0%
B
Total cases Gender T stage N stage M stage Mean age 74 Males 68% T 1 34% N 0 81% M 0 100% 62 Females 32% T 2 53% N 1 12% M 1 0% (34-82) T 3 13% N 2 7%
Table 4.4: Clinical characteristics of the subset of tumor samples present in tissue array 1 (TMA1) which met all the quality control criteria to be considered for the analysis for OLIG1 protein expression. The age range for the sample set is indicated in parenthesis underneath the mean age value. A) Adenocarcinoma samples included in the array. B) Squamous cell carcinoma samples included in the tissue array.
64 4.2.11 Statistical analysis
In order to identify candidate RLGS loci that show frequent methylation in one tumor subtype compared to the other, proportions of methylation in the two groups were compared. The Fisher’s exact test was applied to compare proportions, which avoids any violations of normal assumptions due to smaller sample sizes. Less conservative mid-p values were estimated and 47 RLGS loci with p < 0.06 were used for further
analysis. As methylation events are represented by binary variables, hierarchical cluster
analysis of patient samples was performed by applying Jaccard noninvariant coefficient
similarity metric235, using the 47 RLGS loci with p < 0.06. Cluster analysis was performed
three times, once with the initial group of 25 patients that was used to identify differential
DNA methylation between adenocarcinomas and SCCs, then with a set of 15 new
patients in order to validate the first result, and finally with both sample sets combined.
Real time PCR data were analyzed by applying one-way ANOVA analysis followed by
Scheffe test for multiple comparisons. Comparisons with p < 0.025 (97.5% CI) were
considered significant.
Kruskal-Wallis Rank sum tests and Fisher exact tests were used to compare
differences in baseline characteristics. Univariate and multivariate regression analyses
were performed using the Cox Proportional Hazard Regression Model to determine the
effects of various prognostic variables. Age was used as a dichotomous variable based
on the median age value of the patients in the sample sets. OLIG1 index was used as a
continuous variable comprised of 9 discrete values (1-9). In the multivariate model, the
assumption of proportional hazards was examined for each variable by testing the
significance of correlation coefficient between transformed survival time and the
Schoenfeld residuals of that variable. All statistical analyses were performed using Splus
and R (version 2.0.1) (http://www.r-project.org/) softwares. 65 4.3 Results
4.3.1 Genome-wide DNA methylation analysis of human adenocarcinomas and
SCCs of the lung
RLGS was performed on 11 adenocarcinomas (Adenos 1-11) and 14 SCCs
(SCCs 1-14) to determine if these two lung tumor subtypes could be differentiated based on their aberrant DNA methylation patterns. The samples were selected so that gender, race, age range and tumor differentiation were comparable in both groups. RLGS was performed using both NotI and AscI as restriction landmark enzymes. As previously
reported47, the recognition sequences of these enzymes occur preferentially within CpG
islands (CGIs) as defined by Gardiner-Garden and Frommer98, effectively creating a bias towards the assessment of DNA methylation in promoter sequences226. Additionally, recent bioinformatics analyses indicate that 92.7% of NotI sites fall within the 5’end,
inside or 3’end of transcripts (Dr. Davuluri, personal communication). The DNA
methylation profile from each tumor was scored against a profile generated from tumor-
free lung from the same patient. On average, the methylation status of 3,442 RLGS
fragments (range: 2,590-4,108) was analyzed per sample. The variation in the number of
RLGS fragments analyzed per sample stemmed from individual differences in the quality
of RLGS gels. Low level DNA degradation in specific samples resulted in RLGS
fragments located in the periphery of the gel to become diffuse or not separated well
enough to be analyzed accurately in all specimens. Aberrant DNA methylation was
detected at least once in 395 of the total 4,108 different RLGS loci scored. The average
frequency of CpG island methylation in the adenocarcinomas was 4.82% (range: 3.39%-
6.26%) and 4.23% (range: 3.13%-5.42%) in the SCCs. The methylation level for each
sample was calculated based on the exact number of RLGS loci scored for that sample.
66 Thirty-six RLGS loci, whose methylation frequency was significantly different (p ≤
0.050, Fisher’s exact test) between the adenocarcinomas and the SCCs, were identified.
Eight of these (22%) were methylated in only one of the tumor subtypes. The remaining
sequences were methylated in both subtypes but in varying frequencies (Figure 4.1.A).
Next, hierarchical clustering was performed to determine if the aberrant methylation events detected in our RLGS scan were sufficient to distinguish the adenocarcinomas from the SCCs (Figure 4.1.B). The best segregation of the tumors according to their subtype with the lowest number of misclassifications was achieved when the DNA methylation status of 47 RLGS loci was considered. While the adenocarcinomas clustered into one major group, the SCCs were split in to two groups, one of them branching closer to the adenocarcinomas (SCCs 2, 10 and 14). Also, SCC8 and SCC12 clustered within the adenocarcinoma group. In order to validate if the DNA methylation status of these 47 RLGS loci could be applied to distinguish a new set of adenocarcinomas from a new set of SCCs, RLGS was performed on 15 additional samples (Adenos 12-19, SCCs 15-21). These samples were also selected to ensure that gender, race, age range and tumor differentiation were comparable in both tumor subtypes. Hierarchical clustering of these 15 samples showed a pattern where, again, the adenocarcinomas separated in one major group, while the SCCs were split into two groups (Figure 4.1.C), a segregation pattern also seen in the combined cluster (Figure
4.1.D). Interestingly, most of the SCCs grouping close to the adenocarcinomas (SCC group 1) were moderately differentiated (4/6), while the SCCs clustering entirely separately from the adenocarcinomas (SCC group 2) were predominantly poorly differentiated (8/13). This distribution, though not statistically significant, could indicate a trend that the two aberrant DNA methylation patterns observed in SCCs may reflect, in part, the differentiation state of the tumor.
67
Figure 4.1: Aberrant DNA methylation profile and cluster analysis of adenocarcinomas and SCCs of the lung. (A) DNA methylation patterns of the 47
RLGS fragments that distinguish adenocarcinomas from SCCs. Black boxes indicate
DNA methylation; white boxes indicate absence of DNA methylation; red boxes indicate that the DNA methylation status of that RLGS fragment could not be determined. Each column represents a sample; each row represents an RLGS fragment. (B,C,D)
Hierarchical clustering of adenocarcinoma and SCC samples. (B) Cluster comprised of
25 samples, based on 47 DNA methylation events. (C) Cluster comprised of 15 samples, based on the DNA methylation information of the same 47 sequences as cluster B. (D)
Combined cluster from samples shown in clusters B and C.
68
69 4.3.2 Differentially methylated loci in adenocarcinomas and SCCs
Altogether, 33 of the 47 RLGS loci derived from our analysis were cloned either
previously or in this study63,194. Of those 33 sequences, 28 were associated with a CpG
island and 26 matched an annotated gene locus (Table 4.4). Notably, many of the
identified loci resided in chromosomal bands where loss of heterozygosity (LOH) had
previously been described in lung cancer and/or other neoplasias236. To prioritize the experimental evaluation of the identified genes, SYBR green real-time PCR was performed on a new set of 12 adenocarcinomas (Adenos 20-31) and 12 SCCs (SCCs
22-33). The assay was carried out on 13 genes, those with the highest degree of differential DNA methylation between the two tumor subtypes. The real-time PCR results highlighted that of these 13 genes, BAHD1, DMRTA1 and OLIG1 had the highest
differential mRNA levels between adenocarcinomas and SCCs (p < 0.025, ANOVA
followed by Scheffe) (Figure 4.2.A).
Next, the human lung cancer cell lines A549 and H1299, in which OLIG1, BAHD1
and DMRTA1 are methylated and not expressed, were treated with 1 µM 5-aza-dC for
48 and 72 hs. The mRNA levels of all genes were up-regulated in at least one of the cell
lines by the demethylating agent (Figure 4.2.B). To confirm these results, OLIG1
immunohistochemistry was performed on the H1299 cells. As expected, OLIG1 protein
expression was up-regulated upon treatment with the DNA demethylating agent (Figure
4.2.C). Due to the lack of commercial antibodies for BAHD1 and DMRTA1, COBRA was
performed on both genes (Adenos 20-29 and SCCs 23-33). Our results showed that
partial DNA methylation for BAHD1 was detected in 90% of the samples, while partial
DNA methylation for DMRTA1 was observed in 52% of them. (data not shown). These
observations indicate that expression of BAHD1, DMRTA1 and OLIG1 is directly or
indirectly regulated by DNA methylation.
70
RLGS Chromosomal Gene CGI CGI Reported locus location present location LOH n2E33 1p13.3 ALX3 Yes 5' Yes237 n4E07 1q23.3 Not annotated Yes 5' No a3F21 1q42.11 TP53BP2 Yes 5' No n4D22 2p13.2 EMX1 Yes 5' No n2E68 2p13.2 P450RAI-2 Yes 5' No n4D38 2p21 EPAS1 Yes 5' Yes237 238 n3B55 2q24.2 TBR1 Yes 3' Yes * n4D30 3p21.31 MAPKAPK3 Yes 5' Yes237 n4B44 3q21.3 CHST13 Yes 5' Yes237 n4G92 4q24 NFKB1 Yes 5' Yes239* a3E30 4q34.1 AK125257 Yes 5' Yes240 n2D10 5p15.33 Not annotated Yes No n3C74 5q14.1 SSBP2 Yes 5' Yes241* n4G43 9p21.3 DMRTA1 Yes 5' Yes237 n2D37 10q11.21 ALOX5 Yes 5' Yes242* n3F60 10q26.13 BUB3 Yes 5' No a2C13 11q25 ACAD8 Yes 5' Yes239* 237 n4F58 12q21.1 KCNC2 Yes 5' Yes n3G94 13q31.1 SPRY2 Yes 5' Yes243* n4E44 15q15.1 BAHD1 Yes 5' Yes244a n5E08 15q22.2 Not annotated Yes No n2C39 16p13.12 Not annotated Yes No n3D48 17q11.2 RNF135 Yes 5' Yes245* n5E41 18q21.1 Not annotated Yes Yes237 n3B47 18q23 SALL3 Yes 5' Yes237 n2D43 19q12 KIAA1474 No Yes237 n1D08 19q13.2 Not annotated Yes Yes243* n3E59 19q13.33 SLC17A7 No Yes237 n1F14 20q13.12 C20orf35 Yes 5' No n3D66 20q13.33 TCEA2 No No n2B59 20q13.33 BC052269 No No a4D15 21q22.11 OLIG1 Yes 5' + body Yes241,246 n4G68 22q13.32 Not annotated Yes Yes243*
Table 4.5: Chromosomal location and associated genes for the 33 out of 47 cloned
RLGS loci used to generate the clusters in Figure 4.1. The reported LOH column indicates whether the chromosomal band identified in our study has been previously associated with loss of heterozygosity in lung cancer. Reports of LOH on tumors derived from organs other than lung are indicated with a star (*). CGI: CpG island. Not annotated: No annotated gene is reported for the specified locus. CGI location: Location of the CGI with respect to the transcription start site of its associated gene. 71
Figure 4.2: Real-time PCR analysis of three differentially methylated genes in adenocarcinomas, SCCs and lung cancer cell lines, OLIG1 immunohistochemistry in H1299 cells and OLIG1 deletion analysis. All error bars indicate the SD of 9 different measurements. (A) Real-time PCR expression data for OLIG1, BAHD1 and
DMRTA1. ***: significant at the 97.5% confidence level. N: normal lung, A:
adenocarcinoma, S: SCC (B) OLIG1, BAHD1 and DMRTA1 mRNA expression in A549
and H1299 cell lines treated with 1µM 5-aza-dC for 72 hs. (C) OLIG1
immunohistochemistry on wild-type and 1µM 5-aza-dC treated H1299 cells (400X
magnification). (D) OLIG1 DNA level for adenocarcinomas and SCCs. Samples for
which the level of OLIG1 DNA was significantly lower than that of its matching tumor-free
lung DNA (P<0.050, 1 tail Student’s t-test) are indicated with a star (*).
72
73 4.3.3 OLIG1 in human lung cancer
Our DNA methylation, mRNA expression and 5-aza-dC reactivation data, coupled with literature describing recurrent LOH at chromosome 21q22.1 in SCCs of the lung241,246, led us to select OLIG1 for further study. Two reports described frequent LOH
at microsatellite marker D21S12070 (43.8%) located 2.74Mb upstream and marker
D21S1445 (39.3%) located 0.93Mb downstream of OLIG1241,246. Given the large
distance between the two microsatellite markers, we tested the frequency of OLIG1
deletions by directly assessing the presence of the OLIG1 gene sequence in a subset of
primary tumors. The assay was performed via quantitative real-time PCR on 25
adenocarcinomas (Adenos 20-44) and 34 SCCs (SCCs 22-55). We found that that 36%
(N=9) of the adenocarcinomas and 59% (N=20) of the SCCs showed loss of OLIG1 DNA
compared to tumor-free lung and the frequency of deletion was significantly higher in
SCCs (p = 0.042, 1 tail Z-test) (Figure 4.2.D). This result is in agreement with previously
published studies, reporting significantly higher rates of LOH in SCCs than in
adenocarcinomas241,246. The DNA methylation data generated by RLGS showed the
same trend, with the frequency of OLIG1 DNA methylation being significantly higher in
SCCs.
To determine the location of the OLIG1 promoter, we generated 4 luciferase
constructs (Figure 4.3.A). The constructs were transfected individually into A549 cells
and assayed for luciferase activity. Our results showed that the region 267bp upstream
of the OLIG1 transcription start site (TSS) was sufficient to drive luciferase expression,
and that a putative enhancer element might be located between -267bp and -566bp, due
to the significantly higher luciferase activity of the longer construct (p < 0.001, ANOVA)
(Figure 4.3.A). Thus, we focused on the 560bp region upstream of OLIG1 for further
DNA methylation analysis.
74
Figure 4.3 OLIG1 luciferase promoter assay and bisulfite DNA sequencing. The
OLIG1 CGI is indicated by vertical lines, each of which represents a single CpG. The gene is represented by a grey box with an arrow indicating the transcription start site.
The location of the AscI site is indicated. The gene diagram and constructs are drawn up
to scale. (A) OLIG1 gene diagram and luciferase activity determined for 4 deletion
constructs in A549 cells. The E2F3a promoter was used as a positive control for
luciferase activity. The error bars indicate the SD of 3 independent triplicate
transfections. (B) Bisulfite DNA sequencing of OLIG1 in 2 adenocarcinomas, 2 SCCs
and 4 tumor-free lung samples derived from the same patients. Each line represents an
individual clone and each circle represents a CpG dinucleotide. ● indicate methylated
cytosines; ○ indicate unmethylated cytosines.
75
76 Bisulfite DNA sequencing was performed on 8 human lung samples (2 adenocarcinomas, 2 SCCs and their matching tumor-free lung tissues). A 260bp PCR product spanning from -391bp to -131bp containing 25 CpG dinucleotides was generated. Another 203bp PCR product containing 18 CpG dinucleotides was produced to cover the region from +296bp to +499 bp, where the AscI site (landmark enzyme in
RLGS) is located. In both regions tested, the levels of DNA methylation were significantly higher in SCCs than in adenocarcinomas (p < 0.001, 1 tail Z-test) (Figure
4.3.B).
In order to establish a correlation between OLIG1 DNA methylation, frequency of
deletions at the OLIG1 locus and OLIG1 mRNA expression, Bio-COBRA, a technique
which allows for the rapid and accurate quantification of DNA methylation in a sensitive
and reproducible manner,228,230 was performed on a subset (41 out of 59) of the samples utilized to generate the OLIG1 deletion data already described. The DNA methylation
status of 4 BstUI sites was measured in a 260 bp PCR product extending from -391bp to
-131bp of the OLIG1 locus. DNA methylation was detected in 26 samples, ranging from
7.0% to 100% (mean 54.9%). These DNA methylation values were then combined with
mRNA expression and deletion data. Eleven out of 13 samples in which DNA
methylation alone was detected showed reduced mRNA expression levels compared to
normal lung, as also did 7 out of 9 samples in which OLIG1 deletions alone were
detected. All 13 samples in which concomitant OLIG1 DNA methylation and OLIG1
deletions were detected showed reduced mRNA levels, while 2 out of 6 of the samples
in which no DNA methylation or deletions were assessed showed a reduction in OLIG1
mRNA expression (Figure 4.4). Taken together, these data indicate that DNA
methylation and deletions at the OLIG1 locus in primary human lung tumors can be
correlated with a reduction in OLIG1 at the mRNA level.
77
Figure 4.4 OLIG1 mRNA expression in primary tumor samples in relation to OLIG1
DNA methylation levels and deletions at the OLIG1 locus. Each circle represents a single sample. The presence of DNA methylation and/or deletions is indicated at the bottom of each sample column. mRNA expression levels are indicated in relation to normal lung, which was arbitrarily set as 1. 78 4.3.4 OLIG1 immunohistochemistry on lung tissue arrays
OLIG1 immunohistochemistry was performed on a tissue microarray (TMA1) comprised 59 adenocarcinomas (Adenos 45-103), 74 SCCs (SCCs 56-129), 6 tumor- free lung, and 4 human brain specimens. The immunohistochemical results were scored and an OLIG1 index value was assigned to each sample. The index values ranged from
1 (no expression) to 9 (normal expression). Positive staining was detected in nuclei, indicating the correct localization of the target protein (Figure 4.5A-H). Our analysis determined that 78% (N=46) of adenocarcinomas and 58% (N=42) of SCC were either negative or expressed OLIG1 protein at low levels. In light of the high number of OLIG1 negative and low expressing cases in both lung tumor subtypes, we hypothesized that
OLIG1 protein expression may influence survival in NSCLC patients.
To test this hypothesis, univariate and multivariate analyses were performed. All clinical and geographical variables available for the data set (gender, age, tumor subtype, T and N stages) were included in the models in order to account for potentially confounding factors independent of OLIG1 index which may affect survival. The results of these analyses yielded a hazard ratio of 0.86 for OLIG1 index (95% CI 0.76-0.98, p =
0.023), indicating an association between reduced OLIG1 protein expression and
reduced overall survival. In our analysis the OLIG1 index variable was comprised of 9
discrete values (1-9), where 1 represents lack of protein expression and 9 represents
normal protein levels, as described in the Methods section. Therefore, our results
indicate that for every unit increase in OLIG1 index, there is a risk reduction of 14% in
relation to the risk associated with the lower index. For example, an OLIG1 index of 6 is
associated with a 14% reduction in the risk afforded by an OLIG1 index of 5. By the
same token, an OLIG1 index of 5 is associated with a 14% decrease in the risk afforded
by an OLIG1 index of 4.
79
Figure 4.5 OLIG1 immunohistochemistry on a lung tissue array. (A-H) OLIG1 immunohistochemistry on (A,E) tumor-free lung, (B) an OLIG1 negative adenocarcinoma and (F) an OLIG1 negative SCC; (C) a low OLIG1 expressing adenocarcinoma and (G) a low OLIG1 expressing SCC; (D) a high OLIG1 expressing adenocarcinoma and (H) a high OLIG1 expressing SCC. All images were acquired at 400x magnification. 80 In order to validate our observations, OLIG1 immunohistochemistry was performed on an independent sample set (TMA2), comprised of 74 adenocarcinomas (Adenos 104-
182) and 79 SCCs (SCCs 130-208). The tissue cores were scored as previously described and an OLIG1 index value was assigned to each sample. After completion of the data collection, univariate and multivariate analyses were performed on the data set.
The analyses were carried out in the same manner as for TMA1, including gender, age, tumor subtype, T and N stage variables in the models. For this second data set, the
OLIG1 index hazard ratio was assessed at 0.83 (95% CI 0.74-0.93, p = 0.0012) lending further support to the observation that reduced OLIG1 protein expression is associated with reduced overall survival.
In an effort to improve the precision of the multivariate model, TMA1 and TMA2 were combined and reanalyzed in the same fashion as each individual data set. The rationale for this approach was to increase the sample number, thereby increasing the statistical power and, potentially, the accuracy of the analysis. The OLIG1 index hazard ratio for the combined data was determined at 0.84 (95% CI 0.77-0.91 and p < 0.001).
The complete Cox Proportional Hazard Model for TMA1 and TMA2 combined is shown in Table 4.5.
Finally, we calculated the OLIG1 index hazard ratio for patients positive and negative for OLIG1 protein expression. This hazard ratio was generating by dividing the combined sample sets (N=285) into 2 groups. Samples with an OLIG1 index ≤3 were considered negative, while samples with an index ≥4 were considered positive233. The hazard ratio for this calculation was 0.54 (95% CI 0.38-0.76 and p < 0.001), indicating a
46% lower risk for OLIG1 positive cases. From this multivariate model, the probability of survival at 5 years was calculated for both groups. For OLIG1 positive cases, the probability of survival at 5 years was assessed at 0.62 (95% CI 0.55-0.70), while for
81 OLIG1 negative cases the probability of survival at 5 years was determined at 0.38 (95%
CI 0.94-0.50). The difference between both survival probabilities, 0.24, was statistically significant (95% CI 0.11-0.36), further strengthening our previous observations.
Overall, the comprehensive statistical analysis of our data sets led us to conclude that reduced OLIG1 protein expression is associated with reduced overall survival, and this association is independent of clinical variables such as tumor subtype, T and N stages or geographical variables, such as gender and age. In particular we were able to show that survival at 60 months, a common clinical parameter for assessing lung cancer prognosis, is significantly associated with OLIG1 protein expression.
Multivariate analysis of TMAs 1 and 2 combined (N=285)
Prognostic factor Hazard ratio 95% CI P OLIG1 Index 0.84 0.77 to 0.91 2.4e-05 Female 1.00 Gender Male 1.46 1.04 to 2.06 0.0310 <67 1.00 Age ≥67 1.54 1.08 to 2.19 0.0160 AdenoCa 1.00 Tumor subtype SCC 1.05 0.75 to 1.48 0.7600 T stage 1 1.00 T stage 2 1.21 0.84 to 1.74 0.3000 T stage 3 2.64 1.45 to 4.80 0.0015 N stage 0 1.00 N stage 1 1.54 1.03 to 2.29 0.0340 N stage 2 1.99 0.70 to 5.64 0.2000 TMA1 1.00 TMA2 1.30 0.92 to 1.84 0.1400
Table 4.6: Multivariate analysis of TMAs 1 and 2 combined
82 4.4 Discussion
In this study we have demonstrated that lung adenocarcinomas and SCCs can be distinguished based on the DNA methylation status of 47 discrete loci. This is a remarkable observation, since it not only lends further support to the fact that aberrant
CpG island methylation is non-random47, but it also indicates that different subtypes of neoplasias arising from the same organ can potentially be distinguished based on their aberrant DNA methylation patterns.
One of the 47 aberrantly methylated loci was OLIG1, a basic helix-loop-helix
transcription factor required for oligodendrocyte differentiation but of unknown function in
adult lung247. Immunohistochemical analysis of a large set of adenocarcinomas and
SCCs uncovered lack or reduced OLIG1 protein expression in 68% of the specimens
tested, suggesting that abrogation of OLIG1 might be of clinical relevance in these
subtypes of NSCLC. The impact of OLIG1 protein expression on patient survival was
assessed by univariate and multivariate analyses. Cox Proportional Hazard Models
indicated that lack of OLIG1 protein was strongly associated with poor survival in
NSCLC patients. Validation of these observations in an independent data set mirrored
the results first generated, further strengthening this association. Altogether, our results
suggest that OLIG1 protein expression may provide an additional clinically useful
parameter to determine the utility of supplementary therapy for patients suffering from
lung NSCLC, especially since survival at 60 months is significantly correlated with
OLIG1 protein expression. This finding is potentially of great significance, as
the addition of postoperative adjuvant chemotherapy in T2N0 NSCLC, for example, is
currently a matter of great debate248,249.
Interestingly, the percentage of samples lacking OLIG1 protein was higher than expected within the adenocarcinoma subgroup. Based on the totality of the data
83 collected in this study, it is possible that a post-transcriptional mechanism acting preferentially in the adenocarcinomas may account for either lack of OLIG1 mRNA translation or rapid degradation of the OLIG1 protein product. This scenario reconciles the initial observations of lower DNA methylation and higher mRNA expression in adenocarcinomas compared to SCCs, with the later finding of a higher proportion of
OLIG1 negative adenocarcinomas. Nevertheless, this phenomenon deserves further investigation. The corroboration of a tumor subtype-specific post-translational regulatory mechanism in lung cancer would be an immense contribution towards further understanding the etiology of this disease.
The importance of OLIG1 expression in adult lung may be explained, in part, by extrapolation of known functions of this gene in oligodendrocyte development 250,251. It
has been shown that oligodendrocytes derived from OLIG1-/- mice are unable to
differentiate251, suggesting that at least one of the functions of OLIG1 may pertain to
initiation or maintenance of cellular differentiation. At the same time, sonic hedgehog
(SHH), a secreted ligand of the hedgehog signaling pathway known to be overexpressed in lung cancer252, has been shown to be necessary and sufficient to activate OLIG genes
in oligodendrocytes 253. Thus, abrogation of OLIG1 protein expression may play a role in
inhibiting cellular differentiation, but it could also contribute to the tumor phenotype in
other ways through some of its downstream targets. MAG, a single-pass type I
transmembrane protein involved in brain cellular adhesion254,255 is highly expressed in
adult lung256, and it is also a known target of OLIG1254. In light of these genetic interactions, a growth advantage could be conferred to tumor cells that overexpress
SHH through interaction with currently unknown growth promoting targets, while at the same time abrogating OLIG1 expression concomitant to MAG downregulation. This scenario would explain the high frequency of deletions and DNA methylation observed at
84 the OLIG1 locus. This hypothesis is reinforced by our observation that N0 NSCLC cases
are more likely to be OLIG1 positive than N1 cases. Therefore, this phenotypic
difference could stem, in part, from lack or reduced MAG expression in OLIG1 negative
tumors, which could facilitate detachment of tumor cells form the primary tumor mass.
Taking the relationship between DNA methylation and gene expression into
consideration, our study demonstrates that genome-wide DNA methylation patterns can
be as useful in tumor subtype distinction as gene expression profiling, an approach
which has been successfully utilized in the past to distinguish not only lung tumor
subtypes but also phenotypic differences associated with survival within a lung tumor
subclass257-259. In light of our results, the establishment of differential DNA methylation
patterns could reflect an intrinsic difference in the cellular origin260 of each of the tumor
subtypes, or by distinct oncogenic pathways activated predominantly in one subtype
over the other. It has been well documented that gains in 3q22-q26 where the alpha
catalytic subunit of phosphatidylinositol 3-kinase (PI3K) is located, occurs almost
exclusively in SCCs261. Overexpression of this gene could be correlated with increased activity of its downstream effector, protein kinase B (PKb) in this lung tumor subtype261.
Given the evidence that DNA methylation may be the result of a-priori downregulation of
gene expression176,262, the establishment of differential DNA methylation patterns
between lung tumor subtypes may be the result of distinct oncogenic activities affecting
primarily one type of neoplasia and not the other. Additional studies will be needed to
fully elucidate the mechanisms governing the establishment of tumor subtype specific
DNA methylation patterns.
85 CHAPTER 5
GENOME-WIDE AND TUMOR-TYPE SPECIFIC ABERRANT DNA
METHYLATION IS SIGNIFICANTLY HIGHER IN HUMAN LUNG CANCER
COMPARED TO ELEVEN OTHER PRIMARY HUMAN NEOPLASIAS
5.1 Introduction
Over the past two decades it has become clear that normal DNA methylation
patterns are profoundly disrupted in cancer cells38. A hallmark of the cancer genome is
its overall hypomethylation35, concomitant with hypermethylation of CpG islands39.
Consequently, loss of DNA methylation in heterochromatic regions, such as centromeric repeats and other repetitive sequences, results in genomic instability30,32,263 and
oncogene activation36, whereas gain of DNA methylation in gene-associated CpG
islands generally leads to gene silencing96,264.
Cancer is the leading cause of death worldwide1. In recent years, significant
progress has been made in the realms of cancer detection and treatment, which has
resulted in prolonged life expectancy and better quality of life for cancer patients.
However, in spite of these improvements, the disease still exerts a major toll on human
life. Of all types of neoplasias, lung cancer accounted for 17% (1.3 million) of all cancer
deaths in 20051. Particularly in the United States, lung cancer-associated mortality
accounted for 28% of all cancer deaths, a fraction higher than that of breast, prostate
and colorectal cancers combined (American Cancer Society 2005 report). 86 Almost a decade ago, a major study reported that aberrant CpG island methylation exhibits tumor-type specific patterns47,224, suggesting that genome-wide scans for
aberrant DNA methylation could potentially be utilized to identify genes uniquely silenced
in any given neoplasia. The importance of this observation is that the identification of
such genes could lead to the development of new and possibly more effective
therapeutic strategies, designed to offset tumor-type specific molecular alterations.
In an effort to identify aberrant DNA methylation events characteristic of non-small
cell lung cancer (NSCLC), we reanalyzed RLGS data on 205 primary tumors
representing 12 different types of human cancers225,265. We were able to identify aberrant promoter methylation in all tumor types examined. However, our results indicated that lung neoplasias exhibited the highest level of aberrant DNA methylation per specimen and, as a group, showed the highest percentage of tumor-type specific aberrant DNA methylation events. In addition, we were able to verify downregulation of normal mRNA levels for several of the genes methylated in the lung primary tumors, underscoring the functional connection between DNA methylation and mRNA gene expression.
5.2 Methods
5.2.1 Collection of primary human tumors The following human primary tumors were collected through the Cooperative
Human Tissue Network at The Ohio State University James Cancer Hospital: 33 acute
myeloid leukemia (AML) specimens, 25 lung cancer specimens, 26 colon cancer
specimens, 10 chronic lymphocytic leukemia (CLL) specimens, 14 adult brain tumor
(ABT) specimens (gliomas), 17 cervical cancer specimens, 9 non-seminomatous
testicular tumor (NST) specimens, 14 childhood brain tumor specimens (Brain), 17 head
and neck (H&N) cancer specimens, 25 medulloblastoma (Med) specimens, 8 primitive
87 neuroectodermal tumor (PNT) specimens and 7 seminomatous testicular tumor (SET) specimens. All samples were collected in accordance with Internal Review Board guidelines. Peripheral blood mononuclear cell (PBMCs) DNA collected from random cancer-free donors was utilized as a negative control for DNA methylation. Consent from participants was waived under CFR 46 subpart A.
5.2.2 Restriction landmark genomic scanning (RLGS)
RLGS was performed as previously described63 using NotI as the landmark
enzyme. In order to ensure that the data collected could be compared across samples,
1,194 non-polymorphic RLGS fragments were scored in each specimen47.
5.2.3 Cloning of RLGS Fragments
RLGS fragments of interest were cloned as previously described47,63,194,226,265 with the aid of a human NotI–EcoRV plasmid library. Alternatively, a PCR based method was utilized to identify RLGS fragments not present in the libraries265.
5.2.4 RNA isolation
Total RNA from primary tumors, tumor-free specimens and human lung cancer cell lines was isolated as previously described227. RNA quality was evaluated on RNA 6000
Lab-chips using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto CA).
Samples showing high level of RNA integrity were used for reverse transcription216.
5.2.5 Quantitative real-time PCR
1µg of total RNA was reverse transcribed using oligo dT (Invitrogen, Carlsbad CA), as previously described228. In order to ensure the measurement accuracy of mRNA
88 transcripts derived from intronless genes, regular PCR was performed on DNAseI
treated but not reverse transcribed RNA samples so as to confirm no DNA
contamination was present in the RNA preparations. Quantitative gene expression was
measured using SYBR Green I (Bio-Rad, Hercules CA) in an iCycler (Bio-Rad).
Calcium/calmodulin-dependent protein kinase kinase 2 (CAMKK2) was used as the
normalizing control228. All real-time PCR measurements were formed under the following conditions: 95°C x 2 min (95°C x 30 sec, 60°C x 30 sec) for 35 cycles, followed by a dissociation curve (55°C to 95°C, increasing 0.5°C every 10 sec). The primer sequences utilized for all reactions are listed in Table 5.1.
89
Gene Primer sequence Gene Primer sequence AF119875_RT_F TTTTTGACAAGTGCAAGGTCAG AF119875_RT_R TCGATCTGTAAGAACCGTGATG AK124226_RT_F TCAAAGCAGGGAGAGTTAAAGC AK124226_RT_R AGAGAGGAGAGGATGCTTGATG BC013982_RT_F TTTATGTCAAACAGGGTGCAAG BC013982_RT_R GAAATAGCCAGTTTTGCTCCAG BC039382_RT_F GTTCCTGTGTGGGAATAGCTTG BC039382_RT_R CCGAATCTTCCTTATGGTTCTG BCL2L10_RT_F GGATGGCTTTTGTCACTTCTTC BCL2L10_RT_R TGCTTTCCCTCAGTTCTTGTTC C10ORF78_RT_F CTCAGAGGAAGGTATCCCAATG C10ORF78_RT_R GGGTTAGACATTCACCAGATTG C1ORF164_RT_F GTACCTGCCTCTCTCTCTCCTG C1ORF164_RT_R CTGTCCTCTAGCCCCCTAACTC CYB561_RT_F GATTTCCCATAGTTGGCTTTTG CYB561_RT_R AGTCTAAACAGGAGGCGAACAC DKFZp667B0210 AAAGCCTGAGAAGTCACTTTGG DKFZp667B0210 GAAGTGATGTGGGCATTGACTA DNAJC9_RT_F AAGCACTTGACATTGTGTGAGG DNAJC9_RT_R CAGAAAGCAGCATGAACAGAAC EPB49_RT_F GCTACCAGCTCTCACCTACACC EPB49_RT_R CCAGACCTGGGCAAAAATATAG FLJ39005_RT_F TGGGCGCTATGTACTTGTAGTG FLJ39005_RT_R CAGTCCAGAAGTCAACGATCAG FMN2_RT_F TCCTCCACATCACAGCATTTAG FMN2_RT_R GTTATGGAGAGCAACCCAAGAG FOXF1_RT_F AATAATCAAAACACCGCGTAGG FOXF1_RT_R GTCACAAATGCTGCACTCTAGC HOXA10_RT_F AGGGGACTTCTCTTCCAGTTTC HOXA10_RT_R AGAATTGTGGTGTGCTTGTCAC HOXA9_RT_F ATTTTAAGTGTTCTCGGGGATG HOXA9_RT_R ATAGCTCCGAATTTCCTCACTG HOXC6_RT_F TGGCATTTTACAAACTGTGACC HOXC6_RT_R TGGCTAAACAAACGTCATTCAC IRX2_RT_F GGACAGGACTCTGACATTCTCC IRX2_RT_R CAATTGTGACACCTACCTGTGG KCNC2_RT_F GCAAATGAAGCTTGTACGTGTC KCNC2_RT_R TCACAAAAGGTCCACGATACAG KIAA1622_RT_F TGAATCAGCTGTTATTCCAAGC KIAA1622_RT_R AAGCTGACAATAGCTTCACACG LHFPL4_RT_F GGGAATAACCTTTCTCCAGCTT LHFPL4_RT_R ATCCTAACAGCACACAGCACAG LMX1B_RT_F GAACGACTCCATCTTCCATGAC LMX1B_RT_R CAGGAGGCGAAGTAGGAACTC NGEF_RT_F AATCTACCAGGCACAGATGAGG NGEF_RT_R TCCTTAAGCGGCTAGAAGACAC NKX2-3_RT_F AGCATGAAGGAGAGAAAAATGG NKX2-3_RT_R GAGATCTGGACCGAGGTATCTG NLF2_RT_F GCTGAGTACTGTCCGGGAAC NLF2_RT_R GAAGCAAAAGTCCTGGTCAAAG PELI2_RT_F GTCTTTGCCCTCATGAAGAATC PELI2_RT_R ATGTGACCATCCCCTTAACTTC PTPRN2_RT_F CCCCAGAAATAGGACAATTCAC PTPRN2_RT_R CATCATTCTGTCCGCTCAGTAG RBM9_RT_F GACATTAGGAGCCGATAAATGC RBM9_RT_R ACCAGAATTGCCTGTCAAAGAC SEPT9_RT_F AACGAACCCCTAGAAAGGAGAG SEPT9_RT_R AGCAAGACAGGTAACAGGAAGC SLC5A8_RT_F GATGGCTCATGCTTGTAATCTG SLC5A8_RT_R AAATCCTGGGCTTAAGCTATCC SULT4A1_RT_F TTTGTTTGTAATGGGGAGAAGC SULT4A1_RT_R TTAAAGATGCAAGCAAGCACAG TBX18_RT_F AGCACAGAATGTGAGAGACAGC TBX18_RT_R TCTAGCGGCCTAAAGCATAAAC TLX3_RT_F CAGAAGTACCTGGCCTCTGC TLX3_RT_R GAGCAAAGAGTGACGAGTTGTG UGCU_RT_F TTTGTGAGCCAATTTCAGAATG UGCU_RT_R GCGACTGCATAATCAAGTTTTG
Table 5.1: Primer sequences utilized for measuring mRNA gene expression via real- time PCR
90 5.6 5-aza-2’deoxy-cytidine and trichostatin A treatment of lung cancer cell lines Human lung cancer cell lines A549 and H719 were cultured for 2 days and then treated with 1µM 5-aza-2’deoxycytidine (5-aza-dC) (Sigma-Aldrich, St. Louis MO) for 72 hs or with 5-aza-dC for 72 hs + 300 nM trichostatin A (TSA) for 24 hs as previously described227. After treatment, the cells were collected and total RNA was isolated as
previously described227.
5.2.7 Combined bisulfite restriction analysis (COBRA) COBRA was performed as previously described229. DNA isolated from cell lines and primary tumor samples was bisulfite treated266 and amplified by PCR. All PCR products were purified using affinity columns (Qiagen, Valencia CA) and digested with
10U of BstUI (New England Biolabs, Beverly MA) at 60°C for 4 hs. The digested PCR products were electrophoresed in an 8% polyacrylamide gel and visualized via ethidium bromide staining. All PCR reactions were formed under the following conditions: 95°C x
10 min (95°C x 30 sec, 62°C x 30 sec, 72°C x 30 sec) for 35 cycles, followed by a final extension for 10 min at 72°C. The primer sequences utilized for all PCR reactions are listed in Table 5.2.
91
Gene Primer sequence Gene Primer sequence BAHD1_F GTTTTYGGGGAGTTTGTGGGGGAAT BAHD1_R CRACCCCTAACTCACCAAAAC BT013982_F GAGGAGGAAYGAGGGGAGAAG BT013982_R ACCAAAACCTAAATCTCCCTCC BT039382_F AAGGGGTTGAGGGGGTTGTG BT039382_R CCTAATCACCAACTCTCCTAA C10ORF78_F GGTTGAAAATAAGGTAATAGAAGT C10ORF78_R AAAATACRCAATTAACCAAATTACC C1ORF164_F GTAGTTTAGGAGTTGGAAGGGTT C1ORF164_R ACACRTACAAAAACCCCTAATCC CYB561_F GGGTAGGTTAGGGTAGGGGT CYB561_R RACCCAAATAAACCCTAACTAC DNAJC9_F TATTTGTAGGGATATTTTGTGGTAG DNAJC9_R AACTCRCCACACCTCATTTTAC FLJ39005_F GTTTTATGTAGGGTTGAAGAGG FLJ39005_R CTAATACTCACTATTCTCACCAC FMN2_F GGAAGAGTYGGAGGAGGAGG FMN2_R ACCAATACTAAAAACACCCCACC FOXF1_F AGAGTAGTATTTATTTGGGTTTGTG FOXF1_R CCAAACCAAATTCCTAAAAACAAAC HOXA10_F AAGGGGTYGGGGAGAGTTTTTT HOXA10_R ACACCRCCAAAAACTATAACAAC IRX2_F YGTAGGAGATTTTGGTTTTGTAG IRX2_R CCAACAAACCCAAACTATAATC KIAA1622_F GTTGYGGGAGGTAAAAGGTTTTG KIAA1622_R CAAAACTTTCTCCRACTCTTAACC LHFPL4_F TTTTYGGGGGGTTGGGGAG LHFPL4_R ACCRTACCCAAATTAATCTCCC LMX1B_F TAGGAAGTTTYGGGAGGTGAG LMX1B_R CTCTATCACCTACTTATCAATCC NLF2_F AAGTTTTGGTTAAAGGAGGTTTTG NLF2_R CCTACRCCTAACTACTAAATAC NXK2-3_F GGTTGTAATAAAATTTAGATTTTTAGG NXK2-3_R TACTCCAAATTCAAAATATCTTTAAC PELI2_F GTTTTATTTGTTGTYGGTTTTGATT PELI2_R CCTAACCAAAAAAAAACATAAAACC PTPRN2_F TAYGGAGGAAAAATGTTTTTTGATTT PTPRN2_R CCCAAACTCTAAATCTCAAACCC RBM9_F YGATAGGTTATTTTTTTTTGGGTTT RBM9_R CAAAATAAACAACCCTCCCCAC SEPT9_F GTTTTTTTATTATYGGTTTAGGATTAG SEPT9_R AAAAAACTCRACCTACAATATACCC SLC17A7_F TAGATTTAGGTTTTAGGAGGGTG SLC17A7_R AAAACTCTACAACCACTAAATAAC SLC5A8_F GGAAGTTTTTGGAGGTTTGTTGG SLC5A8_R TACAAAAACRACTACCAACCCTC SULT4A_F TTTTYGGGGTGTTGGGGGTTT SULT4A1_R CCRACAACRAATAAATAACCC TBX18_F GATGTAGGAAGTATAGAGTTGTAT TBX18_R CCTTCCTACCTATAACTTCTCTC UGCG_F GGGGTATYGTTTTTGGGAGAGG UGCG_R CTACRATCTCCCRACTCTAATC
Table 5.2: Primer sequences utilized for PCR amplification of COBRA templates.
92 5.2.8 Statistical Analysis Statistical significance was assessed by applying 2-sided Z-ratio calculations.
Comparisons with p < 0.05 (95% CI) were considered significant. To ensure a normal
distribution of expression values, a log-transformation was applied to the normalized
real-time PCR data. The resulting values were used to generate all heat maps.
5.3 Results
5.3.1 Aberrant DNA methylation levels in twelve primary human neoplasias
RLGS was performed on 33 AML, 25 lung, 26 colon, 10 CLL, 14 ABT, 17 cervical,
9 NST, 14 childhood brain, 17 H&N, 25 medulloblastoma, 8 PNT and 7 SET tumor
specimens in order to determine the levels of aberrant DNA methylation in each tumor
type. The RLGS profile from each tumor was compared and scored against a profile
produced from tumor-free tissue from the same patient. So as to ensure that the RLGS
analyses results could be compared across the sample set, the same 1,194 RLGS
fragments were scored in all samples47. Aberrant DNA methylation was detected in every tumor type; however, tabulation of the RLGS data denoted a substantially wide range in DNA methylation across the sample set. Lung cancer specimens showed, on average, the highest level of DNA methylation per sample: 8.31% (range: 4.77% -
13.55%), while SET specimens showed the lowest: 0.01% (range: 0% - 0.08%) (Figure
5.1.A). Next, we calculated how many different RLGS fragments were aberrant methylated in each tumor type. The results of these calculations indicated DNA methylation occurred at 321 different RLGS fragments within the AML group, while only at a single RLGS fragment within the SET group (Figure 5.1.B).
Tumor-type specific DNA methylation was defined as DNA methylation detected at any RLGS fragment exclusively within a single tumor type and not in any other. As it was
93 the case with the previous analyses, the results of these calculations displayed a considerably wide range in tumor-type specific DNA methylation across the sample set.
The highest level of tumor-type specific DNA methylation was assessed in lung cancer samples: 48.1%, while H&N, Med and SET samples displayed no tumor-type specific
DNA methylation (Figure 5.1.C). This is an important observation, since it suggests that depending on their cell type of origin, the development of some neoplasias may require the inactivation of a given group of genes whose activity is, presumably, a normal trait of that cell type. On the other hand, the development of SET or H&N tumors, for example, maybe be independent of this requirement. Therefore, it is possible that different cell types may require the inactivation of distinct numbers of genes to undergo neoplastic transformation. Based on our results, the transformation of cells giving rise to H&N, medulloblastoma and SET tumors may require fewer genes to be inactivated, compared with cells giving rise to lung and colon tumors, for example.
94
Figure 5.1: RLGS analysis of 12 primary human neoplasias. (A) Average levels of aberrant DNA methylation scored within each tumor type. (B) DNA methylation of unique
RLGS fragments within each tumor type. (C) Tumor-type specific DNA methylation assessed within each sample set. Tumor-type specific DNA methylation was defined as
DNA methylation detected at any RLGS fragment exclusively within a single tumor type and not in any other. All calculations were based on the assessment of the same 1,194
RLGS fragments for each sample. The bars represent the range of DNA methylation detected for each tumor type.
95
96 5.3.2 Identification of aberrantly methylated genes lung cancer
Given that the lung cancer samples showed not only the overall highest level of aberrant DNA methylation, but also the highest proportion of tumor-type specific DNA methylation events, we decided to focus on this tumor type for further studies. Of the 241 methylated RLGS loci scored in this sample set, 142 were cloned and identified63,265. Of these, 88% (125) mapped to an annotated locus and 96% (136) was associated with a
CpG island43,47.
To determine the functional connection between the identified genes and known
cellular processes, the ontology of each gene was mined from a public database267
(http://www.geneontology.org). Surprisingly, our analysis indicated that 16% (23) of the
genes were transcription factors, while 10% (14) were categorized as intracellular
signaling molecules (6% kinases and 5% phosphatases) (Table 5.3). These data
indicate that transcription factors may be a major target of aberrant DNA methylation in
human lung cancer.
97 Percent Molecular Biological Location Gene methylated function process
96% 12q23.2 SLC5A8 Transport Ion transport 96% 17q23.3 CYB561 Electron transport Electron transport 92% 10q26.3 DKFZp667B0210 88% 10p12.2 N/A 84% 3p25.3 LOC375323 80% 10q26.13 N/A 80% 1p34.1 AK056424 Ligase Ubiquitin cycle 80% 1q43 FMN2 Actin binding Development 80% 2q32.1 BC039382 76% 12q21.1 KCNC2 Ion channel Ion transport 76% 14q22.3 PELI2 76% 19q13.43 FLJ39005 76% 2p24.1 BC013982 76% 3q21.3 CHST13 Sulfotransferase Carbohydrate metabolism 76% 5p15.33 IRX2 Transcription factor Development 76% 6q27 N/A 72% 10q22.2 DN/AJC9 72% 16p13.12 N/A 72% 18q.23 SALL3 Transcription factor Transcription regulation 72% 1p13.3 ALX3 Transcription factor Development 72% 5q23.3 N/A 72% 6q14.3 TBX18 Transcription factor Development 72% 8p21.3 EPB49 Actin binding Cytoskeletal organization 72% 9q33.3 LMX1B Transcription factor Differentiation 68% 22q13.31 SULT4A1 Sulphotransferase Steroid metabolism 68% 2q37.1 NGEF 68% 9q31.3 UGCG Transferase Carbohydrate metabolism 64% 10q25.1 C10orf78 64% 15q21.2 BCL2L10 Caspase Apoptotic regulation 64% 1p32.3 AF416921 Esterase Fatty acid metabolism 64% 7p15.2 HOXA9 Transcription factor Development 64% 7p15.2 HOXA10 Transcription factor Development 64% 7q36.3 PTPRN2 Phosphatase Intracellular signaling 60% 10q24.2 NKX2-3 Transcription factor Differentiation 60% 15q15.1 BAHD1 DNA binding 60% 15q22.2 NLF2 60% 16q24.1 FOXF1 Transcription factor Development 60% 19q13.33 SLC17A7 Transport Phosphate transport
Continued
Table 5.3: Percent DNA methylation (25/25 samples = 100%), chromosomal location, name, molecular function and biological process for the 142 genes identified as aberrantly methylated in human lung cancer. N/A: No locus annotated. Blank cell: no molecular function or biological process currently assigned. 98 56% 10q26.2 AK124226 56% 12q13.13 HOXC6 Transcription factor Development 56% 14q32.12 KIAA1622 Protein binding Chromosome condensation 56% 17q25.3 SEPT9 56% 18q21.1 AF119875 56% 22q12.3 RBM9 RNA binding Cell cycle regulation 56% 5q35.1 TLX3 Transcription factor Development 52% 17q11.2 RNF135 Ligase Ubiquitin cycle 52% 2p13.2 EMX1 Transcription factor Development 52% 4p12 ZAR1 Protein binding Development 52% 5q14.1 SSBP2 DNA binding Transcription regulation 52% 5q31.3 HDAC3 Histone deacetylase Chromatin modification 52% 5q33.2 N/A 52% 6q23.2 ENPP1 Endonuclease Nucleotide metabolism 52% 7q36.3 PTPRN2 Phosphatase Intracellular signaling 52% 8p23.3 KIAA0711 Protein binding 48% 10q26.3 AK097335 48% 13q12.2 IPF1 Transcription factor Development 48% 18p11.31 LOC388458 48% 2p13.2 P450RAI2 Monooxygenase Metabolism 48% 3q13.2 BOC Membrane protein Differentiation 48% 9q33.3 LHX2 Transcription factor Differentiation 44% 12q24.31 CDK2AP1 DNA binding Cell cycle regulation 44% 1p36.13 HTR6 G-coupled receptor Intracellular signaling 44% 20q12 N/A 44% 2q12.1 AK096498 44% 2q21.1 FLJ38377 44% 6q21 POPDC3 Membrane protein Development 44% 8p21.2 COE2 40% 16p13.3 ADCY9 Adenylate cyclase Intracellular signaling 40% 16p13.3 UNKL 40% 2q31.1 TLK1 Kinase Chromatin modification 40% 4p16.3 N/A 40% 5q23.2 SNCAIP Protein binding Cellular organization 40% 8q21.13 N/A 40% 9q21.11 KLF9 Transcription factor Transcription regulation 36% 10q26.13 BUB3 Protein binding Spindle organization 36% 11q13.1 C11orf5 36% 12p13.2 DUSP16 Phosphatase Intracellular signaling 36% 12q13.12 FLJ13236 Chaperone Protein folding 36% 17q25.3 CBX2 DNA binding Chromatin modification 36% 18q11.2 LAMA3 Structural molecule Development 36% 19p13.11 FKBP8 Protein binding Intracellular signaling 36% 1p21.2 GPR88 G-coupled receptor Intracellular signaling 36% 20p13 C20orf27 Ligase Protein biosynthesis 36% 20q11.22 AF052211 36% 22q13.2 CGI-96 36% 5q14.1 OTP Transcription factor Development 36% 7q31.1 IPLA2(GAMMA) Phosphlipase Fatty acid metabolism 32% 12q13.11 CCNT1 Transcription factor Cell cycle regulation 32% 13q13.1 13CDN/A73
Continued
99 32% 16p11.2 N/A 32% 20p13 STK35 Kinase Cell Motility 32% 2p15 OTX1 Transcription factor Development 32% 9p23 PTPRD Phosphatase Intracellular signaling 28% 12q15 NUP107 Transport Nucleocytoplasmic transport 28% 19p13.12 PTGER1 G-coupled receptor Intracellular signaling 28% 20q13.12 C20orf35 Protein binding Protein transport 28% 22q13.1 SLC16A8 Transport Ion transport 28% 22q13.31 AI673633 28% 3p21.31 MAPKAPK3 Kinase Intracellular signaling 28% 4q35.1 ANKRD37 Receptor 28% 5q23.3 CSS3 Transferase Carbohydrate metabolism 24% 12q24.21 N/A 24% 16q12.1 N/A 24% 16q24.3 AFG3L1 Peptidase Proteolysis 24% 19p13.3 STK11 Kinase Intracellular signaling 24% 20p11.23 N/A 24% 21q22.12 RUNX1 Transcription factor Development 24% 2p24.3 N/A 24% 5p15.33 N/A 24% 9q22.33 BC002660 Actin binding Cytoskeletal organization 20% 14q21.3 AF068289 20% 15q25.3 SCAND2 Oxidoreductase Protein metabolism 20% 19q13.11 KCTD15 Ion channel Ion transport 20% 3p21.33 ABHD5 Peptidase Proteolysis 20% 6p22.1 N/A 20% 8p21.3 HR Transcription factor Transcription regulation 16% 11p11.2 TP53I11 Cell cycle regulation 16% 11q13.4 BNF1 16% 13q12.11 LATS2 Kinase Cell cycle regulation 16% 15q14 LOC56851 16% 16q13 TM4SF11 Ion channel Ion transport 16% 19q13.42 LENG5 Endonuclease mRNA Processing 16% 1q23.3 THC2117393 16% 1q25.3 AB046834 16% Xp22.22 MID1 Ubiquitin ligase Development 12% 14q24.3 HBLD1 12% 14q32.12 KIAA1622 Protein binding Chromosome condensation 12% 1p36.12 RAP1GA1 GTPase activator Intracellular signaling 12% 1q22 N/A 12% 1q42.12 PARP1 Ribosyltransferase DNA Repair 12% 4q35.1 FLJ30277 12% 5q31.1 PPP2CA Phosphatase Cell cycle regulation 12% 6q15 BACH2 Transcription factor Transcription regulation 12% 8q22.3 YWHAZ Monooxygenase 8% 17q23.2 NOG Protein binding Differentiation 8% 2q35 MGC3035 8% 6p21.33 CLIC1 Transport molecule Ion transport 8% 7q22.3 PRKAR2B Kinase Intracellular signaling 4% 16q24.1 FOXC2 Transcription factor Development 4% 16q24.2 MAP1LC3B Autophagy Development 4% 4q23 TM4SF9 Membrane protein 4% 4q24 CXXC4 Transcription factor Development
100 5.3.3 Identification of lung cancer-specific aberrantly methylated genes
Our analysis indicated that 116 RLGS fragments were aberrantly methylated exclusively in lung cancer. Sixty one percent (71) of these were identified and mapped to the human genome. Of these 71 RLGS fragments, 83% (59) mapped to an annotated locus and 96% (68) was associated with a CpG island43,47. Similar to what was determined for the previous group of 142 genes, 17% (12) were transcription factors and
11% (8) were classified as intracellular signaling molecules (Table 5.4).
Interestingly, the range of tumor-type specific DNA methylation was much narrower across the sample set in comparison with overall DNA methylation levels. As described above, the mean aberrant DNA methylation detected in the lung cancer set was 8.31%, ranging from 4.77% to 13.55% (a 3-fold fluctuation). In contrast the mean tumor-type specific aberrant DNA methylation in these samples was 4.07%, ranging from 3.18% to
5.18% (less than 0.7-fold fluctuation). This observation suggests that the majority of the tumor-type specific DNA methylation events identified in our analysis may be required to concomitantly take place for the development and/or progression of lung neoplasias.
101 Percent Molecular Biological Chromosome Gene methylated function process
88% 10p12.2 N/A 84% 3p25.3 LHFPL4 80% 10q26.13 LHPP Hydrolase Metabolism 80% 1p34.1 C1orf164 Ligase Ubiquitin cycle 80% 1q43 FMN2 Actin binding Development 76% 6q27 N/A 76% 12q21.1 KCNC2 Ion Channel Ion transport 72% 16p13.12 LOC440338 72% 5q31.1 N/A 72% 9q33.3 LMX1B Transcription factor Differentiation 68% 22q13.31 SULT4A1 Sulphotransferase Steroid metabolism 64% 7q36.3 PTPRN2 Phosphatase Intracellular signaling 64% 7p15.2 HOXA9 Transcription factor Development 64% 7p15.2 HOXA10 Transcription factor Development 64% 10q25.1 C10orf78 60% 16q24.1 FOXF1 Transcription factor Development 60% 10q24.2 NKX2-3 Transcription factor Differentiation 60% 19q13.33 SLC17A7 Transport Phosphate transport 60% 15q22.2 NLF2 56% 5q35.1 TLX3 Transcription factor Development 56% 18q21.1 FUSSEL18 56% 22q12.3 RBM9 RNA binding Cell cycle regulation 56% 17q25.3 SEPT9 56% 10q26.2 EST 52% 5q14.1 SSBP2 DNA binding Transcription regulation 52% 7q36.3 PTPRN2 Phosphatase Intracellular signaling 52% 5q33.2 N/A 52% 8p23.3 KBTBD11 Protein binding 52% 4p12 ZAR1 Protein binding Development 52% 2p13.2 EMX1 Transcription factor Development 48% 9q33.3 LHX2 Transcription factor Differentiation 48% 18p11.31 LOC388458 48% 13q12.2 IPF1 Transcription factor Development 44% 12q24.31 CDK2AP1 DNA binding Cell cycle regulation 44% 8p21.2 EBF2 44% 2q12.1 N/A 44% 1p36.13 HTR6 G-coupled receptor Intracellular signaling
Continued
Table 5.4: Percent DNA methylation, chromosomal location, name, molecular function and biological process for the 71 genes identified as aberrantly methylated exclusively in human lung cancer. N/A: No locus annotated. Blank cell: no molecular function or biological process currently assigned.
102 40% 8q21.13 ZNF704 40% 5q23.2 SNCAIP Protein binding Cellular organization 36% 7q31.1 IPLA2γ Phosphlipase Fatty acid metabolism 36% 22q13.2 CGI-96 36% 19p13.11 FKBP8 Protein binding Intracellular signaling 36% 12q13.12 FLJ13236 Chaperone Protein folding 36% 18q11.2 LAMA3 Structural molecule Development 36% 1p21.2 GPR88 G-coupled receptor Intracellular signaling 36% 20q11.22 CBFA2T2 36% 12p13.2 DUSP16 Phosphatase Intracellular signaling 32% 2p15 OTX1 Transcription factor Development 32% 9p23 PTPRD Phosphatase Intracellular signaling 28% 22q13.1 SLC16A8 Transport Ion transport 28% 4q35.1 ANKRD37 Receptor 28% 12q15 NUP107 Transport Nucleocytoplasmic transport 24% 5p15.33 N/A 24% 16q12.1 N/A 24% 19p13.3 STK11 Kinase Intracellular signaling 24% 21q22.12 RUNX1 Transcription factor Development 24% 20p11.23 N/A 20% 14q21.3 N/A 20% 6p22.1 N/A 16% 11p11.2 TP53I11 Cell cycle regulation 16% 15q14 C15orf24 16% Xp22.22 MID1 Ligase Development 16% 16q13 ARL2BP Ion Channel Ion transport 16% 13q12.11 LATS2 Kinase Cell cycle regulation 16% 1q23.3 N/A 12% 14q24.3 NPC2 12% 14q32.12 KIAA1622 Protein binding Chromosome condensation 12% 1q22 N/A 8% 17q23.2 NOG Protein binding Differentiation 8% 6p21.33 CLIC1 Transport Ion transport 4% 4q24 CXXC4 Transcription factor Development
5.3.4 Chromosomal distribution of aberrant DNA methylation in human lung cancer
Various studies have indicated that aberrantly methylated genes are sometimes found in regions of loss of heterozygosity (LOH)58,236,237,239,240. To establish if the genes
identified in our RLGS scan located to known regions of LOH for lung cancer, we
mapped all identified loci to the human genome. Of the 142 mapped loci, only 22.5%
(32) were located in areas where LOH had previously been described (Figure 5.2).
Interestingly, lung-specific DNA methylation events and those DNA methylation events
103 found in lung cancer as well as other neoplasias mapped to regions of LOH with similar frequency (22.5% and 19.4% respectively). Altogether, these data indicate that the majority of the genes identified in our RLGS scan may be preferentially targeted by epigenetic mechanism in human lung cancer.
Figure 5.2: Chromosomal location of aberrant DNA methylation events identified in human lung cancer. DNA methylation events are indicated with horizontal lines across the chromosomes. Orange lines indicate lung-specific DNA methylation, while blue lines indicate DNA methylation observed in lung as well as other neoplasias.
104 Regions of LOH are indicated with red bars alongside each chromosome. Chromosome lengths are drawn to scale.
Next, we sought to determine if aberrant DNA methylation was found in any particular chromosome at a frequency higher than that expected by chance. This evaluation was performed by comparing the proportion of NotI sites methylated per chromosome in relation to the 142 DNA methylation events identified in our RLGS scan, against the proportion of NotI sites per chromosome in relation to the total number of
NotI sites in the human genome (March 2006 assembly). Our analyses indicated that chromosomes 11 and X were aberrantly methylated at a frequency lower than expected
(2-tail Z-ratio p= 0.12 and p= 0.01 respectively), while chromosome 5 was aberrantly
methylated at a frequency higher than expected (2-tail Z-ratio p= 0.06) (Table 5.5). As
indicated, though borderline, two of these cases failed to reach statistical significance. In
addition, our calculations are based on the assumption that each chromosome is
represented in our RLGS profile in proportion to its number of NotI sites. Taken
together, our data suggests that some chromosomes might be susceptible to high levels
of aberrant DNA methylation, regardless of their number of NotI sites, while others might
be refractory to this phenomenon. Biologically, these data could indicate that
hypermethylation of chromosome 5 might confer a growth advantage to lung cancer
cells. By the same token, hypermethylation of the X chromosome might be detrimental,
in light of the unique and complex mechanism operating on this chromosome to regulate
gene dosage268.
105 Expected proportion Observed proportion Chromosome NotI sites P value of DNA methylation of DNA methylation 1 762 7.98% 7.75% 0.92 2 656 6.87% 7.75% 0.68 3 411 4.31% 3.52% 0.65 4 382 4.00% 4.23% 0.89 5 420 4.40% 7.75% 0.06 6 451 4.72% 4.93% 0.91 7 546 5.72% 4.23% 0.45 8 368 3.86% 4.23% 0.82 9 462 4.84% 4.23% 0.74 10 463 4.85% 6.34% 0.41 11 478 5.01% 2.11% 0.12 12 423 4.43% 6.34% 0.28 13 243 2.55% 2.11% 0.74 14 312 3.27% 3.52% 0.87 15 329 3.45% 3.52% 0.96 16 478 5.01% 7.04% 0.27 17 525 5.50% 3.52% 0.30 18 219 2.29% 2.82% 0.68 19 648 6.79% 4.93% 0.38 20 271 2.84% 4.23% 0.33 21 124 1.30% 0.70% 0.55 22 281 2.94% 3.52% 0.69 X 265 2.78% 0.70% 0.01 Y 28 0.29% 0.00% 0.52
Table 5.5: Observed vs. expected aberrant DNA methylation frequency per chromosome in human lung cancer.
5.3.5 Confirmation of DNA methylation in primary human lung cancer samples via combined bisulfite restriction analysis (COBRA)
In order to confirm the DNA methylation data generated by RLGS, COBRA was performed on the 30 genes most frequently methylated in the sample set (Table 5.3).
DNA methylation was detected in 90% (27) of the genes tested. For each gene, 6 tumor- free lung and 6 primary tumor samples were assayed. Figure 5.3 shows 3 representative
COBRA results. These data indicate the DNA methylation at the NotI site, as measured
by RLGS, displays high correlation with DNA methylation in other areas of the same
CpG island where the NotI site is located.
106
Figure 5.3: COBRA analysis of frequently methylated genes. (A) SLC5A8, (B)
CYB561 and (C) FMN2. DNA methylation is detected by the appearance of restriction fragments after enzymatic digestion of PCR products. Normal and Tumor samples are labeled at the top of each panel.
107 5.3.6: Genes identified by RLGS in primary human lung cancers are epigenetically regulated
To confirm that the genes identified in our RLGS scan are subjected to epigenetic regulation, we treated the human lung cancer cell lines A549 and H719 with the DNA demethylating agent 5-aza-dC, or with 5-aza-dC in combination with the histone deacetylase inhibitor TSA. After treatment, the mRNA expression of the top 30 most frequently methylated genes in the lung sample set was measured by quantitative real-time PCR. As expected, the mRNA expression of most of these genes increased after drug treatment (Figure 5.4). In a number of cases, however, mRNA expression could not be restored (grey boxes). This observation could be explained by genomic deletions present in the cell lines. This possibility is reinforced by the fact that most of the genes that failed to be reactivated differ between the two cell lines, which are known to exhibit different genetic abnormalities. Of more interest, however, is the observation that the mRNA expression for number of genes decreased in response to drug treatment. This phenomenon presents an interesting scenario, since it suggests that at least part of the regulatory mechanism of these genes maybe be controlled by other genes which are epigenetically regulated. Similarly, our observation could be explained, at least partially, by the reactivation of miRNAs, whose regulation is known to be altered in human neoplasias269,270.
108
l
at the top of each heat
ed on the left. Gene names are listed .
expression able mRNA ression of the top 30 most frequently methylated genes in A549 and H719 lung cancer cel Cell line names and type of drug treatment are indicat Figure 5.4 mRNA exp lines. map. Grey squares indicate undetect
109 5.3.7 mRNA expression in primary lung tumors of genes identified by RLGS
In order to determine if the genes identified in our RLGS scan showed downregulation at the mRNA level in human primary lung tumors, quantitative real-time
PCR was performed on 3 tumor-free lung and 35 primary tumor specimens. As expected, all genes showed reduced mRNA expression, although at different rates. The frequency of downregulation ranged from 100% for genes such as SLC5A8 to 29% for genes such as NLF2. Overall, our data indicate that the majority of the genes identified in our RLGS scan present reduced mRNA expression in most of the primary samples tested. These data suggest that genome-wide scans for aberrant DNA methylation are an effective tool for the identification of epigenetically regulated genes in primary human lung cancers. Furthermore, as illustrated in Figure 5.2, many of the genes identified in our scan are not located in regions of LOH. This implies that the use of other well- established experimental approaches designed to interrogate the genome for genetic abnormalities, such as comparative genomic hybridization (CGH) and fluorescence in situ hybridization (FISH), among others, would have failed to identify most the genes described in this study. Thus, genetic and epigenetic approaches should be utilized in conjunction in order to maximize the chance of identifying genes involved in the tumorigenic process.
110
mors. ession, u r
man lung t average exp undetectable mRNA
e
g with their
mples, alon Grey squares indicat
tly methylated genes in primary hu
on the right.
at map. Normal lung sa ost frequen names are indicated op of the he e dicated on t ression of the top 30 m lumns. Gen in mRNA exp
. Figure 5.5: Sample identities are comprise the first 4 co expression
111 5.4 Discussion
In this study we have demonstrated that the overall levels of aberrant DNA methylation vary widely in primary human neoplasias. Our data indicate that testicular tumors, both seminomatous and non-seminomatous, are characterized by remarkably low levels of aberrant DNA methylation (0.1% and 0.4% respectively). On the other hand, aberrant DNA methylation levels in brain tumors varied in a tumor subtype-specific manner. This is illustrated by comparing childhood brain tumors, medulloblastomas and primitive neuroectodermal tumors, all of which exhibited average low levels of DNA methylation per specimen (0.3%, 0.5% and 0.5% respectively), with gliomas, which exhibited an average of 3.3% aberrant DNA methylation (~39 loci) per specimen. This is a significant observation, since it indicates that there might be an intrinsic variability in the epigenetic contribution to the development and/or progression of human neoplasias, even those derived from the same organ.
The results of our RLGS scan led us to concentrate on lung tumors for further analysis. Remarkably, lung neoplasias exhibited not only the highest average level of
DNA methylation per specimen, but they also displayed the highest proportion of tumor- type specific DNA methylation events. It should be highlighted that while the overall
DNA methylation levels in lung tumors ranged from 4.77% to 13.55%, tumor-type specific DNA methylation ranged from 3.18% to 5.18%. These results indicate that the development of lung cancer may require a relatively constant “core” of tumor-type specific aberrant DNA methylation.
Mapping of the loci identified in our scan to the human genome denoted that chromosome 5 might be particularly susceptible to aberrant DNA methylation in lung cancer. Even though this result failed to achieve statistical significance (p = 0.06), it indicates a trend that could provide valuable information to further the current
112 understanding of the etiology of the disease. It is possible that chromosome 5 harbors a relatively large number of tumor suppressor genes whose epigenetic inactivation might be required for the initiation of lung tumorigenesis, or whose abrogation may confer a growth advantage to cancer cells. Conversely, chromosome 5 may contain a higher than average number of specific sequence motifs which have been associated with the presence of aberrant DNA methylation271-274. A trend towards absence of DNA
methylation was identified for chromosome 11 (p = 0.12), and it reached statistical
significance for the X chromosome (p = 0.01). The fact that statistical significance was
attained only for the X chromosome deserved special attention. This result may imply
that perturbations of the mechanism operating exclusively on this chromosome to
regulate gene dosage, might either result in cell death, or they may confer the cell a
significant growth disadvantage268.
Verification of RLGS result was performed via COBRA. As expected, aberrant DNA
methylation was observed in 90% of the genes tested. The importance of this validation
resides on the fact that assessment of DNA methylation at a landmark site, such as the
NotI recognition sequence, can be correlated with DNA methylation in neighboring
genomic areas. NotI sites are preferentially found in CpG islands47; therefore our data indicates that methylation at the NotI site generally translates to methylation in the rest of the CpG island. Importantly, this also indicates that, in the context of CpG islands, aberrant DNA methylation does not appear to be confined to specific areas of the island, but that it spreads throughout.
Treatment of lung cancer cell lines with a DNA methylation inhibitor resulted in the up-regulation of mRNA levels for most of the genes tested. This result is evidence that these genes are epigenetically regulated either directly or indirectly. However, the mRNA levels of some genes decreased as a result of drug treatment. This is not surprising,
113 since the current methodology for treatment of cell lines with DNA methylation inhibitors results in a systemic effect over the entire genome. Thus, it is possible that the normal metabolism of other factors involved in the expression control of the down-regulated genes may be altered by the drug treatment. Nevertheless, this observation deserves further study, as it might help elucidate the regulatory mechanism for some of these genes.
Quantitative real-time PCR performed on primary lung tumor specimens indicated that the majority of the genes identified as aberrantly methylated in lung cancer show in fact down-regulation at the mRNA level. These data validated our experimental approach, highlighting the utility of RLGS scans as a discovery tool for the identification of aberrantly methylated genes. Further support for our investigative strategy can be gathered from the current literature, where the expression of some genes detected in this study, such as HOXA9 and HOXA10, had been previously described as being altered in human primary lung cancers275-278.
A significant observation stemming from our work is that transcription factors seem to
be a major target of aberrant DNA methylation in human lung neoplasias. Given that
transcription factors usually regulate the expression of several genes, their concomitant
silencing is likely to affect a large number of genetic pathways. In particular, the majority
of transcription factors identified in this study are known to be involved in cellular
differentiation of lung279,280 as well as other organs281-284. This suggests that the etiology of human lung tumors may be intimately related with alterations in mechanisms governing cellular differentiation.
In this study we have been able to demonstrate that there is an intrinsic variability in the contribution of aberrant DNA methylation to the development and/or progression of human neoplasias. We were also able to identify 142 genes aberrantly methylated in
114 human lung cancer, of which 71 exhibited tumor-type specific DNA methylation. This suggests that abrogation of their expression may play an important role in the etiology of lung cancer. Moreover, our analysis underscored that alterations in the normal expression of transcription factors, particularly those involved in cellular differentiation, may be a hallmark of this type of neoplasia. Overall, our study provides clear evidence that the identification of genes subjected to epigenetic silencing in human neoplasias may provide a viable approach to identifying molecular targets that could become the object of novel therapeutic strategies. Furthermore, the identification of genes epigenetically silenced in a tumor-type specific manner could increase the efficacy of treatments, by devising therapeutic regiments tailored to addressing the specific molecular abnormalities that characterize those tumors.
Another important aspect of this research is its potential to identify disease biomarkers. Put together, our data has provided individual genes that are aberrantly methylated in a tumor-type specific manner. Some of these genes grant further investigation; if, for example, some of them are determined to become aberrantly methylated early in the tumorigenic process, existing technology could be utilized to screen high risk patients. Another viable avenue would be to determine their prognostic value. Tissue arrays provide an excellent platform for this type of studies and have proven successful in the past75,265,285.
Interestingly, other research groups are approaching the biomarker discovery arena
from a fairly different angle. Shames and colleagues72 searched for biomarkers in
several cancer types by identifying genes that could be re-expressed in cell lines treated
with 5-aza-dC. Their experimental strategy also enabled them to compare gene
reactivation in various tumor types, which resulted in the proposal of not only new lung-
cancer specific biomarkers, but also pan-cancer biomarkers. Pan-cancer biomarkers
115 were defined as those genes epigenetically silenced in more than one tumor type, which could be reactivated by a DNA demethylating agent in cancer cell lines. It should be noted that our study has provided comparable data, despite the fact that both experimental approaches were unique. Therefore, by combining these two experimental methodologies, it might be possible to identify biomarkers with high accuracy, since their identification would require that they overcome the various limitations of both strategies.
116 CHAPTER 6
FUTURE DIRECTIONS
6.1 The impact of aberrant DNA methylation in human lung cancer
As we have shown in our studies, aberrant DNA methylation seems to be an integral component in the etiology of lung cancer. Importantly, it is clear that some genes are inactivated only in lung neoplasias and only through DNA methylation. These observations underscore the unique potential of comprehensive epigenetic analyses in furthering our current understanding of lung cancer development and progression.
However, the current gold standard methodology for genome-wide scan of CpG island
DNA methylation, RLGS, requires relatively large amounts of genetic material and is time consuming, precluding the analysis of large sample sets. Furthermore, since RLGS is based on the presence of rare-cutting methylation sensitive restriction enzyme sequences in gene promoters for the interrogation of their DNA methylation status, genes which do not possess those restriction enzyme sequences cannot be evaluated.
Thus, the development of new technical approaches with higher coverage of promoter sequences should undoubtedly become a priority for the epigenetics field. The refinement of currently existing CpG island promoter arrays seems promising. These arrays are able to provide extensive coverage of promoter sequences and require relatively low amounts of tumor DNA. Nevertheless, their use is still limited and several
117 technical difficulties must still be addressed in order for them to become a routine analytical tool.
The discovery of new prognostic factors derived from epigenetic analyses will also shed light on the mechanisms underlying lung tumorigenesis. Once discovered, it will be crucial to try to unravel how those new genes interact with currently known genetic and metabolic pathways in an effort to identify functional redundancies and, most importantly, their unique biological function.
Recent studies have highlighted the functional important of non-coding RNAs, including micro RNAs, in normal cellular homeostasis as well as disease. It is currently unclear if non-coding RNAs interact with the epigenetic machinery. However, given the large number of genes non-coding RNAs have been shown to regulate, it is likely a connection to the epigenome will soon be found. It will be of outmost importance to integrate this emerging science field in DNA methylation studies, as it is likely to help elucidate, at least in part, how aberrant DNA methylation patterns are established and why some sequences exhibit susceptibility of DNA methylation while others seem refractory to it.
Animal models, especially transgenic mice, will also provide a major pillar for in vivo studies. Conditional expression of different components of the epigenetic machinery in a cell and organ-specific manner should help clarify if the deregulation of epigenetic mechanisms is defined by a “universal” series of events or if these events differ significantly in accordance with the cellular context in which they occur. Answering this specific question would enable researchers to maximize the therapeutic efficacy of treatments that alter the epigenome, while at the same time minimizing their potentially harmful effect on uninvolved organs and tissues.
118 6.2 Concluding remarks
The research presented in this thesis has highlighted that DNA methylation mechanisms are profoundly altered in human lung cancer. Specifically, we have been able to show that the genome-wide scanning of CpG islands for aberrant DNA methylation can result in the discovery of novel gene targets, such as oligodendrocyte transcription factor 1, which undergo inactivation during the tumorigenic process.
The discovery of this novel target gene, which we later were able to show could serve as a prognostic factor, is only the beginning. I hope new research conducted by myself and my colleagues will help determine how OLIG1 functions in human lung
cancer. I also hope the small contribution I have made, with the help of my colleagues
and collaborators, towards increasing our current understanding of the etiology of lung
cancer will, in time, lead to improvements in patient treatment. The ultimate goal of
medical research is to cure, or at least lessen, the effect of disease on people in an effort
to improve their quality of life. Keeping this goal in mind should help the research
community make the appropriate decisions for society to truly feel and benefit from the
impact of our work.
119
REFERENCES
1. Jemal, A., Murray, T., Ward, E., Samuels, A., Tiwari, R.C., Ghafoor, A., Feuer, E.J. & Thun, M.J. Cancer statistics, 2005. CA Cancer J Clin. 2005;55:10-30.
2. Fearon, E.R., Hamilton, S.R. & Vogelstein, B. Clonal analysis of human colorectal tumors. Science. 1987;238:193-7.
3. Saracci, R. & Riboli, E. Passive smoking and lung cancer: current evidence and ongoing studies at the International Agency for Research on Cancer. Mutat Res. 1989;222:117-27.
4. Cooper, S. & Spiro, S.G. Small cell lung cancer: treatment review. Respirology. 2006;11:241-8.
5. Rubery, E.D. & Newton, A.A. DNA methylation in normal and tumour virus- transformed cells in tissue culture. I. The level of DNA methylation in BHK21 cells and in BHK21 cells transformed by polyoma virus (PyY cells). Biochim Biophys Acta. 1973;324:24-36.
6. Guseinov, V.A. & Vanyushin, B.F. Content and localisation of 5-methylcytosine in DNA of healthy and wilt-infected cotton plants. Biochim Biophys Acta. 1975;395:229-38.
7. Beranek, D.T., Weis, C.C. & Swenson, D.H. A comprehensive quantitative analysis of methylated and ethylated DNA using high pressure liquid chromatography. Carcinogenesis. 1980;1:595-606.
8. Lapeyre, J.N., Walker, M.S. & Becker, F.F. DNA methylation and methylase levels in normal and malignant mouse hepatic tissues. Carcinogenesis. 1981;2:873-8.
9. Shmookler Reis, R.J. & Goldstein, S. Loss of reiterated DNA sequences during serial passage of human diploid fibroblasts. Cell. 1980;21:739-49.
10. McGhee, J.D. & Ginder, G.D. Specific DNA methylation sites in the vicinity of the chicken beta-globin genes. Nature. 1979;280:419-20.
11. Mandel, J.L. & Chambon, P. DNA methylation: organ specific variations in the methylation pattern within and around ovalbumin and other chicken genes. Nucleic Acids Res. 1979;7:2081-103.
120 12. Jones, P.A. & Taylor, S.M. Cellular differentiation, cytidine analogs and DNA methylation. Cell. 1980;20:85-93.
13. van der Ploeg, L.H. & Flavell, R.A. DNA methylation in the human gamma delta beta-globin locus in erythroid and nonerythroid tissues. Cell. 1980;19:947-58.
14. Shen, C.K. & Maniatis, T. Tissue-specific DNA methylation in a cluster of rabbit beta-like globin genes. Proc Natl Acad Sci U S A. 1980;77:6634-8.
15. Boehm, T.L. & Drahovsky, D. Alteration of enzymatic DNA methylation by chemical carcinogens. Recent Results Cancer Res. 1983;84:212-25.
16. Cheah, M.S., Wallace, C.D. & Hoffman, R.M. Hypomethylation of DNA in human cancer cells: a site-specific change in the c-myc oncogene. J Natl Cancer Inst. 1984;73:1057-65.
17. Razin, A. & Riggs, A.D. DNA methylation and gene function. Science. 1980;210:604-10.
18. Vardimon, L., Neumann, R., Kuhlmann, I., Sutter, D. & Doerfler, W. DNA methylation and viral gene expression in adenovirus-transformed and -infected cells. Nucleic Acids Res. 1980;8:2461-73.
19. Benvenisty, N., Razin, A. & Reshef, L. Developmental changes in the methylation pattern, chromatin conformation and expression of the rat phosphoenolpyruvate carboxykinase gene. Prog Clin Biol Res. 1985;198:185-99.
20. Sanford, J.P., Clark, H.J., Chapman, V.M. & Rossant, J. Differences in DNA methylation during oogenesis and spermatogenesis and their persistence during early embryogenesis in the mouse. Genes Dev. 1987;1:1039-46.
21. Monk, M., Boubelik, M. & Lehnert, S. Temporal and regional changes in DNA methylation in the embryonic, extraembryonic and germ cell lineages during mouse embryo development. Development. 1987;99:371-82.
22. Li, E., Bestor, T.H. & Jaenisch, R. Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell. 1992;69:915-26.
23. Reik, W., Collick, A., Norris, M.L., Barton, S.C. & Surani, M.A. Genomic imprinting determines methylation of parental alleles in transgenic mice. Nature. 1987;328:248-51.
24. Chaillet, J.R., Vogt, T.F., Beier, D.R. & Leder, P. Parental-specific methylation of an imprinted transgene is established during gametogenesis and progressively changes during embryogenesis. Cell. 1991;66:77-83.
25. Li, E., Beard, C., Forster, A.C., Bestor, T.H. & Jaenisch, R. DNA methylation, genomic imprinting, and mammalian development. Cold Spring Harb Symp Quant Biol. 1993;58:297-305.
121 26. Li, E., Beard, C. & Jaenisch, R. Role for DNA methylation in genomic imprinting. Nature. 1993;366:362-5.
27. Ariel, M., Robinson, E., McCarrey, J.R. & Cedar, H. Gamete-specific methylation correlates with imprinting of the murine Xist gene. Nat Genet. 1995;9:312-5.
28. Mohandas, T., Sparkes, R.S. & Shapiro, L.J. Reactivation of an inactive human X chromosome: evidence for X inactivation by DNA methylation. Science. 1981;211:393-6.
29. Riggs, A.D. X inactivation, differentiation, and DNA methylation. Cytogenet Cell Genet. 1975;14:9-25.
30. Tuck-Muller, C.M., Narayan, A., Tsien, F., Smeets, D.F., Sawyer, J., Fiala, E.S., Sohn, O.S. & Ehrlich, M. DNA hypomethylation and unusual chromosome instability in cell lines from ICF syndrome patients. Cytogenet Cell Genet. 2000;89:121-8.
31. Sciandrello, G., Caradonna, F., Mauro, M. & Barbata, G. Arsenic-induced DNA hypomethylation affects chromosomal instability in mammalian cells. Carcinogenesis. 2004;25:413-7.
32. Chen, R.Z., Pettersson, U., Beard, C., Jackson-Grusby, L. & Jaenisch, R. DNA hypomethylation leads to elevated mutation rates. Nature. 1998;395:89-93.
33. Kitamura, E., Igarashi, J., Morohashi, A., Hida, N., Oinuma, T., Nemoto, N., Song, F., Ghosh, S., Held, W.A., Yoshida-Noro, C. & Nagase, H. Analysis of tissue-specific differentially methylated regions (TDMs) in humans. Genomics. 2006.
34. Gyory, I. & Minarovits, J. Epigenetic regulation of lymphoid specific gene sets. Biochem Cell Biol. 2005;83:286-95.
35. Feinberg, A.P. & Vogelstein, B. Hypomethylation distinguishes genes of some human cancers from their normal counterparts. Nature. 1983;301:89-92.
36. Feinberg, A.P. & Vogelstein, B. Hypomethylation of ras oncogenes in primary human cancers. Biochem Biophys Res Commun. 1983;111:47-54.
37. Diala, E.S., Cheah, M.S., Rowitch, D. & Hoffman, R.M. Extent of DNA methylation in human tumor cells. J Natl Cancer Inst. 1983;71:755-64.
38. Jones, P.A. & Baylin, S.B. The fundamental role of epigenetic events in cancer. Nat Rev Genet. 2002;3:415-28.
39. Baylin, S.B., Herman, J.G., Graff, J.R., Vertino, P.M. & Issa, J.P. Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv Cancer Res. 1998;72:141-96.
122 40. Holliday, R. & Jeggo, P.A. Mechanisms for changing gene expression and their possible relationship to carcinogenesis. Cancer Surv. 1985;4:557-81.
41. Baylin, S.B., Makos, M., Wu, J.J., Yen, R.W., de Bustros, A., Vertino, P. & Nelkin, B.D. Abnormal patterns of DNA methylation in human neoplasia: potential consequences for tumor progression. Cancer Cells. 1991;3:383-90.
42. Hatada, I., Hayashizaki, Y., Hirotsune, S., Komatsubara, H. & Mukai, T. A genomic scanning method for higher organisms using restriction sites as landmarks. Proc Natl Acad Sci U S A. 1991;88:9523-7.
43. Larsen, F., Gundersen, G., Lopez, R. & Prydz, H. CpG islands as gene markers in the human genome. Genomics. 1992;13:1095-107.
44. Okazaki, Y., Okuizumi, H., Sasaki, N., Ohsumi, T., Kuromitsu, J., Hirota, N., Muramatsu, M. & Hayashizaki, Y. An expanded system of restriction landmark genomic scanning (RLGS Ver. 1.8). Electrophoresis. 1995;16:197-202.
45. Clark, S.J., Harrison, J., Paul, C.L. & Frommer, M. High sensitivity mapping of methylated cytosines. Nucleic Acids Res. 1994;22:2990-7.
46. Herman, J.G., Graff, J.R., Myohanen, S., Nelkin, B.D. & Baylin, S.B. Methylation- specific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A. 1996;93:9821-6.
47. Costello, J.F., Fruhwald, M.C., Smiraglia, D.J., Rush, L.J., Robertson, G.P., Gao, X., Wright, F.A., Feramisco, J.D., Peltomaki, P., Lang, J.C., Schuller, D.E., Yu, L., Bloomfield, C.D., Caligiuri, M.A., Yates, A., Nishikawa, R., Su Huang, H., Petrelli, N.J., Zhang, X., O'Dorisio, M.S., Held, W.A., Cavenee, W.K. & Plass, C. Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. Nat Genet. 2000;24:132-8.
48. Shiraishi, M. & Sekiya, T. Change in methylation status of DNA of remaining allele in human lung cancers with loss of heterozygosity on chromosomes 3p and 13q. Jpn J Cancer Res. 1989;80:924-7.
49. Makos, M., Nelkin, B.D., Lerman, M.I., Latif, F., Zbar, B. & Baylin, S.B. Distinct hypermethylation patterns occur at altered chromosome loci in human lung and colon cancer. Proc Natl Acad Sci U S A. 1992;89:1929-33.
50. Knudson, A.G., Jr. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci U S A. 1971;68:820-3.
51. Vertino, P.M., Spillare, E.A., Harris, C.C. & Baylin, S.B. Altered chromosomal methylation patterns accompany oncogene-induced transformation of human bronchial epithelial cells. Cancer Res. 1993;53:1684-9.
123 52. Agathanggelou, A., Honorio, S., Macartney, D.P., Martinez, A., Dallol, A., Rader, J., Fullwood, P., Chauhan, A., Walker, R., Shaw, J.A., Hosoe, S., Lerman, M.I., Minna, J.D., Maher, E.R. & Latif, F. Methylation associated inactivation of RASSF1A from region 3p21.3 in lung, breast and ovarian tumours. Oncogene. 2001;20:1509-18.
53. Grote, H.J., Schmiemann, V., Geddert, H., Rohr, U.P., Kappes, R., Gabbert, H.E. & Bocking, A. Aberrant promoter methylation of p16(INK4a), RARB2 and SEMA3B in bronchial aspirates from patients with suspected lung cancer. Int J Cancer. 2005;116:720-5.
54. Hsu, H.S., Wen, C.K., Tang, Y.A., Lin, R.K., Li, W.Y., Hsu, W.H. & Wang, Y.C. Promoter hypermethylation is the predominant mechanism in hMLH1 and hMSH2 deregulation and is a poor prognostic factor in nonsmoking lung cancer. Clin Cancer Res. 2005;11:5410-6.
55. Zochbauer-Muller, S., Fong, K.M., Virmani, A.K., Geradts, J., Gazdar, A.F. & Minna, J.D. Aberrant promoter methylation of multiple genes in non-small cell lung cancers. Cancer Res. 2001;61:249-55.
56. Nagatake, M., Osada, H., Kondo, M., Uchida, K., Nishio, M., Shimokata, K., Takahashi, T. & Takahashi, T. Aberrant hypermethylation at the bcl-2 locus at 18q21 in human lung cancers. Cancer Res. 1996;56:1886-91.
57. Toyooka, S., Toyooka, K.O., Miyajima, K., Reddy, J.L., Toyota, M., Sathyanarayana, U.G., Padar, A., Tockman, M.S., Lam, S., Shivapurkar, N. & Gazdar, A.F. Epigenetic down-regulation of death-associated protein kinase in lung cancers. Clin Cancer Res. 2003;9:3034-41.
58. Smith, L.T., Lin, M., Brena, R.M., Lang, J.C., Schuller, D.E., Otterson, G.A., Morrison, C.D., Smiraglia, D.J. & Plass, C. Epigenetic regulation of the tumor suppressor gene TCF21 on 6q23-q24 in lung and head and neck cancer. Proc Natl Acad Sci U S A. 2006;103:982-7.
59. Dai, Z., Popkie, A.P., Zhu, W.G., Timmers, C.D., Raval, A., Tannehill-Gregg, S., Morrison, C.D., Auer, H., Kratzke, R.A., Niehans, G., Amatschek, S., Sommergruber, W., Leone, G.W., Rosol, T., Otterson, G.A. & Plass, C. Bone morphogenetic protein 3B silencing in non-small-cell lung cancer. Oncogene. 2004;23:3521-9.
60. Otterson, G.A., Khleif, S.N., Chen, W., Coxon, A.B. & Kaye, F.J. CDKN2 gene silencing in lung cancer by DNA hypermethylation and kinetics of p16INK4 protein induction by 5-aza 2'deoxycytidine. Oncogene. 1995;11:1211-6.
61. Kohno, T., Kawanishi, M., Inazawa, J. & Yokota, J. Identification of CpG islands hypermethylated in human lung cancer by the arbitrarily primed-PCR method. Hum Genet. 1998;102:258-64.
124 62. Dammann, R., Takahashi, T. & Pfeifer, G.P. The CpG island of the novel tumor suppressor gene RASSF1A is intensely methylated in primary small cell lung carcinomas. Oncogene. 2001;20:3563-7.
63. Dai, Z., Lakshmanan, R.R., Zhu, W.G., Smiraglia, D.J., Rush, L.J., Fruhwald, M.C., Brena, R.M., Li, B., Wright, F.A., Ross, P., Otterson, G.A. & Plass, C. Global methylation profiling of lung cancer identifies novel methylated genes. Neoplasia. 2001;3:314-23.
64. Bowman, R.V., Yang, I.A., Semmler, A.B. & Fong, K.M. Epigenetics of lung cancer. Respirology. 2006;11:355-65.
65. Tsou, J.A., Shen, L.Y., Siegmund, K.D., Long, T.I., Laird, P.W., Seneviratne, C.K., Koss, M.N., Pass, H.I., Hagen, J.A. & Laird-Offringa, I.A. Distinct DNA methylation profiles in malignant mesothelioma, lung adenocarcinoma, and non- tumor lung. Lung Cancer. 2005;47:193-204.
66. Swank, P.R., Greenberg, S.D., Hunter, N.R., Spjut, H.J., Estrada, R., Trahan, E.B. & Montalvo, J. Identification of features of metaplastic cells in sputum for the detection of squamous-cell carcinoma of the lung. Diagn Cytopathol. 1989;5:98- 103.
67. Bunn, P.A., Jr. Molecular biology and early diagnosis in lung cancer. Lung Cancer. 2002;38:S5-8.
68. Belinsky, S.A. Gene-promoter hypermethylation as a biomarker in lung cancer. Nat Rev Cancer. 2004;4:707-17.
69. Fruhwald, M.C. DNA methylation patterns in cancer: novel prognostic indicators? Am J Pharmacogenomics. 2003;3:245-60.
70. Kim, H., Kwon, Y.M., Kim, J.S., Lee, H., Park, J.H., Shim, Y.M., Han, J., Park, J. & Kim, D.H. Tumor-specific methylation in bronchial lavage for the early detection of non-small-cell lung cancer. J Clin Oncol. 2004;22:2363-70.
71. Belinsky, S.A., Nikula, K.J., Palmisano, W.A., Michels, R., Saccomanno, G., Gabrielson, E., Baylin, S.B. & Herman, J.G. Aberrant methylation of p16(INK4a) is an early event in lung cancer and a potential biomarker for early diagnosis. Proc Natl Acad Sci U S A. 1998;95:11891-6.
72. Shames, D.S., Girard, L., Gao, B., Sato, M., Lewis, C.M., Shivapurkar, N., Jiang, A., Perou, C.M., Kim, Y.H., Pollack, J.R., Fong, K.M., Lam, C.L., Wong, M., Shyr, Y., Nanda, R., Olopade, O.I., Gerald, W., Euhus, D.M., Shay, J.W., Gazdar, A.F. & Minna, J.D. A genome-wide screen for promoter methylation in lung cancer identifies novel methylation markers for multiple malignancies. PLoS Med. 2006;3:e486.
73. Colella, S., Shen, L., Baggerly, K.A., Issa, J.P. & Krahe, R. Sensitive and quantitative universal Pyrosequencing methylation analysis of CpG sites. Biotechniques. 2003;35:146-50. 125 74. Ehrich, M., Nelson, M.R., Stanssens, P., Zabeau, M., Liloglou, T., Xinarianos, G., Cantor, C.R., Field, J.K. & van den Boom, D. Quantitative high-throughput analysis of DNA methylation patterns by base-specific cleavage and mass spectrometry. Proc Natl Acad Sci U S A. 2005;102:15785-90.
75. Lu, Y., Lemon, W., Liu, P.Y., Yi, Y., Morrison, C., Yang, P., Sun, Z., Szoke, J., Gerald, W.L., Watson, M., Govindan, R. & You, M. A gene expression signature predicts survival of patients with stage I non-small cell lung cancer. PLoS Med. 2006;3:e467.
76. Sun, Z. & Yang, P. Gene expression profiling on lung cancer outcome prediction: present clinical value and future premise. Cancer Epidemiol Biomarkers Prev. 2006;15:2063-8.
77. Guo, L., Ma, Y., Ward, R., Castranova, V., Shi, X. & Qian, Y. Constructing molecular classifiers for the accurate prognosis of lung adenocarcinoma. Clin Cancer Res. 2006;12:3344-54.
78. Ichihara, S., Toyooka, S., Fujiwara, Y., Hotta, K., Shigematsu, H., Tokumo, M., Soh, J., Asano, H., Ichimura, K., Aoe, K., Aoe, M., Kiura, K., Shimizu, K., Date, H. & Shimizu, N. The impact of epidermal growth factor receptor gene status on gefitinib-treated Japanese patients with non-small-cell lung cancer. Int J Cancer. 2006.
79. Ahmed, S.M. & Salgia, R. Epidermal growth factor receptor mutations and susceptibility to targeted therapy in lung cancer. Respirology. 2006;11:687-92.
80. Jeon, Y.K., Sung, S.W., Chung, J.H., Park, W.S., Seo, J.W., Kim, C.W. & Chung, D.H. Clinicopathologic features and prognostic implications of epidermal growth factor receptor (EGFR) gene copy number and protein expression in non-small cell lung cancer. Lung Cancer. 2006;54:387-98.
81. Mendelsohn, J. & Baselga, J. Epidermal growth factor receptor targeting in cancer. Semin Oncol. 2006;33:369-85.
82. Christman, J.K. Separation of major and minor deoxyribonucleoside monophosphates by reverse-phase high-performance liquid chromatography: a simple method applicable to quantitation of methylated nucleotides in DNA. Anal Biochem. 1982;119:38-48.
83. Chiang, P.K., Gordon, R.K., Tal, J., Zeng, G.C., Doctor, B.P., Pardhasaradhi, K. & McCann, P.P. S-Adenosylmethionine and methylation. Faseb J. 1996;10:471- 80.
84. Franchina, M. & Kay, P.H. Evidence that cytosine residues within 5'-CCTGG-3' pentanucleotides can be methylated in human DNA independently of the methylating system that modifies 5'-CG-3' dinucleotides. DNA Cell Biol. 2000;19:521-6.
126 85. Malone, C.S., Miner, M.D., Doerr, J.R., Jackson, J.P., Jacobsen, S.E., Wall, R. & Teitell, M. CmC(A/T)GG DNA methylation in mature B cell lymphoma gene silencing. Proc Natl Acad Sci U S A. 2001;98:10404-9.
86. Clark, S.J., Harrison, J. & Frommer, M. CpNpG methylation in mammalian cells. Nat Genet. 1995;10:20-7.
87. Ramsahoye, B.H., Biniszkiewicz, D., Lyko, F., Clark, V., Bird, A.P. & Jaenisch, R. Non-CpG methylation is prevalent in embryonic stem cells and may be mediated by DNA methyltransferase 3a. Proc Natl Acad Sci U S A. 2000;97:5237-42.
88. Schmitt, F., Oakeley, E.J. & Jost, J.P. Antibiotics induce genome-wide hypermethylation in cultured Nicotiana tabacum plants. J Biol Chem. 1997;272:1534-40.
89. Blanchard, F., Tracy, E., Smith, J., Chattopadhyay, S., Wang, Y., Held, W.A. & Baumann, H. DNA methylation controls the responsiveness of hepatoma cells to leukemia inhibitory factor. Hepatology. 2003;38:1516-28.
90. Egger, G., Liang, G., Aparicio, A. & Jones, P.A. Epigenetics in human disease and prospects for epigenetic therapy. Nature. 2004;429:457-63.
91. Fruhwald, M.C., O'Dorisio, M.S., Dai, Z., Rush, L.J., Krahe, R., Smiraglia, D.J., Pietsch, T., Elsea, S.H. & Plass, C. Aberrant hypermethylation of the major breakpoint cluster region in 17p11.2 in medulloblastomas but not supratentorial PNETs. Genes Chromosomes Cancer. 2001;30:38-47.
92. Jones, P.A. & Laird, P.W. Cancer epigenetics comes of age. Nat Genet. 1999;21:163-7.
93. Ballabio, A. & Willard, H.F. Mammalian X-chromosome inactivation and the XIST gene. Curr Opin Genet Dev. 1992;2:439-47.
94. Heard, E., Clerc, P. & Avner, P. X-chromosome inactivation in mammals. Annu Rev Genet. 1997;31:571-610.
95. Allaman-Pillet, N., Djemai, A., Bonny, C. & Schorderet, D.F. Methylation status of CpG sites and methyl-CpG binding proteins are involved in the promoter regulation of the mouse Xist gene. Gene Expr. 1998;7:61-73.
96. Herman, J.G. & Baylin, S.B. Gene silencing in cancer in association with promoter hypermethylation. N Engl J Med. 2003;349:2042-54.
97. Colot, V. & Rossignol, J.L. Eukaryotic DNA methylation as an evolutionary device. Bioessays. 1999;21:402-11.
98. Gardiner-Garden, M. & Frommer, M. CpG islands in vertebrate genomes. J Mol Biol. 1987;196:261-82.
127 99. Takai, D. & Jones, P.A. Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc Natl Acad Sci U S A. 2002;99:3740-5.
100. Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev. 2002;16:6-21.
101. Jaenisch, R. & Bird, A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat Genet. 2003;33 Suppl:245-54.
102. Dai, Z., Zhu, W.G., Morrison, C.D., Brena, R.M., Smiraglia, D.J., Raval, A., Wu, Y.Z., Rush, L.J., Ross, P., Molina, J.R., Otterson, G.A. & Plass, C. A comprehensive search for DNA amplification in lung cancer identifies inhibitors of apoptosis cIAP1 and cIAP2 as candidate oncogenes. Hum Mol Genet. 2003;12:791-801.
103. Okano, M., Takebayashi, S., Okumura, K. & Li, E. Assignment of cytosine-5 DNA methyltransferases Dnmt3a and Dnmt3b to mouse chromosome bands 12A2-A3 and 2H1 by in situ hybridization. Cytogenet Cell Genet. 1999;86:333-4.
104. Santos, F., Hendrich, B., Reik, W. & Dean, W. Dynamic reprogramming of DNA methylation in the early mouse embryo. Dev Biol. 2002;241:172-82.
105. Gaudet, F., Rideout, W.M., 3rd, Meissner, A., Dausman, J., Leonhardt, H. & Jaenisch, R. Dnmt1 expression in pre- and postimplantation embryogenesis and the maintenance of IAP silencing. Mol Cell Biol. 2004;24:1640-8.
106. Grandjean, P.W., Crouse, S.F. & Rohack, J.J. Influence of cholesterol status on blood lipid and lipoprotein enzyme responses to aerobic exercise. J Appl Physiol. 2000;89:472-80.
107. Gringras, P. & Chen, W. Mechanisms for differences in monozygous twins. Early Hum Dev. 2001;64:105-17.
108. Cardno, A.G., Rijsdijk, F.V., Sham, P.C., Murray, R.M. & McGuffin, P. A twin study of genetic relationships between psychotic symptoms. Am J Psychiatry. 2002;159:539-45.
109. Fraga, M.F., Ballestar, E., Paz, M.F., Ropero, S., Setien, F., Ballestar, M.L., Heine-Suner, D., Cigudosa, J.C., Urioste, M., Benitez, J., Boix-Chornet, M., Sanchez-Aguilera, A., Ling, C., Carlsson, E., Poulsen, P., Vaag, A., Stephan, Z., Spector, T.D., Wu, Y.Z., Plass, C. & Esteller, M. Epigenetic differences arise during the lifetime of monozygotic twins. Proc Natl Acad Sci U S A. 2005;102:10604-9.
110. Cantoni, G.L. The role of S-adenosylhomocysteine in the biological utilization of S-adenosylmethionine. Prog Clin Biol Res. 1985;198:47-65.
128 111. Yi, P., Melnyk, S., Pogribna, M., Pogribny, I.P., Hine, R.J. & James, S.J. Increase in plasma homocysteine associated with parallel increases in plasma S- adenosylhomocysteine and lymphocyte DNA hypomethylation. J Biol Chem. 2000;275:29318-23.
112. Wainfan, E. & Poirier, L.A. Methyl groups in carcinogenesis: effects on DNA methylation and gene expression. Cancer Res. 1992;52:2071s-2077s.
113. Pogribny, I.P., Basnakian, A.G., Miller, B.J., Lopatina, N.G., Poirier, L.A. & James, S.J. Breaks in genomic DNA and within the p53 gene are associated with hypomethylation in livers of folate/methyl-deficient rats. Cancer Res. 1995;55:1894-901.
114. Pogribny, I.P., James, S.J., Jernigan, S. & Pogribna, M. Genomic hypomethylation is specific for preneoplastic liver in folate/methyl deficient rats and does not occur in non-target tissues. Mutat Res. 2004;548:53-9.
115. Shivapurkar, N. & Poirier, L.A. Tissue levels of S-adenosylmethionine and S- adenosylhomocysteine in rats fed methyl-deficient, amino acid-defined diets for one to five weeks. Carcinogenesis. 1983;4:1051-7.
116. Smiraglia, D.J., Rush, L.J., Fruhwald, M.C., Dai, Z., Held, W.A., Costello, J.F., Lang, J.C., Eng, C., Li, B., Wright, F.A., Caligiuri, M.A. & Plass, C. Excessive CpG island hypermethylation in cancer cell lines versus primary human malignancies. Hum Mol Genet. 2001;10:1413-9.
117. Herman, J.G., Latif, F., Weng, Y., Lerman, M.I., Zbar, B., Liu, S., Samid, D., Duan, D.S., Gnarra, J.R., Linehan, W.M. & et al. Silencing of the VHL tumor- suppressor gene by DNA methylation in renal carcinoma. Proc Natl Acad Sci U S A. 1994;91:9700-4.
118. Momparler, R.L., Eliopoulos, N. & Ayoub, J. Evaluation of an inhibitor of DNA methylation, 5-aza-2'-deoxycytidine, for the treatment of lung cancer and the future role of gene therapy. Adv Exp Med Biol. 2000;465:433-46.
119. Esteller, M. Cancer epigenetics: DNA methylation and chromatin alterations in human cancer. Adv Exp Med Biol. 2003;532:39-49.
120. Han, S.Y., Iliopoulos, D., Druck, T., Guler, G., Grubbs, C.J., Pereira, M., Zhang, Z., You, M., Lubet, R.A., Fong, L.Y. & Huebner, K. CpG methylation in the Fhit regulatory region: relation to Fhit expression in murine tumors. Oncogene. 2004;23:3990-8.
121. Kim, J.S., Lee, H., Kim, H., Shim, Y.M., Han, J., Park, J. & Kim, D.H. Promoter methylation of retinoic acid receptor beta 2 and the development of second primary lung cancers in non-small-cell lung cancer. J Clin Oncol. 2004;22:3443- 50.
129 122. Maruyama, R., Sugio, K., Yoshino, I., Maehara, Y. & Gazdar, A.F. Hypermethylation of FHIT as a prognostic marker in nonsmall cell lung carcinoma. Cancer. 2004;100:1472-7.
123. Sathyanarayana, U.G., Padar, A., Huang, C.X., Suzuki, M., Shigematsu, H., Bekele, B.N. & Gazdar, A.F. Aberrant promoter methylation and silencing of laminin-5-encoding genes in breast carcinoma. Clin Cancer Res. 2003;9:6389- 94.
124. Sorm, F., Piskala, A., Cihak, A. & Vesely, J. 5-Azacytidine, a new, highly effective cancerostatic. Experientia. 1964;20:202-3.
125. Goffin, J. & Eisenhauer, E. DNA methyltransferase inhibitors-state of the art. Ann Oncol. 2002;13:1699-716.
126. Issa, J.P., Garcia-Manero, G., Giles, F.J., Mannari, R., Thomas, D., Faderl, S., Bayar, E., Lyons, J., Rosenfeld, C.S., Cortes, J. & Kantarjian, H.M. Phase 1 study of low-dose prolonged exposure schedules of the hypomethylating agent 5- aza-2'-deoxycytidine (decitabine) in hematopoietic malignancies. Blood. 2004;103:1635-40.
127. Yan, L., Nass, S.J., Smith, D., Nelson, W.G., Herman, J.G. & Davidson, N.E. Specific inhibition of DNMT1 by antisense oligonucleotides induces re-expression of estrogen receptor-alpha (ER) in ER-negative human breast cancer cell lines. Cancer Biol Ther. 2003;2:552-6.
128. Chuang, J.C., Yoo, C.B., Kwan, J.M., Li, T.W., Liang, G., Yang, A.S. & Jones, P.A. Comparison of biological effects of non-nucleoside DNA methylation inhibitors versus 5-aza-2'-deoxycytidine. Mol Cancer Ther. 2005;4:1515-20.
129. Juttermann, R., Li, E. & Jaenisch, R. Toxicity of 5-aza-2'-deoxycytidine to mammalian cells is mediated primarily by covalent trapping of DNA methyltransferase rather than DNA demethylation. Proc Natl Acad Sci U S A. 1994;91:11797-801.
130. Ghoshal, K., Datta, J., Majumder, S., Bai, S., Kutay, H., Motiwala, T. & Jacob, S.T. 5-Aza-deoxycytidine induces selective degradation of DNA methyltransferase 1 by a proteasomal pathway that requires the KEN box, bromo-adjacent homology domain, and nuclear localization signal. Mol Cell Biol. 2005;25:4727-41.
131. Byrd, J.C., Stilgenbauer, S. & Flinn, I.W. Chronic lymphocytic leukemia. Hematology (Am Soc Hematol Educ Program). 2004:163-83.
132. 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;249:135-64.
133. Issa, J.P. & Byrd, J.C. Decitabine in chronic leukemias. Semin Hematol. 2005;42:S43-9. 130 134. Issa, J.P. CpG-island methylation in aging and cancer. Curr Top Microbiol Immunol. 2000;249:101-18.
135. Richardson, B. Impact of aging on DNA methylation. Ageing Res Rev. 2003;2:245-61.
136. Ahuja, N. & Issa, J.P. Aging, methylation and cancer. Histol Histopathol. 2000;15:835-42.
137. Freitas, M.A., Sklenar, A.R. & Parthun, M.R. Application of mass spectrometry to the identification and quantification of histone post-translational modifications. J Cell Biochem. 2004;92:691-700.
138. Cosgrove, M.S. & Wolberger, C. How does the histone code work? Biochem Cell Biol. 2005;83:468-76.
139. Fischle, W., Wang, Y. & Allis, C.D. Binary switches and modification cassettes in histone biology and beyond. Nature. 2003;425:475-9.
140. Fischle, W., Wang, Y. & Allis, C.D. Histone and chromatin cross-talk. Curr Opin Cell Biol. 2003;15:172-83.
141. Strahl, B.D. & Allis, C.D. The language of covalent histone modifications. Nature. 2000;403:41-5.
142. Issa, J.P., Ahuja, N., Toyota, M., Bronner, M.P. & Brentnall, T.A. Accelerated age-related CpG island methylation in ulcerative colitis. Cancer Res. 2001;61:3573-7.
143. Sidransky, D. Emerging molecular markers of cancer. Nat Rev Cancer. 2002;2:210-9.
144. Oakeley, E.J., Schmitt, F. & Jost, J.P. Quantification of 5-methylcytosine in DNA by the chloroacetaldehyde reaction. Biotechniques. 1999;27:744-6, 748-50, 752.
145. Gonzalgo, M.L. & Jones, P.A. Rapid quantitation of methylation differences at specific sites using methylation-sensitive single nucleotide primer extension (Ms- SNuPE). Nucleic Acids Res. 1997;25:2529-31.
146. Xiong, Z. & Laird, P.W. COBRA: a sensitive and quantitative DNA methylation assay. Nucleic Acids Res. 1997;25:2532-4.
147. Stanssens, P., Zabeau, M., Meersseman, G., Remes, G., Gansemans, Y., Storm, N., Hartmer, R., Honisch, C., Rodi, C.P., Bocker, S. & van den Boom, D. High-throughput MALDI-TOF discovery of genomic sequence polymorphisms. Genome Res. 2004;14:126-33.
148. Ehrich, M., Bocker, S. & van den Boom, D. Multiplexed discovery of sequence polymorphisms using base-specific cleavage and MALDI-TOF MS. Nucleic Acids Res. 2005;33:e38. 131 149. Eads, C.A., Danenberg, K.D., Kawakami, K., Saltz, L.B., Blake, C., Shibata, D., Danenberg, P.V. & Laird, P.W. MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res. 2000;28:E32.
150. Zeschnigk, M., Bohringer, S., Price, E.A., Onadim, Z., Masshofer, L. & Lohmann, D.R. A novel real-time PCR assay for quantitative analysis of methylated alleles (QAMA): analysis of the retinoblastoma locus. Nucleic Acids Res. 2004;32:e125.
151. Afonina, I., Zivarts, M., Kutyavin, I., Lukhtanov, E., Gamper, H. & Meyer, R.B. Efficient priming of PCR with short oligonucleotides conjugated to a minor groove binder. Nucleic Acids Res. 1997;25:2657-60.
152. Galm, O., Rountree, M.R., Bachman, K.E., Jair, K.W., Baylin, S.B. & Herman, J.G. Enzymatic regional methylation assay: a novel method to quantify regional CpG methylation density. Genome Res. 2002;12:153-7.
153. Cottrell, S.E., Distler, J., Goodman, N.S., Mooney, S.H., Kluth, A., Olek, A., Schwope, I., Tetzner, R., Ziebarth, H. & Berlin, K. A real-time PCR assay for DNA-methylation using methylation-specific blockers. Nucleic Acids Res. 2004;32:e10.
154. Jahr, S., Hentze, H., Englisch, S., Hardt, D., Fackelmayer, F.O., Hesch, R.D. & Knippers, R. DNA fragments in the blood plasma of cancer patients: quantitations and evidence for their origin from apoptotic and necrotic cells. Cancer Res. 2001;61:1659-65.
155. Sozzi, G., Conte, D., Mariani, L., Lo Vullo, S., Roz, L., Lombardo, C., Pierotti, M.A. & Tavecchio, L. Analysis of circulating tumor DNA in plasma at diagnosis and during follow-up of lung cancer patients. Cancer Res. 2001;61:4675-8.
156. Ronaghi, M., Karamohamed, S., Pettersson, B., Uhlen, M. & Nyren, P. Real-time DNA sequencing using detection of pyrophosphate release. Anal Biochem. 1996;242:84-9.
157. Ronaghi, M., Uhlen, M. & Nyren, P. A sequencing method based on real-time pyrophosphate. Science. 1998;281:363, 365.
158. Uhlmann, K., Brinckmann, A., Toliat, M.R., Ritter, H. & Nurnberg, P. Evaluation of a potential epigenetic biomarker by quantitative methyl-single nucleotide polymorphism analysis. Electrophoresis. 2002;23:4072-9.
159. Tost, J., Dunker, J. & Gut, I.G. Analysis and quantification of multiple methylation variable positions in CpG islands by Pyrosequencing. Biotechniques. 2003;35:152-6.
160. Dupont, J.M., Tost, J., Jammes, H. & Gut, I.G. De novo quantitative bisulfite sequencing using the pyrosequencing technology. Anal Biochem. 2004;333:119- 27.
132 161. Ronaghi, M. & Elahi, E. Pyrosequencing for microbial typing. J Chromatogr B Analyt Technol Biomed Life Sci. 2002;782:67-72.
162. Kuppuswamy, M.N., Hoffmann, J.W., Kasper, C.K., Spitzer, S.G., Groce, S.L. & Bajaj, S.P. Single nucleotide primer extension to detect genetic diseases: experimental application to hemophilia B (factor IX) and cystic fibrosis genes. Proc Natl Acad Sci U S A. 1991;88:1143-7.
163. Singer-Sam, J., LeBon, J.M., Dai, A. & Riggs, A.D. A sensitive, quantitative assay for measurement of allele-specific transcripts differing by a single nucleotide. PCR Methods Appl. 1992;1:160-3.
164. Szabo, P.E. & Mann, J.R. Allele-specific expression and total expression levels of imprinted genes during early mouse development: implications for imprinting mechanisms. Genes Dev. 1995;9:3097-108.
165. Greenwood, A.D. & Burke, D.T. Single nucleotide primer extension: quantitative range, variability, and multiplex analysis. Genome Res. 1996;6:336-48.
166. Thomassin, H., Kress, C. & Grange, T. MethylQuant: a sensitive method for quantifying methylation of specific cytosines within the genome. Nucleic Acids Res. 2004;32:e168.
167. Agrelo, R., Setien, F., Espada, J., Artiga, M.J., Rodriguez, M., Perez-Rosado, A., Sanchez-Aguilera, A., Fraga, M.F., Piris, M.A. & Esteller, M. Inactivation of the Lamin A/C Gene by CpG Island Promoter Hypermethylation in Hematologic Malignancies, and Its Association With Poor Survival in Nodal Diffuse Large B- Cell Lymphoma. J Clin Oncol. 2005.
168. Raval, A., Lucas, D.M., Matkovic, J.J., Bennett, K.L., Liyanarachchi, S., Young, D.C., Rassenti, L., Kipps, T.J., Grever, M.R., Byrd, J.C. & Plass, C. TWIST2 Demonstrates Differential Methylation in Immunoglobulin Variable Heavy Chain Mutated and Unmutated Chronic Lymphocytic Leukemia. J Clin Oncol. 2005.
169. Song, F., Smith, J.F., Kimura, M.T., Morrow, A.D., Matsuyama, T., Nagase, H. & Held, W.A. Association of tissue-specific differentially methylated regions (TDMs) with differential gene expression. Proc Natl Acad Sci U S A. 2005;102:3336-41.
170. Yu, L., Liu, C., Vandeusen, J., Becknell, B., Dai, Z., Wu, Y.Z., Raval, A., Liu, T.H., Ding, W., Mao, C., Liu, S., Smith, L.T., Lee, S., Rassenti, L., Marcucci, G., Byrd, J., Caligiuri, M.A. & Plass, C. Global assessment of promoter methylation in a mouse model of cancer identifies ID4 as a putative tumor-suppressor gene in human leukemia. Nat Genet. 2005;37:265-74.
171. Lewin, J., Schmitt, A.O., Adorjan, P., Hildmann, T. & Piepenbrock, C. Quantitative DNA methylation analysis based on four-dye trace data from direct sequencing of PCR amplificates. Bioinformatics. 2004;20:3005-12.
133 172. Rakyan, V.K., Hildmann, T., Novik, K.L., Lewin, J., Tost, J., Cox, A.V., Andrews, T.D., Howe, K.L., Otto, T., Olek, A., Fischer, J., Gut, I.G., Berlin, K. & Beck, S. DNA methylation profiling of the human major histocompatibility complex: a pilot study for the human epigenome project. PLoS Biol. 2004;2:e405.
173. Adorjan, P., Distler, J., Lipscher, E., Model, F., Muller, J., Pelet, C., Braun, A., Florl, A.R., Gutig, D., Grabs, G., Howe, A., Kursar, M., Lesche, R., Leu, E., Lewin, A., Maier, S., Muller, V., Otto, T., Scholz, C., Schulz, W.A., Seifert, H.H., Schwope, I., Ziebarth, H., Berlin, K., Piepenbrock, C. & Olek, A. Tumour class prediction and discovery by microarray-based DNA methylation analysis. Nucleic Acids Res. 2002;30:e21.
174. Gitan, R.S., Shi, H., Chen, C.M., Yan, P.S. & Huang, T.H. Methylation-specific oligonucleotide microarray: a new potential for high-throughput methylation analysis. Genome Res. 2002;12:158-64.
175. Shi, H., Maier, S., Nimmrich, I., Yan, P.S., Caldwell, C.W., Olek, A. & Huang, T.H. Oligonucleotide-based microarray for DNA methylation analysis: principles and applications. J Cell Biochem. 2003;88:138-43.
176. Leu, Y.W., Yan, P.S., Fan, M., Jin, V.X., Liu, J.C., Curran, E.M., Welshons, W.V., Wei, S.H., Davuluri, R.V., Plass, C., Nephew, K.P. & Huang, T.H. Loss of estrogen receptor signaling triggers epigenetic silencing of downstream targets in breast cancer. Cancer Res. 2004;64:8184-92.
177. Rein, T., DePamphilis, M.L. & Zorbas, H. Identifying 5-methylcytosine and related modifications in DNA genomes. Nucleic Acids Res. 1998;26:2255-64.
178. Kuo, K.C., McCune, R.A., Gehrke, C.W., Midgett, R. & Ehrlich, M. Quantitative reversed-phase high performance liquid chromatographic determination of major and modified deoxyribonucleosides in DNA. Nucleic Acids Res. 1980;8:4763-76.
179. Gomes, J.D. & Chang, C.J. Reverse-phase high-performance liquid chromatography of chemically modified DNA. Anal Biochem. 1983;129:387-91.
180. Morris, R.G. Improved liquid chromatographic fluorescence method for estimation of plasma sotalol concentrations. Ther Drug Monit. 1989;11:63-6.
181. Ramsahoye, B.H. Measurement of genome wide DNA methylation by reversed- phase high-performance liquid chromatography. Methods. 2002;27:156-61.
182. del Gaudio, R., Di Giaimo, R. & Geraci, G. Genome methylation of the marine annelid worm Chaetopterus variopedatus: methylation of a CpG in an expressed H1 histone gene. FEBS Lett. 1997;417:48-52.
183. Huang, T.H., Perry, M.R. & Laux, D.E. Methylation profiling of CpG islands in human breast cancer cells. Hum Mol Genet. 1999;8:459-70.
134 184. Yan, P.S., Chen, C.M., Shi, H., Rahmatpanah, F., Wei, S.H., Caldwell, C.W. & Huang, T.H. Dissecting complex epigenetic alterations in breast cancer using CpG island microarrays. Cancer Res. 2001;61:8375-80.
185. Lippman, Z., Gendrel, A.V., Colot, V. & Martienssen, R. Profiling DNA methylation patterns using genomic tiling microarrays. Nat Methods. 2005;2:219- 24.
186. Nouzova, M., Holtan, N., Oshiro, M.M., Isett, R.B., Munoz-Rodriguez, J.L., List, A.F., Narro, M.L., Miller, S.J., Merchant, N.C. & Futscher, B.W. Epigenomic changes during leukemia cell differentiation: analysis of histone acetylation and cytosine methylation using CpG island microarrays. J Pharmacol Exp Ther. 2004;311:968-81.
187. Kuromitsu, J., Kataoka, H., Yamashita, H., Muramatsu, M., Furuichi, Y., Sekine, T. & Hayashizaki, Y. Reproducible alterations of DNA methylation at a specific population of CpG islands during blast formation of peripheral blood lymphocytes. DNA Res. 1995;2:263-7.
188. Motiwala, T., Ghoshal, K., Das, A., Majumder, S., Weichenhan, D., Wu, Y.Z., Holman, K., James, S.J., Jacob, S.T. & Plass, C. Suppression of the protein tyrosine phosphatase receptor type O gene (PTPRO) by methylation in hepatocellular carcinomas. Oncogene. 2003;22:6319-31.
189. Smiraglia, D.J., Smith, L.T., Lang, J.C., Rush, L.J., Dai, Z., Schuller, D.E. & Plass, C. Differential targets of CpG island hypermethylation in primary and metastatic head and neck squamous cell carcinoma (HNSCC). J Med Genet. 2003;40:25-33.
190. Rush, L.J. & Plass, C. Restriction landmark genomic scanning for DNA methylation in cancer: past, present, and future applications. Anal Biochem. 2002;307:191-201.
191. Costello, J.F., Smiraglia, D.J. & Plass, C. Restriction landmark genome scanning. Methods. 2002;27:144-9.
192. Rush, L.J., Dai, Z., Smiraglia, D.J., Gao, X., Wright, F.A., Fruhwald, M., Costello, J.F., Held, W.A., Yu, L., Krahe, R., Kolitz, J.E., Bloomfield, C.D., Caligiuri, M.A. & Plass, C. Novel methylation targets in de novo acute myeloid leukemia with prevalence of chromosome 11 loci. Blood. 2001;97:3226-33.
193. Kremenskoy, M., Kremenska, Y., Ohgane, J., Hattori, N., Tanaka, S., Hashizume, K. & Shiota, K. Genome-wide analysis of DNA methylation status of CpG islands in embryoid bodies, teratomas, and fetuses. Biochem Biophys Res Commun. 2003;311:884-90.
194. Zardo, G., Tiirikainen, M.I., Hong, C., Misra, A., Feuerstein, B.G., Volik, S., Collins, C.C., Lamborn, K.R., Bollen, A., Pinkel, D., Albertson, D.G. & Costello, J.F. Integrated genomic and epigenomic analyses pinpoint biallelic gene inactivation in tumors. Nat Genet. 2002;32:453-8. 135 195. Nagai, H., Kim, Y.S., Yasuda, T., Ohmachi, Y., Yokouchi, H., Monden, M., Emi, M., Konishi, N., Nogami, M., Okumura, K. & Matsubara, K. A novel sperm- specific hypomethylation sequence is a demethylation hotspot in human hepatocellular carcinomas. Gene. 1999;237:15-20.
196. Konishi, N., Tao, M., Nakamura, M., Kitahaori, Y., Hiasa, Y. & Nagai, H. Genomic alterations in human prostate carcinoma cell lines by two-dimensional gel analysis. Cell Mol Biol (Noisy-le-grand). 1996;42:1129-35.
197. Lindsay, S. & Bird, A.P. Use of restriction enzymes to detect potential gene sequences in mammalian DNA. Nature. 1987;327:336-8.
198. Ishkanian, A.S., Malloff, C.A., Watson, S.K., DeLeeuw, R.J., Chi, B., Coe, B.P., Snijders, A., Albertson, D.G., Pinkel, D., Marra, M.A., Ling, V., MacAulay, C. & Lam, W.L. A tiling resolution DNA microarray with complete coverage of the human genome. Nat Genet. 2004;36:299-303.
199. Ching, T.T., Maunakea, A.K., Jun, P., Hong, C., Zardo, G., Pinkel, D., Albertson, D.G., Fridlyand, J., Mao, J.H., Shchors, K., Weiss, W.A. & Costello, J.F. Epigenome analyses using BAC microarrays identify evolutionary conservation of tissue-specific methylation of SHANK3. Nat Genet. 2005.
200. Weber, M., Davies, J.J., Wittig, D., Oakeley, E.J., Haase, M., Lam, W.L. & Schubeler, D. Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat Genet. 2005;37:853-62.
201. Misawa, A., Inoue, J., Sugino, Y., Hosoi, H., Sugimoto, T., Hosoda, F., Ohki, M., Imoto, I. & Inazawa, J. Methylation-associated silencing of the nuclear receptor 1I2 gene in advanced-type neuroblastomas, identified by bacterial artificial chromosome array-based methylated CpG island amplification. Cancer Res. 2005;65:10233-42.
202. Jabbari, K. & Bernardi, G. Cytosine methylation and CpG, TpG (CpA) and TpA frequencies. Gene. 2004;333:143-9.
203. Szyf, M., Pakneshan, P. & Rabbani, S.A. DNA demethylation and cancer: therapeutic implications. Cancer Lett. 2004;211:133-43.
204. Bastian, P.J., Palapattu, G.S., Lin, X., Yegnasubramanian, S., Mangold, L.A., Trock, B., Eisenberger, M.A., Partin, A.W. & Nelson, W.G. Preoperative serum DNA GSTP1 CpG island hypermethylation and the risk of early prostate-specific antigen recurrence following radical prostatectomy. Clin Cancer Res. 2005;11:4037-43.
205. Friedrich, M.G., Chandrasoma, S., Siegmund, K.D., Weisenberger, D.J., Cheng, J.C., Toma, M.I., Huland, H., Jones, P.A. & Liang, G. Prognostic relevance of methylation markers in patients with non-muscle invasive bladder carcinoma. Eur J Cancer. 2005.
136 206. Levenson, V.V. DNA methylation biomarkers of cancer: moving toward clinical application. Pharmacogenomics. 2004;5:699-707.
207. Brock, M.V., Gou, M., Akiyama, Y., Muller, A., Wu, T.T., Montgomery, E., Deasel, M., Germonpre, P., Rubinson, L., Heitmiller, R.F., Yang, S.C., Forastiere, A.A., Baylin, S.B. & Herman, J.G. Prognostic importance of promoter hypermethylation of multiple genes in esophageal adenocarcinoma. Clin Cancer Res. 2003;9:2912-9.
208. Usadel, H., Brabender, J., Danenberg, K.D., Jeronimo, C., Harden, S., Engles, J., Danenberg, P.V., Yang, S. & Sidransky, D. Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Res. 2002;62:371-5.
209. Maruyama, R., Toyooka, S., Toyooka, K.O., Harada, K., Virmani, A.K., Zochbauer-Muller, S., Farinas, A.J., Vakar-Lopez, F., Minna, J.D., Sagalowsky, A., Czerniak, B. & Gazdar, A.F. Aberrant promoter methylation profile of bladder cancer and its relationship to clinicopathological features. Cancer Res. 2001;61:8659-63.
210. Maruyama, R., Toyooka, S., Toyooka, K.O., Virmani, A.K., Zochbauer-Muller, S., Farinas, A.J., Minna, J.D., McConnell, J., Frenkel, E.P. & Gazdar, A.F. Aberrant promoter methylation profile of prostate cancers and its relationship to clinicopathological features. Clin Cancer Res. 2002;8:514-9.
211. Crocitto, L.E., Korns, D., Kretzner, L., Shevchuk, T., Blair, S.L., Wilson, T.G., Ramin, S.A., Kawachi, M.H. & Smith, S.S. Prostate cancer molecular markers GSTP1 and hTERT in expressed prostatic secretions as predictors of biopsy results. Urology. 2004;64:821-5.
212. Lu, C.Y., Tso, D.J., Yang, T., Jong, Y.J. & Wei, Y.H. Detection of DNA mutations associated with mitochondrial diseases by Agilent 2100 bioanalyzer. Clin Chim Acta. 2002;318:97-105.
213. Panaro, N.J., Yuen, P.K., Sakazume, T., Fortina, P., Kricka, L.J. & Wilding, P. Evaluation of DNA fragment sizing and quantification by the agilent 2100 bioanalyzer. Clin Chem. 2000;46:1851-3.
214. Fuller, R.A., Clark, J., Kretzner, L., Korns, D., Blair, S.L., Crocitto, L.E. & Smith, S.S. Use of microfluidics chips for the detection of human telomerase RNA. Anal Biochem. 2003;313:331-4.
215. Funes-Huacca, M., Regitano, L.C., Mueller, O. & Carrilho, E. Semiquantitative determination of Alicyclobacillus acidoterrestris in orange juice by reverse- transcriptase polymerase chain reaction and capillary electrophoresis--laser induced fluorescence using microchip technology. Electrophoresis. 2004;25:3860-4.
137 216. Auer, H., Lyianarachchi, S., Newsom, D., Klisovic, M.I., Marcucci, G. & Kornacker, K. Chipping away at the chip bias: RNA degradation in microarray analysis. Nat Genet. 2003;35:292-3.
217. Boyd, V.L. & Zon, G. Bisulfite conversion of genomic DNA for methylation analysis: protocol simplification with higher recovery applicable to limited samples and increased throughput. Anal Biochem. 2004;326:278-80.
218. Toyooka, K.O., Toyooka, S., Maitra, A., Feng, Q., Kiviat, N.C., Smith, A., Minna, J.D., Ashfaq, R. & Gazdar, A.F. Establishment and validation of real-time polymerase chain reaction method for CDH1 promoter methylation. Am J Pathol. 2002;161:629-34.
219. Parkin, D.M., Bray, F., Ferlay, J. & Pisani, P. Estimating the world cancer burden: Globocan 2000. Int J Cancer. 2001;94:153-6.
220. Jemal, A., Clegg, L.X., Ward, E., Ries, L.A., Wu, X., Jamison, P.M., Wingo, P.A., Howe, H.L., Anderson, R.N. & Edwards, B.K. Annual report to the nation on the status of cancer, 1975-2001, with a special feature regarding survival. Cancer. 2004;101:3-27.
221. Greene, F.L. & Sobin, L.H. The TNM system: our language for cancer care. J Surg Oncol. 2002;80:119-20.
222. Vielh, P., Spano, J.P., Grenier, J., Le Chevalier, T. & Soria, J.C. Molecular prognostic factors in resectable non-small cell lung cancer. Crit Rev Oncol Hematol. 2005;53:193-7.
223. Potti, A., Mukherjee, S., Petersen, R., Dressman, H.K., Bild, A., Koontz, J., Kratzke, R., Watson, M.A., Kelley, M., Ginsburg, G.S., West, M., Harpole, D.H., Jr. & Nevins, J.R. A genomic strategy to refine prognosis in early-stage non- small-cell lung cancer. N Engl J Med. 2006;355:570-80.
224. Esteller, M., Corn, P.G., Baylin, S.B. & Herman, J.G. A gene hypermethylation profile of human cancer. Cancer Res. 2001;61:3225-9.
225. Brena, R.M., Huang, T.H. & Plass, C. Quantitative assessment of DNA methylation: potential applications for disease diagnosis, classification, and prognosis in clinical settings. J Mol Med. 2006:1-13.
226. Dai, Z., Weichenhan, D., Wu, Y.Z., Hall, J.L., Rush, L.J., Smith, L.T., Raval, A., Yu, L., Kroll, D., Muehlisch, J., Fruhwald, M.C., de Jong, P., Catanese, J., Davuluri, R.V., Smiraglia, D.J. & Plass, C. An AscI boundary library for the studies of genetic and epigenetic alterations in CpG islands. Genome Res. 2002;12:1591-8.
227. Tada, Y., Brena, R.M., Hackanson, B., Morrison, C., Otterson, G.A. & Plass, C. Epigenetic modulation of tumor suppressor CCAAT/enhancer binding protein alpha activity in lung cancer. J Natl Cancer Inst. 2006;98:396-406.
138 228. Brena, R.M., Auer, H., Kornacker, K., Hackanson, B., Raval, A., Byrd, J.C. & Plass, C. Accurate quantification of DNA methylation using combined bisulfite restriction analysis coupled with the Agilent 2100 Bioanalyzer platform. Nucleic Acids Res. 2006;34:e17.
229. Eads, C.A. & Laird, P.W. Combined bisulfite restriction analysis (COBRA). Methods Mol Biol. 2002;200:71-85.
230. Brena, R.M., Auer, H., Kornacker, K. & Plass, C. Quantification of DNA methylation in electrofluidics chips (Bio-COBRA). Nat Protocols. 2006;1:52-58.
231. Abdel-Rahman, M.H., Yang, Y., Zhou, X.P., Craig, E.L., Davidorf, F.H. & Eng, C. High frequency of submicroscopic hemizygous deletion is a major mechanism of loss of expression of PTEN in uveal melanoma. J Clin Oncol. 2006;24:288-95.
232. Goethals, L., Perneel, C., Debucquoy, A., De Schutter, H., Borghys, D., Ectors, N., Geboes, K., McBride, W.H. & Haustermans, K.M. A new approach to the validation of tissue microarrays. J Pathol. 2006;208:607-14.
233. Allred, D.C., Harvey, J.M., Berardo, M. & Clark, G.M. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol. 1998;11:155-68.
234. Poola, I., DeWitty, R.L., Marshalleck, J.J., Bhatnagar, R., Abraham, J. & Leffall, L.D. Identification of MMP-1 as a putative breast cancer predictive marker by global gene expression analysis. Nat Med. 2005;11:481-3.
235. Kaufman, L. & Rousseeuw, P. Finding Groups in Data - An Introduction to Cluster Analysis, (John Wiley & Sons, New York, 1990).
236. Breuer, R.H., Postmus, P.E. & Smit, E.F. Molecular pathology of non-small-cell lung cancer. Respiration. 2005;72:313-30.
237. Virmani, A.K., Fong, K.M., Kodagoda, D., McIntire, D., Hung, J., Tonk, V., Minna, J.D. & Gazdar, A.F. Allelotyping demonstrates common and distinct patterns of chromosomal loss in human lung cancer types. Genes Chromosomes Cancer. 1998;21:308-19.
238. Aghmesheh, M., Suo, Z., Friedlander, M., Nesland, J.M., Kaern, J., Stewart, M., Kconfab, Dorum, A., Tucker, K.M. & Buckley, M.F. Chromosome 2q24.2 is lost in sporadic but not in BRCA1-associated ovarian carcinomas. Pathology. 2006;38:145-51.
239. Yuan, E., Li, C.M., Yamashiro, D.J., Kandel, J., Thaker, H., Murty, V.V. & Tycko, B. Genomic profiling maps loss of heterozygosity and defines the timing and stage dependence of epigenetic and genetic events in Wilms' tumors. Mol Cancer Res. 2005;3:493-502.
139 240. Girard, L., Zochbauer-Muller, S., Virmani, A.K., Gazdar, A.F. & Minna, J.D. Genome-wide allelotyping of lung cancer identifies new regions of allelic loss, differences between small cell lung cancer and non-small cell lung cancer, and loci clustering. Cancer Res. 2000;60:4894-906.
241. Sato, S., Nakamura, Y. & Tsuchiya, E. Difference of allelotype between squamous cell carcinoma and adenocarcinoma of the lung. Cancer Res. 1994;54:5652-5.
242. Dumur, C.I., Dechsukhum, C., Ware, J.L., Cofield, S.S., Best, A.M., Wilkinson, D.S., Garrett, C.T. & Ferreira-Gonzalez, A. Genome-wide detection of LOH in prostate cancer using human SNP microarray technology. Genomics. 2003;81:260-9.
243. Nowak, N.J., Gaile, D., Conroy, J.M., McQuaid, D., Cowell, J., Carter, R., Goggins, M.G., Hruban, R.H. & Maitra, A. Genome-wide aberrations in pancreatic adenocarcinoma. Cancer Genet Cytogenet. 2005;161:36-50.
244. Kee, H.J., Shin, J.H., Chang, J., Chung, K.Y., Shin, D.H., Kim, Y.S., Kim, S.K. & Kim, S.K. Identification of tumor suppressor loci on the long arm of chromosome 15 in primary small cell lung cancer. Yonsei Med J. 2003;44:65-74.
245. Fukumoto, M. & Nakayama, K. Ovarian epithelial tumors of low malignant potential: are they precursors of ovarian carcinoma? Pathol Int. 2006;56:233-9.
246. Lee, E.B., Park, T.I., Park, S.H. & Park, J.Y. Loss of heterozygosity on the long arm of chromosome 21 in non-small cell lung cancer. Ann Thorac Surg. 2003;75:1597-600.
247. Zhou, Q., Wang, S. & Anderson, D.J. Identification of a novel family of oligodendrocyte lineage-specific basic helix-loop-helix transcription factors. Neuron. 2000;25:331-43.
248. Winton, T., Livingston, R., Johnson, D., Rigas, J., Johnston, M., Butts, C., Cormier, Y., Goss, G., Inculet, R., Vallieres, E., Fry, W., Bethune, D., Ayoub, J., Ding, K., Seymour, L., Graham, B., Tsao, M.S., Gandara, D., Kesler, K., Demmy, T. & Shepherd, F. Vinorelbine plus cisplatin vs. observation in resected non- small-cell lung cancer. N Engl J Med. 2005;352:2589-97.
249. Arriagada, R., Bergman, B., Dunant, A., Le Chevalier, T., Pignon, J.P. & Vansteenkiste, J. Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med. 2004;350:351-60.
250. Ligon, K.L., Fancy, S.P., Franklin, R.J. & Rowitch, D.H. Olig gene function in CNS development and disease. Glia. 2006;54:1-10.
251. Arnett, H.A., Fancy, S.P., Alberta, J.A., Zhao, C., Plant, S.R., Kaing, S., Raine, C.S., Rowitch, D.H., Franklin, R.J. & Stiles, C.D. bHLH transcription factor Olig1 is required to repair demyelinated lesions in the CNS. Science. 2004;306:2111-5.
140 252. Watkins, D.N., Berman, D.M., Burkholder, S.G., Wang, B., Beachy, P.A. & Baylin, S.B. Hedgehog signalling within airway epithelial progenitors and in small- cell lung cancer. Nature. 2003;422:313-7.
253. Alberta, J.A., Park, S.K., Mora, J., Yuk, D., Pawlitzky, I., Iannarelli, P., Vartanian, T., Stiles, C.D. & Rowitch, D.H. Sonic hedgehog is required during an early phase of oligodendrocyte development in mammalian brain. Mol Cell Neurosci. 2001;18:434-41.
254. Xin, M., Yue, T., Ma, Z., Wu, F.F., Gow, A. & Lu, Q.R. Myelinogenesis and axonal recognition by oligodendrocytes in brain are uncoupled in Olig1-null mice. J Neurosci. 2005;25:1354-65.
255. Sow, A., Lamant, M., Bonny, J.M., Larvaron, P., Piaud, O., Lecureuil, C., Fontaine, I., Saleh, M.C., Garcia Otin, A.L., Renou, J.P., Baron, B., Zakin, M. & Guillou, F. Oligodendrocyte differentiation is increased in transferrin transgenic mice. J Neurosci Res. 2006;83:403-14.
256. Shmueli, O., Horn-Saban, S., Chalifa-Caspi, V., Shmoish, M., Ophir, R., Benjamin-Rodrig, H., Safran, M., Domany, E. & Lancet, D. GeneNote: whole genome expression profiles in normal human tissues. C R Biol. 2003;326:1067- 72.
257. Yamagata, N., Shyr, Y., Yanagisawa, K., Edgerton, M., Dang, T.P., Gonzalez, A., Nadaf, S., Larsen, P., Roberts, J.R., Nesbitt, J.C., Jensen, R., Levy, S., Moore, J.H., Minna, J.D. & Carbone, D.P. A training-testing approach to the molecular classification of resected non-small cell lung cancer. Clin Cancer Res. 2003;9:4695-704.
258. Beer, D.G., Kardia, S.L., Huang, C.C., Giordano, T.J., Levin, A.M., Misek, D.E., Lin, L., Chen, G., Gharib, T.G., Thomas, D.G., Lizyness, M.L., Kuick, R., Hayasaka, S., Taylor, J.M., Iannettoni, M.D., Orringer, M.B. & Hanash, S. Gene- expression profiles predict survival of patients with lung adenocarcinoma. Nat Med. 2002;8:816-24.
259. Bhattacharjee, A., Richards, W.G., Staunton, J., Li, C., Monti, S., Vasa, P., Ladd, C., Beheshti, J., Bueno, R., Gillette, M., Loda, M., Weber, G., Mark, E.J., Lander, E.S., Wong, W., Johnson, B.E., Golub, T.R., Sugarbaker, D.J. & Meyerson, M. Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses. Proc Natl Acad Sci U S A. 2001;98:13790- 5.
260. Lee, J.S., Heo, J., Libbrecht, L., Chu, I.S., Kaposi-Novak, P., Calvisi, D.F., Mikaelyan, A., Roberts, L.R., Demetris, A.J., Sun, Z., Nevens, F., Roskams, T. & Thorgeirsson, S.S. A novel prognostic subtype of human hepatocellular carcinoma derived from hepatic progenitor cells. Nat Med. 2006;12:410-6.
141 261. Massion, P.P., Kuo, W.L., Stokoe, D., Olshen, A.B., Treseler, P.A., Chin, K., Chen, C., Polikoff, D., Jain, A.N., Pinkel, D., Albertson, D.G., Jablons, D.M. & Gray, J.W. Genomic copy number analysis of non-small cell lung cancer using array comparative genomic hybridization: implications of the phosphatidylinositol 3-kinase pathway. Cancer Res. 2002;62:3636-40.
262. Frigola, J., Song, J., Stirzaker, C., Hinshelwood, R.A., Peinado, M.A. & Clark, S.J. Epigenetic remodeling in colorectal cancer results in coordinate gene suppression across an entire chromosome band. Nat Genet. 2006;38:540-9.
263. Kazazian, H.H., Jr. & Goodier, J.L. LINE drive. retrotransposition and genome instability. Cell. 2002;110:277-80.
264. Nephew, K.P. & Huang, T.H. Epigenetic gene silencing in cancer initiation and progression. Cancer Lett. 2003;190:125-33.
265. Brena, R., Morrison, C., Liyanarachchi, S., Jarjoura, D., Davuluri, R., Otterson, G., Reisman, D., Glaros, S., Rush, L. & Plass, C. Global DNA Methylation Profiling of Non-small Cell Lung Cancer Identifies a Differentially Methylated Gene, OLIG1, as a Novel Prognostic Factor. PLoS Medicine, in press. 2007.
266. Frommer, M., McDonald, L.E., Millar, D.S., Collis, C.M., Watt, F., Grigg, G.W., Molloy, P.L. & Paul, C.L. A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proc Natl Acad Sci U S A. 1992;89:1827-31.
267. Harris, M.A., Clark, J., Ireland, A., Lomax, J., Ashburner, M., Foulger, R., Eilbeck, K., Lewis, S., Marshall, B., Mungall, C., Richter, J., Rubin, G.M., Blake, J.A., Bult, C., Dolan, M., Drabkin, H., Eppig, J.T., Hill, D.P., Ni, L., Ringwald, M., Balakrishnan, R., Cherry, J.M., Christie, K.R., Costanzo, M.C., Dwight, S.S., Engel, S., Fisk, D.G., Hirschman, J.E., Hong, E.L., Nash, R.S., Sethuraman, A., Theesfeld, C.L., Botstein, D., Dolinski, K., Feierbach, B., Berardini, T., Mundodi, S., Rhee, S.Y., Apweiler, R., Barrell, D., Camon, E., Dimmer, E., Lee, V., Chisholm, R., Gaudet, P., Kibbe, W., Kishore, R., Schwarz, E.M., Sternberg, P., Gwinn, M., Hannick, L., Wortman, J., Berriman, M., Wood, V., de la Cruz, N., Tonellato, P., Jaiswal, P., Seigfried, T. & White, R. The Gene Ontology (GO) database and informatics resource. Nucleic Acids Res. 2004;32:D258-61.
268. Straub, T. & Becker, P.B. Dosage compensation: the beginning and end of generalization. Nat Rev Genet. 2007;8:47-57.
269. Slack, F.J. & Weidhaas, J.B. MicroRNAs as a potential magic bullet in cancer. Future Oncol. 2006;2:73-82.
270. Calin, G.A. & Croce, C.M. MicroRNA signatures in human cancers. Nat Rev Cancer. 2006;6:857-66.
271. Feltus, F.A., Lee, E.K., Costello, J.F., Plass, C. & Vertino, P.M. DNA motifs associated with aberrant CpG island methylation. Genomics. 2006;87:572-9.
142 272. Feltus, F.A., Lee, E.K., Costello, J.F., Plass, C. & Vertino, P.M. Predicting aberrant CpG island methylation. Proc Natl Acad Sci U S A. 2003;100:12253-8.
273. Bock, C., Paulsen, M., Tierling, S., Mikeska, T., Lengauer, T. & Walter, J. CpG island methylation in human lymphocytes is highly correlated with DNA sequence, repeats, and predicted DNA structure. PLoS Genet. 2006;2:e26.
274. Das, R., Dimitrova, N., Xuan, Z., Rollins, R.A., Haghighi, F., Edwards, J.R., Ju, J., Bestor, T.H. & Zhang, M.Q. Computational prediction of methylation status in human genomic sequences. Proc Natl Acad Sci U S A. 2006;103:10713-6.
275. Calvo, R., West, J., Franklin, W., Erickson, P., Bemis, L., Li, E., Helfrich, B., Bunn, P., Roche, J., Brambilla, E., Rosell, R., Gemmill, R.M. & Drabkin, H.A. Altered HOX and WNT7A expression in human lung cancer. Proc Natl Acad Sci U S A. 2000;97:12776-81.
276. Choi, Y.W., Choi, J.S., Zheng, L.T., Lim, Y.J., Yoon, H.K., Kim, Y.H., Wang, Y.P. & Lim, Y. Comparative genomic hybridization array analysis and real time PCR reveals genomic alterations in squamous cell carcinomas of the lung. Lung Cancer. 2007;55:43-51.
277. Choi, J.S., Zheng, L.T., Ha, E., Lim, Y.J., Kim, Y.H., Wang, Y.P. & Lim, Y. Comparative genomic hybridization array analysis and real-time PCR reveals genomic copy number alteration for lung adenocarcinomas. Lung. 2006;184:355- 62.
278. Lechner, J.F., Wang, Y., Siddiq, F., Fugaro, J.M., Wali, A., Lonardo, F., Willey, J.C., Harris, C.C. & Pass, H.I. Human lung cancer cells and tissues partially recapitulate the homeobox gene expression profile of embryonic lung. Lung Cancer. 2002;37:41-7.
279. Maeda, Y., Dave, V. & Whitsett, J.A. Transcriptional control of lung morphogenesis. Physiol Rev. 2007;87:219-44.
280. Biben, C., Wang, C.C. & Harvey, R.P. NK-2 class homeobox genes and pharyngeal/oral patterning: Nkx2-3 is required for salivary gland and tooth morphogenesis. Int J Dev Biol. 2002;46:415-22.
281. Cammenga, J., Niebuhr, B., Horn, S., Bergholz, U., Putz, G., Buchholz, F., Lohler, J. & Stocking, C. RUNX1 DNA-binding mutants, associated with minimally differentiated acute myelogenous leukemia, disrupt myeloid differentiation. Cancer Res. 2007;67:537-45.
282. Shinozaki, K., Yoshida, M., Nakamura, M., Aizawa, S. & Suda, Y. Emx1 and Emx2 cooperate in initial phase of archipallium development. Mech Dev. 2004;121:475-89.
283. Ding, Y.Q., Yin, J., Kania, A., Zhao, Z.Q., Johnson, R.L. & Chen, Z.F. Lmx1b controls the differentiation and migration of the superficial dorsal horn neurons of the spinal cord. Development. 2004;131:3693-703. 143 284. Hammond, K.L. & Whitfield, T.T. The developing lamprey ear closely resembles the zebrafish otic vesicle: otx1 expression can account for all major patterning differences. Development. 2006;133:1347-57.
285. Ishikawa, N., Daigo, Y., Takano, A., Taniwaki, M., Kato, T., Tanaka, S., Yasui, W., Takeshima, Y., Inai, K., Nishimura, H., Tsuchiya, E., Kohno, N. & Nakamura, Y. Characterization of SEZ6L2 cell-surface protein as a novel prognostic marker for lung cancer. Cancer Sci. 2006;97:737-45.
144