PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2105 p:/ClRC/DateDue.indd-p.1

PROGRESSIVE, NON-RANDOM ALTERED PATTERNS OF METHYLATION
IN GENE-SPECIFIC AND GC-RICH REGIONS OF DNA UNDERLIE
TUMORIGENESIS
By

Ammie Norene Bachman

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Pharmacology and Toxicology

2006

ABSTRACT

PROGRESSIVE, NON-RANDOM ALTERED PATTERNS OF METHYLATION
IN GENE-SPECIFIC AND GC-RICH REGIONS OF DNA UNDERLIE
TUMORIGENESIS
By

Ammie Norene Bachman

Epigenetics is broadly defined as processes that establish heritable states of gene
expression without altering the DNA sequence. DNA methylation, (i.e. 5-methylcytosine
content of DNA), is a well characterized epigenetic mark. Altered patterns of DNA
methylation can lead to the aberrant expression of genes. My central hypothesis states
that the ability to maintain patterns of methylation is inversely related to susceptibility to
tumorigenesis. This hypothesis is tested in the context of the central paradigm for
explaining events leading to tumorigenesis. The model describes a multi-stage and multi-
step (i.e. multi-mechanism) model for the development of precancerous lesions and their
evolution into frank carcinomas. The stages defined by this model are initiation,
promotion and progression. The promotion stage of tumorigenesis involves the step-wise
accumulation of heritable changes which are critical for the selection and clonal
expansion of initiated cells. Therefore changes in methylation which accumulate during
the promotion stage are not a result of the neoplastic state, but key contributors to the
process. With the development of a novel method for measuring changes in methylation
in GC-rich regions of the genome with high reproducibility, I have characterized patterns

of DNA methylation during the promotion stage of carcinogenesis in three separate

model systems. Specifically I have demonstrated highly similar changes in methylation,
predominantly hypomethylation, with three different promoting compounds. In addition,
I observed hypomethylation of the promoter region of the Ha-ras oncogene with
simultaneous stability of patterns of methylation in the promoter region of LINE-1
elements which are retrotransposable elements that comprise ~ 30% of the mouse
genome. These findings exemplify the selective nature of promoter-induced (i.e.
phenobarbital) disruption and indicate that changes in methylation are not entirely
random. Detection of hypomethylation, occurring simultaneously with hypermethylation,
in GC-rich regions is demonstrated as a principal contributor to the development and
progression of tumors. Consistent with the working model of carcinogenesis, regions of
altered methylation are seen to persist from early time points to later precancerous and
cancerous time points. This lends strong experimental support for the accumulation of
critical changes in DNA methylation, both increases and decreases, during tumor
promotion. The extent and frequency (i.e. more changes accumulating in a shorter period
of time increases susceptibility) with which changes in methylation accrued seems to
strongly relate to the relative susceptibilities of B6C3Fl and C57BL/6 mice to liver
tumorigenesis, a key point supporting the overall hypothesis. Further evidence to support
the contribution of progressive changes in methylation to the neoplastic state is presented
using the SENCAR 2-stage mouse skin initiation/promotion model. Reversibility of
altered methylation, a hallmark of promotion, is also observed and discussed. Taken
together, my findings from three distinct model systems were complementary and
consistent in supporting the notion that progressive, non-random changes (i.e. instability

of the epigenome) in methylation underlie tumorigenesis.

This dissertation is dedicated to my husband Todd M. Bachman

iv

ACKNOWLEDGEMENTS

My educational career has been filled with countless encouraging words and
numerous intellectually challenging moments. As I reflect upon the people who have
brought me to this point, I must acknowledge with high regard the guidance and
assistance of Dr. Jay I. Goodman. My development as a professional scientist has been
accelerated and my critical thinking skills will be forever grateful. The support and
advice of my guidance committee members, Dr. James Trosko, Dr. Norbert Kaminski,
and Dr. John LaPres is greatly appreciated. In addition, the collaborative efforts of Dr.
Lisa Kamendulis from the Indiana University School of Medicine and Drs. Geoff Curtin
and Dave Doolittle from RJ. Reynolds Tobacco Company are recognized and respected.
Credit is due to my longtime undergrad lab mate, Ling Zhong, who was an integral factor
to the success of my research. Others who have also contributed include Miriam
Kleinman and Emily Eckenrode. A very special thank you is due to Jennifer Phillips who
has, within only a short period of time, made me realize the value of sharing a common
goal in a research setting and doing it with laughter. From my undergraduate alma mater
Bloomsburg University of Pennsylvania, I must thank Drs. Judith Kipe-Nolt and George
Davis for inspiring me to take on graduate school and explore my love of science.

Most importantly, I must thank with all my heart, my parents Jane and Richard
Carnell who have supported me in every way imaginable from 500 miles away. They
will always be a driving force in my desire to achieve. I must also thank my brother who
has always been a source of motivation, competition and brain power when I needed it
the most. Finally, my husband deserves all the credit in the world for truly living and

understanding what this all means to me.

TABLE OF CONTENTS

PAGE(S)

LIST OF TABLES .............................................................................. viii-ix
LIST OF FIGURES ............................................................................. x-xiii
INTRODUCTION ................................................................................ 1-48
Cancer ....................................................................................... 1-2
Multistage Carcinogenesis and Epigenetics ........................................... 2-7

The Origin of the Cancer Stem Cell .................................................... 7-8
Regulation of Methylation Patterns: DNA methyltransferases .................... 9-13
5-Methyl Cytosine Distribution and Abundance ................................... 13-15

DNA Methylation and Nutrition ...................................................... 15-19

Gene Specific Patterns of Methylation and Regulation of Gene Transcription. . ..19
Silencing of Tumor Suppressor Genes ..................................... 19-21

Activation of Oncogenes ..................................................... 22-24

Activation of Transposable Elements ....................................... 24-28

Altered Methylation as a Precursor to Disease ..................................... 28-31

DNA Methylation and Promotion of Tumorigenesis: Rodent Models .............. 31

Phenobarbital and Liver Tumorigenesis .................................... 32-36

SENCAR mouse skin initiation-promotion model ........................ 36-38

A novel approach to measuring global changes in DNA methylation .......... 3 8-43
Hypothesis and Objectives ............................................................ 44-46
Specific Aims ........................................................................... 47-48
References for the Introduction and Discussion ................................ 231-241

CHAPTER 1 — Diethanolamine and Phenobarbital Produce an Altered Pattern of
Methylation in GC-Rich Regions of DNA in B6C3F 1 Mouse Hepatocytes Similar to that

Resulting from Choline Deficiency ........................................................... 49-94
Abstract ...................................................................................... 50
Introduction ............................................................................. 5 1-53
Materials and Methods ................................................................ 54-57
Results ................................................................................... 58-72
Discussion ............................................................................... 73-78
Acknowledgement ......................................................................... 78
References .............................................................................. 79-82
APPENDIX I ........................................................................... 83-94

CHAPTER 2 — Phenobarbital Induces Progressive Patterns of GC-Rich and Gene-

Specific Altered Methylation in the Liver of Tumor-Prone B6C3F 1 Mice ............ 95-158
Abstract .................................................................................. 96-97
Introduction ........................................................................... 98-100
Materials and Methods ............................................................. 101-111
Results ................................................................................ 112-137
Discussion ............................................................................ 13 8- 145

vi

References ........................................................................... 146-149

APPENDIX II ........................................................................ 150-158
Materials and Methods .................................................... 151-153
Results ....................................................................... 153-156
Discussion ................................................................... 157-158
CHAPTER 3 — Altered Methylation in Gene-specific and GC-rich regions of DNA is
Progressive and Non-random During Promotion of Skin Tumorigenesis ............ 159-210
Abstract .................................................................................... 160
Introduction .......................................................................... 161-164
Materials and Methods ............................................................. 164-167
Results ................................................................................ 168-185
Discussion ............................................................................ 186-191
Acknowledgement ........................................................................ 1 91
References ............................................................................ 1 92-1 94
APPENDIX III ...................................................................... 195-210
Materials and Methods .................................................... 196-199
Results ....................................................................... 200-207
Discussion .................................................................. 208-209
SUMMARY .................................................................................. 211-213
Specific Aims Addressed .......................................................... 214-219
Support for Hypothesis ............................................................ 220-221
DISCUSSION ................................................................................ 222-228

Significance of Research
1. AP-PCR/CE has proven to be a consistent and reproducible method for
detecting changes in methylation in GC-rich regions regardless of organ
type or model system ....................................................................... 222-223
2. Treatment induced changes in methylation are not purely random, but
show defined and reproducible patterns of disruption that accrue with
time ......................................................................... 223-224
3. Hypomethylation, occurring simultaneously with hypermethylation, is
a predominant and highly significant contn'butor to the development and
progression of tumors .................................................... 225-226
4. As outlined by the classic multi-stage model of carcinogenesis, the
progressive accumulation of critical changes in DNA methylation, both
increases and decreases, appears to be a universal feature of tumor
promotion .................................................................. 226-228

Conclusions ......................................................................... 228-230

LIST OF REFERENCES USED IN INTRODUCTION, SUMMARY, AND
DISCUSSION... ............................................................................................... 231-241

vii

LIST OF TABLES

CHAPTER 1
Table 1: Summary of Replicative DNA Synthesis ............................................. 59
Table 2: Summary of GC-Rich Regions of Altered Methylation (RAMs) .................. 65

Table 3: Common Regions of Altered Methylation (RAMS): Comparison of
Diethanolamine (DEA) or Phenobarbital (PB) with Choline Deficient (Choline

 

Def) Treatment ....................................................................................... 69
Table 4: Unique Regions of Altered Methylation (RAMS): Diethanolamine (DEA)

or Phenobarbital (PB) as Compared to Choline Deficient Treatment ........................ 72
CHAPTER 1 APPENDIX

[Quendix Table 1: Excel Workbook illustrating the Organization of Raw Data into
PCR Product Size and Corresponding Peak Area ............................................... 87

Appendix Table 2: Excel Workbook illustrating the removal of background and
adjustment of data according to the minimum data requirements and the outlined

 

guidelines ............................................................................................. 90
Appendix Table 3: Excel Workbook demonstrating the alignment of a Treated peak

area data points to the control PCR product size reporting peak area data points .......... 91
CHAPTER 2

Table 1: Example of Comparisons for Determining whether a CSC-induced Change
in Methylation Reversed Following the Recovery Period .................................... 105

Table 2: Methylation Sensitive Digestion with RsaI/MspI or RsaI/HpaII: Summary
Of GC-Rich Regions of Altered Methylation (RAMs) in the Liver of B6C3F1 and
C57BL/6 Mice in Response to 2 or 4wk 0.05% PB ........................................... 113

 

Table 3: Dissimilarity Between PB-Induced Methylation Changes and Control
Methylation Patterns in B6C3F1 and C57BL/6 Mouse Liver ............................... 114

 

Table 4: Dissimilarity Between PB-induced Methylation Changes in B6C3F 1 and
C57BL/6 Mouse Liver ............................................................................ 123

Table 5: Methylation Sensitive Digestion with RsaI/Mspl or RsaI/HpaII: Summary
Of GC-Rich Regions of Altered Methylation (RAMS) in Liver and

viii

Kidney ofB6C3F1 and C57BL/6 Mice in Response to Treatment with 0.05%PB
For 4wks ............................................................................................ 125

Table 6: Dissimilarity Between Methylation Changes in the LIVER as Compared
to the KIDNEY following 4 wks of Treatment with PB ..................................... 127

Table 7: Methylation Sensitive Digestion with BfaI/BSSHII: Summary of GC-Rich
Regions of Altered Methylation (RAMs) in the Liver of B6C3F] and C57BL/6

Mice Following Treatment with PB for 2 or 4 wks ........................................... 128
CHAPTER 3

Table 1: Summary of GC-Rich Regions of Altered Methylation: Comparison of
DMBA Initiated , CSC 8wk Promotion and Tumor Tissue to Control ..................... 169

Table 2: Measure of the Percent Dissimilarity of 8wk, 3, 9, 18, 27mg CSC or Tumor
To DMBA/Acetone Control ..................................................................... 171

Table 3: Summary of GC-Rich Regions of Altered Methylation: Comparison of
DMBA Initiated, CSC 4 and 8wk Promotion to Control ..................................... 174

ix

LIST OF FIGURES
INTRODUCTION

Figare 1: Multistage Carcinogenesis .............................................................. 3
Figure 2: Methylation, A Heritable Feature of DNA .......................................... 10
Figure 3: DNA methylation and l-carbon Metabolism ....................................... 17
F 'gure 4: Transposable elements encompass both transposons and retrotransposons. . . ..25
Figure 5: Altered Methylation and LINE-1 Elements ......................................... 27
Figure 6: B6C3F] Control Animals: Peak Areas of PCR products generated

following digestion with Mspl as a Percent of those generated following the HpaII

Digest ................................................................................................. 41

Figure 7: B6C3F1 PB Treated Animals: Peak Areas of PCR products generated
following digestion with MspI as a Percent of those generated following HpaII

Digest ................................................................................................. 42
CHAPTER 1
Figge 1: Global Methylation Status in DNA from Primary Mouse Hepatocytes. . ....60

F igare 2: GC-Rich DNA Methylation Status in Primary B6C3F 1 Mouse
Hepatocytes ...................................................................................... 62-64

F igge 3: Comparison of Diethanolamine and Choline deficiency Induced
Aberrant GC-rich Methylation Patterns ...................................................... 66-67

Figge 4: Comparison of Phenobarbital and Choline deficiency Induced
Aberrant GC-rich Methylation Patterns ...................................................... 70-71

CHAPTER 1 APPENDIX

Appendix Figure 1: Individual Animal Average Peak Area Chart For
Visualization of Data Reproducibility and Consistency ........................................ 93

CHAPTER 2

F igare 1: Progressive Changes in Methylation: Changes Which Carry Forward
From 2wk to 4wk in B6C3F1 and C57BL/6 Mice ....................................... 116-117

Figge 2: Progressive Changes in Methylation: Reversibility of Regions of
Altered Methylation Induced by PB ........................................................ 118-119

Figure 3: Identification of PB-Induced Unique and Carry Forward RAMs in

B6C3F1 Liver. 121-122
Figure 4: Methylation Status of the Promoter Region of Ha-ras ........................... 130

Figge 5: Methylation Status of the Promoter Region of LINE-1 in B6C3F 1
and C57BL/6 Mouse Liver ....................................................................... 131

Figure 6: Effect of 2wk 0.05% Phenobarbital Promotion on the Expression of
Ha-ras in B6C3F1 Mouse L1ver133

Figure 7: Effect of 4wk 0.05% Phenobarbital Promotion on the Expression of
Ha-ras in B6C3F1 Mouse L1ver134

F igge 8: Reversal of Increased Expression Following 4sk Phenobarbital Exposure
In B6C3F1 Mouse Liver ........................................................................... 135

Figare 9: Effect of 2 and 4wk Phenobarbital on the Expression of Ha-ras in
C57BL/6 Mouse Liver ............................................................................ 136

xi

Figure 10: Effect of 4 wk Phenobarbital Promotion on the Expression Level of
LINE-1 Elements in B6C3F1 and C57BL/6 Mouse Liver ................................... 137

CHAPTER 2 APPENDIX

Appendix Figure 1: Analysis of Altered Methylation Near to the Transcriptional
Start Site of Ha-ras in B6C3F1 and C57BL/6 Mice at 2 and 4wks ................... 155-156

CHAPTER 3

Figure 1: Increases (Hypermethylations and New Methylations) in GC-Rich

Methylation are Dose-Dependent ............................................................... 170
Figure 2: Tumor Incidence Increases with Dose of CSC .................................... 173

F igare 3: Progressive Changes in Methylation: Changes Which Persist From
4 to 8wk and from 8wks to Tumor ............................................................... 175

Figure 4: Progressive Changes in Methylation: Changes Induced by 8wk,
9, 18, and 27mg CSC and Persist to Tumor .............................................. 177-178

Figure 5: Progressive Changes in Methylation: Reversible Methylation Changes

Following 4 and 8wk Recovery ............................................................. 180-181
Figge 6: Methylation Status of the Promoter Region of Ha—ras ........................... 182
Figare 7: SmaI Restriction Digest Analysis of Ha-ras ....................................... 184
Figare 8: Expression of Ha-ras .................................................................. 185
CHAPTER 3 APPENDIX

Appendix Figure 1: Reversible Methylation Changes Following 8wk Promotion ....... 201

xii

Appendix Figure 2: Reversible Methylation Changes Following 4 and 8wk
Promotion with 27mg CSC ....................................................................... 202

Appendix Figure 3: Targeted Region of LINE-1 .............................................. 204

 

Appendix Figure 4: Methylation Status of the Tan Site Within a Targeted
Region of LINE-1 ................................................................................. 205

Appendix Figure 5: Methylation Status of the thI Site Within a Targeted

Region of LINE-1 ................................................................................. 206
Appendix Figure 6: Expression of LINE-1 .................................................... 207

 

xiii

INTRODUCTION
Cancer

According to the American Cancer Society, cancer accounts for 23% of all deaths,
second only to cardiovascular disease. Understanding the cause and pathogenesis of
cancer through research is necessary for treating the millions of people affected
worldwide. To do this, research endeavors must build on existing concepts in addition to
developing new and different approaches to treatment and diagnosis. Potential and
known carcinogens must be logically and rationally assessed in terms of risk posed to
humans. This will require a comprehensive look at the underlying molecular
mechanisms of cancer with a focus on the mode of action of the chemical(s) of interest.
This can enhance the scientific basis for three key aspects of safety assessment: 1)
selection of doses for testing, including rational selection of the high dose; 2) evaluation
of the dose-response relationship, and 3) rational species-to-species extrapolation.

Cancer is characterized by six fundamental changes in cell physiology including
self-sufficiency in growth signals, insensitivity to growth-inhibitory signals, evasion of
apoptosis, limitless replicative potential, sustained angiogenesis, and the ability to invade
and metastasize (Hanahan and Weinberg, 2000). The genetic pathways giving rise to
these classic hallmarks have complex origins and can involve the mis-regulation or
mutation of critical cell cycle proteins, transcription factors and signal transduction
proteins among others. This adds to the difficulty of discerning absolute cause and
constrains the notion of one encompassing method to stop or prevent cancer from

occurring. However, a working model of carcinogenesis has been proposed and serves as

a unifying factor in determining the steps and targets involved in the progression of

precancerous tissue to tumor tissue.

Multi-Stage Carcinogenesis and Epigenetics

A multistage, multi-step (i.e. multi-mechanism) process of carcinogenesis
provides a framework for explaining events leading to cell proliferation and
tumorigenesis (Figure 1). The stages defined by this process are initiation, promotion
and progression (Dragan et al., 1993). During initiation, a heritable change occurs in the
genome. A heritable change is often associated with direct mutation of the DNA
sequence. However, epigenetic modifications (6. g. DNA methylation, histone acetylation)
which do not affect the base sequence of DNA, can also be heritable alterations.
Promotion of the selected cells, or those that were initiated, occurs when an agent (e. g.
phenobarbital, peroxisome proliferators) allows for preferential growth over neighboring
cells. Agents acting through mechanisms not involving direct DNA damage are termed
non-genotoxic and can be thought of as acting through a secondary mechanism of
carcinogenesis (Goodman and Watson, 2002). This stage is reversible, in that, if the
promoting stimulus is withdrawn, the altered cells possessing advantageous grth
characteristics stop proliferating and altered foci can “remodel”. The promoting stimulus
continues to foster the growth of initiated cells and new subsets of cells arising from
those that are undergoing clonal expansion. Subsequent to the iterative nature of this
process is progression. At this point, cells are clonally expanding even in the absence of

the promoting stimulus. In addition, cells at this stage typically exhibit marked

INITIATION

   
    
  
  
   
 

Clonal expansion

Initiating
® —>-—>—> -—> —>

Event

2nd Critical Event

PROMOTION

  

Clonal Expansion
~ ~
~ ~ ~

PROGRESSION

Figre 1 Multistage Carcinogenesis The three stages of carcinogenesis are initiation,
promotion, and progression. During initiation a cell (represented by a circle) acquires
some heritable change (e. g. mutation or altered DNA methylation) within the genome.
Each line through a circle represents a heritable event. In the presence of a promoting
stimulus, initiated cells possessing a growth advantage over neighboring cells proliferate.
The process repeats until the cells reach a state of autonomous, clonal expansion; this is
termed progression.

karyotypic instability including chromosomal damage and changes in ploidy (Dragan et
aL,1993)

The focal point during the promotion stage of carcinogenesis is the accumulation
of heritable changes within the genome. These are fundamental to the initiation and
development of cancer. As mentioned above, mutation is the obvious and standard
example of a heritable change which can seed the development of tumorigenesis.
Importantly, epigenetics has also taken a parallel role to mutations. Epigenetics is
broadly defined as processes that establish heritable states of gene expression without
altering the DNA sequence. This includes DNA methylation and histone acetylation each
of which alter the regulation of gene expression but do not affect the base sequence of
DNA (Feinberg, 2001).

Methylation of cytosines to produce S-methyl cytosine is a well characterized
epigenetic mark. Because both cytosine and 5-methyl cytosine base pair with guanine,
this epigenetic modification is not a mutation. The majority of S-methylation cytosine
occurs at cytosines 5’ to guanine although methylation of non-CpG dinucleotides such as
CpA, and CpT have been reported (J abbari and Bemardi, 2004) in addition to CprG
methylation (Jackson et al. , 2002). Therefore, altered patterns of methylation can
potentially effect a large majority of the genome and evoke widespread consequences.
Furthermore, DNA methylation can be a precursor to mutation. Spontaneous
deamination of 5-methyl cytosine yields thymine and this can base-pair with adenine
resulting in a CG to TA transition mutation (Cooper and Krawczak, 1989). In addition,
DNA adduct formation due to oxidative stress or agents such as dimethylsulfate and

ethylnitrosourea can result in altered methylation. Under conditions of oxidative stress,

 

the common DNA adduct, 8-hydroxyl-2’-deoxyguanosine, has been shown to interfere
with the ability of the human DNA methyltransferase to methylate target cytosines
nearby (Turk et al., 1995). Additionally Tan and Li, 1990, demonstrated that 6-O-
methylguanine located 5’ to cytosine can affect the maintenance methylation of the
opposite strand in a hemimethylated duplex. The presence of this adduct might
destabilize the hemi-methylated site and cause the methylase to detach, or the adduct
could both enhance or decrease the site as a substrate for the methylase depending on its
position within the genome. In this manner, DNA adducts can either increase or decrease
the methylation of neighboring cytosines leaving an abnormal, yet heritable, epigenetic
pattern.

Histone modifications are also considered reversible epigenetic processes.
Nuclear DNA is packaged into nucleosomes. The core histone octamer consists of an
H3-H4 tetramer (H32-H42) and two H2A-H2B dimers. Around this histone core,
approximately 200bp of DNA is wrapped. Each histone has a flexible N-terminal tail
which can be reversibly modified by acetylation, methylation, ubiquitination,
biotinylatioin, and phosphorylation (Spotswood and Turner, 2002; Petterson and Laniel,
2004). The modification of these histone tails can destabilize higher order chromatin
structure. Transcriptionally active chromatin, or euchromatin, is associated with
methylation of lysine 4 and 9 in addition to acetylation of lysine 9 and 14 (Espino et al.,
2005). Transcriptionally repressed chromatin, or heterochromatin, is associated only
with the methylation of lysine 9 (Espino et al., 2005). Each of these mechanistically

contribute to the transcriptional regulation of euchromatic genes (Richards and Elgin,

2002). Histone acetylation leads to a more relaxed chromatin conformation while histone
deacetylation results in a tighter packaging of the DNA.

There is a tightly regulated relationship between histone modifications, chromatin
structure, and DNA methylation (Szyf et al., 2004). The order of events by which
chromatin is modified to yield a transcriptionally active or inactive state is not well
characterized. However, three routes to epigenetic silencing have been proposed. These
include the possibility that DNA methylation dictates histone modification, histone
modification mediates DNA methylation, or nucleosome remodeling facilitates DNA
methylation (Lund and van Lohuizen, 2004). Evidence supporting aspects of all three of
these possibilities has been shown (Fuks et al., 2000; Chaumeil et al., 2004; Dennis et al. ,
2001)

Although the timeline of events is still being elucidated, the cooperation between
histone modifications and DNA methylation is undisputed. A “histone code” hypothesis
has been developed to describe the role of histone acetylases and deacetylases in
conjunction with ATP-dependent remodeling factors (Strahl and Allis, 2000). These
ATP-dependent factors (e. g. SWI-SNF, Mi-2, and ISWI families) cause the disruption
and sliding of nucleosomes along the helical path of DNA to facilitate transcription
(Ballestar and Esteller, 2002). The presence of DNA methylation elicits histone
deacetylation and prevents methylation of lysine 4 on histone 3. The removal of 5-
methyl cytosine allows for the methylation of lysine 4 on histone 3 and without the
underlying repression mechanisms, histones undergo acetylation (Lande-Diner and Cedar,
2005). The link between factors affecting histone conformation and DNA methylation

involves methyl-DNA binding proteins (Ballestar and Esteller, 2002). A possible order

of events leading to gene inactivation begins with a low level of DNA methylation at the
promoter. This methylation signal recruits the methylated DNA-binding protein MBD2,
which recruits histone deacetylases and Dnmtl, one maintenance methylase responsible
for regulating the status of DNA methylation. Histone deacetylation and subsequent
methylation of a promoter region of a gene by Dnmtl results in the recruitment of the
methyl DNA-binding protein MeCP2. MeCP2 in turn recruits a histone H3, lysine 9
methyltransferase for methylation of lysine 9 and condensation of chromatin (Espino et
al., 2005). Deacetylation and methylation reactions coupled to the recruitment of
numerous proteins largely prevents transcription factors from gaining access to the DNA.
In support of this, a combined administration of 5’-AZA and trichostatin A (histone
deacetylase inhibitor) resulted in activation of a cytomegalovirus promoter-driven
reporter gene construct (Grassi et al., 2003). Notably, each treatment alone reactivated
the reporter gene construct, however, differing enzyme kinetics were reported (Grassi et
al., 2003). These reversible reactions clearly cooperate as integrative epigenetic
mechanisms for gene regulation and illustrate their significance during the process of
tumorigenesis.
The Origin of the Cancer Stem Cell

The origin of the “cancer stem cell” is a highly debated topic and each side of this
larger issue deserves mention. The cancer stem cell is, in essence, the cell that acquires
the first heritable change to its genome that sets it on a potential path leading to tumor
development. These cells that acquire the first critical initiating event(s) could be derived
from normal stem cells, early stem cell progenitor cells, or differentiated cells (Bjerkvig

et al., 2005). The self- renewing properties of cancer stem cells are characteristic of stem

cells which are an obvious and likely origin. However, de-differentiation of normal cells
to a stem-cell like state is also possible. In support of this, differentiated astrocytes in
which the epidermal growth factor receptor pathway is activated and tumor suppressors
p16 and p19 are inactivated lead to a common high-grade glioma phenotype in vivo
(Bachoo et al., 2002). The expression of a critical transcription factor, Oct4, is associated
with pluripotency and the downregulation of Oct4 has been linked to the differentiation
of somatic cell lineages (Tai et al., 2005). Ectopic expression of this transcription factor
in certain somatic cells has also been implicated in active dedifferentiation (Shimazaki et
al., 1993; Hochedinger et al., 2005). Altered methylation resulting in heritable genomic
changes could potentially contribute to the evolution of normal cells into the cancer stem
cell-like state. For example, murine embryonic stem cells were induced to differentiate
in vitro to embryoid bodies and then treated with 5’-aza-2’-deoxycytidine (5’-AZA).
During DNA synthesis 5’-AZA is incorporated into newly synthesized DNA in place of
deoxycytosine and covalently binds to DNA methyltransferases. This linkage effectively
depletes the cell of fimctional DNMTs and leads to hypomethylation or “demethylation”
afier successive rounds of replication (J utterrnann et al., 1994). The embryoid bodies
exhibited stem cell-like characteristics including stem cell like morphology with unclear
cell-to-cell boundary and proliferative responsiveness. In addition, increased expression
of embryonic stem cell markers such as Oct4, Nanog, and Sox2 were measured
suggesting that the differentiated state of these cells was reversed (Ysuji-Takayama et a1 .,
2004). Therefore, both the stem cell theory and dedifferentiation theory for the origin of
the cancer-stem cell are not mutually exclusive in the overall context of measuring

progressive changes in DNA methylation.

Regulation of Methylation Patterns: DNA methyltransferases

Methylation, as a heritable feature of DNA, is mainly dependent on the
maintenance methyltransferase (DNMTl) during DNA replication, however, cycling
through various states of methylation is accomplished via de novo and demethylases
(Figure 2) (Pradhan and Esteve, 2003). After one round of replication in which a new
daughter strand has been synthesized, the DNA exists in a hemi-methylated state where
one strand is methylated and the newly synthesized daughter strand is wholly
unmethylated. To return to a fully methylated state Dnmtl , a maintenance methylase,
recognizes hemimethylated DNA and methylates CpG sites accordingly (Bestor, 2000).
Proper maintenance methylation is critical in that a completely hypomethylated state of
DNA can occur when two rounds of replication ensues without proper maintenance
methylation. Dnmtl is thought to be essential for the survival of an organism. Mice
deficient for Dnmtl die in mid-gestation with significantly reduced levels of DNA
methylation (Bestor and J aenisch, 1992). Acting as an integral part of cell cycle control,
interaction of Dnmtl with PCNA, p21 WAF 1 (inhibitor of cyclin-dependent kinases
(CDKS) and the processivity factor of the replication fork has been demonstrated
(Chuang et al., 1997). The close association between Dnmtl and PCNA has also led to
the suggestion that Dnmtl serves as a Signal for mis-match repair during replication and
methylation of the hemi-methylated state (Wang and Shen, 2002; Mortusewicz et al.,
2005) Interestingly, inhibition of Dnmtl has been shown to negatively affect DNA
synthesis and progression through the cell cycle in human non-small cell lung carcinoma,
A549 cells (Knox et al., 2000). Furthermore, a network of connections between Dnmtl

and histone modifying enzymes, methyl binding proteins and heterochromatin binding

Methylated DNA

CH, CH,

4' 5’-GGCCACGTGACCGG-3’
3’-CCGGTGCACTGGCC-5’

 

 

3

CH CH

. . Maintenance

DNA Replication! I Methylation
0H CH

3 . 3
5’-GGCCACGTGACCGG-3’
3’-CCGGTGCACTGGCC-5’

de novo methylation

DNA replication with no
maintenance methylation

 

 

 

Demethylation without DNA replication

5’-GGCCACGTGACCGG-3’ <
3’-CCGGTGCACTGGCC-5’

Hypomethylated DNA

 

F'gare 2 Methylation, A Heritable Feature of DNA Following DNA replication, the
DNA exists in a hemi-methylated state; one strand of the DNA is methylated while the

newly synthesized strand is not yet methylated. Maintenance methylation will return
hemi-methylated DNA to fully methylated DNA. If a second round of replication ensues
without proper maintenance methylation, a hypomethylated state of DNA occurs; Both
strands of the DNA are unmethylated. In addition, fully methylated DNA can become
hypomethylated via demethylation without DNA replication and hypomethylated DNA
can be returned to the fully methylated state through de novo methylation.

(Adapted from Hergersberg, Experientia, 1991)

10

protein all point to Dnmtl ’s involvement in gene regulation and epigenetic signaling
(Hermann et al., 2004).

Just as failure to methylate hemi-methylated DNA results in hypomethylation, so
too can demethylation of fully methylated DNA. Mechanistically, removing a methyl
group from cytosine would involve cleavage of a carbon-carbon bond, making this an
unlikely reaction due to the high energy requirement (Bhattacharya et al., 1999).
Therefore, indirect mechanisms involving base excision and repair have also been
proposed (V airapandi, 2004). Thermodynamically, direct demethylation became feasible
with the identification of methanol as the leaving group and water as a possible reactant
(Ramchandani et al., 1999). The demethylase might act to stabilize an intermediate state
so that a hydroxide ion can then attack the C5 methyl group (Ramchandani et al., 1999).
In either case, the main consequence of DNA demethylation is a hypomethylated state of
DNA which could have functional consequences. Demethylation-induced
hypomethylation has been linked to enhanced transcription of the T-cell growth hormone
interleukin-2 gene. This gene is actively demethylated in T lymphocytes and allows for
proliferation and the production of other cytokines including interferon y and IL-4
(Bruniquel and Schwartz, 2003). Demethylation by DNA demethylase has also resulted
in the up-regulation of the c-myc oncogene in human gastric cancer (Fang et al., 2004).
Therefore removing methyl groups from cytosines within DNA has various implications
and raises the question of the specificity of DNA demethylase for DNA in normal cells
and cancerous cells. The demethylase activity has been shown to associate with PCNA

during replication in normal cells and target hemi-methylated CpG sites. However, in

11

cancer cell lines, fully methylated CpG islands are the substrate for the DNA demethylase
activity (V airapandi, 2004).

A hypomethylated state of DNA can be returned to the fully methylated state via
de novo methylation, associated with both Dnmt3a and Dnmt3b enzymes (Pradhan and
Esteve, 2003). Both proteins are essential for mouse development. Their close
association is demonstrated by the fact that double knockouts in murine embryonic stem
cells have a more severe phenotype than each individual deletion mutant indicating there
is some compensatory activity by each (Okano et al., 1999). Dnmt3a is ubiquitous while
Dnmt3b is normally present at low levels (Xie et al., 1999). Even though Dnmt3a and 3b
are highly related , they are encoded by separate genes and do exhibit somewhat
specialized roles (Hermann et al., 2004). Dnmt3b is processive supporting its ability to
methylate pericentromeric repeats carrying high CG content. One cause of
immunodeficiency, centromeric instability, facial abnormalities, (ICF), is mutation of the
Dnmt3b gene, however, the Dnmt3a gene is unaffected. Therefore, the characteristic
hypomethylation at pericentromeric satellite regions in this rare recessive autosomal
disorder is solely attributed to Dnmt3b (Xu et al., 1999). Interestingly, over-expression
of a the Dnmt3b4 splice variant has been associated with DNA hypomethylation on
pericentromeric satellite region in human hepatocellular carcinomas (Saito et al., 2002).
Elevation of the ratio of Dnmt3b4 to a second splice variant, Dnmt3b3, could cause
competition for the targeted region upsetting the balance needed to properly maintain
methylation of pericentromeric satellite regions (Saito et al., 2002). The higher intrinsic
methylation activity of Dnmt3b over Dnmt3a coupled with its frequent over-expression

in various tumors, supports a role for Dnmt3b in tumorigenesis (Robertson et al., 1999).

12

Dnmt3a on the other hand is more specific, showing preference to methylate sites
that are flanked by pyrimidines rather than purines and therefore, methylation events are
more controlled (Lin et al., 2002). In line with this, the establishment of methylation
patterns at single copy genes has been attributed to Dnmt3a, in cooperation with Dmnt3L
(Hata et al., 2002). Dnmt3a shows strong interactions with a number of proteins
including histone H3, lysine 9 methyltransferase Suv39, Dnmtl , and histone deacetylases
(Kim et al., 2002; Fuks et al., 2003). A very detailed network of co-operativity between
these enzymes and the methylation machinery including methyl binding proteins and
histone acetylases and deacetylases point to compensatory mechanisms which could
preserve the cyclic balance of the methylation states of DNA. Therefore, the fidelity of
endogenous mechanisms maintaining the proper state of methylation throughout the
genome is crucial especially during times when a high percentage of cells are

proliferating.

5-Methyl Cytosine Distribution and Abundance

In mammalian genomes, there is a positive correlation between gene density and
(G + C) content where 75-80% of genes reside in the (G + C)-richest half of the genome
(Waterston, R.H. et al., 2002). Therefore, the distribution of methylated cytosines within
CpG dinucleotides has important meaning in understanding the regulation of gene
expression. CpG dinucleotides are not evenly distributed throughout the genome (Bird,
2002). An estimated 70% of all CpG sites are methylated; however, completely
unmethylated CpG islands regions account for ~1% of the genome and an estimated 15%

of the total genomic CpG sites (Roberston and Wolffe, 2000). CpG islands are short

13

stretches of DNA, at least 200bp in length, possessing 50% or greater GC content and a
higher proportion of CpG dinucleotides than expected. In total about 15,500 CpG islands
are estimated (Waterson et al., 2002), of which, the majority are mainly found within the
promoter regions or first exons of genes (Gardiner-Garden and F rommer, 1987). The
normal status of methylation of each individual region varies although the majority of
CpG islands are normally unmethylated allowing for transcriptional activity of the
respective gene (Antequera, 2003). Many CpG dinucleotides are also located in GC-rich
promoter and promoter-like regions of transposable elements (Liang et al., 2002) which
comprise approximately 33% of the human and mouse genomes (Yoder et al., 1997)
indicating that cytosine methylation could significantly contribute to the regulation of
non-coding regions (1'. e. repetitive regions) as well as the coding regions throughout the
genome.

Non-CpG methylation (eg. CpA, CpT, CpC) has also been reported and expands
the total proportion of the genome potentially affected by DNA methylation (Dodge et al.,
2002). For example, one early report based on the nearest neighbor technique estimated
that 55% of all methylation in human spleen DNA could be at dinucleotides other than
CpG (Woodcock et al., 1997). Since then, more accurate representations and roles for
non-CpG methylation have been proposed. The significance for non-CpG methylation
during early development has been questioned due to the fact that 15-20% of total
cytosine methylation content of embryonic stem cells is at sequences other than CpG
(Ramsahoye et al., 2000). This non-CpG methylation is associated with the activity of
the de novo methyltransferase Dnmt3a which is highly expressed in embryonic stem cells

(Ramsahoye er al., 2000). The functional role of methylated CpA and CpT sites, the

14

most frequent form of non-CpG methylation identified, during development, is hard to
determine due to the fact that Dnmt3a'/' mice die shortly after birth (Okano et al., 1999).
However, the CpA methylation was again associated with Dnmt3a and or Dnmt3b in a
model of de novo methylation of murine Maloney leukemia virus provirus DNA in virus-
infected embryonic stem cells (Dodge 2! al., 2002). With this model, CpA methylation
was detected at ~l .4% of all sites in infected wild-type and ~1.0% in Dnmtl knockout
cells. However, in Dnmt3a and 3b knockout cells, only 0.2% of all sites exhibited CpA
methylation demonstrating the relationship between Dnmt3 enzymes and non-CpG
methylation.

A large portion of my research involved analysis of the methylation status of the
external cytosine within CpCpG sites. Very few studies report methylation at CprG
sites and even less have proposed a role for CpCpG methylation. In Arabidopsis, CprG
methylation plays a role in gene silencing and is mediated by histone H3 lysine 9
methylation through interaction of the DNA methyltransferase gene with methylated
chromatin (Jackson et al., 2002). Evidence for significance in mammalian systems is
limited. Methylation at both cytosines with in CpCpG sites has been reported to prevent
binding of Spl , an important transcription factor, to its target cis element (Inoue and
Oishi, 2005). However, the functional significance of methylation at solely the external

cytosine has not been investigated.

DNA Methylation and Nutrition

The importance of understanding how diet influences carcinogenesis is discussed

in the context of altering DNA methylation. Maintaining patterns of methylation is

15

highly dependent on the availability of methyl groups from S-adenosylmethionine (SAM).
This methyl donor is derived from methionine and serves as the main methyl donor in
methylation reactions involving DNA, RNA, hormones, neurotransmitters, membrane
lipids, and proteins (Ross and Poirier, 2002). SAM is directly synthesized from its
precursor methionine, an essential amino acid (Figure 3) (Van den Veyver, 2002).
Choline and folate interact with the metabolism of methionine at its precursor
homocysteine. Choline is metabolized to betaine which serves as the methyl donor to
regenerate methionine (Van den Veyver, 2002). Alternatively, methyl-tetrahydrofolate
derived from folate and l-carbon metabolism can donate a methyl group to homocysteine
to form methionine (Van den Veyver, 2002). When SAM donates a methyl group, it is
converted to S-adenosylhomocysteine (SAH). The conversion of SAH back to
homocysteine leads to the recycling of homocysteine and methionine. The balance of
these inter-dependent factors determines the availability and utility of SAM.

Indications for the tight inter-relationships of these factors is apparent when
knockout mice are studied. Methylene tetrahydrofolate reductase catalyzes the transfer of
a methyl group from methyl-tetrahydrofolate to homocysteine. Mice lacking this enzyme
show depleted choline and betaine levels as the maintenance of methionine synthesis is
stressed (N iculescu and Zeisel, 2002). In addition, cystathionine beta-synthase knockout
mice accumulate homocysteine and must convert it to methionine to remove it. In doing
so, choline and betaine pools are depleted (Niculescu and Zeisel, 2002). Importantly,
homozygous mutants for cystathionine beta-synthase Show a lower genomic DNA
methylation status in liver as compared to wild-type. This effect was also observed in

kidney tissue but not brain (Choumenkovitch et al., 2002).

16

DNA Synthesis DNA Methylation

dUMP Dihydrofolate\
THF E Methionine \
/ SAM Cytosine-DNA
dTMP 5 DNA Methyltransferase

Methionine
C 2 ’ '
10 H THF C“°""" 5-methyl Cytosine-DNA

Synthase
Betaine / SAH
Homocysteine /

Fi ure 3 DNA meth lation and l-carbon Metabolism Schematic representation of the
cyclic interplay between DNA synthesis and DNA methylation. DNA methyltransferase
transfers a methyl group from S-adenosylmethionine (SAM) to cytosine to form S-methyl
cytosine and S-adenosylhomocysteine (SAH). Other dietary factors involved in this
process are outlined for their roles in DNA synthesis or DNA methylation cycling.
(Adapted from Choi and Mason, 2002)

 

   
  

 

 

 

 

 

 

 

B12
Methyl tetrahydrofolate
reductase
5-methyl THF

 

 

17

Forms of methyl deficiency are induced via Choline, methionine, choline and
methionine, or folate deficient diets. Hypomethylation is frequently observed with
methyl-deficient diets. Deficiencies in methionine and choline have been shown to lead
to global hypomethylation in the livers of mice (Counts et al., 1996) as well as over
expression of oncogenes in the livers of rats (Wainfan and Poirier, 1992). These diets
serve as effective promoting agents in multistage hepatocarcinogenesis. A key
component to their effects is the induction of a hypomethylated state of DNA. Male
F344 rats fed a methyl-deficient diet for 9, 18, 24 and 36wks showed decreased levels of
SAM, SAM/SAH ratios, and global DNA hypomethylation (Pogribny et al., 2005). Re-
feeding the rats a methyl-adequate diet restored all parameters except the hypomethylated
state of DNA and did not prevent the expansion of initiated foci. This suggests that
stable DNA hypomethylation induced by methyl-deficiency is a promoting factor in
stimulating initiated cells (Pogribny et al., 2005).

Absolute levels of critical factors such as SAM and SAH might not be as
important as the ratio between the two (Shivapurkar and Poirier, 1983). The ratio of
SAM to SAH has been suggested to contribute in the mis-regulation of DNA methylation
reactions. SAH is the intermediate formed during the recycling of SAM to homocysteine,
the direct precursor of methionine. In effect, methyl deficiency decreases SAM and
increases SAH shifting the proportionality towards SAH which is a feedback inhibitor of
DNA methyltransferases. Therefore, the SAM/SAH ratio indirectly serves as a
determinant of the extent of methylation (Shivapurkar and Poirier, 1983). SAM, on the
other hand, acts as a feedback stimulator for the formation of 5-methyl tetrahydrofolate

which donates a methyl group to homocysteine and facilitates the maintenance of

18

methionine levels (James et al., 2003). This increases the intracellular requirement of
folate which when depleted can compromise the de novo synthesis of deoxynucleotides in
addition to further impairing the synthesis of SAM (James et al., 2003). Therefore, the
balance of these factors is critical to keeping up with the demand for maintaining the
status of genomic methylation patterns.
Gene-Specific Patterns of Methylation and Regulation of Gene Transcription
Altered methylation as a precursor to toxicity is largely centered on regulating the
expression of genes, most importantly, oncogenes, tumor suppressor genes, and
transposable elements, which can be either increased or decreased resulting in toxic
outcomes. Oncogene expression can be up-regulated via hypomethylation while tumor
suppressor genes can be silenced when methylated. Both are classic contributors to the
initiation and progression of cancer (Jones and Baylin, 2002). With my research I have
consistently identified both increases and decreases in the methylation status of DNA
occurring simultaneously in both precancerous and cancerous tissue. Therefore, it is

imperative to consider the direct consequences of each distinct type of alteration.

Gene-Specific Methylation: Silencing of Tumor Suppressor Genes
Hyperrnethylation of promoter regions which are most commonly CpG island
regions decrease expression levels of the corresponding gene. This silencing event has
important consequences in the context of tumorigenesis and has been well characterized
and demonstrated in tumors. Frequently cited genes which are observed to be
hyperrnethylated include, p16, MGMT, CDKNZB, and RASSF 1A (Jones and Baylin,

2002). Silencing of the tumor suppressor gene, p16, was demonstrated in gastric

l9

carcinoma tissue (Chong et al., 2003). P16 inhibits the activity of cyclin-dependent
kinase 4 (cdk4) or cdk6. When inhibited, cdk4 and cdk6 can not phosphorylate
regulatory proteins. For example, the retinoblastoma protein (Rb) must be
phosphorylated in order to trigger a series of events transitioning the G1 to S phase of the
cell cycle. Therefore, if p16 expression decreases, cdk4 and cdk6 activity will increase
resulting in phosphorylation of the Rb protein and progression of the cell cycle (Byeon et
al., 2004). Aberrant promoter methylation of RASSF 1A is frequently detected in tumors
of bladder, breast, colon, kidney, liver, and lung among others indicating its very
common involvement in the progression of cancer (Pfeifer and Darnmann, 2005).
Knockout mice in which exon 1 of RASSFlA was deleted resulted in a more severe
tumor susceptibility phenotype in mice supporting its tumor suppressive role. The
biological role of RASSF 1A is unknown but is hypothesized to be involved in several
grth regulatory and apoptotic pathways (Pfeifer and Dammann, 2005).

In analyzing skin precancerous and cancerous tissue, one of my consistent
findings was that hyperrnethylation is a more frequent occurrence than hypomethylation
when promoting with cigarette smoke condensate. This indicates a predominant role for
silencing of tumor suppressor genes in advancing skin tumorigenesis, and correlates to
the hyperrnethylation of tumor suppressor promoter regions p16, MGMT, and HOXAS
observed using the same model (Watson et al., 2004). In addition, exposure of B6C3F1
mice to mainstream cigarette smoke has been reported to silence the Death Associated
Protein (DAP)-kinase and Retinoic Acid Receptor (RAR)-B genes via promoter
hyperrnethylation and an increase the incidence of primary lung neoplasms (Hutt et al.,

2005).

20

Because DNA methylation is a reversible epigenetic mark, restoring the normal
methylation status of these tumor suppressor genes becomes clinically relevant.
Experimentally, reversal of a methylated state has been demonstrated. For example,
microarray profile analysis revealed the silencing of 30 genes within an analyzed panel of
expressed CpG island sequence tags in breast cancer cells, The re-expression of these
silenced genes was confirmed by treatment with 5-AZA (Shi et al., 2002). In a clinical
setting re-expression of silenced genes via 5-AZA has limited success due to its non-
specific effects. However, with proleukemic myelodysplastic syndrome, promising
results have been obtained and could be related to the reactivation of the cyclin-
dependent kinase inhibitor gene p15 (Herman and Baylin, 2003).

In addition to reactivating silenced tumor suppressor genes, specific methylation
profiles of these genes could be used as unique biological and clinical parameters to
identify different risk groups among patients (Banelli et al, 2005). The CpG island
methylator phenotype was originally described in a subset of sporadic colorectal cancer
with microsatellite instability (Toyota et al., l999). This phenotype refers to the
Simultaneous hypermethylation of multiple genes and this concept has been expanded for
the purposes of identifying methylation profiles of silenced genes. On a small scale, gene
hyperrnethylation profiles have been created for neuroblastic tumors (Banelli et al., 2005).
On a much larger scale, over 600 primary tumor samples representing 15 major tumors
types were categorized and characterized for association with abnormal gene silencing
illustrating the potential utility of using tumor suppressor genes as biomarkers for

predicting and diagnosing cancer (Estellar et al., 2001).

21

Gene-Specific Methylation: Activation of Oncogenes

Hypomethylation of promoter regions of genes has been linked to an increase in
transcriptional activity. Ha-ras is a classic oncogene which plays a central role in signal
transduction pathways, specifically the SOS-Ras-Raf-MAP kinase mitogenic cascade
which transfers signals from growth factor receptors and integrins to the nucleus, leading
to cell proliferation (Hanahan and Weinberg, 2000). It has commonly been implicated in
tumorigenesis due to the high rate of mutation observed. Specifically, codons 12 and 61
are frequently mutated in spontaneous and mutagen-induced C3H/HE and B6C3F lmouse
liver tumors (Whysner et al., 1996). When phenobarbital, a non—genotoxic rodent liver
tumor promoter is administered without previous initiation, findings show that an
increase in the number of tumors is observed in susceptible mice. For example, in
B6C3F1 mice exposed to 0.05% PB for 1 year, 100% of mice had 3-8 tumors; for those
mice not exposed to PB, 29% had only 1-2 tumors per mouse. Additionally, the
frequency of mutation in Ha-ras is lower than that found in spontaneous tumors
(Whysner et al., 1996). To extend the previous example, only 7% of PB-induced liver
tumors showed point mutations in codon 61 as compared to 64% of spontaneous tumors
(Maronpot et al., 1995). This leads to the possibility that Ha-ras is regulated by an
epigenetic mechanism such as methylation.

In much the same way, methyl deficiency elicited through diet or via agents such
as arsenic can lead to a deficit in available stores of methyl groups resulting in global
hypomethylation and gene specific altered methylation as observed with Ha-ras. In
relatively sensitive and resistant mice, a choline-devoid methionine-deficient diet resulted

in hypomethylation of Ha-ras in liver after 12wks of administration (Counts er al. , 1997).

22

In 2002, Okoji eta1., showed that a methyl deficient diet administered to male C57BL/6J
mice in conjunction with arsenic leads to hepatic DNA global hypomethylation and a
reduced frequency of methylation at cytosine sites within the promoter region of the Ha-
ras gene. Therefore, it is plausible that altered DNA methylation by either a deficiency
induced state or as a consequence of non-genotoxic agents might result in increased
expression of cell cycle control genes and hence aid tumor development.

Although numerous studies have focused on the occurrence of Ha-ras
hypomethylation and activation in liver, regulation of the methylation status of Ha—ras, c-
myc, c-jun, cyclin D and r-ras have also been implicated in various other cancers and
models. Digestion with MspI, HpaII, and Hhal has shown that Ha-ras is hypomethylated
in some human colon and lung cancers (F einberg and Vogelstein, 1983). In addition, site
specific hypomethylation of a single CCGG site in the third exon of the c-myc oncogene
was correlated to malignancy in vitro Wachtenheim et al., 1994). Hypomethylation and
over-expression of c-jun and c-myc protooncogenes was demonstrated in liver tumors
initiated by N-methyl-N-nitrosourea and promoted with dichloroacetic and trichloroacetic
in female B6C3F1 mice (Toa et al., 2000). Similarly, cyclin D, a protein involved in
triggering the onset of the S phase in the cell cycle is over-expressed in a subset of gastric
carcinoma. Hypomethylation of the promoter region of this gene was observed in 71% of
gastric carcinomas analyzed and was correlated to an increase in expression of that gene
(Oshimo et al., 2003). Also associated with gastric cancer, was is thought to inhibit BC]-
2 mediated rescue of apoptosis. This gene was seen to be silenced in normal gastric
mucosa but activated via hypomethylation in more than half of gastric cancers (N ishigaki

et al., 2005). These studies demonstrate that methylation as an epigenetic mechanism

23

contributes to the activation of oncogenes during the process of tumorigenesis in both

murine and human models.

Gene-Specific Methylation: Activation of Transposable Elements

This section is a brief synopsis of the more extensive review which was published in
Toxicological Sciences in 2003. Please refer to the Camel] and Goodman, 2003
reference to obtain this review.

Evidence for methylation as a contributor to transcriptional control has been
implicated in oncogenes and tumor suppressor genes as well as transposable elements
(Jones and Baylin, 2002; Yoder et al., 1997). Transposable elements account for
approximately one third of the human genome and are distributed in a non-random
fashion (Yoder etal., 1997). The term transposable element encompasses both
transposons and retrotransposons (Figure 4). Transposons have inverted terminal repeats,
encode a transposase activity, and move from one site to another through a "cut and
paste" mechanism (Smit and Riggs, 1996). Retrotransposons (e. g. LlNE elements),
which move by a "copy and paste" mechanism, proceed through an RNA intermediate
largely dependent on their encoded reverse transcriptase activity. However, they might
utilize the host’s reverse transcriptase (Ostertag and Kazazian Jr., 2001). In this manner,
a copy of the original can be integrated into a new genomic location. Therefore, stability
of the genome depends upon keeping these movable and amplifiable elements
transcriptionally repressed.

It is instructive to consider the role of altered methylation as an epigenetic

mechanism for the activation of retrotransposable elements leading to their expression

24

 

TRANSPOSABLE ELEMENTS

 

 

 

1) TRANSPOSONS
LITRIORFII ORF2 ] ORFn llTR]

 

 

 

 

 

 

2
) RETROTRANSPOSON S
Autonomous Nonautonomous
[5' LTR [ GAG I POL l 3* LTR ] l l Prompter I ALMAAAAN
LTR Retrotransposon SINE

 

[ Promoter [UTR l ORFI I ORF2 I AAIAAAAAAN

Non-LTR Retrotransposon (LINE)

F'gpre 4 Transposable elements encompass both transposons and retrotransposons
(l) Transposons have inverted terminal repeats (ITR), which act as cis elements in the

integration process, and two (or more) open reading frames (ORF), one of which encodes
a transposase activity. These elements move by a “cut and paste” mechanism. (2)
Retrotransposons are divided into autonomous and nonautonomous elements.
Autonomous elements include long terminal repeat (LTR) and non-LTR subgroups.
LTR-containing elements are structurally similar to retroviruses although they lack a
functional env gene. Non-LTR elements contain an internal promoter for RNA
polymerase II, a 5’ untranslated region (UTR) and a 3’ deoxyadenosine (A)-rich tract.
Nonautonomous elements (SINES) contain an internal promoter for RNA polymerase III
and a 3’ A-rich tract (Carnell and Goodman, 2003).

25

and possible retrotransposition (Figure 5) (Carnell and Goodman, 2003). Given the sheer
number and distribution of these elements, both their movement and expression can lead
to unstable conditions within the genome. lnsertional mutagenesis, being the obvious end
result of integration can be linked to chromosomal rearrangements leading to numerous
diseases (Ostertag and Kazazian, 2001). Along with colon cancer and leukemia, other
human diseases associated with insertional mutagenesis include hemophilia A,
Duchenne’s muscular dystrophy, and Huntingdon’s disease. In addition to this, deletions
and duplications can arise from unequal crossing over and mis-pairing of homologous
sequences (Kazazian and Goodier, 2002). Furthermore, altered expression of genes
harboring the integrant have been cited (Britten, 1997).

Aberrant transposable element activity presents a clear risk to the stability and
integrity of the genome. Therefore, maintenance of these elements in a transcriptionally
silent state is essential. Methylation has been suggested as a likely source of regulation.
Hence, the mutagenicity and altered stability created by these elements might be a
product of disturbed methylation patterns. Global hypomethylation, a common feature of
tumor cells, has been associated with hypomethylation of LINE elements (Kaneda et al.,
2004). Here again, the interplay between epigenetics and mutagenesis takes shape. In
this manner, altered DNA methylation might lead to the aberrant transcriptional
activation of retrotransposons which could occur by a secondary, threshold exhibiting
mechanism (Goodman and Watson, 2002). Further research into the timing and
mechanism of altered methylation as a precursor to toxicity must then include oncogenes

and tumor suppressor genes as classic contributors, the mis-regulation of retrotransposons,

26

a, _. - -—‘ ‘8‘
.-- ~-._‘
.2 an“ ._

,...-v”‘

//“" NUCLEUS CYTOPLASM

, O O ‘x
/ \
,

L1 element \

\

/
{/1. Epigenetic Changel W\

, Original L1 Element \
I 9. integration )
" 8. Reverse Transcryition

\ O; O:

\\ 2,1, ‘MAN /
2. EM“ maximum /

 

  

 

 

 

 

 

Activity ‘\ AM" .Entry into nucleus
\4 RNA Processing ”Jr/.5.“
6 ORF1 encoded protein MIN“ ““ (f)
i" ORF2 encoded proteinsj U AAA"
L1 element 4mm 9X90" 1
1' Methyl Group MA»
I Loss or Methyl Group 5. Translation . . .
L1 Promoter Region 6. Post- translational modification

 

 

 

Figare 5 Altered Methylation and LINE-1 Elements Schematic representation of an
epigenetic change as a precursor to expression and movement of retrotransposable
elements. (1) An epigenetic change, e. g., hypomethylation of the retrotransposable
elements allows for (2) enhanced transcriptional activity. (3) RNA processing and (4)
mRN A export ensue. (5) Translation and (6) posttranslational modification precede the
formation of a ribonucleoprotein particle in which ORFl and ORF2 encoded proteins are
associated with the original mRNA. (7) Once entry into the nucleus has occurred, (8 and
9) reverse transcription and integration are achieved via the encoded reverse transcriptase
and endonuclease through a mechanism termed target primed reverse transcription
(TPRT) (Carnell and Goodman, 2003).

27

and the confounding interaction of both working together to create genomic instability

and disrupt homeostatic mechanisms.

Altered Methylation As A Precursor To Disease:

Although the intense focus of my research was on the role of DNA methylation in
cancer, it is important to put into context the power of altered DNA methylation in non-
cancer outcomes. Numerous disorders have been associated with aberrant regulation of
DNA methylation patterns. These include, but are not limited to, imprinting disorders,
repeat instability diseases, mental disorders and syndromes resulting from defects of the
methylation machinery. Imprinting is the variable phenotypic expression of a gene which
is dependent on whether it is of paternal or maternal origin. Approximately 80 genes are
known to be imprinted and loss of this programming has been linked to human diseases
(Roberston, 2005). An important regulator of imprinted gene expression is the CCCTC-
binding factor (CTCF) which regulates the ability of distant enhancers to access
promoters and has been shown to only bind to the unmethylated parental allele
(Roberston, 2005). Beckwith-Wiedemann Syndrome (BWS) is a maternally transmitted
disorder resulting in a predisposition to embryonic tumors (Robertson, 2005). BWS
arises from a loss of imprinting at two imprinting control regions. Hypermethylation of
the maternal allele is commonly seen at the first imprinting control region and loss of
DNA methylation occurs within region 2 to produce the characteristic anatomical
malformations (Robertson, 2005). Similarly, Prader-Willi syndrome results from the loss
of paternally expressed genes and is characterized by hyperphagia, obesity during

childhood and mental retardation. Other documented imprinting disorders include

28

Angelman syndrome, Albright hereditary osteodytrophy and transient neonatal diabetes
mellitus.

Repeat instability diseases arise from the expansion of trinucleotide repeats which
leads to mutation or silencing of genes (Roberston, 2005). Fragile X syndrome is the
most common example and is a form of inherited mental retardation. Normally, the
fragile X mental retardation-1 (F MR1) gene contains a highly polymorphic CGG repeat,
between 6 and 52 repeats within its 5’-untranslated region of exon 1. In fragile X
patients, the copy number of the repeats increase dramatically to 200-600 which is also
accompanied by de novo methylation and histone deacetylation of the CpG island
upstream of the gene. This aberrant methylation and histone modification result in
silencing of the gene (Robertson and Wolffe, 2000). Aberranthypomethylation is the
causal factor in facioscapulohumeral muscular dystrophy (F SHD). Affected individuals
show repeat contraction 11-150 copies down to 1-10. This contraction results in loss of
methylation and the aberrant expression of proximal genes (Robertson, 2005).

Altered methylation might be a contributing factor in neurological and
developmental disorders. Mutations in the DNMT3b gene which codes for a de novo
methylase, leads to ICF or immunodeficiency, centromeric instability and facial
anomalies (Jones and Baylin, 2002). Here brain development is disrupted when
methylation patterns are not maintained due to improper gene expression patterns and
chromosomal structure. Aberrant methylation resulting from folate deficiency or vitamin
B12 deficiency is important in regards to neuropsychiatric symptoms. Methylation
changes seen in the CNS might be linked to low levels of SAM (Singh et al., 2003). In

light of this, SAM supplementation has been used as an antidepressant (Bottiglieri et al.,

29

1994). Alterations in the folate and methyl-group metabolism including increased
methionine adenosyltransferase activity, increased incidence of methyl-tetrahydrofolate
reductase mutations and increased homocysteine levels have been reported in patients
with schizophrenia. All these factors indicate that schizophrenia might in part arise from
deficiencies in methylation and therefore altered methylation which would begin to
explain the various nonmendelian irregularities of schizophrenia (Petronis, 2004).
Mutations in MeCP2, are the primary cause of the neurodevelopmental disorder Rett
syndrome which affects approximately 1 out of every 15,000 females worldwide and is
associated with mental retardation and autism (Van den Veyver, 2002). Missense
mutations have been identified in the MeCP2 gene which cluster in the methyl binding
domain resulting in a decreased affinity of the MeCP2 for its target methylated CpG
dinucleotide (Shahbazian and Zoghi, 2002). Similarly, mice lacking MBD2 have a
neurobehavioral phenotype (Van den Veyver, 2002).

Cardiovascular disease has also been explored for connections with altered
patterns of DNA methylation. Patients with vascular disease showed increased plasma
total homocysteine and S-adenosylhomocyteine (SAH) and lower SAM/SAH ratios.
Altered global DNA methylation status in white blood cells from male atherosclerotic
vascular patients was correlated with increases in total homocysteine and SAH providing
a cursory link between the two (Castro et al., 2003). Because of cellular proliferation and
monoclonality of some cells within atherosclerotic lesions, atherosclerosis has been
compared to benign vascular tumors (Hiltunene and Yla-Herttuala, 2003). Therefore,
changes in DNA methylation observed during atherogenesis might contribute to lesion

development in a similar way to tumor development (Hiltunene and Yla-Herttuala, 2003)

30

Interestingly, global DNA methylation, a common feature of tumor DNA, has been
measured in murine, human and rabbits with advanced atherosclerosis (Zaina er al.,
2005). Genes at least partially regulated by DNA methylation that play a role in
atherosclerosis include IFN-y, PDGF, and the human estrogen receptor (Hiltunene and
Yla—Herttuala, 2003). Studies points to the hyperrnethylation of specifically the human
estrogen receptor as an early predisposing factor (Zaina et al., 2005). Therefore, the role
of altered DNA methylation extends far beyond cancer and disrupts various homeostatic

mechanisms leading to a range of human disorders.

DNA Methylation and Promotion of Tumorigenesis: Rodent Models

Rodent models of carcinogenesis are normally the first, line of experimentation in
trying to understand the underlying molecular mechanisms so that ultimately, concepts
can be applied to prediction, diagnosis and clinical treatment of human cancer. In
addition, deducing the relative risk to humans through the use of murine models is
commonplace in reproductive and carcinogenicity testing. However, inducing malignant
transformation in human and mouse cells bears some basic differences (Hahn and
Weinberg, 2002). These differences should not diminish the value of mouse models but
serve as a cautionary reminder of the need for critical assessment of the significance of
the results. Both the murine liver tumorigenesis model and SENCAR 2-stage mouse skin
model were employed in testing the potential of three very different compounds to act as

tumor promoting agents capable of disrupting DNA methylation patterns.

31

Phenobarbital and Liver Tumorigenesis

Differences in the susceptibility of strains and stocks of mice to develop liver
tumors is extremely valuable in that it allows for the comparison of molecular events
occurring in highly susceptible and more resistant mice which might shed light on
specific critical events involved in tumorigenesis. Theoretically, use of these models
could provide insight regarding the basis for variable levels of human susceptibility to the
formation of cancer. In addition, the liver is the primary target site of carcinogenesis for
more than 200 chemicals as identified by the National Toxicology Program data
(Haseman et al., 1984). B6C3F 1 are particularly sensitive to the formation of liver
tumors as they are derived from a cross between the relatively resistant maternal strain,
C57BL/6J, and the highly susceptible paternal strain, C3H/He. The incidence of
spontaneous liver tumors in B6C3F1 mice over an 18month period was reported to be
29%, with 1-2 tumors per mouse (Becker, 1982). This relative sensitivity seems to be
enhanced during Chemically induced hepatocarcinogenesis. B6C3F 1 mice administered
0.05% PB in their drinking water for 18 months exhibited a 100% liver tumor incidence
with 3-8 tumors observed per mouse. (Becker, 1982). Owing to the high sensitivity of
these mice, promoting agents have been tested alone or following exposure to an
initiating agent. The fact that tumor formation is induced in mice only exposed to a
promoting agent suggests that a few “spontaneously” initiated hepatocytes are present in
these mice. In addition, due to the heritability of changes in methylation, epigenetic
initiation is also possible (Goodman and Watson, 2002). Initiation and promotion events
in the mouse liver tumor model correlate to the formation of altered hepatic foci. These

foci represent the clonal expansion of a single initiated cell in response to a tumor

32

promoting agent and supports the framework of multi-stage carcinogenesis (Klaunig et
al., 1990).

The sensitivity of the B6C3F1 mouse to liver tumorigenesis is a good
experimental model for assessing mechanisms of carcinogenicity, however, in using the
relative resistant C57BL/6 mouse strain as a parallel experiment additional value is added
to the results obtained. C57BL/6 mice are considered relatively resistant to liver
tumorigenesis because they rarely develop spontaneous liver tumors even when exposed
to the promoting agent PB (Becker, 1982). Emphasis has been placed on discerning the
genetic and molecular differences between these mice concerning their relative
sensitivities to hepatocarcinogenesis. The hepatocarcinogen sensitive locus (hcs) has
been implicated in the variable sensitivities of these mice (Drinkwater et al., 1989). The
intermediate susceptibility of B6C3F] are due to the semi-dominant alleles of the highly
susceptible C3H/He and resistant C5 7BL/6 parents of origin. This locus appears to affect
the grth rate of pre-neoplastic foci during promotion (Manenti et al., 1994). The
grth and development of preneoplastic foci in response to initiation with N,N-
diethylnitrosamine (DEN) and promotion with PB was mouse strain-dependent (C3H/He
< B6C3F1 < C57BL/6) (Goldsworthy and Fransson-Stern, 2002). The time to conversion
of foci to masses also correlated with strain susceptibility (Goldsworthy and Fransson-
Stem, 2002).

The majority of cellular and molecular differences between the tumor-prone and
tumor-resistant mice are generally seen during the promoting stage of tumorigenesis. For
example, initiation of C3H/He and C57BL/6 mice with DEN did not reveal any

differences in the persistence of hepatic DNA adducts or DNA repair (Drinkwater and

33

Ginsler, 1986). However, differences in the rate and induction of apoptosis during
promotion, once thought to play a major role in strain susceptibilities, has recently been
contested. Hepatocarcinogenesis was induced by a single dose of DEN followed by
promotion with PB for 90 wks. Growth rates of preneoplastic foci and tumors were
largely determined by the relative rates of cell proliferation in C3H/He and C57BL/6
mice. Importantly, apoptotic activity in preneoplastic foci was low in both mouse strains
and appears to play only a minor role in susceptibility differences (Bursch et al., 2005a).
In support of this, a follow-up study showed that apoptotic activity in both strains is
comparable and does not increase following cessation of the promotion stimulus (i. e. PB)
(Bursch et al., 2005b). Therefore, the focus for susceptibility differences remains on
rates of cell proliferation. In line with this theory, parenchymal cells were tested for
responsiveness to signals inducing replication or apoptosis. Hepatocytes from C57BL/6
mice possess a low basal rate of DNA synthesis and low inducibility by epidermal growth
factor, but a higher sensitivity to induction of apoptosis by TGF-Bl than hepatocytes of
the C3H/He strain (Parzefall et al., 2002). in addition, based on work with chimeric mice,
Lee (1991) provided convincing evidence that the susceptibility differences lie within the
hepatocytes themselves and not within the micro-environment. This indicates that
grth potential and possibly clues to susceptibility are manifested at the cellular and
molecular level.

Phenobarbital, as demonstrated, has somewhat served as the compound of choice
for the investigation of multistage carcinogenesis in the context of liver tumorigenesis in
mice. Fundamentally, PB is used as a sedative and anticonvulsant and is known for its

ability to induce expression of P450 genes, specifically CYB2B1 (Whysner, 1996). In

34

addition PB has been shown to block gap junctional intracellular communication (GJIC),
a frequent finding in cancer cells, with interesting strain specific differences (Ito et al.,
1998; Warner et al., 2003; Trosko and Chang, 2001). However, the most relevant finding
to my research involving PB has been the observed differences of strains and stocks of
mice to maintain patterns of methylation (Ray et al., 1994; Counts et al., 1996; Watson
and Goodman, 2002). Previous studies in our lab have shown that PB induces a greater
extent of global hypomethylation at 1, 2, and 4 weeks in the B6C3F 1 mice as compared
to C57BL/6 (Counts et al., 1996). Critical to this finding was the fact that cell
proliferation was enhanced to a greater extent in the C57BL/6 mice as compared to the
B6C3F 1 mice. Therefore, global hypomethylation levels in the B6C3F 1 mice can not be
solely attributed to decreased fidelity of the maintenance methyltransferase in the face of
PB-induced increased cellular proliferation (Counts et al., 1996). Refined analysis of
changes in methylation has revealed that GC-rich hyperrnethylation exists concurrently
with global hypomethylation (Jones and Laird, 1999). This was demonstrated using a
global approach to specifically measuring Changes in GC-rich regions of DNA which are
normally associated with the promoter regions of genes. A 2wk exposure of PB induced
a greater degree of altered methylation, specifically hypermethylation, in GC-rich regions
in B6C3F1 mice than C57BL/6 mice (Watson and Goodman, 2002).

Hypomethylation of the promoter regions of oncogenes has also been a result of
PB promotion. Increased levels of raf and Ha-ras expression in liver were observed
following a 2wk promoting dose of PB. Increased expression of raf was associated with
hypomethylation and was only observed in B6C3F1 (Ray et al., 1994). With a lower

dose of PB, only increases in Ha-ras mRNA were observed indicating that lower doses of

35

PB promote hepatocytes with increased Ha—ras expression while higher doses select for
cells exhibiting increased expression of both Ha—ras and raf (Counts et al., 1997).
Collectively, these studies indicate a reduced capacity of the B6C3F 1 mouse to maintain
hepatic patterns of methylation and hence shows enhanced sensitivity to tumorigenesis.
However, there is still a need to refine and Specifically identify which changes in
methylation are in common or are different between B6C3F1 and C57BL/6 mice in order
to assess more accurately the importance of increases and decreases in methylation. In
addition, the specificity of PB to alter methylation in its target tissue needs to be tested.
Thus I have extended previous studies by employing an advanced and highly sensitive
technique for measuring increases, decreases, and new methylations in GC-rich regions
in response to 2 or 4 wk, 0.05% dose of PB in B6C3F1 and C57BL/6 mice. In addition
transposable elements as well as Ha-ras were examined for altered methylation in
response to PB promotion. Chapter Two focuses on the design and experimental results
of this study.
SENCAR Mouse Skin Initiation-Promotion Model

One of the best defined experimental in vivo models for epithelial cancer
development is the chemically induced tumor model of mouse skin. The outbred
SENCAR (acronym for SENsitive to skin CARcinogenesis) mouse was developed in the
1960’s and 1970’s when mice, sensitive to papilloma formation in response to
administration of initiating and promoting agents were selected for via breeding (Stern
and Conti, 1996; Boutwell, 1964). These mice also have a rather high spontaneous
incidence of tumors (965.6% and 369.4%) of which the majority are papillary tumors of

the lung (Melchionne et al., 1986). However, it is the sensitivity to initiating and

36

promoting agents which is highly advantageous to discerning the role of altered DNA
methylation in tumorigenesis during defined stages of carcinogenesis, particularly
initiation and promotion Treatment of SENCAR mouse skin with a single application of
an initiating agent followed by repeated application of a promoter results in the formation
of benign papillomas and malignant carcinomas. Initiation by 7, 12-
dimethylbenz[oi]anthracene (DMBA) and promotion with 12-O-tetradecanoylphorbol 13-
acetate (TPA) is a standard regimen for a consistent response to the induction of skin
tumors (Hennings et al., 1997; Coghlan et al., 2000). DMBA is a polycyclic aromatic
hydrocarbon that requires metabolism by the mouse epidermal aryl hydrocarbon
hydroxylase enzyme system (AHH) of which the highest activity is found in the
epidermal layer of mouse Skin (DiGiovanni, 1992). DMBA is genotoxic and it is known
to bind extensively to DNA creating DNA adducts (DiGiovanni, 1992). In addition, an
important part of initiation is thought to be associated with somatic mutation of c-Ha—ras
as a high percentage of DMBA-induced mouse skin carcinomas and papillomas have
activated c-Ha—ras (Balmain et al., 1984).

In using this model, the timing and appearance of regions of altered methylation
in both GC-rich regions and in gene-specific promoter regions in response to skin
promotion can be studied in addition to altered patterns of methylation in tumor tissue.
Altered patterns of methylation have been identified in response to initiation with DMBA
and promotion with cigarette smoke condensate (CSC). Both increases and, less
frequently, decreases were observed (Watson et al., 2003). Gene promoter
hyperrnethylation was also observed for the tumor suppressors p16, MGMT, and HoxAS

(Watson et al., 2004). Importantly, repression of HoxAS was linked to an increased

37

amount of methylation in the HoxAS promoter region. Methylation of the promoter
region was reversible and correlated to restoration of normal expression. The
fundamental theory of multistage carcinogenesis during the promotion stage is that
heritable, “critical” changes in the genome progressively accumulate, but are reversible
upon the cessation of the promoting agent. A clear demonstration of progressive, non-
random Changes in methylation was accomplished by combining the novel technique to
measure GC-rich changes in methylation (arbitrarily primed PCR and capillary
electrophoresis) with the multistage model of carcinogenesis. In doing so I was able to
track changes in methylation from very early time points through to neoplastic stages.

Details are provided in Chapter 3.

A Novel Approach to Measuring Global Changes in DNA Methylation

My experimental results concerning altered methylation in GC-Rich regions have
been obtained via methylation sensitive restriction digestion, arbitrarily primed PCR and
capillary electrophoretic separation of PCR products. Development of this method has
provided a more refined and quantitative approach to assessing altered methylation in
response to treatment. Experimental details of this methodology are found in Chapters 1,
2 and 3. However, outlining the advantages over earlier versions of this technique in
addition to explaining the theoretical “proof of concept” is necessary in light of its central
importance to my experimental findings.

The precursor version of this technique as described in Watson and Goodman,
2002 and Watson et al., 2003, was successfully employed to measure changes in GC-

rich regions. The technique involved PCR of digested genomic DNA in which PCR

38

products were radiolabeled with a-P33-dATP nucleotides. Therefore, the higher the
adenine content of the PCR product, the more label that was incorporated. PCR products
were separated on a 6% polyacrylamide sequencing gel. Following electrophoresis the
gel was stored with a phosphoimage screen and exposure of that screen was detected with
the phosphorimager. The intensities of each separate band were determined using the
NIH image analysis program and statistical differences between band intensities
identified changes in methylation when comparing control and treated samples.

Numerous limitations of this method have been overcome with the optimization
of capillary electrophoretic separation of PCR products. This newly developed method
provides multiple advantages. One of the most valuable features is the quantitative data
that are produced. The ability to perform statistical calculations adds credibility and
confidence to the results. A higher level of reproducibility was gained in addition to a 10
fold expansion of the number of analyzable PCR products. The volume of information
obtained from a single experiment created an in depth means of answering questions
concerning treatment related disruption of control methylation patterns in various model
systems and organs.

The “proof of concept” for this methodology was based on the well documented
observation that the majority of 5-methyl cytosine occurs at cytosines which are 5’ to
guanine. Both MspI and HpaII restriction enzymes recognize the 5’ CCGG 3’ sequence
and out between the internal cytosine and guanine. Hpall will not cut the DNA if the
internal cytosine is methylated, however MspI will. Therefore, based on the above
statement, MspI should digest the DNA more thoroughly than HpaII. In order to test this,

a sample of control DNA from 3 animals was digested with MspI and HpaII separately.

39

PCR was performed on these digested samples. An average peak area of the three
animals from each digest was calculated for each PCR product size formed. The average
peak area of each Mspl PCR product was compared to the area peak area of the
corresponding HpaII PCR product (i. e. the PCR products were of the same size in base
pairs). Therefore, the amount of PCR product of a particular size formed following
digestion with Mspl was compared to the amount of PCR product formed of the same
size following digestion with HpaIl. The Mspl digest was calculated as a percent of the
HpaII digest. With this calculation, all negative numbers indicate more restriction by
Mspl whereas all positive numbers indicate less restriction by Mspl. Three control
mouse liver DNA samples from 3 animals and 3 DNA samples from mice treated with
0.05% PB for 2wks were used in two separate tests (Figures 6, and 7). Importantly,
comparison of both control and treated samples from each digest shows a greater amount
of restriction by Mspl (i. e. the majority of calculated values fell below the x-axis)
indicating that the method is credible. Fundamentally, capillary electrophoresis as a
method of detection of PCR product following arbitrarily primed PCR shows conceptual
validity.

With this “proof of concept” in hand, control patterns of methylation could be
compared to treatment-induced altered methylation patterns. Four types of altered
methylation can be detected: 1) complete hypomethylation, a 100% loss of methylation
within a region of DNA 2) partial hypomethylation, a statistically significant decrease as
compared to control 3) hypermethylation, a statistically significant increase as compared
to control and 4) new methylation, a gain of methylation within a completely

unmethylated region following treatment. The simultaneous detection of increases,

40

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

I

l ’ - l

80 L . Indicates less restriction by Mspl DJ

i 0 Indicates more restriction by Mspl ;

I l

-._-L_.L_L_---_L_LL_ i

60 if i

‘63 4o - I

0 .

.9 l

o 20 —. — ___, ______-_ ~~W w he fl WA,

To l

£- 0 r a _ 100 200 300 400 50%
“5 iii i 200 sic ml sin

a ' i

c -20 l l

0 l I

§ i

0. 4° t f i

l a i

-60 J.

0 l

'80 1 .

l l e. 3 . ., l C l

-100 ’ ‘

 

 

PCR Product Size (Base Pairs)

flgpre 6 B6C3F] CMol Animals: Peak Areas of PCR products generated
following digestion with Mspl as a Percent of those generated following the HpaII
Digest Liver DNA from 3 Control B6C3F 1 animals was digested with Mspl and HpaII
separately. PCR was performed on the digested samples and the average peak area of the
3 animals from each digest was calculated for each PCR product size. The average peak
area formed following digestion with Mspl was calculated as a percent of the average
peak area of the same PCR product formed following digestion with HpaII. All negative
numbers indicate more restriction by Mspl and all positive numbers indicate less
restriction by Mspl.

41

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
. . Indicates less restriction by Mspl
80
0 Indicates more restriction by Mspl
60
"JS
a, 40 L...__._ _,_ __ L L I”, 7 7 V L_ __ .____-L_._._._. __.__ _ , L. L La” L.L_..L “L
D)
D 20 6
(:5 100 150 200 250 300 350 400 450
a 0 l T I IJ
I ll 50 110 1150 200 250 3 00 3150 40 450
la— ,20 ._._.WW , A» —__~~_.__ _. -_- ._ __ _ ._.__._
O
H
c -40 v
0
2 0
w .60 - - ——~—-——’ -+—- A, H—- ,L i L- ~-— ~— ~~ ~‘ -— —~r-——L—-—— »-—— 'l -—-»-——-—i
O.
O
-80
.100 O-O-O~—--O—--OO<> A O~<>~~-<L~O O

 

 

PCR Product Size (Base Pairs)

Figure 7 B6C3F] PB Treated Animals: Peak Areas of PCR products generated

following digestion with Mspl as a Percent of those generated following the HpaII
Digest Liver DNA from 3 B6C3F 1 animals treated with 0.05% PB for 2wks was

digested with Mspl and HpaII separately. PCR was performed on the digested samples
and the average peak area of the 3 animals from each digest was calculated for each PCR
product size. The average peak area formed following digestion with Mspl was
calculated as a percent of the average peak area of the same PCR product formed
following digestion with HpaII. All negative numbers indicate more restriction by Mspl
and all positive numbers indicate less restriction by Mspl.

42

decreases and new methylations is a very important to assessing genome wide altered
methylation. This methodology was applied in three different model systems including
hepatocytes in culture, mouse liver and kidney and mouse skin. The data obtained were
reproducible and highly consistent across models. This methodology is further discussed

in Chapters 1, 2, and 3.

43

Hypothesis and Objectives

Research projects were designed and developed to test aspects directly and indirectly

related to my overall hypothesis. Susceptibility to carcinogenesis is related inversely to
the capacity to maintain normal DNA methylation patterns. This broad hypothesis

includes the possibility of altered methylation as a precursor to toxicity and allows for the
interpretation that susceptibility can largely denote physiological states and events that
are conducive to fostering the process of carcinogenesis. One extension of this overall
hypothesis is that altered DNA methylation is a cause and not an effect of tumorigenesis.
The following three sub-hypotheses lend strong support for this concept.

1. Changes in methylation are progressive during the promotion stage of tumorigenesis

2. gauges in methylation are non-random during the promotion stage of tumorigenesis

 

3. _C_l_r_anges in the promoter regions of genes correlate to changes in gene expression.
Diethanolamine (DEA) is an alkanolamine found in many consumer products and
is widely used in industry. In 1999, the National Toxicology Program applied DEA in
95% ethanol to the skin of mice and rats for 2yrs which led to significant increases in the
incidence and multiplicity of liver tumors in mice but not rats. In a follow up study using
B6C3F 1 mice, dermal application of DEA resulted in decreased levels of SAM, increased
levels of SAH, and a reduction in phosphocholine (McKeeman et al., 2002). All of these
factors are indicative of choline deficiency which has also shown to disrupt cellular
growth and division, induce a methyl-deficient state and cause liver tumors in B6C3F 1
mice (Zeisel, 1996; Newbeme, 1982). In addition, a methyl-deficient diet alters
methylation of DNA in target tissue (Counts et al., 1996). Given that DEA is not DNA

reactive (Knaak et al., 1997), it possibly acts by disrupting choline homeostasis to induce

44

choline deficiency and consequently alters DNA methylation patterning which facilitates
tumorigenesis. By using B6C3F] mouse hepatocytes in culture I was interested in
measuring DEA-induced changes in methylation to establish that DEA is working, at
least in part, through an epigenetic mechanism. Importantly, patterns of altered
methylation induced by choline deficiency are compared to DEA to determine the extent
of similarity between them. Similarities would support the notion that changes in
methylation are not simply a random process. In addition, treatment with phenobarbital
acted as a positive control for inducing altered patterns of methylation in mouse
hepatocytes.

The extent of altered methylation in response to PB has previously been
investigated in mice which are considered susceptible (B6C3F1) or resistant (C57BL/6)
to liver tumorigenesis (Counts et al., 1996, Watson and Goodman, 2002). The findings
strongly indicate that when compared to the relatively resistant C57BL/6 mouse, the
tumor-prone B6C3F1 mouse is less able to maintain its normal methylation patterns, and
therefore exhibits a greater degree of hypomethylation throughout the genome (i. e. global
hypomethylation) in addition to more extensive disruption of patterns of methylation in
GC-rich regions. In order to further characterize the patterns of altered methylation
induced in the B6C3F 1 and C57BL/6 mice, I examined PB-induced changes in
methylation on a region by region basis for similarities and, more importantly,
differences between them. Specifically PB-induced regions of altered methylation
unique to B6C3F1 liver were identified and hypothesized to be important for the
development of hepatocarcinogenesis. Kidney, a non-target tissue, was used to assess

the specificity of PB for altered methylation in liver. Progressive, non-random changes in

45

methylation were identified and tested for reversibility. In addition, methylation of the
promoter regions of the Ha-ras oncogene and LINE-1 elements were examined.
Activation of the Ha-ras oncogene and or LIN E-l elements via hypomethylation could
lead to disruption of cell cycle regulation and genomic instability.

The progressive accumulation of heritable changes is fundamental to the process
of carcinogenesis. Increasingly aberrant subclone populations arise from the clonal
expansion of initiated cells during promotion. The SENCAR mouse 2-stage
initiation/promotion model of carcinogenesis has been successfully used to observe the
contribution of epigenetic mechanisms to this process. Specifically, altered patterns of
methylation in GC-rich regions in addition to gene-Specific promoter regions of genes in
precancerous and cancerous tissue were measured in response to promotion with cigarette
smoke condensate (CSC). Dose and time dependent changes were observed and were
largely reversible, a hallmark of tumor promotion (Watson et al., 2003; Dragan et al.,
1993). In addition hypermethylation of three tumor suppressor genes was demonstrated
(Watson et al., 2004). I extended these initial studies to investigate more clearly the step-
wise accumulation of changes in methylation over time which then “carry forward” to
tumor tissue. These specific and non-random methylation changes could be critical to
regulating the expression of oncogenes, tumor suppressors, and possibly
retrotransposable elements. Therefore the methylation status of the Ha-ras oncogene and
the LIN E-l element were analyzed and compared to their expression. Finally,
reversibility of changes in methylation were assessed to fully evaluate the stability of

altered DNA methylation following cessation of the promoting stimulus.

46

Each of the three chapters in this dissertation focus on testing the outlined hypotheses
and objectives by addressing the following gecific aims.
A) Examination of DEA-induced altered methylation in GC-rich regions as a
contributor to the development of liver tumors in B6C3F1 mice.
1) To assess global and GC rich methylation alterations in response to
DEA, CD, and PB in B6C3F1 mice.
2) Compare changes in methylation in GC-rich regions following DEA and PB
treatment to those observed after choline deficiency.
B) Effects of phenobarbital (PB) on gene specific and GC rich regional methylation
status of hepatic DNA in tumor prone and tumor-resistant mice.
1) To determine if cancer susceptibility in mice is related to differences in the
ability to maintain normal patterns of methylation in response to PB
2) Characterize progressive changes in methylation by specifically identifying
regions of altered methylation that carry forward with time in the B6C3F] and
C57BL/6 mice.
3) Identify changes in methylation which are unique to B6C3F1 liver when
compared to altered DNA methylation observed in C57BL/6 liver or B6C3F]
kidney.
4) To determine the effects of PB on the methylation status of the promoter
regions of Ha-ras and LINE-1 elements and subsequently analyze changes in their
gene expression.
5) To test the reversibility of PB-induced altered methylation in GC-Rich

regions and gene-specific promoter regions (i.e. Ha-ras and LINE-1 elements).

47

C) Characterization of GC-rich and gene specific methylation changes in tumor and
precancerous skin tissue during the promotion stage of the 2-stage,
initiation/promotion SENCAR mouse model.
1) Evaluate the methylation status of GC-rich regions following 8wk promotion
with increasing doses (3, 9, 18, 27mg) of the promoting agent, CSC.
2) Evaluate the methylation status of GC-rich regions following 4wk and 8wk
promotion with 27mg of CSC.
3) Evaluate the methylation status of GC-rich regions in tumor tissue (29wk) and
compare to precancerous tissue.
4) Characterize progressive changes in methylation by specifically identifying
regions of altered methylation that carry forward with time from 4wk to 8wk and
from 8wk to tumor tissue (29wk)
5) Assess the reversibility of changes in methylation in GC-rich regions upon
cessation of CSC application.
6) Evaluate changes in the methylation status of the promoter regions of Ha-ras
and LINE-1 elements to changes in gene expression in both precancerous and

tumor tissue.

REFERENCES FOR INTRODUCTION, SUMMARY, AND DISCUSSION
SECTIONS ARE LISTED ON PAGES 231-241.

48

CHAPTER 1

DIETHANOLAMINE AND PHENOBARBITAL PRODUCE AN ALTERED
PATTERN OF METHYLATION IN GC-RICH REGIONS OF DNA IN B6C3F]
MOUSE HEPATOCYTES SIMILAR TO THAT RESULTING FROM CHOLINE
DEFICIENCY

This chapter represents a manuscript that was submitted to Toxicological Sciences in
November, 2005. Authors include: Bachman, Ammie N. Kamendulis, Lisa M. and
Goodman , Jay I.

49

ABSTRACT

DNA methylation is an epigenetic mechanism regulating transcription, which
when disrupted, can alter gene expression and contribute to carcinogenesis.
Diethanolamine (DEA), a non-genotoxic alkanolamine, produces liver tumors in mice.
Studies suggest DEA inhibits choline uptake and causes biochemical changes consistent
with choline deficiency (CD). Rodents fed methyl-deficient diets exhibit altered
methylation of hepatic DNA and an increase in liver tumors, e.g., CD causes liver tumors
in B6C3F1 mice. We hypothesize that DEA-induced CD leads to altered methylation
patterns which facilitates tumorigenesis. B6C3F 1 hepatocytes in primary culture were
grown in the presence of either 4.5mM DEA, 3mM Phenobarbital (PB) or CD media for
48hrs. These concentrations induced comparable increases in DNA synthesis. PB, a
nongenotoxic rodent liver carcinogen known to alter methylation in mouse liver, was
included as a positive control. Global, average, DNA methylation status was not
affected. The methylation status of GC-rich regions of DNA, which are often associated
with promoter regions, were assessed via methylation-sensitive restriction digestion and
arbitrarily primed PCR with capillary electrophoretic separation and detection of PCR
products. DEA, PB and CD treatments resulted in 54, 63, and 54 regions of altered
methylation (RAMS), respectively, and the majority were hypomethylations. A high
proportion of RAMs (72%) were identical when DEA was compared to CD. Similarly,
70% were identical between PB and CD. Altered patterns of methylation in GC-rich
regions induced by DEA and PB resemble that of CD and indicate that altered DNA
methylation is an epigenetic mechanism involved in the facilitation of mouse liver

tumorigenesis.

50

INTRODUCTION

Diethanolamine (DEA), an alkanolamine, is used in industrial applications such as
textile processing, industrial gas purification, and preparation of agricultural chemicals.
In addition, fatty acid condensates synthesized from DEA are found in numerous
consumer products such as cosmetics, soaps and detergents (Knaak et al., 1997).
Widespread human exposure to DEA prompted the National Toxicology Program (N TP)
to examine its carcinogenic potential. Denna] applications of DEA in 95% ethanol for 2
years led to significant increases in the incidence and multiplicity of liver tumors in male
and female B6C3F 1 mice, but not F344 rats. Recently, DEA- induced increases in liver
cell proliferation were observed in vitro. Importantly, this effect was specific to F344
rats and B6C3F1 mice and not observed with human hepatocytes (Kamendulis and
Klaunig, 2005). Based on in vitro genetic toxicity studies DEA and/or its metabolites are
not mutagenic (N TP, 1999), suggesting that it induces a tumorigenic response via a
secondary, non-genotoxic mechanism(s).

Similar in structure to ethanolamine and choline, two essential precursors for the
synthesis of phospholipids, DEA is incorporated into hepatic phospholipids, perhaps
disrupting regulation of Choline and l-carbon metabolism. Furthermore, DEA can
inhibit the uptake of choline leading to intracellular deficiency, even if there is an
adequate amount of choline in the diet (Lehman-McKeeman and Gamsky, 1999).
Deficiencies in the major dietary sources of methyl groups, specifically, choline and
methionine, lead to hepatocarcinogenesis in rodents (Poirier, 1994, Henning and
Swendseid, 1996). Choline deficiency (CD) causes hepatocyte proliferation and

apoptosis (Albright et al., 1996, Ziesel, 1996). In particular, CD in rodents, including

51

B6C3F1 mice, in the absence of known carcinogens, increases liver tumor development
(Newbeme et al., 1982, Newbeme and Rodgers, 1986.)

Diets lacking in choline and methionine result in altered levels of S-adenosyl
methionine (SAM), and S-adenosyl homocysteine (SAH). SAM is the main methyl
donor for a variety of methylation reactions including DNA methylation (Ziesel, 1996).
In effect, methyl deficiency decreases SAM and increases SAH shifting the
proportionality towards SAH which is a feedback inhibitor of DNA methyltransferases
and, therefore, the SAM/SAH ratio is a determinant of the extent of methylation
(Shivapukar and Poirier, 1983). In B6C3F1 mice, dermal application of DEA resulted in
decreased levels of SAM, increased levels of SAH, and a reduction in phosphocholine,
the intracellular storage form of choline, which are all consistent with previous reports of
biochemical changes associated with CD which leads to methyl deficiency (Lehman-
McKeeman et al., 2002). Indeed, deficiencies in methionine and choline have been
shown to lead to global, average hypomethylation of DNA in the livers of B6C3F1 mice
(Counts et al., 1996).

It has been hypothesized that alteration of the epigenome, specifically DNA
methylation is a mechanism underlying DEA-induced tumorigenesis in B6C3F 1 mouse
liver (Lehman-McKeeman and Gamsky, 1999; Kammendulis and Klaunig, 2005).
Methylation of cytosines to produce 5-methyl cytosine is a well characterized, heritable,
epigenetic mark (Feinberg, 2001). The majority of 5-methyl cytosine occurs at cytosines
5’ to guanine. These CpG dinucleotides are not evenly distributed throughout the genome
(Bird, 2002), but are concentrated in GC-rich promoter regions of genes and transposable

elements typically being located within CpG islands which are stretches of DNA, at least

52

200bp in length that possess a 50% or greater GC content and a higher proportion of CpG
dinucleotides than expected (Gardiner-Garden and F rommer, 1987). Decreases in
methylation are associated with increases in gene transcription while increases in
methylation are associated with decreases in gene transcription (Jones and Laird, 1999).

Phenobarbital (PB) is a non-genotoxic promoter of rodent liver tumors (Whysner
et al., 1996). Increased cell proliferation and altered DNA methylation are likely
involved in tumor promotion (Goodman and Watson, 2002). Following PB
administration increases in DNA synthesis occur in B6C3F1 liver, indicating enhanced
cell proliferation, as early as 1-2 weeks (Klaunig, 1993). Additionally, PB induces more
global hypomethylation in the liver tumor-prone B6C3F] mouse, as compared to the
relatively resistant C57BL/6, mouse (Counts et al., 1996). A more critical look at this
has shown that PB induces hyperrnethylation in selected GC-rich regions of DNA in
addition to global hypomethylation demonstrating a non-random disruption of the
epigenome (Watson and Goodman, 2002).

Using B6C3F1 mouse hepatocytes in primary culture, we have examined GC-rich
regions of the genome for changes in methylation in response to treatment with DEA,
choline deficient media or PB. The hypothesis being tested is that DEA-induced CD
leads to altered methylation patterns which facilitate mouse liver tumorigenesis. The
effects of DEA and PB on DNA methylation status was ascertained and compared with
changes produced by CD. Specifically, we have assessed global (average) methylation
and evaluated the methylation status of GC-rich regions of the genome using an

arbitrarily primed PCR approach.

53

MATERIALS AND METHODS

Mouse Hepatocytes

Male B6C3F1 mice, 6-8 weeks old, obtained from Harlan Sprague-Dawley were housed
in a facility at Indiana University School of Medicine (IUSM) and cared for in
accordance with the University’s animal use and care guidelines. Hepatocytes were
isolated by a 2-Step in Situ collagenase perfusion (Klaunig et al., 1981), cultured, and
treated with 4.5mM DEA, 0.0898 mg/l choline, or 3mM PB for 48hrs at IUSM. Isolated
hepatocytes from each of 3 animals per dosing group were divided and cultured in two
plates. DNA was obtained from 8-10x106 cells using TRIzol Reagent, following the

manufacturer’s guidelines.

DNA Synthesis

Replicative DNA synthesis was measured according to the method of James and Roberts,
(1996). BrdU (20mM final concentration) was added to cell cultures during the last 16
hours of culture. Cells,l x 106 hepatocytes/60mm culture dish, were washed and fixed
with methanol. Incorporated BrdU was localized using an anti-BrdU antibody followed
by a peroxidase linked secondary antibody and a DAB substrate. Replicative DNA
synthesis was measured by scoring the percentage of BrdU positive nuclei in a minimum
of 1000 hepatocytes. Statistical significance was determined via a Randomized Complete

Block Design ANOVA, post-hoe test, Tukey’s, p<0.05.

SssI Global (Average) Methylation Assay
This assay allows for methylation at the 5’ position of cytosine at every unmethylated

CpG site in DNA via the enzyme SssI methylase using [Methyl-3 H] S-adenosyl

54

methionine (SAM) as the methyl donor, as described previously (Counts et al., 1996).
Global DNA methylation can be determined by the amount of 3H-methyl groups
incorporated into DNA, since there is an inverse relationship between incorporation of
radioactivity and the degree of methylation. Each DNA sample was incubated with

0.75 pg of DNA per 5 replicates with 2.25 units 8531 Methylase, 1.5 uCi [3H-methyl]
SAM and reaction buffer (10mM Tris, 120mM NaCl, 10mM EDTA, lmM DTT, pH 7.9)
to voltune. Reactions were spotted onto DE81 ion exchange filters and washed with
25ml 0.5M phosphate buffer, 2ml 70% ethanol and 2ml 100% ethanol and allowed to dry

before scintillation counting. All results are expressed as cpm/ug DNA.

Arbitrarily Primed PCR and Capillary Electrophoresis

We have developed an arbitrarily primed PCR procedure (AP-PCR) that provides a
thorough, overall evaluation of the methylation status of GC-rich regions of DNA.
Importantly, a comparison of data obtained from DNA isolated from control and treated
tissue permits the simultaneous detection of treatment-related increased methylation
(hyperrnethylation, more methylation in a region that was methylated in control),
decreased methylation (hypomethylation, less methylation in a region that was
methylated in control) and new methylations (methylation in regions that were not
methylated in control). Therefore, an in depth picture of treatment related altered
methylation is provided. This technique employs methylation sensitive restriction
digestion, arbitrarily primed PCR and capillary electrophoretic separation of PCR

products.

55

Restriction Digests:

DNA samples are subjected to double digests with restriction enzymes: a) a methylation
insensitive enzyme, and b) a methylation sensitive enzyme. RsaI is the methylation
insensitive enzyme which is used initially to cut DNA into fragments in order to facilitate
complete digestion by the second enzyme, a methylation-sensitive restriction enzyme.
The methylation sensitive enzymes used in this study were MspI and HpaII. Both
recognize 5’CCGG 3’ sites, and cut between the cytosine and guanine. Mspl will not
restrict DNA if the external cytosine is methylated, while HpaII will not restrict DNA if
the internal cytosine is methylated. Both Rsal/Mspl and RsaI/HpaII double digests were

employed.

Arbitrarily Primed PCR (AP-PCR) and Capillary Electrophoresis:

PCR is performed on restriction digests using a single arbitrary primer 5’
AACCCTCACCCTAACCCCGG 3’ (Gonzalgo et al., 1997), that was modified by
having it fluorescently labeled at the 5’ end with HEXTM (purchased from Integrated
DNA Technologies). This primer was designed to bind well to GC-rich regions and the
5’CCGG 3’ sequence at its 3’ end increases the probability of primer annealing to the
HpaII and Mspl restriction Site. This allows for detection of methylation at the site of
primer annealing and between sites of primer annealing. Each PCR product is viewed as
representing a GC-rich region of the genome. PCR products were purified, using a
sephadex G50 superfine matrix, and separated via capillary electrophoresis, using a ABI
3700 Genetic Analyzer (Genomics Technology Support Facility (GTSF) at Michigan
State University). Base pair markers are run Simultaneously with the samples in order to

accurately size the PCR products. The results represented as size of PCR products, in

56

base pairs, and their corresponding peak areas are analyzed using the Excel® program. A
consensus, average, peak area for each PCR product reporting in control and treated
groups is prepared, and the consensus control and treated peak areas at a specific PCR
product are compared. This permits us to detect treatment-related: a) hypomethylations
which include both 100% decreases and decreases which are statistically significant when
compared to control, b) hypermethylations which are increases which are statistically
significant when compared to control, and c) new methylations which are indicated by
the formation of a PCR product following treatment which was not formed under control
conditions. Significance is determined via a Student’s t-test, p < 0.05. Analysis of the
data includes the following assumptions: 1) each separate PCR product of a defined size
represents a distinct region of the genome, 2) a region can include one or more
recognition sequences for the specific methylation-sensitive restriction enzyme employed
located between the annealing sites of the up- and down-stream primers; thus, the amount
of each PCR product formed can be viewed as representing an “average” of the
methylation status of the particular recognition sequences located between the up- and
down-stream primers, and 3) changes in the amount of each PCR product represents the
altered methylation status of a particular GC-rich region of DNA. A detailed account of
the AP-PCR, capillary electrophoresis method, including the data analysis steps are

provided as supplementary data in Appendix 1.

57

RESULTS

In order to provide an equivalent baseline from which we could compare the
effects of DEA, PB or CD on the methylation status of DNA in B6C3F1 hepatocytes, we
selected concentrations (4.5 mM, 3 mM and 0.098 mg/l for DEA, PB and CD media,
respectively) that produced equivalent increases in DNA synthesis during the 48 hr
culture period (Table 1).

DEA or PB treatment as well as culture in CD media did not affect global,
average methylation status (Figure 1).

Analysis of GC-rich regions of DNA provided a more detailed picture of altered
methylation patterns than simply evaluating global, average methylation. DEA treatment
resulted in 43 regions of hypomethylation, which composed 80% of the total aberrant
regions detected within GC-rich areas of DNA (Figure 2a). Of these, 26 (60%) exhibited
a 100% decrease (i.e. a complete loss of methylation) at those regions. The large degree
of significant decreases in methylation (both partial and complete hypomethylation) was
approximately equal in number at both the external and internal cytosine of 5’-CCGG-3’
regions based upon the results of the RsaI/Mspl and RsaI/HpaII digests. In comparison,
relatively few regions of methylation increased with only 1 hypermethylation and 10 new
regions of methylation (Figure 2a and 2d). Here increases were mainly detected via the
RsaI/HpaII digest indicating a preference for altered methylation of the internal cytosine
within the recognition sequence.

PB produced a pattern of altered methylation similar to DEA. The largest
proportion of altered regions, 75%, were hypomethylations (Figure 2b) with 49% of the

total decreases exhibiting a complete loss of methylation. Increases in methylation

58

Table 1. Summary of Replicative DNA Synthesis

 

 

 

 

 

Treatment Labeling Indexa’b
Control 1.84i0.09 ,
Diethanolamine 7.51:l:0.21c
Choline Deficiency 7.14i0.17c
Phenobarbital 7.6liO.24°

 

 

 

 

" Labeling Index: percentage of BrdU positive nuclei in a minimum of 1000 hepatocytes
b Labeling index is expressed as mean (n=3) percent i standard error

‘ Statistically different from control. Statistical significance was determined via ANOVA,
Tukey’s p<0.05

59

2CKNMDO

18CKXWD ~

1(30CN10

14(XNDO

12CKXND ~

Cpm/ug DNA
8 8
8 8
O O

ENDOIKJ

4(KMDO

2(NJOfla

 

 

109851
126083

 

 

 

 

"*”’ ei673"*’”7“_*”
114974

1

 

 

 

 

 

 

 

 

 

 

 

 

79499
115815

“F

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Control 4.5mM DEA 3mM PB Choline-Deficient

Figure 1: Global Methylation Status in DNA from Primagy Mouse Hepatocytes.
Global methylation of DNA isolated from primary mouse hepatocytes treated for 48hrs.

with 4.5mM DEA, choline deficient media or 3mM phenobarbital is presented. Each bar
represents the mean CPM/ug DNA of 3 animals, +/- standard error. DEA, choline
deficiency, and phenobarbital treatment were statistically (p<0.05) no different from

control.

60

included 3 regions of hypermethylation and 13 regions of new methylation (Figure 2b
and 20). Similar to the results obtained from DEA treatment, there was a bias towards
increased methylation at the internal cytosine within the 5’CCGG 3’ recognition
sequence.

DNA isolated from hepatocytes maintained in CD media exhibited the greatest
number of regions where 5’Me-C content was either partially or completely decreased
(Figure 2C). Methylation was lost completely in 37 of the 49 (76%) total hypomethylated
regions. Very few increases in methylation were observed; 1 site of hyperrnethylation
was identified via the RsaI/HpaII digest and 4 regions of new methylation were identified
via the RsaI/MspI digest (Figure 2c and 2d) indicating that of the small number of
increases, most occurred at the external cytosine in contrast to increases induced by DEA
or PB which occurred mainly at the internal cytosine. The predominate alteration in
methylation patterns was a decrease in methylation at multiple regions within GC-rich
regions. PB produced the greatest degree of altered methylation with 63 total altered
regions. DEA and CD treatment were strikingly similar with 54 total altered regions
(Table 2).

Due to the overall similarity in patterns of altered methylation among the different
treatments, a more refined approach to analyzing and comparing the data was employed.
Changes occurring at identical PCR product sizes between two treatments were
considered common regions of altered methylation. Figure 3 depicts the 39 regions of
altered methylation in common between DEA and CD treatments. The magnitudes of
change at only 2 regions of the 39 total regions were statistically different (Figure 3 and

Table 3). Of the 44 common regions of altered methylation between PB and CD

61

Figure 2. GC-Rich DNA Methylation Status in Primagy B6C3F1 Mouse Hepatocytes
RsaI/HpaII (closed symbols) and RsaI/MspI (open symbols) digestion and subsequent
AP-PCR was performed on DNA isolated from B6C3F1 mouse hepatocytes treated with
either DEA (A), phenobarbital (B), or choline deficient media (C) for 48hrs. Regions of
hypomethylation were prevalent across all treatments. (D) Regions of new methylation
resulting from treatment are shown in terms of the peak area for each PCR product size.

F our regions of new methylation whose peak areas exceeded the scale of the chart were
labeled above the chart with their corresponding peak area values.

Tables tallying the regions of altered methylation for each treatment are shown as an inset
in each Chart. Regions of hypo-, hyper-, and new methylation determined by the data are
expressed in terms of the treated mean for each PCR product size as a percent of the
control mean for each PCR product size. All changes projecting below the x-axis
represent decreases in methylation (hypomethylation) while all those above the x-axis
represent increases in methylation (hyperrnethylation). All 100% hypomethylations are
considered to be significant, and only the hyperrnethylations and partial

hypomethylations that were statistically significantly different from control values
(Student’s t-test, p<0.05) are depicted.

62

A. Diethanolamine

 

Sites of HypoM Sites of HyperM
19 1
24 0
43

% Consensus Control

 

PCR Product Size (bp)

B. Phenobarbital

1500
Sites of HypoM Sites of
19 3
28
mo
47
a
c
8
”mo
:3
In
a
e
to
c
8 no
32

€10

-1(I10

PCR Product Size (bp)

63

  

Figure 2 (cont’d)

C. Choline Deficiency

810
Sites of HypoM Sites of HyperM

(no Hpall ' 25
Mspl D 24
‘10 Total 49

I
I
I
I

 

210

" l
l
70o

00

47:10

“la Consensus Control

4.10

610

810

 

41110
PCR Product Size (bp)

D. Sites of New Methylation

0 106092

’3 96372
101618' A88078

1 8000 . .
Digest Sites of New
DEA Hpall . 7
Mspl

3
14000 , j - . IPB Hpall 9
Mspl 4
0
4

 

16000

 

12000 Choline Def Hpall
Mspl
1 0000

8000

Peak Area

6000

4000

2000

 

O
100 200 300 400 500 600 700

PCR Product Size (bp)

64

Table 2. Summary of GC-Rich Regions of Altered Methylation (RAMS)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

, RAMs RAMs RAMs “New”
Treatment DIgCSt Hypomethylation' Hypermethylationb Methylationc TOTAL

Diethanolamine HpaII l 9 l 7
Mspl 24 0 3

Totald 43 1 10 54
Phenobarbital Hpall l 9 3 9
Mspl 28 0 4

Totald 47 3 13 63
0533.13; Hpall 25 1 0
Mspl 24 0 4

Totald 49 1 4 54

 

" Hypomethylated RAMS include both statistically significant (p<0.05) decreases and

100% decreases.

b Hyperrnethylated RAMS are only those increases which are statistically significant

(p<0.05).

cNew methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer

annealing which was not formed under control conditions.

dTotal RAMS including hypomethylations, hypermethylation, and new methylations for
the combined digests are reported for each treatment

65

 

Fiflre 3. Comparison of Diethanolamine and Choline deficiency Induced Aberrant
GC-rich Methylation Patterns Regions of altered methylation induced by DEA and
choline deficiency are compared for the RsaI/HpaII (A) and Rsal/Mspl (B) digestion.
PCR products of identical size formed in both the control and treatment groups were
considered to be common regions of altered methylation. These common regions of
hypo- hyper- and new methylations are represented. For the majority of common regions
of aberrant methylation, the magnitude and direction of change induced by DEA and
choline deficiency were statistically no different (One-Way ANOVA, p<0.05). At only
two common regions identified by the RsaI/HpaII digest, were the magnitudes of change
statistically different. In each case, choline deficiency induced a greater loss of
methylation than DEA.

All changes projecting below the x-axis represent decreases in methylation
(hypomethylation) while all those above the x-axis represent increases in methylation
(hyperrnethylation). All 100% hypomethylations are considered to be significant, and
only the hyperrnethylations and partial hypomethylations that were statistically
significantly different from control values (Student’s t-test, p<0.05) are depicted

66

A. Hpall Digest

% Consensus Control

% Consensus Control

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100.0 3
o Diethanolamine I
800 I Choline Deficiency II
600 n DEA and Choline Deficiency _
U Change Statistically Different
40.0
20.0
0.0 1 ' . 30 , 4 , 500 60
190 150 zIo 50 300 350 400 450 500 £50
-2o.o
40.0 44% ~— _ t.“ ._... _ L L LLL L L. LL
430.0 —< j <>
it
4300 . — .——— iii—“I60 _ .L_ _ ..... L L L L _L
.1000 “Mk—Ll L A L LL“.-.L..L L
U 0
PCR Product Size (bp)
B. Mspl Digest
0 106092
508070
1WD ;/ m
: o Diethanolamine
00.0 - —— -- —— _— —-~A -—- ~~ 16000
: I Choline Deficiency
60.0 — : ..—__-_ l DEA and Choline Deficiency 12000 g
I ......... New Methylation I'
40.0 : I °°°° E
l 9 in
I I i
20.0 e——-—-— -—— ————-— — , 1 4 -_ ____ .L.4ooo
9 l :
0.0 1 0 zoo so ' ' 490 00 e l 700 o
100 260 400 obi: 790
.200
. I . I
.4004 ~— .L,
j l. l. I
«60.0 A
-eo.o - -— —— —~ — 4
'100-0 I I: m I III I I
PCR Product Size (bp)

67

treatments, only 5 regions differed statistically in magnitude (Figure 4, and Table 3). The
patterns of altered methylation produced by DEA and PB were 72% and 70% similar,
respectively, to that of CD demonstrating the high degree of similarity (Table 3). Unique

changes elicited by DEA and PB were few in numbers (Table 4).

68

Table 3. Common Regions of Altered Methylation (RAMS): Comparison of
Diethanolamine (DEA) or Phenobarbital (PB) with Choline Deficient (Choline Def)

Treatment“

 

Common RAMs Common RAMs

 

 

 

 

DEA vs. Choline With Decreased With Increased Totcal RAMs In SPerTent
mm It
Def Methylationb Methylationc O o m“ arIty
Com lete
p . d 25 -
Hypomethylation
Partial 9
Hypomethylatione - - 0
, 3 9/54 72 /0
New Methylation - 3
Magnitude of Change . 2

Statistically Differenth"

 

 

Common RAMs Common RAMs

 

 

PB vs Choline Def With Decreased With Increased Total RAMS In 1,63th
' . b . c Common Similarity
Methylation Methylation
Corn lete
p . d 23 -
Hypomethylation
Partial
12 -

Hypomethylatione

 

New Methylationf - 3 44/ 6 3j 7 0%

 

Hypermethylationg - 1

 

 

Magnitude of Change .

Statistically Differenth" 5 -

 

 

 

 

 

 

" Data are summarized from Figures 2 and 3.

b Total RAMs exhibiting decreased methylation (i.e. complete hypomethylations and
partial hypomethylations) that are in common between DEA and Choline Def or PB and
Choline Def treatments

6 Total RAMS exhibiting increased methylation (i.e. hypermethylations and new
methylations) that are in common between DEA and Choline Def or PB and Choline Def.
“Complete hypomethylation indicates a complete or 100% loss of methylation.

‘ Partial Hypomethylations are statistically significant (p<0.05) decreases as compared to
control

f New methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not formed under control conditions.

3 Hyperrnethylations are statistically significant (p<0.05) increases as compared to
control.

h Significance was based on a One Way ANOVA, p<0.05.

l'Choline Def showed a significantly greater extent of hypomethylation than DEA and PB
treatment at the number of RAM indicated.

j Total RAMs are reported from Table 2.

69

Figpre 4. Comparison of Phenobarbital and Choline deficiency induced aberrant
GC-rich methylation patterns Regions of altered methylation induced by PB and
choline deficiency are compared for the Rsal/HpaII (A) and RsaI/Mspl (B) digestion.
PCR products of identical size formed in both the control and treatment groups were
considered to be common regions of altered methylation. These common regions of
hypo- hyper- and new methylations are represented. For most regions, the magnitude and
direction of change induced by PB and choline deficiency were statistically no different
(One-Way ANOVA, p<0.05). The magnitudes of decrease for five common regions,
identified by the RsaI/HpaII digest, were statistically different. In each case, choline
deficiency induced a greater loss of methylation than PB.

All changes projecting below the x-axis represent decreases in methylation
(hypomethylation) while all those above the x-axis represent increases in methylation
(hyperrnethylation). All 100% hypomethylations are considered to be significant, and
only the hypermethylations and partial hypomethylations that were statistically
significantly different from control values (Student’s t-test, p<0.05) are depicted.

70

% Consensus Control

% Consensus Control

A. Hpall Digest

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100 l
a Phenobarbital
so I Choline Deficiency ‘4
60 I u PB and Choline Deficiency
I 0 Change Statistically Different
40
20«—— -—- A- —ie_-4mfie——-» “7—
o 100 . 200 300 400 500 ‘ 600
tho 150 2b0 250 300 350 400 450 500 i550 silo
.20
A
‘° 1
i i I
m t
. I
-80
40° ‘W 1:: I I II F n l“ """"
‘01] 0 U 0
PCR Product Size (bp)
B. Mspl Digest
A $372
I 88078
100.0 -— _ --- -— -——--~-» - -- 20000
d
: APhenobarbital
80° _ _ 7 ‘ 77d! E u __ I I Choline Deficiency ’ 10000
50.0 L L L ____ i— _ _, UPS and Choline Deficiency 412000
: ----- New Methylation 3
40.0 ——~- #—~ — : a — - 0000 :-
' ll
.. -= ...:
. ‘ fi‘ r— T —— —‘— " v I —— - —-<t
I— i :
0° jooL _L__. if} 3 ' 490 590 sip.-. 70 o
1 2 30C 400 5C!) 700
-20.0
40.0
A I it
00.0 A ' s
l A
-eo.0 #— —~ — —- — —
‘1‘”.0 .l “.... .... ......I- .....IML. I I mm-......_._.-.......... ...u u .—
PCR Product Size (bp)

71

Table 4. Unique“ Regions of Altered Methylation (RAMs): Diethanolamine (DEA)
or Phenobarbital (PB) as Compared to Choline Deficient Treatment

 

 

 

 

 

 

 

 

 

 

 

DEA PB
Complete Hypomethylationb 1 1
Partial Hypomethylationc 6 7
Hypermethylationd 1 2
New Methylatione 7 9
Total RAM NOT in Common 15/54f 19/63f
Percent Difference 28% 300/0

 

“ Unique RAMs denotes all RAM which were .not in common between DEA and choline
deficiency or PB and choline deficiency.

b Complete hypomethylation indicates a complete or 100% loss of methylation.

0 Partial Hypomethylations are statistically significant (p<0.05) decreases as compared to
control

“(Hypermethylations are statistically significant (p<0.05) increases as compared to
control.

e New methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not formed under control conditions.

f Total RAMS are reported from Table 2.

72

DISCUSSION

We have developed and applied a novel procedure for analyzing altered
methylation in GC-rich regions of the genome, including CpG islands. Simple in design,
this technique employs methylation sensitive restriction digestion of DNA, arbitrarily
primed PCR amplification, and electrophoretic separation of PCR products to provide a
detailed, quantitative overview of the extent of treatment-related disruption of
methylation throughout the genome. Comparably, the strength and utility of our
technique lies in its ability to simultaneously identify increases, decreases and new
methylations within multiple, distinct regions of the genome. This provides a sensitive,
quantitative method which reproducibly detects the extent of treatment-related altered
patterns of methylation.

There are a variety of techniques for analyzing changes in methylation within a
particular gene. Methylation specific PCR, including variations such as MethyLight and
HM Methyl Light can be effectively employed for these applications (Cottrell and Laird,
2003). Other procedures include combined bisulfite restriction analysis (COBRA) which
assesses the methylation status of particular CpG sites (Xiong and Laird, 1997) and the
enzymatic regional methylation assay for determining changes in methylation between
two primers designed for a targeted region (Galm et al., 2002). These are excellent
methods for evaluating specific genes. However, their utility is limited when one wants
to discern the extent to which a particular treatment might disrupt normal methylation
patterns, e. g., in this situation a gene-by-gene approach would be too cumbersome.

There are other approaches for assessing methylation status of the genome. For

example, the combination of sodium bisulfite conversion of DNA followed by PCR

73

amplification of alu and LINE elements can be used to estimate methylation changes
(Yang et al., 2004); however, the focus is only on repetitive elements, i.e., “junk DNA”
and not genes. Also, the methylation-sensitive amplified fragment length polymorphism
(AF LP) technique allows for comparative genome wide scanning of methylation status
via fingerprinting techniques and has recently been adapted to a DNA microarray
hybridization technique (Yamamoto and Yamamoto, 2004). This procedure requires a
custom microarray panel and a complex approach to data analysis. Global, average
methylation analysis via SssI methyltransferase (Balaghi and Wagner, 1993) is
straightforward, but limited in scope; increases in methylation in one portion of the
genome may balance out decreases in other areas. The combined AP-PCR capillary
electrophoresis technique described in this paper affords the ability to assess altered DNA
methylation (increases, decreases and new methylations) in multiple GC-rich regions of
the genome simultaneously and quantitatively. Furthermore, it is highly appropriate under
situations when the research question being asked is, “Does a particular treatment cause
disruption of normal patterns of DNA methylation and to what extent does this occur?”
With this methodology we have assessed DEA, CD, and PB induced alteration of
methylation in B6C3F1 mouse hepatocytes.

The ability of DEA to alter methylation in vitro in B6C3F1 mouse hepatocytes
was investigated as a proposed non-genotoxic mode of action of the compound’s ability
to cause carcinogenesis in mouse liver. Treatment-induced increases in cell proliferation,
as measured by increases in DNA synthesis, have been reported and proposed to facilitate
tumorigenesis in B6C3F1 mice (Klaunig, 1993, Zeisel, 1996). Specifically, genetic and

epigenetic alterations of the DNA are possible contributors (Counts et al., 1996). Cell

74

proliferation fosters the occurrence and accumulation of spontaneous mutations (Schulte-
Hermann, 1987). More importantly, high rates of DNA synthesis might compromise the
capacity to maintain normal methylation patterns leading to mis-regulated gene
expression patterns. Therefore, by selecting doses based on induction of comparable
increases in cell proliferation, we were able to directly compare and analyze changes in
methylation.

Several factors work in concert to sustain normal methylation levels. These
include the maintenance and de novo DNA methyltransferases (e. g. Dnmtl , Dnmt3a and
3b), demethylases and the availability of both SAM and methyl groups. For example,
Dnmtl is the maintenance methyltransferase responsible for methylating newly
synthesized daughter strands of DNA; this ensures the heritability of the methylation
pattern (Hermann et al., 2004). Altered patterns of methylation, specifically
hypomethylation, may arise when the activity of Dnmtl does not increase with enhanced
rates of DNA synthesis. Alternatively, the same effect could be observed if SAM does
not provide a sufficient supply of methyl groups, (i.e. methyl deficiency depletes the
availability of methyl groups), to maintain the up regulated Dnmtl activity. DNA
methylation patterns are also under the influence of demethylases (e. g., MBD2) which
can decrease the level of 5-methyl cytosine when cells are not synthesizing DNA (Detich
et al., 2003). Thus, indicating that DNA methylation is reversible. Importantly, SAM
directly inhibits MBD2 and, therefore, diminished formation of SAM during a state of
methyl deficiency could relieve the inhibition of demethylase activity and facilitate
hypomethylation of DNA (Detich et al., 2003). As hypothesized, DEA, by inducing

cellular choline depletion, contributes to perturbation of l-carbon metabolism, leading to

75

decreased availability of methyl groups, impaired formation of SAM, and disruption of
normal DNA methylation patterns.

Assessment of global (average) methylation status and methylation of GC-rich
regions of DNA were performed. Global, average levels of methylation following
treatment with DEA, CD media and PB were comparable to control (Figure 1). This
could indicate that, 1) global methylation levels are unaffected by DEA, CD, and PB or
2) approximately equal levels of methylation increases and decreases are occurring
simultaneously in multiple regions of the genome. This second possibility underlies the
importance of specifically examining GC—rich regions for a more detailed picture of
overall altered methylation.

Within GC-rich regions of DNA, hypomethylation was the predominant alteration
induced by DEA, PB and CD. Hypomethylation in the promoter regions of genes is
associated with increased gene expression (Jones and Baylin, 2002). Critical losses of
methylation in the promoter regions of oncogenes such as c-jun and c-myc, CDNK3
(cyclin-dependent kinase inhibitor 3), and c-Ha-ras, have been demonstrated (N iculescu
et al., 2004; Tao et al., 2000). Hypomethylation associated overexpression of c-jun and
c-myc was observed in livers promoted with dichloroacetic and trichloroacetic acid, both
of which are considered non-genotoxic carcinogens (Tao et al., 2000). Human
neuroblastoma cells cultured in CD media showed loss of methylation in the promoter
region of the CDNK3 gene, an important regulator of cell cycle progression, and up-
regulation of expression. In addition, genetic instability via activation of transposable
elements (Roman-Gomez, 2005), elevated mutation rates (Chen et al., 1998) and

chromosomal instability (Eden et al., 2003) have all been associated with

76

hypomethylated DNA. Methyl deficiency in rats induced irreversible global DNA
hypomethylation in rat liver which supported a role for loss of methylation during the
cancer initiation and or promotion stages of hepatocarcinogenesis (Pogribny et al., 2005).
These studies emphasize and support our view that DNA hypomethylation is a
mechanism involved in tumor promotion (Counts and Goodman, 1994) and the data
presented in the current paper support the hypothesis that DEA, CD and PB treatment act
by this mechanism to produce mouse liver tumors.

Altered methylation status of cytosines within the CpG dinucleotide is most
commonly investigated; however, methylation of CprG and non-CpG sites also exists.
In particular, the role of altered methylation at CpCpG sites has not been thoroughly
investigated. There are three possible states of methylation of the CpCpG sites analyzed.
These include: 1) mCpCpG, methylation of the external cytosine, 2) CmepG,
methylation of the internal cytosine and 3) memeG, methylation of both the internal
and external cytosine. Our results show that loss of methylation status at both mCpG and
"‘CpCpG sites occurs with approximately equal frequency suggesting that factors
affecting the methylation status of mCpG sites also act on mCpCpG sites. Studies
evaluating non-CpG methylation have mainly focused on CpA, CpT, and CpC
methylation. However, one particular study proposed a biological role for methylation of
both cytosines within CpCpG sites. Methylation of both cytosines within CpCpG sites
has been reported to prevent binding of Sp] , an important transcription factor, to its target
cis element thereby contributing to abnormal regulation of gene expression (Clark et al.,

1997; Inoue and Oishi, 2005). Effects due to methylation of only the external cytosine

77

were not reported. This stresses the importance of a broad and critical analysis of both
CpG and non-CpG methylation during the promotion stage of tumorigenesis.

We have demonstrated remarkable similarities between the DEA, CD, and PB
treatment related disruption of methylation patterns in B6C3F1 mouse hepatocytes grown
in vitro during a short 48 hour exposure. This indicates that a common mechanism is
shared by all three treatments. The extreme similarity between patterns of altered
methylation in GC-rich regions due to DEA and CD supports the notion that DEA
indirectly depletes the pool of methyl groups needed for methylation of cytosine by
inhibiting choline uptake into cells (Lehman-McKeeman and Gamsky, 1999). The
resulting hypomethylation mimics that of dietary CD. Dietary PB has been shown to
cause global hypomethylation (Counts et al., 1996), and hypermethylation, along with
some decreased methylation, in GC-rich regions of DNA (Watson and Goodman, 2002)
in the livers of B6C3F 1 mice afier 2 and 4 wk of administration. Therefore, continued
exposure to the promoting stimuli, may lead to progressive changes in methylation
including hypomethylations, hypermethylations, and new methylations which accrue in a
stepwise manner to contribute to tumorigenesis. This is consistent with the view that a
variety of alterations in methylation contribute to carcinogenesis (Counts and Goodman,
1995), and that there are progressive alterations of methylation during the transformation
process (Watson et al., 2003). Hence, altered methylation, initially hypomethylation, is a
likely epigenetic, non-genotoxic, mode of action underlying the abilities of DEA, PB and
CD to promote the development of mouse liver tumors.

ACKNOWLEDGEMENT

Support provided by American Chemistry Council is gratefully acknowledged.

78

REFERENCES

Albright CD, Liu R, Bethea TC, Da Costa KA, Salganik RI, Zeisel SH. (1996). Choline
deficiency induces apoptosis in SV40-immortalized CWSV-l rat hepatocytes in culture.
J. FASEB. 10, 510-6.

Balaghi, M. and Wagner, C. (1993). DNA methylation in folate deficiency: use of CpG
methylase. Biochem. Biophys. Res. Commun. 193, 1184-1190.

Bird, A. (2002). DNA methylation patterns and epigenetic memory. Genes and Dev.
16, 6-21.

Chen, R.Z., Pettersson, U., Beard, C., Jackson-Grusby, L., and Jaenisch, R. (1998).
DNA hypomethylation leads to elevated mutation rates. Nature 395, 89-93.

Clark, S.J., Harrison, J ., and Molloy, PL. (1997). Spl binding is inhibited by
(m)Cp(m)CpG methylation. Gene 195, 67-71.

Cottrell, SE. and Laird, PW. (2003). Sensitive detection of DNA methylation. Ann.
NY. Acad. Sci. 983: 120-130.

Counts, J .L. ad Goodman, J .I. (2004). Hypomethylation of DNA: An epigenetic
mechanism involved in tumor promotion. Mol. Carcinogen. 1 1: 185-188.

Counts, J .L. and Goodman, J .I. (1995). Alterations in DNA methylation may play a
variety of roles in carcinogenesis. Cell 83, 13-15.

Counts, J .L., Sarrniento, J .I., Harbison, M.L., Downing, J .C., McClain, R.M. and
Goodman, J .I. (1996). Cell proliferation and global methylation status changes in mouse
liver after phenobarbital and/or choline-devoid, methionine-deficient diet administration.
Carcinogenesis 17, 1251-1257.

Detich, N., Hamm, 8., Just, G., Knowx, J .D., and Szyf, M. (2003). The methyl donor S-
adenosylmethionine inhibits active demethylation of DNA. J. Biol. Chem. 278, 20812-
20820.

Eden, A., Gaudet, F ., Waghmare, A., and Jaenisch, R. (2003). Chromosomal instability
and tumors promoted by DNA hypomethylation. Science 300, 455.

Feinberg, AP. (2001). Cancer epigenetics takes center stage. Proc Natl Acad Sci USA
98, 392-394.

Galm, 0., Rountree, M.R., Bachman, K.E., Jair, K.W., Baylin, SB. and Herman, J .G.

(2002). Enzymatic regional methylation assay: a novel method to quantify regional CpG
methylation density. Genome Res. 12, 153-157.

79

Gardiner-Garden, M. and F rommer, M. (1987). CpG islands in vertebrate genomes. J
M01 Bio 196, 261-282.

Gonzalgo, M.L., Liang, G., Spruck, CH. 111, Zingg, J-M., Rideout, W. M. III, and Jones,
RA. (1997). Identification and characterization of differentially methylated regions of

genomic DNA by methylation-sensitive arbitrarily primed PCR. Cancer Res. 57, 594-
599.

Goodman, J .I. and Watson, RE. (2002). Altered DNA methylation: a secondary
mechanism involved in carcinogenesis. Annu. Rev. Pharmacol. T oxicol. 42, 501-525.

Henning, SM. and Swendseid, ME. (1996). The role of folate, choline, and methionine
in carcinogenesis induced by methyl-deficient diets. Adv. Exp. Med. Biol. 399, 143-155.

Hermann, A., Gowher, H., and Jeltsch, A. (2004). Biochemistry and biology of
mammalian DNA methyltransferases. Cell Mol. Life Sci. 61, 2571-2587.

Inoue, S. and Oishi, M. (2005). Effects of methylation of non-CpG sequence in the

promoter region on the expression of human synaptotagmin XI (sytl 1). Gene 348, 123—
134.

James, NH, and Roberts, RA. (1996). Species differences in response to peroxisome
proliferators correlate in vitro with induction of DNA synthesis rather than suppression of
apoptosis. Carcinogenesis 17, 1623-1632.

Jones, RA. and Baylin, SB. (2002). The fundamental role of epigenetic events in
cancer. Nature Rev. Genet. 3, 415-428.

Jones, P.A., and Laird, PW. (1999). Cancer epigenetics comes of age Nature Genet.
21, 163-167.

Kamendulis L.M. and Klaunig, J .E. (2005). Species Differences in the induction of
hepatocellular DNA synthesis by diethanolamine. Toxicol. Sci. 87, 328-336.

Klaunig, J .E. (1993). Selective induction of DNA synthesis in mouse preneoplastic and
neoplastic hepatic lesions after exposure to phenobarbital. Environ. Health Persp. 101
(Suppl. 5), 235-239.

Klaunig, J .E., Goldblatt, P.J., Hinton, D.E., Lipsky, M.M., Chacko, J ., and Trump, BR
(1981). Mouse liver cell culture. I. Hepatocyte isolation. In Vitro. 17, 913-925.

Knaak, J .B., Leung, H.-W., Stott, W.T., Busch, J., and Biski, J. (1997). Toxicology of
mono- di- and tn'ethanolamine. Rev. Environ. Contam. T oxicol. 146, 1-86.

80

Lehman-McKeeman, L.D., and Gamsky, EA. (1999). Diethanolamine inhibits choline
uptake and phosphatidylcholine synthesis in Chinese Hamster Ovary cells. Biochem.
Biophys. Res. Commun. 262, 600-604.

Lehman-McKeeman, L.D., Gamsky, E.A., Hicks, S.M., Vassallo, J .D., Mar, M.-H., and
Zeisel, SH. (2002). Diethanolamine induces hepatic choline deficiency in mice. T oxicol.
Sci. 67, 38-45.

National Toxicology Program (1999). Toxicology and carcinogenesis studies of
diethanolamine in F344/N and B6C3F1 mice (Dermal Studies). NTP TR 478. US.
Department of Health and Human Services, National Institutes of Health.

Newbeme, P.M., deCamagro, J. L.V., and Clark, A.J. (1982). Choline deficiency,
partial hepatectomy, and liver tumors in rats and mice. T oxicol. Path. 10, 95-106.

Newbeme, RM. and Rodgers, AB (1986). Labile methyl groups and the promotion of
cancer. Annu Rev Nutr. 6, 407-432.

Niculescu, M.D., Yamamuro, Y. and Zeisel, SH. (2004). Choline availability modulates
human neuroblastoma cell proliferation and alters the methylation of the promoter region
of the cyclin-dependent kinase inhibitor 3 gene. J. Neurochem. 89, 1252-1259.

Progribny, 1., et al. (2005). Irreversible global DNA hypomethylation as a key step in
hepatocarcinogenesis induced by dietary methyl deficiency. Mutat. Res. Sept 3. E-pub
ahead of print.

Poirier, LA. (1994). Methyl group deficiency in hepatocarcinogenesis. Drug Metab.
Rev. 26, 185-199.

Roman-Gomez, J. et al. (2005). Promoter hypomethylation of the LINE-1
retrotransposable elements activates sense/antisense transcription and marks the
progression of chronic myeloid leukemia. Oncogene Sept 17. E-pub ahead of print.

Schulte-Hermann, R. (1987). Initiation and promotion in hepatocarcinogenesis. Arch
Toxicol. 60, 179-181.

Shivapurkar, N. and Poirier, LA. (1983). Tissue levels of S-adenosylmethionine and S-
adenosylhomocysteine in rats fed methyl-deficient, amino acid-defined diets for one to
five weeks. Carcinogenesis 4, 1051-1057.

Tao, L., Yang, S., Xie, M., Kramer, P.M., and Pereira, MA. (2000). Hypomethylation
and overexpression of c-jun and c-myc protoncogenes and increased DNA
methyltransferase activity in dichloroacetic and trichloroacetic acid-promoted mouse
liver tumors. Cancer Lett. 158, 185-193.

81

Watson, RE and Goodman, J .I. (2002). Effects of phenobarbital on DNA methylation
in GC-rich regions of hepatic DNA from mice that exhibit different levels of
susceptibility to liver tumorigenesis. Toxicol. Sci. 68, 51-58.

Watson, R.E., Curtin, G.M., Doolittle, DJ. and Goodman, J .I. (2003). Progressive
alterations in global and GC-rich DNA methylation during tumorigenesis. Toxicol. Sci.
75, 289-299.

Whysner, J ., Ross, P.M., and Williams, GM. (1996). Phenobarbital mechanistic data
and risk assessment: Enzyme induction, enhanced cell proliferation, and tumor
promotion. Pharmacol Ther. 71, 153-191.

Xiong, Z. and Laird, PW. (1997). COBRA: A sensitive and quantitative DNA
methylation assay. Nucleic Acids Res. 25, 2532-2534.

Yamamoto, F. and Yamamoto, M. (2004). A DNA microarray-based methylation-
sensitive (MS)-AF LP hybridization method for genetic and epigenetic analyses. Mol.
Gen. Genomics 271, 678-686.

Yang, A.S., Estecio, M.R.H., Doshi, K., Kondo, Y., Tajara, E., and Issa, J-P. J. (2004). A
simple method for estimating global DNA methylation using bisulfite PCR of repetitive
DNA elements. Nucleic Acids Res. 32, e38.

Ziesel, SH. (1996). Choline: A nutrient that is involved in the regulation of cell
proliferation, cell death and transformation. Adv. Exp. Med. Biol. 399, 131-141.

82

CHAPTER 1 APPENDIX
This appendix is a detailed description of the materials and methods for the arbitrarily
primed PCR and capillary electrophoretic approach employed to assess methylation
status in GC-rich regions throughout the genome. In addition, data organization and
analysis, including statistical calculations performed using the Excel® program, are
explained in detail. This appendix can also serve as supplementary information on the

materials and methods for Chapters 2 and 3.

83

ARBITRARILY PRIMED PCR AND CAPILLARY ELECTROPHORESIS

Restriction Digests:

DNA samples, of which duplicates are prepared, are subjected to double digests with
restriction enzymes: a) a methylation insensitive enzyme, and b) a methylation sensitive
enzyme. RsaI, is the methylation insensitive enzyme which is used initially to cut DNA
into fragments in order to facilitate complete digestion by the second enzyme, a
methylation-sensitive restriction enzyme. DNA samples are subjected to double digests
with restriction enzymes: a) a methylation insensitive enzyme, and b) a methylation
sensitive enzyme. RsaI, is the methylation insensitive enzyme which is used initially to
cut DNA into fragments in order to facilitate complete digestion by the second enzyme, a
methylation-sensitive restriction enzyme. The methylation sensitive enzymes used in this
study were Mspl and Hpall. Both recognize S’CCGG 3’ sites, and out between the
cytosine and guanine. Mspl will not restrict DNA if the external cytosine is methylated,
while HpaII will not restrict DNA if the internal cytosine is methylated. Both RsaI/Mspl
and RsaI/HpaII double digests were employed.

The Mspl and Hpall restriction enzymes allow for analysis of methylation at both CpG
sites and CpCpG sites. Methylation of cytosines, not S’to guanine is less commonly
examined. Restriction digests contain lug DNA and 5.0 units RsaI in Roche Buffer L.
Samples are incubated for 1hr. at 37°C before addition of 2.5 units of either Mspl or
HpaII. A second 2.5 unit aliquot of the respective enzyme is added after an additional
2hr, and in order to insure complete digestion, total incubation time is 18hr. The
enzymes were inactivated by incubating at 650C for 10min. Samples were stored at 4°C

until analyzed.

84

Arbitrarily Primed PCR (AP-PCR):

AP-PCR is performed on restriction digests using a single arbitrary primer, 5’ AAC CCT
CAC CCT AAC CCC GG 3’ (modified from Gonzalgo et al., 1997), fluorescently
labeled at the 5’ end with HEXTM (purchased from Integrated DNA Technologies). This
primer was designed to bind well to GC-rich regions and the 5’CCGG 3’ sequence at its
3’ end increases the probability of primer annealing to the Mspl and Hpall restriction
site, allowing for the detection of methylation at the site of primer annealing and between
sites of primer annealing. Each PCR reaction was composed of 5.0ul of the restriction
digest, 0.8 uM primer, 1.0 unit Taq polymerase, 1X MasterAmpTM PCR PreMix L, and
glass distilled water (GDW) to 10ul. In order to obtain a sufficient quantity of PCR
product for capillary electrophoresis, duplicate PCR reactions were prepared for every
one restriction digest and are combined prior to purification. The Taq polymerase was
added to each reaction following a 5min incubation at 80°C. Cycling conditions were as
follows: 940C for 2 min, 5 cycles of 94°C for 30 s, 40°C for 1 min, and 72°C for 1 min
30 s, 40 cycles of 940C for 15 s, 55°C for 15 s, and 72°C for 1 min, and a single time
delay of 5 min at 72°C followed by 3 40C soak. Combined PCR reactions were desalted
and purified at the Genomics Technology Support Facility (GTSF) at Michigan State

University, using a sephadex G50 superfine matrix.
Capillary Electrophoretic Separation of Products:

Ten nanograms of each purified, desalted PCR product were added to a solution of

forrnamide and a carboxy-X-rodamine (ROXTM)-labeled size marker for sizing and

85

normalization of the results. Size marker fragments increased incrementally by 200bp up
to 1000bp. Using a 103cc injection time, a 2n] aliquot was injected into the GTSF
Applied Biosystems 3700 Genetic analyzer. Data were collected using Genescan 3.7
which compiles the results as size of PCR product in base pairs with a corresponding
peak area representative of the amount of PCR product generated. Only fragments
greater than 100 bp and peak areas with corresponding peak heights greater than 100
units were analyzed to minimize incorporating background noise and/or primer-dimers

into the data set.

Raw Data Organization

The data are organized into rows and columns within the excel workbook where a list of
all PCR product sizes (base pairs) from every analyzed sample are placed in one column.
The peak areas with their corresponding PCR product size from each sample are placed
into their corresponding row. The excel file will appear as PCR product sizes in column
A and across the rows of each size will be all the samples reporting a peak area for that
size PCR product as shown in Table 1.

Following this raw data organization, the RsaI/HpaII data are placed in a separate

workbook from RsaI/MspI data.

Removal if Baigroynd Data
The following data analysis steps are performed on each digest separately. Using the

following rules, the data are aligned across treatment groups and “background” data, (i.e.,

86

Appendix Table 1. Excel workbook illustrating the Organization of Raw Data into
PCR Product Size and Corresponding Peak Area

 

PCR Product
Size (base pairs)

Control 1

Control 1

Treatment 1

Treatment 1

 

56“

2185b

4321

 

75

3018

1466

3274

 

79 330

99 I705
108

1 14 34695 1528

126 1683
148

 

 

 

 

 

 

 

 

 

 

 

 

" PCR product sizes in base pairs are reported.

b Peak area values representing the amount of PCR product are reported for each
corresponding PCR product size.

87

reported peak areas that do not “fit” within our rules and are deemed to be spurious) are
removed from futher calculations.

A. Data that have a corresponding PCR product size of 99bp or less are omitted. For
example, in Table 1, the 4 rows containing 56, 75, 79, and 99bp would be deleted. This
is to ensure that the final data analysis does not include PCR product sizes and peak areas
that may be attributed to background fluorescence or primer dimers.

B. Minimum Data Requirement for each PCR Product Size. Each row (each PCR
product size) is examined individually within a defined control/treatment group, and LIE
pga_k area data points that do not meet the following criteria age deleted. In order for the
data to be considered valid, peak area data points must be reported for at least 50% of the
number of animals and at least 50% of the total number of replicates. For example if
there are 4 control animals, each with 2 replicates, 2 of the 4 animals and 4 of the 8
replicates must report a peak area for the given PCR product size. For odd numbers of
animals such as 5 animals with 2 replicates each, 3 animals and 5 replicates must report a
peak area.

C. Due to the possibility of occasional slight variation with regard to determining the size
of PCR products, peak area points which are within 2 base pairs are, under the

circumstances outlined below, combined to complete (or add to) a row.

The goal of these occasional adjustments is to move selected peak area data points only if

the move is one or two base pairs away and there is a clear fit with other replicates.

 

Extensive manipulation is avoided.

88

An example of this is illustrated in Table 2. With the presented data, a complete row of
peak area points must contain at least 4 peak area points because there are 4 animals (#1-
4) and 2 replicates of each animal. The numbers with strikethrough font would be
deleted and those peak area data points within 2 bp of neighboring data would be moved
to fill in the data set for that PCR product size.

Some of these data point adjustments are somewhat subjective, thus two people
organizing the same data set might end up with slightly different alignments, however,

their calculated outcomes should be virtually identical.

Alignment of the Control and Treated Data

Once data are aligned between control/treatment groups, the results of two groups,

 

experimental and control, are compared at each PCR product size. Without altering the
control data, the treatment data are aligned with the control, ONLY if the PCR product
sizes are within lbp of the corresponding control PCR product size.

Before moving the treated data, the raw data are taken into account. If all the peak area
points that were moved originally are within 2bp of the control PCR product size it is
being aligned to then a move is deemed to be appropriate.

Table 3 demonstrates the alignment of a row of treated data with the control data.

At this point the data adjustments are complete.

89

Appendix Table 2. Excel workbook illustrating the removal of background and
adjustment of data according to the minimum data requirements and the outlined
guidelines

 

 

 

 

 

 

 

 

 

 

 

 

 

PCR
Product Sample 1 Sample 1 Sample 2 Sample 2 Sample 3 Sample 3 Sample 4 Sample 4
(base Replicate 1 Replicate 2 Replicate l Replicate 2 Replicate 1 Replicate 2 Replicate 1 Replicate 2
pairs»)

263 +939”

266 4349”

268 29+6b

27o 5939b

271

272 5874‘ 7868 6679 7351 9824 , 6222 5837

273 14059c

274

275 84§b

276 65145d

277 68353 16983 29863 72469 58139 38949

 

 

 

 

 

 

 

 

 

" Peak area values representing the amount of PCR product are reported for each
corresponding PCR product size.

b With a total of 4 samples, 2 replicates per samples, at least 2 of the 4 animals and 4 of
the eight replicates must be reporting a peak area. A single peak area for the
corresponding size does not meet the requirements and is deleted (represented by the
strikethrough font).

CA peak area data point that is within 2 base pairs of a row of data points that meet all
requirements, can be added to that row to complete the row. The 14059 data point would
be added to the 272bp PCR product size row.

dA peak are data point that is within 2 base pairs of a row of data points that meet all
requirements, can be added to that row even though it will not complete the row. The
65145 data point would be added to the 277bp PCR product size row.

90

 

Appendix Table 3. Excel Workbook demonstrating the alignment of a Treated
peak area data points to the control PCR product size reporting peak area data

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

points
PCR Control Control Control Control PCR Treated Treated Treated Treated
Product Sample 1 Sample 1 Sample 2 Sample 2 Product Sample 1 Sample 1 Sample 2 Sample 2
Size R lic ate 1 Replicate Replicate Replicate Size Replicate Repl1cate Replicate Replicate
(bp) 3" 2 1 2 (bp) 1 2 1 2

Data Arrangement Prior to Aligning
343 2525358 166902 156223 286118 343
344 344 315897 274255 35888 80682

Aligned Datab

343 252585 166902 156223 286118 343 315897 274255 35888 80682
344 344

 

 

 

 

 

 

 

 

 

 

" Peak area values representing the amount of PCR product are reported for each
corresponding PCR product size.
b Performing this alignment is appropriate only if in the raw data (from the treated group)

all the peak areas fell within the 342. 343. 344 OR 343. 344‘. 345 base pair range.

91

 

Visualization of Data Consistenm Creation of Individual Animal Avergge Peak Area
ELM;

In order to visualize the consistency of the data across the animals within a particular
experimental group, the average peak areas for each animal at each PCR product size are

plotted (Figure 1).

Data Calculations and Analysis
The results as size of PCR products, in base pairs, and their corresponding peak areas are
analyzed using the Exce1® program. To compare changes between treated and control
groups within a digest, first, a consensus control is created. For each distinct PCR
product, an average peak area is determined across all the control samples. The standard
deviation, standard error, and 95% confidence interval are also calculated. A consensus
treated is created using the same analysis steps indicated for creating a consensus control.
The following equations are used to calculated the consensus treated peak areas for each
PCR product size as a percent of the corresponding consensus control peak areas:
% MspI consensus control = ((MspI consensus treated — Mspl consensus control)/Msp1
consensus control) x 100

% HpaII consensus control = ((HpaII consensus treated — HpaII consensus

control)/HpaII consensus control) x 100
These results are plotted using the Excel® program as size of PCR product in base pairs,
on the x-axis vs. percent consensus control (as calculated above). In this manner,
positive values indicate regions where treatment resulted in increased methylation
(hyperrnethylation), i.e., more of a specific size PCR product formed as compared to

control. Negative values represent regions where treatment resulted in decreased

92

40000

 

0 Animal 1's Average Peak Area

A Animal 3's Average Peak Area l

i

I
35000 - i l
g l DAnimai 2's Average Peak Area ‘

1

30000 i l

25000 . .... ..__ .* - _. ’ .-.._. _. ..__ ___ ___.__ __$__. A—____. -..—‘1: *" “ :T" "-f—_ "——"“

20000 «- ~— -~ ’-—— — —.l -— _—__.-_-.<i._- .# _#-.+ A, _ ‘ -fi, __ _
i T <5
15000 , F E

10000 $
5000 -- --—v _———- § —- - -
0 .

0 100 200 300 400 500 600 700
PCR Product Size (bp)

PEAKAREA

 

 

 

 

 

 

 

 

 

 

 

 

Appendix Figure 1. Individual Animal Average Peak Area Chart For Visualization
of Data Reproducibility and Consistency. Peak areas are compared across all the
animals for each individual PCR product size within a treatment group. Each animal’s
average peak area is plotted on the y-axis and the PCR product size (bp) is plotted on the
x-axis. Peak area is representative of the amount of PCR product amplified.

93

methylation (hypomethylation), i.e., less of a specific size PCR product formed as
compared to control. Individual t-tests (p<0.05) were performed to compare the average
peak area for a given PCR product size for the treated group with the average peak area
for the same PCR product size for the control group. In this fashion we are able to
discern three distinct types of changes in methylation: hyperrnethylation,
hypomethylation and new methylation. Hypermethylations are increases which are
statistically significant when compared to control. Hypomethylations include both 100%
decreases and decreases which are statistically significant when compared to control.
The formation of a PCR product following treatment which was not observed under
control conditions is deemed a new methylation. Each PCR product of a distinct size is
assumed to represent a distinct region of the genome.

The extent of similarity of changes across treatment groups was compared on a region by
region basis. PCR products of identical size between the control and treatment groups
were considered to be common regions of altered methylation. At each of these
corresponding regions, a One-Way ANOVA (p<0.05) was used to determine statistical

significance.

94

CHAPTER 2

PHENOBARBITAL INDUCES PROGRESSIVE PATTERNS OF GC-RICH AND
GENE-SPECIFIC ALTERED METHYLATION IN THE LIVER OF TUMOR-
PRONE B6C3F] MICE

This chapter represents a draft version of a manuscript that will be submitted to
Toxicological Sciences in December, 2005. Authors include: Bachman, Ammie N.,
Phillips, Jennifer, M. and Goodman, Jay I.

95

ABSTRACT

Altered DNA methylation contributes to tumorigenesis in that changes in
methylation can alter gene expression in a heritable fashion. Phenobarbital (PB) is a non-
genotoxic rodent carcinogen which induces global hypomethylation and regions of
hyperrnethylation in mouse liver. Liver tumor sensitive (B6C3F1) and resistant
(C57BL/6) male mice were administered 0.05% (w/w) PB in drinking water for 2 or
4wks. In addition, 2wk recovery groups were included to assess reversibility. DNA was
isolated from both liver (target) and kidney (non-target) tissue. The methylation status of
GC-rich regions of DNA was quantitatively assessed via methylation-sensitive restriction
digestion, arbitrarily primed PCR and capillary electrophoretic separation of PCR
products in order to discern PB-induced regions of altered methylation (RAMs). Those
PB-induced RAMs which carry forward from an early to a later time point are more
likely to be mechanistically relevant as compared to those that do not. In this context, it
is important to note that 12 of the 69 RAMs (17%) present in B6C3F 1 liver at 2 wk were
also seen at 4 wk while only 1 of the 123 RAMs (<1%) present in C57BL/6 liver was
seen at this later time point. In the B6C3F1 mice, 57 unique (as compared to the
C57BL/6) regions of altered hepatic methylation (RAMs), predominantly
hypomethylation, were observed after 2 wk of treatment with PB and this increased to 86
at 4wk. Changes in methylation were largely reversible. Comparatively, altered
methylation in the liver was highly dissimilar to kidney. Following 4 wk of treatment
with PB, bisulfite sequencing revealed hypomethylation of the 5’ promoter region of Ha-
ras in B6C3F1 , but not C57BL/6, which correlated with an increase in gene expression.

These data support the 3 key points: 1) progressive, non-random changes in methylation

96

play an important role in tumorigenesis; 2) altered DNA methylation is an epigenetic
mechanism underlying the ability of PB to cause liver tumorigenesis; and 3)
susceptibility to tumorigenesis is related inversely to the capacity to maintain normal

patterns of methylation.

97

INTRODUCTION

Phenobarbital (PB) is a rodent liver carcinogen which lacks the ability to damage
DNA in a direct fashion (Whysner et al., 1996 and 1998). Therefore, it is considered to
be a nongenotoxic compound. PB is frequently used as a model compound in
carcinogenesis studies. Susceptibility to liver cancer varies widely with genetic
background in mice. B6C3F1 mice are particularly sensitive to the formation of liver
tumors. They are derived from a cross between the relatively resistant maternal strain,
C57BL/6, and the highly susceptible paternal strain, C3H/He. Spontaneous rates of liver
tumor formation are high in the sensitive B6C3F 1 mice (29%) as compared to the
resistant C57BL/6 mice (0%) and administration of PB (0.05% w/w in drinking water)
results in a 100% hepatic tumor incidence in B6C3F1 mice while the incidence in the
C57BL/6 strain is 0% (Becker, 1982). Additionally, N-nitrosodiethylarnine (DEN), a
classic genotoxic carcinogen, is far more potent with regard to producing liver tumors in
B6C3F1 as compared to C57BL/6 mice (Drinkwater and Ginsler, 1986; Buchmann et al.,
1991). Interestingly, the activation of the Ha—ras proto-oncogene via a point mutation in
codon 61 occurs at a different frequency in spontaneous, DEN-induced and PB-induced
B6C3F] liver tumors. The incidence is approximately 50-60% in spontaneous (Stowers et
al., 1988; Fox etal., 1990), and 30-35% in DEN-induced (Stowers et al., 1988;
Buchmann et al., 1991) while it is only 7% in the PB-induced tumors (Fox et al., 1990).
This indicates that the mechanism by which PB causes these tumors might be somewhat
different than that of DEN and is not simply a “magnification” of what occurs

spontaneously.

98

Operationally, carcinogenesis can be described experimentally as having three
defined stages: initiation, promotion and progression (Dragan et al., 1993). Cells
possessing heritable changes in the genome, acquired during initiation, are selected for
and proliferate, given an appropriate promoting stimulus, and eventually are able to
progress to frank carcinomas. Exceedingly aberrant subclone populations arise with the
progressive accumulation of either mutations or, importantly, epigenetic changes.
Specifically, DNA methylation is an epigenetic modification which, when altered, can
affect the normal expression of genes in a heritable fashion (Goodman and Watson,
2002). Therefore, altered patterns of methylation are thought to play a causative role in
all stages of tumorigenesis, and this is not incompatible with a role for mutations, too
(Goodman and Watson, 2002).

Increased DNA methylation (hypermethylation) might silence the expression of
tumor suppressor genes (Goodman and Watson, 2002; Jones and Baylin, 2002) while
decreased methylation (hypomethylation) might facilitate the expression of oncogenes
(Costello and Plass, 2001; Goodman and Watson, 2002). While most attention was
previously focused on hypermethylation, the contribution of hypomethylation is gaining
more interest. Retrotransposable elements, e.g., LlNE-1 elements, are normally kept in a
highly methylated state in order to keep them transcriptionally silenced.
Hypomethylation may lead to their expression resulting in genomic instability which
could play a role in tumorigenesis (Carnell and Goodman, 2003). Mice with decreased
Dnmtl expression exhibit genome-wide DNA hypomethylation, chromosomal instability,
and activation of oncogenes. Additionally, they develop aggressive T cell lymphomas

(Gaudet et al., 2003). Fibroblasts, derived from embryonic stem cells (ES) which were

99

transiently demethylated in order to cause loss of imprinting, were tumorigenic when
placed into immunosuppressed mice (Holm et al., 2005). Furthermore, chimeric mice
derived from the ES which lacked imprinting developed tumors in multiple tissues which
arose from the ES cells (Holm et al., 2005). Actually, there are multiple roles for altered
DNA methylation in carcinogenesis and this likely involves a combination of selected
hypo- and hypermethylations of key genes (Counts and Goodman, Cell, 1995).

A global decrease in hepatic DNA methylation which occurs simultaneously with
hyperrnethylation of GC-rich regions is more pronounced in the liver tumor-prone
B6C3F1 mouse as compared to the resistant C57BL/6 mice treated with PB (Counts et
a1 ., 1996; Watson and Goodman, 2002). Thus, it has been hypothesized that, sensitivity
to tumorigenesis might be related inversely to the capacity to maintain normal patterns of
DNA methylation (Counts et al., 1996; Goodman and Watson, 2002; Watson and
Goodman, 2002).

Altered patterns of DNA methylation in GC-rich regions of DNA were measured
in response to a tumor promoting dose of PB in target (i.e. liver) and non-target (i.e.
kidney) tissues. Reversibility, a hallmark of tumor promotion was assessed for the
observed changes in liver. The methylation status of the promoter regions of Ha-ras and
LINE-1 elements were measured and correlated to changes in gene expression.
Relatively resistant, C57BL/6 mice are used as a “control” comparison to the tumor prone
B6C3F1 mice in testing the hypothesis that progressive, non-random changes in

methylation underlie susceptibility to tumorigenesis.

100

MATERIALS AND METHODS

Man

Male B6C3F1 (C57BL/6 X C3H/He) and C57BL/6 mice (ages 29-32 days) were obtained
from Charles River Laboratories (Wilmington, MA). Animals were allowed to acclimate
for 7 days prior to being randomly assigned to treatment groups. B6C3F1 mice were
house 5/cage and C57BL/6 mice were housed individually, in a temperature controlled
environment and given food and water ad libitum. Care was given in accordance with the
All Use and Animal Care Guidelines of Michigan State University. Mice, 6-7 animals
per group, were administered PB at a concentration of either 0.05% (w/w) or 0.002%
(w/w) in the drinking water for 2 or 4 weeks. Recovery groups were given control water
for two weeks subsequent to dosing. In this manner, reversibility of alterations in
methylation induced by phenobarbital can be assessed. Mice were euthanized by C02

asphyxiation, and the livers and kidneys were snap-frozen at -80°C.

DNA and RNA Isolation

In order to isolate DNA, lml of TRIzol Reagent (Invitrogen) per 100mg frozen sample,
was added to a dounce homogenizer. Frozen liver or kidney tissue was added to the
TRIzol Reagent and thoroughly homogenized. RNA and DNA was isolated according to

the manufacturer’s protocol.

Arbitrarily Primed PCR and Capillafl Electrophoresis
We have developed an arbitrarily primed PCR procedure (AP-PCR) that provides a

thorough, overall evaluation of the methylation status of GC-rich regions of DNA.

101

Importantly, a comparison of data obtained from DNA isolated from control and treated
tissue permits the simultaneous detection of treatment-related increased methylation
(more methylation in a region that was methylated in control), decreased methylation
(less methylation in a region that was methylated in control) and new methylations
(methylation in regions that were not methylated in control). Therefore, an in depth
picture of treatment related altered methylation is provided. This technique employs
methylation sensitive restriction digestion, arbitrarily primed PCR and capillary
electrophoretic separation of PCR products. Please refer to Charger 1 for detgil_s
concerning basic experimental methods and data calculations. All new informgtion
related to the method follows.

Restriction Digestion — BfaI/BssHII Double Digest

BssHII recognizes GCGCGC and cuts between the 5’ guanine and cytosine. These
sequences are predominantly found within CpG islands and therefore cutting is less
frequent (Shiraishi et al., 1995). Generally, BssHII restricts its target sequence only if all
cytosines are unmethylated. Restriction digests contain lug DNA and 5.0 units BfaI in
1X New England Biolabs Buffer 4. Samples are incubated for 1hr. at 37°C before
addition of 2.5 units BssHII. Following addition of BssHII, the incubation temperature is
raised to 50°C. A second 2.5 unit aliquot of BssHII is added after an additional 2hr, and
in order to insure complete digestion, total incubation time is 18hr. The enzymes were
inactivated by incubating at 80°C for 10min. Samples were stored at 4°C until analyzed.
Reversibility of altered methylation: Calculations

The extent to which altered methylation observed following specific periods of

promotion was reversible was calculated by using each 2 or 4wk 0.05% PB promotion

102

group as the working “control” pattern of methylation. The recovery data are calculated
as a % of this “control”. The 2wk, 0.05% PB promotion group served as the “control” for
the 2wk 0.05% PB promotion, 2wk recovery group. The 4wk, 0.05% PB promotion
group served as the “control” for the 4wk 0.05% PB promotion, 2wk recovery group.

A consensus, average, peak area for each PCR product reporting in 2 or 4wk,
0.05% PB promotion groups is prepared and designated as the “control” consensus. In
addition a consensus, average, peak area for each PCR product reporting in 2 or 4wk
0.05% PB promotion, 2wk recovery is prepared and designated as the “reversal”
consensus. “Control” consensus and “reversal” consensus peak areas at each specific
PCR product are compared. The following equations are used to calculate the “reversal”
consensus peak areas for each PCR product size as a percent of the corresponding
“control” consensus peak areas.
% Mspl “control” consensus = ((Mspl “reversal” consensus — Mspl “control”
consensus)/Mspl “control” consensus)) x 100
% Hpall “control” consensus = ((HpaII “reversal” consensus — Hpall “control”
consensus)/Hpa11 “control” consensus)) x 100
As described previously, hypomethylations, hypermethylations and new methylations are

tallied.

Steps to determining whether a CSC-induced change in methylation reversed following

the recovery period

103

l. A list of all the PCR product sizes exhibiting 2 or 4wk 0.05% PB-induced hypo-,
hyper-, and new methylations are compiled.

2. A list of the PCR product sizes exhibiting changes in methylation that occurred during
recovery as calculated above, are compiled.

3. The two lists are aligned by PCR product size so that changes in methylation due to
promotion and changes in methylation which occur during recovery are directly
comparable. An example of this appears in Table 1.

4. Changes in methylation induced by PB promotion are only considered recoverable if a
change in methylation with opposite direction (i.e. hypomethylations are opposite to

hypermethylations and new methylations) is observed following recovery.

Assumptions the AP-PCR Data Analysis

Analysis of the data includes the following assumptions: 1) each separate PCR product of
a defined size represents a distinct region of the genome, 2) a region can include one or
more recognition sequences for the specific methylation-sensitive restriction enzyme
employed located between the annealing sites of the up- and down-stream primers; thus,
the amount of each PCR product formed can be viewed as representing an “average” of
the methylation status of the particular recognition sequences located between the up- and
down-stream primers, and 3) changes in the amount of each PCR product represents the

altered methylation status of a particular GC-rich region of DNA.

104

Table 1. Example of Comparisons for Determining whether a PB-induced Change
in Methylation Reversed Following the Recovery Period

 

 

 

 

 

 

 

 

4wk Promotion Group 4wk, Promotion, Zwk Recovery
Group
Changes in Promotion
PCR product size Change in PCR product Change in methylation are induced change
(base pairs) Methylation size Methylation Opposite in in methylation ‘5
Direction? feverscd?a
245 New Meth 245 HypoM Yes Yes
3 15 HyperM 3 15 HyperM No No
456 HypoM 456 New Meth Yes Yes
505 New Meth 505 HypoM Yes Yes
5 15 HyperM 5 15 HypoM Yes Yes

 

 

 

 

 

 

 

" Changes in methylation induced by CSC promotion are only considered recoverable if a
change in methylation with opposite direction (i.e. hypomethylations (HypoM) are
opposite to hypermethylations (HyperM) and new methylations (N ewMeth)) are
observed following recovery.

105

Common and Unique Regions of Altered Methylation

PCR products of identical size that occur in two treatment groups (e. g. 2wk and 4wk
0.05% PB) were considered to be common regions of altered methylation (RAMs). The
methylation changes associated with RAMs in common between 2 treatment groups are
considered equivalent and persistent if the changes in methylation are in the same
direction and the extents of change are statistically no different as determined by 2-way
ANOVA, p=0.05. RAMs in common between 2 treatment groups are considered unique
RAMs if, 1) the changes in methylation are opposite in direction (i.e. a hypomethylation
is elicited by one treatment and a new methylation is elicited in the same region by the
comparison treatment) or 2) the changes in methylation are in the same direction but the
extents of change are statistically different as determined by 2-way AN OVA, p=0.05.

RAMs are deemed to be unique if they are only observed in a particular treatment group.

Evaluating Total RAMs

The RsaI/Mspl and RsaI/HpaII digests were considered separate experiments to
determine regions of the genome which exhibit altered methylation in response to
treatment. Although Hpall and Mspl both restrict CCGG sites, Hpall identifies altered
methylation at the internal cytosine while Mspl identifies altered methylation at the
external cytosine. When we look at the total number of RAMs identified by HpaII and
Mspl there is the unavoidable possibility that a slight amount of double counting might
occur. This would be the case if, methylation were altered at both the internal and

external cytosine within CCGG sites of particular genomic region. Therefore, we would

106

be considering differences in methylation within a given region even if there were some

“double counting” of RAMS.

% Dissimilarity Calculations

Control vs. Treatment

The dissimilarity between control patterns of methylation and altered patterns resulting
from PB administration was calculated. The total number of PCR products reporting in
control was added to the total number of unique PCR products (i.e. those PCR product
sizes that were not formed under control conditions) reporting from treatment to get the
total number of combined PCR products. This represents the total number of regions
(PCR products) analyzed between the two groups. The total number of combined PCR
products divided by the total number of regions (PCR products) exhibiting a statistically
significant change (hypomethylations, hypermethylations, and new methylations) times

100 equals the percent dissimilarity from control.

PB Treated B6C3F] vs. PB Treated C5 7BL/6

The dissimilarity between the extent of altered methylation due to PB in C57 in
comparison to B6 at 2 and 4 wks was calculated. Unique RAMs were identified for the
B6 mouse. These include 1) RAMs in common with C5 7, but the methylation change is
opposite in direction, 2) RAMs in common with C57, but the extent of change is
statistically greater (2-way AN OVA, p<0.05) in B6, and 3) RAMs which are only
observed in the B6 mouse. The unique RAMs were divided by the total RAMs (i.e. all

hypomethylations, hypermethylations, and new methylations in B6 mouse) and

107

multiplied by 100 to get the percent dissimilarity. In addition, the dissimilarity between
the extent of altered methylation due to PB in liver in comparison to kidney for B6 and
C57 was calculated. Unique RAMs were identified for the liver. These include: 1)
RAMS in common with kidney, but the methylation change is opposite in direction, 2)
RAMs in common with kidney, but the extent of change is statistically greater (2-way
ANOVA, p<0.05) in liver, and 3) RAMs which are only observed in the liver. The
unique RAMs were divided by the total RAMs (i.e. all hypomethylations,
hypermethylations, and new methylations in liver) and multiplied by 100 to get the

percent dissimilarity.

Gene-Specific Methylation Analysis: Bisulfite Seguencing of 5’ Promoter Region of

Bugs

Bisulfite Conversion and PCR Amplification

Bisulfite conversion of DNA effectively deaminates all un-methylated cytosines to uracil
leaving methylated cytosines unaffected. 2ug DNA was bisulfite converted using the EZ
DNA Methylation Kit (Zymo Research, Orange, CA). PCR is performed with bisulfite
converted DNA which allows for the replacement of uracil with thymine and 5-
methylcytosine with cytosine. Consequently, only cytosines that were originally
methylated remain in the DNA sequence. PCR is carried out using primers specific for
bisulfite converted DNA and containing no CpG sites. The two Ha-ras primers, 5’ GGT
GGG TTA GAG TGT TTA AGA TTT G 3’ and 5’ CTC TTA CTC TAA AAA ACA
TTT CCA C 3’ were used to amplify the -950nt to -1232nt (283bp) region of the Ha—ras

promoter relative to the transcriptional start site. Primers were designed based on

108

sequence information obtained from Brown et al., 1988 and Neades et al., 1991. Each
PCR reaction contained 0.51lg bisulfite converted DNA, 1X FailsafeTM Buffer G
(Epicentre®; Madison, WI) 0.311M each primer, 1.5 units Taq Polymerase (InvitrogenTM)
and GDW to a final volume of 25ul. Cycling conditions were: 95°C for 3 min, 38 cycles
of 95°C for 45 s, 58°C for 45 s, and 72°C for 1 min, followed by 1 time delay cycle of
72°C for 3min and a 40C soak. Amplification of the target region was verified by gel
electrophoresis on a 3% agarose gel. Duplicate Ha-ras PCR reactions were combined and
purified using the Qiagen Qiaquick PCR Purification Kit. Samples were quantified

fluorometrically.

Sequencing

Automated sequencing of purified Ha-ras PCR products was carried out at the Genomics
Technology Support Facility at Michigan State University using an ABI PRISM®3100
Genetic Analyzer. Two separate sequencing reactions are performed for each gene.
Sequencing reactions are composed of 20ng PCR product, 30 pmol of either the forward

primer or reverse primer, and

Reverse Transcription of RNA

RNA samples were treated with DNaseI (Invitrogen) to purify the RNA from
contaminating DNA remaining after isolation. Each reaction contained 2ug RNA, 1X
DNaseI reaction buffer, 2 units DNaseI, and DEPC-treated GDW to a final volume of
20ul. Samples were incubated at room temp. for 15min followed by addition of MgClz to

a final concentration of 2.27mM. RNA was heated to 65°C for 10min to inactivate the

109

DNaseI enzyme. The TaqMan Reverse Transcription Kit (Applied Biosystems; Foster
City, CA) was used to reverse transcribed the DNaseI treated RNA. Each reverse
transcription reaction contained, 1X Reverse Transcription Reaction Buffer, 5.5mM
MgC12, 200uM of each dNTP, 2.5uM random hexarner, 20 units RNase Inhibitor, 62.5
units Multiscribe Reverse Transcriptase and DEPC-treated GDW to a final volume of
50ul. The reactions are incubated at 25°C for 10min, 42°C for 1hr, and 95°C for 5 min.

All samples were stored at 4°C until needed.

Expression of Ha-ras and LINE-1
Real Time PCR: Primers

A custom TaqMan assay including primers (For 5’ TGG TGG GCA ACA AGT GTG
A3’ and Rev 5’ GGC CTG CCG AGA CTC A 3’) and probe (5’ F AM CTG GCT GCT
CGC ACT GT 3’) specific for Exon 3 of the Ha—ras gene was purchased from Applied
Biosystems. The assay was designed based on sequence information obtained from
Brown et al., 1988 and Neades et al., 1991. A custom TaqMan assay, including primers
(For 5’ GGT CAA ATC TAA GTG GAT CAA GGA ACT 3’ and Rev 5’ GCT TTT CCC
CAC TTT CTC CTC TAT 3’) and probe (5’ FAM CAG AGA CAC TGA AAC TT 3’),
specific for ORF2 of the LIN E-l element (Accession M13002) was purchased from
Applied Biosystems. In addition an Applied Biosystems custom TaqMan assay,
including primers (For 5’ CTA CTA CCG ATT GGA TGG TTT AGT GA 3’ and Rev 5’
GTC AAG TTC GAC CGT CTI‘ CTC A 3’ and probe (FAM 5’ CCG TGG GCC GAC

CC3’), was used for the control gene, 188 rRNA (Accession X00686).

110

Real Time PCR

Triplicate reactions for both the gene of interest (Ha-ras or LINE-l) and the control gene
(18S) were prepared per sample. Standards were also prepared for 188, Ha-ras and
LINE-1 and ranged from 5 x 101 copies/ul to 5 x 107 copies/u]. Each Ha-ras or LINE
reaction contained 1 x Custom Assay Mix (Applied Biosystems; Foster City, CA), 1X
TaqMan Universal PCR MasterMix, 8ul cDNA and GDW to a final volume of 25ul.
Each 188 reaction contained 1 x Custom Assay Mix (Applied Biosystems; Foster City,
CA), lX TaqMan Universal PCR MasterMix, 2ul 1:100 diluted cDNA and GDW to a
final volume of 25ul. Reactions for each standard contained 1 x Custom Assay Mix
(Applied Biosystems; Foster City, CA), 1X TaqMan Universal PCR MasterMix, 2ul
standard and GDW to a final volume of 25ul. Cycling conditions were as follows: 50°C
for 2 min, 95°C for 10min, and 40 cycles of 95°C for 153ec. and 60°C for 1 min. The

absolute standard curve method for quantifying fold change over control was employed.

111

RESULTS

Genome-wide analysis of altered methylation in GC-rich regions in response to 2
or 4wk PB, 0.05% (w/w) in the diet, discerned numerous regions of altered methylation
(RAMs) and established the occurrence of hypomethylations, hypermethylations, and
new methylations in B6C3F1 and C57BL/6 mice. B6C3F1 mice exhibited 69 total
RAMs, primarily hypomethylations at 2wks (Table 2). With 4wks treatment,
hypomethylated, hyperrnethylated, and new RAMs in the B6C3F1 mice increased to 98,
a 42% increase in total RAMs. In contrast, while a large numbers of RAMs (123 total)
were observed in the liver of C57BL/6 mice at 2wks, the total decreased to 88, a 28%
decrease, primarily due to a lower number of hypermethylations and new methylations at
the later time point (Table2).

Quantifying hypomethylated, hypermethylated and newly methylated RAMs
allows for the identification and comparison of patterns and trends of PB-induced altered
methylation between the liver turnor-prone B6C3F1 mice and the relatively resistant
C57BL/6 mice. Additionally, assessing the extent to which 2 or 4 week PB disrupted
methylation patterns as compared to controls is important, too. An evaluation of percent
dissimilarity provided an initial, overall assessment of PB’s effects in B6C3F1 and
C57BL/6 mice. Methylation patterns arising from 2 wks PB were 45% and 67%
dissimilar to methylation patterns of controls in B6C3F1 and C57BL/6 mice, respectively
(Table 2). At 4 wks, dissimilarity increased to 61% and 79% (Table 3).

Overall dissimilarities revealed that PB-induced patterns of methylation in
C57BL/6 deviated from controls slightly more than in B6C3F1 emphasizing the need for

a more refined approach to identify, evaluate and prioritize changes in methylation which

112

Table 2. Methylation Sensitive Digestion with RsaI/MspI or RsaI/HpaII: Summary
of GC-Rich Regions of Altered Methylation (RAMs) in the Liver of B6C3F] and
C57BL/6 Mice in Response to 2 or 4wk 0.05% PB

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. RAMs RAMs RAMs
Treatment Digest HypoMa HyperMb “New” Methylationc TOTAL
2 wk 0.05% PB
B6C3Fl Hpall 20 1 6 27
Mspl 26 6 10 42
Totald 46 7 16 69
C57BL/6 Hpall 19 6 17 42
Mspl 18 33 3o 81
Total“ 37 39 - 47 123
4 wk 0.05% PB
B6C3F 1 alaII 13 4 21 3 8
Mspl 46 7 7 6O
Totald 59 11 28 98
C57BL/6 Hpall 36 8 9 53
Mspl 35 0 0 35
Totald 71 8 9 88

 

 

 

 

 

 

 

" Hypomethylated (HypoM) RAMs include both statistically significant (p<0.05)
decreases and 100% decreases.

b Hypermethylated (HyperM) RAMs are only those increases which are statistically
significant (p<0.05).

CNew methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not present under control conditions.

dTotal RAMs including hypomethylations, hypermethylation, and new methylations for
the combined digests are reported for each treatment

113

Table 3. Dissimilarity Between PB-Induced Methylation Changes and Control
Methylation Patterns in B6C3F1 and C57BL/6 Mouse Liver

 

 

 

 

 

 

 

 

Dissimilari t Be tween° Total Total RAM as b Percent Dissimilarity
y ' Regionsa Compared to Control To Control
B6C3F]
2 weeks
Control 005% PB 153 69 . 69/153 45%
Control 0%:ng 202 123 123/202 61%
C57BL/6
Control 03023:; 147 98 98/147 67%
Control :0?ng 11 1 88 88/11 1 79%

 

 

 

 

 

 

 

 

“ Total regions includes every PCR product size reporting between control and treated
groups. Each PCR product represents a region of the genome.

Total RAMs represents the number of regions exhibiting a statistically significant
(p<0.05) change in methylation including complete hypomethylations (100%), partial
hypomethylations, hypermethylations, or new methylations.

114

might mechanistically be important regarding promotion of tumorigenesis. An important
feature of the promotion stage of tumorigenesis is the progressive accumulation of
heritable alterations to the genome. Therefore, we evaluated the number of changes in
methylation which “carry forwar ” over time during PB-treatment. “Carry forward”
RAMs, identified in the B6C3Fl and C57BL/6 mice, included all RAMs that were in
common between the 2 and 4 week time points which exhibited equivalent changes in
methylation (Figure 1a). RAMs identified only at 4 weeks plus those RAMs which were
in common with those identified at 2 weeks but the methylation change was either 1)
opposite in direction or 2) changing to a greater extent at 4 weeks were classified as
unique RAMs (Figure 1a). A total of 12 PB-induced RAMs (17% of the total RAMs
observed at 2 wk) including 9 hypomethylations, 1 hypermethylation, and 2 new
methylations, carried forward from 2 to 4 wks in B6C3F1 liver (Figure 1b). Strikingly,
only one hypomethylation out of 123 total RAMs (<1%) in C57BL/6 liver was seen to
carry forward to 4 wk (Figure 1b).

Two-and four-week periods of promotion with PB were followed by 2 wk
recovery periods. Figure 2 illustrates that with 2 wks PB exposure and 2wks recovery in
B6C3F1 mouse liver 100% (7/7) of the hypermethylations and 75% (12/16) of the new
methylations reversed. In addition, a large number of hypermethylations (73%, 8/11)
and new methylations (61%, 17/28) which included two of the persistent RAMs were
seen to reverse during the recovery period following 4wk PB administration. Recovery
was similar in C57BL/6 mice in that 31% (12/39) and 100% (8/8) of hypermethylations
and 51% (24/47) and 89% (8/9) of new methylations reversed during 2wks recovery

following 2 and 4wks PB exposure, respectively. This illustrates that increases in

115

F'gpre 1 Progressive Changes in Methylation: Changes Which Carpy Forward
From 2 to 4wk in B6C3F1 and C57BL/6 Mice The flow chart illustrates the steps

necessary to determine progressive changes in methylation. Separate comparisons
between B6C3F] and C57BL/6, 2 and 4wk regions of altered methylation (RAMs) were
performed (a). Each RAMs induced by 2wk, 0.05% PB promotion are represented (O)
for B6C3F 1 and C57BL/6 mice. RAMs induced by 2wk PB were compared to those
resulting from 4wk, 0.05% PB in B6C3F 1 and C57BL/6 mice. Common RAMs in which
the magnitudes of change were equivalent (2-way AN OVA, p<0.05) were considered
carry forward changes. RAMs unique to 4wk, 0.05% PB (0) included common RAMs
in which the magnitudes of change were different or the RAMs were only observed with
4wk 0.05% PB (b). Hypomethylations (HYPOM), hypermethylations (HYPERM), and
new methylations (N EWM) are segregated. Total unique changes (minus carry forward
changes) are tallied and reported for each category of methylation change.

116

. LIVER

C57BU6
2wk, 0.05% PB

B6C3F1

2wk, 0.05% PB

Methylation Change /
is Equivalent

Carry Forward Changes

 

HY POM

 

 

NEWM HYPERM

 

 

 

2wk 0.05% PB
B6C3F1 LIVER

Common Regions of
Altered Methylation

LIVER

.EEEEEQEEx C57BUB
4wk, 0.05% PB
4wk. 0.05% PB

(RAMs)

Methylation Change is
Statistically Different

\.

Unique RAM: at
4 weeks

 

Total Unique Methylation
Changes at 4wks

)5

1. M BBIC57 methylation d'tange is
opposite to 4wk 86/057 methylation change
2. 4wk 86/657 changes to a greater extent
than M BG/C57

5

NEW HYPERM HYPOM

 

eoooooo
6%3§3%P§%§9

37

 

HYPERM HY POM

1:5

 

NEWM

 

 

 

2wk 0.05% PB
C57BL/6 LIVER

 

G

00
G

0 0 50
0Q 00000;030:020303080202030
0000000000000

9

98

 

0

0000000

000 ”

 

 

0

0 0 0 0 0 0 0 0
_ °00000000000000000

26

0

 

Carry Forward
RAM

4wk 0.05% PB
BéC3Fl LIVER

 

Q

HYPOM

0° 70
9&Bi:d%?d%fid%flf3%fid%9
%%%%%%%%%%%%%%
ooooooooooooo

 

o°o°o°oo 3

 

E
E

 

0 0 0
000000

 

(:arry

Fomrd

RAMs

4wk 0.05% PB
C57BL/6 LIVER

117

 

 

Fi re 2. Pro ressive Chan es in Meth lation: Reversibili of Re “ms of Altered
Methylation Induced by PB Hypomethylations (HYPOM), hypermethylations
(HYPERM), and new methylations (N EWM) are represented for 2wk (O) and 4wk (0)
0.05% PB promotion. RAMs which carried forward from 2 to 4wks are also represented
(O). RAMs induced by 2 or 4wk PB which reversed during the 2 week recovery periods
in B6C3F1 (a) and C57BL/6 (b) are boxed. Total unique changes (minus carry forward
RAMs and RAMs which reversed) are tallied and reported for each category of
methylation change.

   

118

 

10
26

0
000000

°o°o°o°o°o°o°o

O
4wk 0.05% PB
B6C3F1 LIVER

0

0
0

0
o0
0
0
O0

0
00

o
'o
°o

‘ 0

e 99
9009 0
RAM.

0

 

 

 

 

Carry Forward

 

 

2wk 0.05% PB
B6C3F1 LIVER

 

 

 

 

 

 

 

 

    

 

  

 

 

 

 

 

 

 

 

 

o ooo 1
0 0
m 0 0 0
. 0 0
0 0 0
0 0
0 0 0
. 0 0
0 0 0
0 0 . 0 0
0 0 Y 00000 BR W
oooom M R o o 0 pm w
com aw oo ./.L no
“v ooooM mm ooooo M 6 mm
m . R
ooooW _ L n ooooo 0 mm I Lm
O 0 m k 0 O o 0 WV]. E
o o w o o o C I 2
0 0 2 O 00
0 OW 0 O O o E
o o ooooo o o E
o om m o o o o 000 I
m 0000 0 0 0 E ._
9W 0% 0 O 00 mi
899W 0 W%F -0. . . . ., ,. m .. .. . . . n
so m 0 mm E mmm
22>: €232 22>: :xmaEZBmz ... 22>: EmmaEEBmz F

 

 

 

 

 

 

 

 

 

 

 

R7
w... as
u so
moss
Y 399
R e7399999
w coco
0 000
C P000
u %00
uses
k 0.80089
n “so
2908
a 22E

 

119

UVIR
MK RECOVERY

methylation which accumulate in response to the promoting stimuli are largely reversible
following a relatively brief period of promoter treatment. However, the reversibility of
hypomethylated RAMs in both B6C3F1 and C57BL/6 mice was rather low and might
indicate that 2wks are insufficient for “re-methylation” of unmethylated cytosines (Figure
3).

In light of the extreme difference between B6C3F1 and C57BL/6 mice in terms of
changes in methylation which carry forward over time, the prime focus of the
investigation was on identifying specific differences as well as similarities in RAMs
between the liver tumor-prone B6C3F1 and the relatively resistant C5 7BL/6 mice. We
focused on ascertaining which PB-induced increases (i.e. hypermethylations and new
methylations) or decreases (i.e. partial and complete hypomethylations) in methylation
occurred in the same regions of the genome. For each specific region, changes in
methylation identified in B6C3F1 and C57BL/6 hepatic DNA were compared and
classified as 1) equivalent, 2) opposite in direction (e.g., hypomethylation in the B6C3F1
and hyperrnethylation in the C57BL/67), or 3) changing in the same direction, but to a
greater extent in the B6C3F1. Additionally, those RAMs which were unique to the
B6C3F1 mouse are emphasized. Initially, by using this information, we were able to
calculate an overall dissimilarity between PB-induced patterns of altered methylation in
B6C3F1 mice as compared to those in the C57BL/6 mice. With 2wk and 4wks of PB
exposure, the patterns of altered methylation in the livers of B6C3F1 mice were 81% and
85% dissimilar to those in C57BL/6, respectively (Table 4).

By subtracting out the individual RAMs that were both in common and equivalent

between B6C3F1 and C57BL/6 (i.e. the 19% and 15% similarities) at the 2 and 4 wk time

120

F 'gpre 3. Identification of PB-Induced Unique and Carpy Forward RAMs in
B6C3F1 Liver. The flow chart illustrates the four steps necessary in defining the unique

and carry forward PB-induced RAMs in B6C3F1 liver. Four separate comparisons were
performed (a). 1. B6C3F1 unique RAMs induced by 2wk, 0.05% PB promotion are
represented (O). 2. B6C3F1 unique RAMs induced by 2wk PB were compared to total
RAMs resulting from 4wk, 0.05% PB. Common RAMs in which the magnitudes of
change were equivalent (2-way ANOVA, p<0.05) were considered carry forward
changes. RAMs unique to 4wk, 0.05% PB (0) included common RAMs in which the
magnitudes of change were different or the RAMs were only observed with 4wk 0.05%
PB 3. B6C3F1 total RAMs were compared to C57BL/6 total RAMs. RAMs which are in
common with C57BL/6 and exhibit equivalent changes in methylation are identified (©).
4. B6C3F1 total RAMs minus C57BL/6 common and equivalent changes were compared
to B6C3F1 RAMs in kidney. RAMs which are in common with C57BL/6 (©) and
kidney (CI) and exhibit equivalent changes in methylation are identified.
Hypomethylations (HYPOM), hypermethylations (HYPERM), and new methylations
(NEWM) are segregated. Total unique changes (minus carry forward changes and
changes in common and equivalent with C57BL/6 or kidney) are tallied and reported for
each category of methylation change. (+) RAMs are considered less critical if the RAMs
are in common with C57BL/6 and the change in methylation is equivalent. Carry
Forward RAMs are only determined with comparison (2.) in which 2wk B6C3F1 total
unique RAMs are compared to 4wk B6C3F1 RAMs. (*) RAMs which are in common
and equivalent in B6C3F1 and C57BL/6 at 2wks but only observed in B6C3F1 at 4wks.

121

BSC3F1 LIVER
M. 0.05% PB

B6C3F1 LIVER
2wk, 0.05% PB
Total Unique Changes

86C3F 1 LIVER
4wk, 005% PB
Total Unique Changes

B6CSF1 LIVER
4wk. 005% PB
Total Unique Changes Minus
CS7BL/6 Common and
Equivalent Changes

 

 

 

 

1.
COMPARE C57BL/6 LIVER

2wk, 0.05% PB

2.
COMPARE 86C3F1 LIVER
4wk, 0.05% PB
3

M. C57BL/6 LIVER

4wk, 005% PB

4.
EM. BSC3F1 KIDNEY
\ 4wk. 005% PB

Common Regiona of RAMs Unique to
Altered Methylation
(RAMs)

Methylation Change ‘/ \
is Equivalent . Total Unique Methylation
Methylation Change is C a

 

 

 

 

 

 

 

 

 

 

 

RAMs are not Critical QB "‘998
Carry Forward RAMe‘ _ . Statistically Different
1. Methylation change is opposite
2. Methylation change is in the same direction. but
is changing to a greater extent
38.888: 2-
9 9 Q 0 G O 43 RAM! 36C3Fl LIVER
2 o o e o o e E o o 54
8 é *9o°o°§°2°Z°3030332030203
“0000098000098 2 0 *00000000000000
9 e o o o e Il=> g -o°o°o°o° -
E as 3 E o o o ”
E 8 5 0 0 0 0 0 O 0 0
0
g 9 9 °o°o 11 E o oo‘’o"o°o"o°o“’o"o°o ’6
g 000° 8 0‘

 

 

 

Mlnue RAMs ln Common and Equivalent with 657

3' Carry Forward
RAMs

NEWM HYPERM HYPOM

4- Carry Forward
RAMs

 

 

09 o°o°o°o°o°o°ia ..

CB {15
* ’3 \
-o°o°o°o°o°o CLO“

G

     
 

*o o o o o o Og© 1:.

o 0 o o 0 © ‘3
2 8 re.
GGG °o°o°o°o°o°o§cte©
*o o o o o o for
a (’9 Q
9o°o°o°o°o°o°o°clo©

HYPO
o
6
o
o
o
o
o

ké)

 

 

‘*0 0 0 @J
‘ 0000000

 

 

0 , 00000000000000000

 

o , ooooooooo

10 g -o o o c w
A:
E °o°o°o°

16 2 oooooooat"
3 mos.
g as °o°o°o°o°o°o°o .4 m

 

 

 

 

 

Mlnue RAMs In Common and Equlvalent with CS7

Mlnue RAMs In Common and Equivalent with C57 or
Kidney

122

Table 4. Dissimilarity Between PB-Induced Methylation Changes in B6C3F1 and
C57BL/6 Mouse Liver

 

 

 

 

 

 

D, , 1 , B M . Total RAMs Total Unique RAMs As 11' Ferrel? T
isSImIarlty e een. in 36“ Compared to C57BL/6b 152131;]; 122/21 0
2 week 0.05% PB
B6C3F1 ] C57BL/6 69 56 56/69 81%
4 week 0.05% PB
B6C3F1 [C57BL/6 98 83 83/98 85%

 

 

 

 

 

 

“ Total RAMs in B6C3F1 is reported from Table 1
b Total Unique RAMs represents the number of regions exhibiting a statistically
significant (p<0.05) change in methylation including complete hypomethylations (100%),

partial hypomethylations, hypermethylations, or new methylations that were only
observed in B6C3F1 mice.

123

points, we were able to further prioritize which RAMs are most unique to the B6C3F1
mice and, therefore, likely to contribute to liver tumor formation (Figure 3a). Figure 3b,
comparison 1, represents the total unique RAMs (i.e. all RAMs that were in common and
equivalent to C57BL/6 are not represented) in the B6C3F1 mice at 2 wks. This pool of
B6C3F1 total unique RAMs (i.e. total RAMs minus RAMs in common with C57BL/6)
was compared to the total B6C3F1 RAMs identified at 4wks and established that 7 of the
57 unique RAMs observed at 2 wk carried forward to 4 wks (Figure 3b, comparison 2.).
Additionally, 54 hypomethylations, 11 hypermethylations, and 26 new methylations were
all unique to the B6C3F1 mouse at 4 wks, and this adds up to a total of 98 RAMs in
B6C3F1 at the 4 wk time point (Figure 3b, comparison 2.). A comparison of 4wk PB-
induced RAMs in B6C3F1 and C57BL/6 mice identified 12 common RAMs with
equivalent changes in methylation (Figure 3b, comparison 3.) These RAMs were
subtracted from the total changes observed in the B6C3F1 mice at 4 wk because they are
considered less likely to contribute to tumor formation, and this results in a total of 86
unique RAMs in B6C3F1 at the 4 wk time point (this includes 5 RAMs which were
observed in both B6C3F1 and C57BL/6 mice at 2 wk and only in the B6C3F1 at 4 wk,
Figure 3b, comparison 3).

Methylation status of the DNA of kidneys of B6C3F1 and C57/BL/6 mice
allowed for a comparison between liver and kidney (a non-target for tmnorigenesis)
tissue. A substantial number of PB-induced RAMs were observed in kidney DNA.
Changes in methylation were induced in B6C3F1 kidney at 4wks. However, in
comparison to liver, they were fewer in number and, importantly, did not include

hypomethylated RAMs, a distinct feature of PB-induced alterations in the liver (Table 5).

124

Table 5. Methylation Sensitive Digestion with RsaI/MspI or RsaI/HpaII: Summary
of GC-Rich Regions of Altered Methylation (RAMs) in Liver and Kidney of B6C3F1
and C57BL/6 Mice in Response to Treatment with 0.05% PB for 4 wk

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. RAMs RAMs RAMs
Treatment DlgCSt HypoMa HyperMb “New” Methylationc TOTAL
B6C3F1
LIVER Hpall 13 4 2 l 3 8
Mspl 46 7 7 6O
Totald 59 11 28 98
KIDNEY Hpall 0 8 33 41
Mspl 0 2 23 25
Totald 0 10 56 66
C57BL/6
LIVER Hpall 36 8 9 53
Mspl 35 0 O 35
Totald 71 8 9 88
KIDNEY Hpall 4 0 l 5
Mspl 1 8 2 2 22
rotord 22 2 3 27

 

 

 

 

 

 

 

 

" Hypomethylated (HypoM) RAMs include both statistically significant (p<0.05)
decreases and 100% decreases.

b Hypermethylated (HyperM) RAMs are only those increases which are statistically
significant (p<0.05).

CNew methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not present under control conditions.

dTotal RAMs including hypomethylations, hypermethylation, and new methylations for
the combined digests are reported for each treatment group

125

Altered methylation in the kidney of C57BL/6 mice mirrored the pattern of altered
methylation in the liver; however, the total number of RAMs were much lower in kidney
(Table 5). It was instructive to determine the degree of similarity in altered patterns of
methylation between the kidney and the liver of each of the different mice. In B6C3F1
mice RAMs induced by PB in the liver were 94% dissimilar to those in kidney whereas in
C57BL/6 mice RAMs in the liver were 85% dissimilar to those in kidney (Table 6).

In much the same way that C57BL/6 was used as a “control” for trying to discern
critical RAMs in B6C3F1 liver, patterns of altered methylation in kidney can be
compared to liver to further distinguish which changes in methylation are likely to be
critical in B6C3F1 liver. This analysis identified 5 common RAMs with equivalent
changes in methylation (Figure 3b, comparison 4.) These RAMs were subtracted from
the total unique changes because they are considered less likely to contribute to tumor
formation. Therefore, of the 98 total RAMs detected in the liver of B6C3F1 mice at 4
wks, 74 unique RAMs plus the 7 RAMs which carried forward from 2 to 4 weeks
represent the PB-induced changes in methylation observed solely in B6C3F1 mouse liver.

Analysis of altered methylation was also assessed using the methylation
insensitive enzyme Bfal in conjunction with the methylation sensitive enzyme BssHII.
BfaI restricts CpG islands less than Rsal (Shiraishi et al., 1995) and BssHII has a 6 base
recognition sequence (Shiraishi et al., 1995) as compared to the 4 base recognition
sequence of Mspl and Hpall. Patterns of methylation detected at 2 and 4 wks in B6C3F1
and C57BL/6 mice using BssHII were similar to those detected with Mspl and Hpall.
Less total changes were observed at 2wks in both B6C3F1 and C57BL/6 mice (Table 7).

At 4wks, B6 exhibited 25 total RAMs as compared to 12 RAMs in C57BL/6 mice (Table

126

Table 6. Dissimilarity Between Methylation Changes in the LIVER as compared to
the KIDNEY following 4 wk of Treatment with PB

 

 

 

 

 

 

Dissimilarity Between: Total RAN? Total Unique RAMS A: Dissflfiilzer'iltty To
in LIVER Compared to KIDNEY KIDNEY
B6C3F1
LIVER I KIDNEY 98 92 92/98 94%
C57BL/6
LIVER I KIDNEY 88 75 75/88 85%

 

 

 

 

 

 

“ Total RAMs in liver is reported from Table 1

b Total Unique RAMs represents the number of regions exhibiting a statistically
significant (p<0.05) change in methylation including complete hypomethylations (100%),
partial hypomethylations, hypermethylations, or new methylations that were only
observed in the liver.

127

Table 7. Methylation Sensitive Digestion with BfaI/BssHII: Summary of GC-Rich
Regions of Altered Methylation (RAMs) in the Liver of B6C3F1 and C57BL/6 Mice
Following Treatment with PB for to 2 or 4wk

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. RAMs RAMs RAMs
Treatment Digest HypoMa HyperMb “New” Methylationc TOTAL
2 wk 0.05% PB
B6C3F1 BssHII 2 1 1 l 14
C57BL/6 BssHII 14 1 O l 5
4 wk 0.05% PB
B6C3F1 BssHII 18 2 5 25
C57BL/6 BssHII 3 3 6 12

 

“ Hypomethylated (HypoM) RAMs include both statistically significant (p<0.05)
decreases and 100% decreases.

b Hypermethylated (HyperM) RAMs are only those increases which are statistically
significant (p<0.05).

CNew methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not present under control conditions.

128

7). This analysis served to re-enforce the results obtained with RsaI/MspI and
RsaI/HpaII digests.

GC-rich regions are associated with the promoter regions of genes and altered
methylation within these regions could affect gene expression. Therefore, the
methylation status of a 5 ’ promoter region of Ha-ras oncogene was analyzed in addition
to an upstream region of the LINE-1 element. The methylation status of a 283bp region
of the Ha-ras promoter, containing 22CpG dinucleotides, which was 950nt upstream of
the transcriptional start site was evaluated (Figure 4a). Bisulfite sequencing revealed 3
methylated cytosines within the targeted region under control conditions in both the
B6C3F1 and C57BL/6 mice. One additional cytosine was methylated in 50% and 67% of
the control animals in B6C3F1 and C57BL/6 mice respectively(Figure 4b). With 2wk
0.05% PB the methylation status of one cytosine at -1174nt decreased only in the
B6C3F1 mice. This site of altered methylation (-1174nt) was seen to reverse following
2wks recovery (Figure 4b). Following 4 wk of PB treatment, the B6C3F1, but not
C57BL/6 mice, also exhibited hypomethylation of the cytosine at position -1174. In
addition, 2/6 animals were hypomethylated at a second cytosine (-1204nt) (Figure 40).
Following the 2wk recovery period the cytosine at -1204 of all 6 animals, and 3/6 at -
1174nt reversed to a methylated status (Figure 4c). Bisulfite sequencing was also
performed on a 5’ region of the LINE-1 element containing 11CpG dinucleotides 975nt
upstream of the transcriptional start site. All 11 cytosines were methylated in control
animals as well as animals treated with 2 or 4wk 0.05% PB (Figure 5).

With evidence for hypomethylation within a limited region of the Ha-ras

promoter only in the B6C3F1 mice at 2 and 4wk, the effect on gene expression levels was

129

A.

4°“ 1“
A "N .950
yw .-
0
4232
B.
'09“ (0"

5' W 3’ C57BU6 and B6C3F1 Control
9W 3' (3573116 0.05% PB
5‘ W 3' B6C3F1 0.05°/o PB

S‘W 3' B6C3F1 0.05% PB RECOVERY

5* W 3‘ C57BU6 and B6C3F1 Control

5' W 3' C57BU6 035% PB

5’ w 3’ BGC3F1 0,05% PB RECOVERY

Farm 4. Methylation Status of the Promoter Region of Ha-ras. A diagram of the
Ha-ras promoter indicating location of PCR primers and CpG sites (gray lollipops) in

relation to the transcriptional start site is presented (a). Bisulfite sequencing analysis
within the region spanning 283bp revealed that 3 cytosines are methylated (black
lollipops) in both B6C3F1 and C57BL/6 control mice. In addition, the cytosine at
position -1033 was methylated in 50% and 65% of B6C3F1 and C57BL/6 mice,
respectively. This is represented by a half black and half white lollipop. PB-induced
hypomethylation (striped lollipop) of the cytosine at -1174nt was only seen the B6C3F1
mice at 2wks. This hypomethylated state reversed following 2wks recovery (b). At 4wks,
3 cytosines were methylated in control animals. PB induced hypomethylation of the
cytosine at position -1174nt in B6C3F1 mice and hypomethylation of the cytosine at
position -1204nt in 30% of B6C3F1 mice. With 2wks recovery, the methylation status of
the cytosine at -1204nt fully reversed while reversal was only seen in 50% of the B6C3F1
mice at position -1174nt (c). Black = methylated, White = unmethylated, Striped =
Hypomethylated in 100% of animals.

130

A.

-975
EE EEEE E E: 9‘ F ORF1 F ORF2
5' 31
.1191

#0 +1128

TM 3' C57BL/6 and 86C3F1 Control
9% 3' C57BL/6 and B6C3F1 “-05% PB

- F'gpre 5. Methylation Status of the Promoter Region of LINE-11 in B6C3F1 and
C57BL/6 Mouse Liver. A diagram of the LINE-l promoter indicating location of PCR

primers and CpG sites (gray lollipops) in relation to the transcriptional start site of the 1gt
open reading frame (ORF) is presented (a). Bisulfite sequencing analysis within the
outlined region revealed that all 11 cytosines are methylated (black lollipops) in both
B6C3F1 and C57BL/6 control and treated mice. (b).

131

investigated. Fold change of Ha—ras expression over normalized control levels was
measured. No change in gene expression was observed between control and PB treated
animals at 2wks (Figure 6). At 4wks selective increases in expression were observed in
response to PB (Figure 7). Three animals exhibited a level of expression which exceeded
the upper 95% confidence limit of the controls (Figure 7). Following two weeks
recovery, 4/6 animals expressed basal levels of Ha-ras expression (Figure 8). In
comparison, Ha—ras expression levels in C57BL/6 were unaffected by 2 or 4wks PB
treatment (Figure 9). The methylation status of cytosines within the targeted region of
the LINE-1 promoter was unchanged by treatment with PB in both B6C3F1 and
C57BL/6 mice. Consistent with this, the levels of expression of LINE-l in control and

treated animals at 2 and 4wks were comparable (Figure 10).

132

         

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.00 -
2.62
2.50 3
2(1)!
1.54
o 1.52 1'63
E’ i l
5 I 50 -l 4%;
P 1.00 :92};
'5 1‘21 0.97
,4
Pi“:
0.72 064 33%:
38 fl
0504 0.26 «III
1 9'3 ....... l
r i .-
000 4 r 4 4 4 4 "
861 862 863 86-4 865 86-6 Control 86-1 862 863 86-4 B6-5 0.05%
Mean PB
Control Animals 2wk 0.05% PB Treated Animals M”

Fig1_rre 6. Effect of 2wk 0.05% Phenobarbital Promotion on the Expression of Ha-

ras in B6C3F1 Mouse Liver. PB-lnduced changes in the gene expression of Ha-ras at
2wks in B6C3F1 mice as detected by real-time PCR are expressed as fold change over
control. Six control animals and 5 PB treated animals were assayed. Treatment did not

increase the expression of Ha-ras at 2wks.

133

5.007 '

4.50-
4.004
3.50-

3004

210'

Fold Change

1.50‘

1.1” a

* Fold increase above
upper 95% Confidence

limit of controls

1.30

L

 

 

1.37

 

 

0.87
0.77

 

 

0.1!)

“-1

0.91
H 0.79
“-2 “-3

K4

ll

1.68

1.49

1.00

 

$5 m Control 0.05% “-1 “-2 m3 “-4 M “-6

 

Control Animals

liken

 

PB
he... 4wk 0.05% PB Treated Animals

Figpre 7. Effect of 4wk 0.05% Phenobarbital Promotion on the Expression of Ha-
ras in B6C3F1 Mouse Liver. PB-Induced changes in the gene expression of Ha-ras at

4wks in B6C3F1 mice as detected by real-time PCR are expressed as fold change over
control. Six control animals and 6 PB treated animals were assayed. Increases in Ha-ras
expression which exceeded the upper limit of the 95% confidence interval of the control
animals are noted with an asterisk C“).

134

3.50'l

tool

E
to
8

1.94

Fold Change

ZW‘ 1.50

”“ ' 5;??in
m ”5 ”1
0.61 ‘ _.

to. r

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mlvm-va-S'W'K-S'M'Comml 0.05% 861862
When PB
Control Animals Mm 4wk 0.05%/2wk Recovery Animals

 

 

Figure 8. Reversal of Increased expression following 4wk Phenobarbital Exposure

_ip_B6C3F1 Mouse Liver. PB-Induced changes in the gene expression of Ha-ras
following 4wks PB treatment and a 2wk recovery period in B6C3F1 mice was detected
by real-time PCR and expressed as fold change over control. Six control animals and 6
treatment/recovery animals were assayed. Increases in Ha-ras expression which
exceeded the upper limit of the 95% confidence interval of the control animals are noted
with an asterisk (*).

135

 

 

1.33

           

1.09

\\\\\\\ \\\\\\\\\w

    
   

2wk 0.05% PB Treated Animals

Treelled
Mean

4wk 0.05% PB Treated Animals

 

1.40

 

\\\\\\\\\\ \\\\\\\\\\\\ I\...

i
.6... .\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\\\m .\\\\\\\\\ \\\\\\\\I\\\\\\\
.. \\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\ \. w o. \\\\\\\\I\I\\\\\\\\\\\

      

 

1.03

 

0.95

 

 

 

 

 

 

 

 

 

1.60

 

1.40

 

355 Eng

60
.40 ,
020“

 

0.00

Control Animals

057-1 C57-2 C57-3 C57-4 C57-5 057-6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

D.

 

 

140

120

1.00—
080~

060

040—- E— - # ~— E

020
000

35:0 so.“

057-1 C57-2 C5745 C57-4 0576 057-6 C57

C57

C57-1 C57-2 C57-3 057—4 C57-5 C57-6

 

Control Animals

 

Fi ure 9. Effect of 2 and 4wk Phenobarbital on the Ex ression of Ha-ras in
C57BL/6 Mouse Liver. PB-Induced changes in the gene expression of Ha-ras at 2 wks
(a) and 4wks (b) in C57BL/6 mice was detected by real-time PCR and expressed as fold
change over control. Six control animals and 6 PB treated animals were assayed at each
time point. Treatment did not increase the expression of Ha-ras at 2 or 4wks.

136

m§\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\x

mV\\\\\\\\\\\\\\\\\\\\\\\\\\

\\\\\\\\\\\\\\\\\.

ms\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\.
mx\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\x
mV\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

m§§§§m

 

 

 

 

 

 

 

 

 

136-2 86-3 B64 136-5 36-6 Control

 

 

 

 

1.20 a

355 son.

0.40-

0.“)

1

136-1

liken

 

Mean

 

4wk 0.05% PB Treated Animals

Control Animals

m.
mV\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\s

m\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
m§\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\.

m§§§§

l

mV\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\§
m§§§§

 

 

 

 

 

 

 

 

 

 

 

 

 

C57-S C57-6

C57-l C57-2 C57-3 C57-4

 

1 20
00
80
60

0.40 '

0.20

0.00

00:20 20“.

C57-3 C57-4 C57-5 C57-6 0.05%

Control C57-l C57-2

PB

Mean

 

 

4wk 0.05% PB Treated Animals

Fi re 10. Effect of 4wk Phenobarbital Promotion on the Ex ression Level of

Control Animals

gene expression of LINE-1 elements at 4 wks in B6C3F 1 (a) and C57BL/6 (b) mice was detected

by real-time PCR and expressed as fold change over control. Six control animals and 6 PB
treated animals were assayed. Treatment did not increase the expression of LIN E-l elements at

LINE-1 Elements in B6C3F1 and C57BL/6 Mouse Liver. PB-Induced changes in the
4wks in either B6C3F1 or C57BL/6 mouse liver.

137

DISCUSSION

The initiation stage of tumorigenesis involves an irreversible alteration of the
genome via a mutation or possibly epi genetic event (Dragan et al., 1993; Goodman and
Watson, 2002). The cells that acquire the first critical initiating event(s) could,
theoretically, be derived from stem cells, early stem cell progenitor cells, or differentiated
cells (Bjerkvig et al., 2005). These cells which have acquired a growth advantage over
neighboring cells within the context that they might be selected for and clonally expand
during promotion; thus, tumorigenesis involves the progressive, clonal expansion of cell
populations which accumulate heritable changes to their genomes leading to them
becoming increasingly abnormal (Nowell, 1976). Cell self-renewal is an important
feature of the liver as evidenced by the regenerative response hepatocytes following
partial hepatectomy, i.e., the hepatocyte is a differentiated cell that remains capable of
replication (F austo, 1997). Adult mouse liver cells (i.e. mature hepatocytes) transplanted
into transgenic mice have the capacity to divide at least 12 times (Tateno and Yoshizato,
1996). This capacity for cell proliferation indicates that hepatocytes themselves can be
the functional stem cells of the liver (Forbes et al., 2002). It follows that hepatocytes
might also serve as the progenitor cells for liver tumors, and this is not incompatible with
a role for stem cells, too. Thus, the production of altered methylation in hepatocytes and
the progressive accumulation of these heritable epigenetic changes might contribute to
hepatocarcinogenesis

During DNA replication stable patterns of methylation are mainly dependent on
the maintenance methyltransferase, DNMTl. However, methylation is also regulated by

de novo methyltransferases (e. g. Dnmt3a and 3b), demethylases (Pradhan and Esteve,

138

2003), and the availability of both SAM and methyl groups (Zeisel, 1996). Altered
patterns of methylation, specifically hypomethylation, might arise when the levels of
Dnmtl and/or SAM are low in replicating cells. Alternatively, aberrant activity of
demethylases and de novo methylases could induce hypo- and hypermethylated states in
quiescent cells. In this context, the activity of PB could disrupt multiple factors. One
mediator of PB responses is the nuclear receptor CAR (constitutive active/androstane
receptor). This relationship is important to consider in that CAR knockout mice do not
develop eosinophilic foci or advanced liver tumors when promoted with PB (Yamamoto
et al., 2004). Still, of 138 hepatic genes increased or decreased by PB only about 50%
are regulated by CAR which implies that PB has numerous CAR-independent actions
(Ueda et al., 2002).

The actions of PB result in complex patterns of altered methylation in many if not
all of the cells in the liver, with multiple unique changes produced in the hepatic DNA of
liver tumor-prone B6C3F1 mice. However, within individual hepatocytes, these changes
are not simply random as demonstrated by the fact that PB-induced altered methylation
was detected within the promoter region of the Ha-ras oncogene whereas the methylation
status of LIN E-l elements, a very large portion of the genome (~33%), was unaffected
over the course of treatment with PB. Indicating that normal patterns of methylation are
maintained in a more stable fashion in some portions of the genome as compared to
others. Importantly, the fact that methylation of LINE-1 elements, a portion of the “junk
DNA” which makes-up approximately 30% of the genome did not change indicates that
the PB-induced altered methylation detected over a four wk course of treatment was

concentrated within the pool of “regular” genes. When this occurs within critical genes,

139

e.g., those controlling portions of genes that regulate cell proliferation, it could facilitate
their aberrant expression, providing the cells affected in this fashion with a growth
advantage. Thus, the accumulation of aberrant RAMs, particularly those which progress
during early times after the start of PB treatment, might be involved in initiation and/or
promotion of tumorigenesis. The progressive accumulation of changes during PB
promotion was substantiated by the fact that 17% of the RAMs in B6C3F 1 mice carried
forward from 2 to 4 weeks. Strikingly, less than 1% of the C57BL/6 RAMs carried
forward (Figure lb). These RAMs which carried forward likely represent the
aforementioned critical epigenetic changes which contribute to the clonal proliferation of
subsets of initiated cells. C57BL/6 PB-induced RAMs were used as the “control”
comparison to PB-induced RAMs in B6C3F1 mice at 2 and 4wks. At 2wks 12 RAMs
were in common and equivalent between B6C3F1 and C57BL/6. By subtracting out
these particular RAMs 57 RAMs (43 hypomethylations, 3 hypermethylations, and 11
new methylations) were identified as unique to the B6C3F1 mice. Comparison at 4wks,
identified 86 RAMs (48 hypomethylations, 10 hypermethylations, and 28 new
methylations) unique to B6C3F1 mouse liver. A comparison of the 86 unique RAMs in
B6C3F1 liver to RAMs in the kidney further refined the number of unique RAMs to 81,
in that only 5 new methylations were in common and equivalent to kidney. These 81
RAMs, specific to B6C3F 1 mouse liver are likely critical to the development of tumors.
In addition, they could be important in defining the inherent susceptibility of the B6C3F1
mice to liver tumorigenesis. The ability to associate each unique RAM with a specific

gene will be key to addressing this issue and is presently being pursued.

140

Tmnors arising from promotion with PB have defined characteristics.
Specifically, activation of Ha—ras via point mutation is a not a common feature of PB-
induced tumors (Fox et al., 1990), but is frequently seen in spontaneously arising tumors
in B6C3F1 mice (Maronpot et al., 1995). In addition, the down regulation of TGF-B, a
cell cycle inhibitor, is dependent on promotion with PB (Reisenbichler et al., 1994). Pre-
TGF-B is metabolized to its active form by the mannose 6-phosphate/insulin growth
factor II receptor (M6P-IGF2r) and, therefore, the gene encoding M6P-IGF2r can be
viewed as a tumor suppressor gene (DeSouza et al., 1995). M6P-IGF 2r is an imprinted
gene in rodents, but is biallelically expressed in humans (Xu et al., 1993), which is
regulated by methylation. Methylation of a CpG site in an intron of the expressed allele
is the imprinting single which is required, in addition to hypomethylation of the 5’
flanking region, for its normal expression (Stoger et al., 1993). Two key observations of
the current study are directly related to these findings: 1) PB induces both increases and
decreases of methylation in GC-rich regions, and 2) PB-induced hypomethylation of the
Ha—ras gene was associated with selective increases in gene-expression. At 4wks, PB
induced 48 unique hypomethylated RAMs in addition to 10 unique hypermethylated
RAMs and 23 new methylations in the B6C3F1 mice. The simultaneous
hypermethylation of tumor suppressor genes and hypomethylation of oncogenes could
facilitate the formation of tumors. For instance, either hyperrnethylation of the promoter
region of the M6P-IGF2r gene or hypomethylation leading to removal of the imprinting
signal could silence this “tumor-suppressor” and lead to decreased levels TGF-B which
would remove the growth inhibition and enhance cell proliferation. In this context, it is

instructive to a recent report indicating that hypomethylation leading to disruption of

141

imprinting can lead to tumorigenesis (Holm, et al., 2005). In addition, hypomethylation
might result in up-regulation of oncogenes. This has been suggested earlier for Ha-ras
and raf (Ray et al., 1994; Counts et al., 1996) and with regard to Ha-ras more
convincing evidence is provided in the current study.

In the working model of carcinogenesis reversibility is characteristic of tumor
promotion as continued exposure to the promoting agent is necessary for progressive
clonal expansion of cells exhibiting increasingly abnormal phenotypes. In light of this,
reversibility of changes in methylation in response to a 2 or 4wk exposure to PB were
assessed following 2wk recovery periods. The large majority of hypermethylations and
new methylations were reversible however, very few hypomethylations reversed. De
novo methylation of hypomethylated CCGG sites could restore the normal methylation
pattern. However Dnmt3a and 3b prefer AT-rich flanking sequences surrounding the
CpG dinucleotide (Handa and Jeltsch, 2005). Biologically, this difficulty in reversing
hypomethylation could be necessary for the continued promotion of initiated cells.
Global hypomethylation of CCGG sites induced by a methyl-deficient diet in rats was
stable during re-exposure to a methyl-adequate diet and correlated to the persistence of
altered hepatic foci (Pogribny et al., 2005). In our model, progressive changes in
methylation were only seen in the B6C3F1 mice; however, minimal reversibility of
hypomethylated RAMs was a common feature of both the sensitive and resistant mice.
This indicates that irreversible hyomethylation may be necessary but not sufficient for the
continued expansion of clones of initiated cells in response to PB.

Genetic differences likely contribute to the observed divergence in the ability to

maintain patterns of DNA methylation during promotion with PB. The hepatocarcinogen

142

sensitive locus (hcs) has been suggested to account for approximately 85% of the
difference in strain susceptibilities (Drinkwater and Ginsler, 1986). Furthermore, this
locus appears to affect the growth rate of pre-neoplastic foci during the promotion stage
of tumor induction in a strain dependent manner (Hanigan et al., 1988; Goldsworthy and
F ransson-Stem, 2002). Therefore, tumor induction facilitated by altered patterns of
methylation could be enhanced by the pre-existing genetic predisposition. For instance,
strain differences in the activity of one or more de novo methyltransferase enzymes could
potentially account for the extreme difference between B6C3F1 and C57BL/6 mice at 2
and 4wks. B6C3F 1 RAMs with increased methylation, were few in number at 2wks (23
RAMs) but increased by 4wks (39 RAMs). This response was in direct opposition to
C57BL/6 where high numbers where seen at 2wks (86 RAMs), and dramatically less at
4wks (17 RAMs). Therefore, the hcs might functionally involve a compromised ability
to maintain normal methylation patterns in liver tumor-prone mice, e. g., the B6C3F1
mouse.

Methylation sensitive restriction digestion, arbitrarily primed PCR combined with
capillary electrophoretic detection of PCR products is a novel approach to simultaneously
measuring increases, decreases, and new methylations in multiple GC-rich regions
throughout the genome. GC-rich regions, including both G + C rich and CpG island
regions are closely associated with gene promoters. CpG islands are short stretches of
DNA, at least 200bp in length, possessing 50% or greater GC content and a higher
proportion of CpG dinucleotides than expected (Gardiner-Garden and Frommer, 1987).
Basically, three complementary sets of analyses were performed: RsaI/HpaII, RsaI/Mspl

and BfaI/BssHII. Overall, altered methylation at ~7.45% of all CpG dinucleotides

143

estimated for the mouse genome is detected by the 4-base cutter isoschizomers, Mspl and
Hpall which complement each other’s sensitivity to inhibition by methylation within
their common recognition sequence (Frazarri, and Greally, 2004). Analysis of CpG
islands is more directly targeted by using the BfaI/BssHII combination. The 6-base
recognition site of BssHII is rare, as compared to the 4-base recognition site of Mspl and
Hpall, and tightly associated with CpG islands (Shiraishi et al., 1995). Furthermore,
concerning the non-methylation sensitive restriction enzymes employed there are fewer
restriction sites within CpG islands for Bfal as compared to RsaI (Shiraishi et al., 1995).
With each analysis, an increase in the total number of RAMs from 2 to 4 wks in B6C3F1
mouse was identified (Tables 1 and 6). In addition, less RAMs were observed in the
C57BL/6 mice at 4wks. Therefore, these restriction enzyme combinations are
complementary and reinforce the notion that our method is capable of providing insight
regarding the methylation status of the genome.

The differential ability to maintain patterns of DNA methylation is hypothesized
to contribute to the variable susceptibilities of B6C3F1 and C57BL/6 mice. As an
extension of previous reports (Counts et al., 1996; Watson and Goodman, 2002), we have
demonstrated that patterns of methylation are more susceptible to disruption in the
sensitive mouse. lmportantly a large proportion of the regions of altered methylation in
B6C3F1 mouse liver carried forward in comparison to C57BL/6. This is a highly
significant observation and shows that B6C3F1 mice accumulate changes much quicker
and earlier than C57BL/6 which is in direct agreement with their relative sensitivities to
tumor formation. This is strong experimental evidence which indicates that the

progressive accumulation of heritable changes are key to the promotion stage and

144

facilitate tumorigenesis. Gene-specific analysis in B6C3F1 mice revealed a non-random
pattern of altered methylation in which Ha-ras hypomethylation correlated to selective
increases in gene expression and the heavily methylated LINE-1 elements were
unaffected by PB. Collectively, these data indicate that: 1) the progressive, non-random
changes in methylation play an important role in tumorigenesis; 2) altered DNA
methylation is an epigenetic mechanism underlying the ability of PB to cause liver
tumorigenesis; and 3) susceptibility to tumorigenesis is related inversely to the capacity

to maintain normal patterns of methylation.

145

REFERENCES

Becker, RF. (1982). Morphological classification of mouse liver tumors based on
biological characteristics. Cancer Res. 42, 3918-3923.

Bjerkvig, R., Tysnes, B.B., Aboody, D.S., Najbauer, J. and Terzis, A.J.A. (2005). The
origin of the cancer stem cell: current controversies and new insights. Nat. Rev. Cancer
5, 899-904.

Brown, K., et al. (1988). Isolation and characterization of the 5’ flanking region of the
mouse c-Ha-ras gene. Mol. Carcinogenesis 1, 161-170.

Buchmann, A., Bauer-Hofmann, R., Mahr, J ., Drinkwater, N.R., Luz, A. and Schwarz,
M. (1991). Mutational activation of the c-Ha-ras gene in liver tumors of different rodent

strains: Correlation with susceptibility to hepatocarcinogenesis. Proc. Natl. Acad. Sci.
USA 88, 911-915.

Carnell AN. and Goodman, J .I. (2003). The long (LINES) and the short (SINEs) of it:
Altered methylation as a precursor to toxicity. Toxicol. Sci. 75, 229-235.

Costello, J .F . and Plass, C. (2001). Methylation matters. J. Med Genet. 38, 285-303.

Counts, J .L. and Goodman, J.I. (1995). Alterations in DNA methylation may play a
variety of roles in carcinogenesis. Cell 83, 13-15.

Counts, J .L., Sarmiento, J .I., Harbison, M.L., Downing, J .C., McClain, R.M. and
Goodman, J .I. (1996). Cell proliferation and global methylation status changes in mouse

liver after phenobarbital and/or choline-devoid, methionine-deficient diet administration.
Cell 17, 1251-1257.

DeSouza, A.T., Hankins, G.R., Washington, M.K., Fine, R.L., Orton, TC, and Jirtle,
R.L. (1995). Frequent loss of heterozygosity of 6q at the mannose 6-phosphate/insulin-
like growth factor II receptor locus in human hepatocellular tumors. Oncogene 10,

1725-1729.

Dragan Y.P., et al., (1993). The initiation-promotion-progression model of rat
hapatocarcinogenesis. Proc Soc Exp Biol Med. 202, 16-24.

Drinkwater, NR. and Ginsler, J .J . (1986). Genetic control of hepatocarcinogenesis in
C57BL/6J and C3H/HeJ inbred mice. Carcinogenesis 7, 1701-1707.

Fausto, N. (1997). Hepatocytes break the rules of senescence in serial transplantation
studies. Am. J. Pathol. 151, 1187-1189.

146

Fazzari, MJ. and Greally, J .M. (2004). Epigenomics: Beyond CpG islands. Nat. Rev.
Genet. 5, 446- 455.

Forbes, S., Vig, P., Poulsom, R., Thomas, H., and Alison, M. (2002). Hepatic stem cells.
J. Path. 197, 510-518.

Fox, T.R., Schumann, A.M., Watanabe, P.G., Yano, B.L., Maher, V.M., and McCormick,
J .J . (1990). Mutational analysis of the H-ras oncogene in spontaneous C57BL/6 x
C3H/He mouse liver tumors and tumors induced with genotoxic and nongenotoxic
hepatocarcinogens. Cancer Res. 50, 4014-4019.

Gardiner-Garden, M. and Frommer, M. (1987) CpG islands in vertebrate genomes. J.
Mol. Biol. 196, 261-282.

Gaudet, F ., Hodgson, J .G., Eden, A., Jackson-Grusby, L., Dausman, J ., Gray, J .W.,
Leonhardt, H., and J aenisch, R. (2003). Induction of tumors in mice by genomic
hypomethylation. Science 300, 489-492.

Goldsworth, TL. and F ransson-Steen, R. (2002). Quantitation of the cancer process in
C57BL/6J, B6C3F1 and C3H/HeJ Mice. T oxicol. Path. 30, 97-105.

Gonzalgo, M.L., Liang, G., Spruck, CH. 111, Zingg, J-M., Rideout, W. M. 111, and Jones,
RA. (1997). Identification and characterization of differentially methylated regions of

genomic DNA by methylation-sensitive arbitrarily primed PCR. Cancer Res. 57, 594-
599.

Goodman, J .I. and Watson, RE. (2002). Altered DNA methylation: A secondary
mechanism involved in carcinogenesis. Annu. Rev. Pharmacol. T oxicol. 42, 501-525.

Handa, V. and Jeltsch, A. (2005). Profound flanking sequence preference of Dnmt3a
and Dnmt3b mammalian DNA methyltransferases shape the human epigenome. J. Mol.
Biol. 348, 1103-1112.

Hanigan, M.H., Kemp, C.J., Ginsler, J .J . and Drinkwater, NR. (1988). Rapid growth of
preneoplastic lesions in hepatocarcinogen-sensitive C3H/HeJ male mice relative to
C57BL/6J male mice. Carcinogenesis 9, 885-890.

Holm, T.M., Jackson-Grusby, L. Brambrink, T., Yamada, Y., Rideout III, W.M., and
Jaenisch, R. (2005). Global loss of imprinting leads to widespread tumorigenesis in
adult mice. Cancer Cell 8, 275-285.

Jones, RA. and Baylin, SB. (2002). The fundamental role of epigenetic events in
cancer. Nat. Rev. Genet. 3, 415-428.

147

Maronpot, R.R., Fox, T., Malarkey, DE, and Goldsworthy, TL. (1995). Mutations in
the ras proto-oncogene: clues to etiology and molecular pathogenesis of mouse liver
tumors. Toxicology 101, 125-156.

Neades, R., et al. (1991). Transient expression of the cloned mouse c-Ha—ras 5’
upstream region in transfected primary SENCAR mouse keratinocytes demonstrates its
power as a promoter element. M01. Carcinog. 4, 369-375.

Nowell, PC. (1976). The clonal evolution of tumor cell populations. Science 194, 23-
28.

Pogribny, I.P. et al. (2005). Irreversible global DNA hypomethylation as a key step in
hepatocarcinogenesis induced by dietary methyl deficiency. Mutat. Res. In Press.

Pradhan, S. and Esteve, P.-O. (2003). Mammalian DNA (cytosine-5) methyltransferases
and their expression. Clin Immun. 109, 6-16.

Ray, J .S., Harbison, M.L., McClain, R.M., and Goodman, J .I. (1994). Alterations in the
methylation status and expression of the raf oncogene in phenobarbital-induced and
spontaneous B6C3F1 mouse liver tumors. Mol. Carcino. 9, 155-166.

Reisenbichler, H., Chari, R.S., Boyer, I.J., and J irtle, R.L. (1994). Transforming growth
factor-beta receptors type I, II and III in phenobarbital-promoted rat liver tumors.
Carcinogenesis 15, 2763-2767.

Shiraishi, M., Lerman, LS, and Sekiya, T. (1995). Preferential isolation of DNA
fragments associated with CpG islands. Proc. Natl. Acad. Sci. USA 92, 4229-4233.

Stoger, R., Kubicka, P., Liu, C.G., Kafri, T., Razin, A., Cedar, I-I., Barlow, DP. (1993).
Matemal-specific methylation of the imprinted mouse Igf2r locus identifies the expressed
locus as carrying the imprinting signal. Cell 73, 61-71.

Stowers, S.J., Wiseman, R.W., Ward, J .M., Miller, E.C., Miller, J .A., Anderson, M.W.
and Eva, A. (1988). Detection of activated proto-oncogenes in N-nitrosodiethylamine-
induced liver tumors: a comparison between B6C3F1 mice and Fischer 344 rats.
Carcinogenesis 9, 271-276.

Tateno, C. and Yoshizato, K. (1996). Growth and differentiation in culture of

clonogenic hepatocytes that express both phenotypes of hepatocytes and biliary epithelial
cells. Am. J. Path. 149, 1-13

Ueda, A. et al. (2002). Diverse roles of nuclear orphan receptor CAR in regulating
hepatic genes in response to phenobarbital. Mol. Pharmacol. 61, 1-6.

148

Watson, RE. and Goodman, J .I. (2002). Effects of phenobarbital on DNA methylation
in GC-rich regions of hepatic DNA from mice that exhibit different levels of
susceptibility to liver tumorigenesis. T oxicol. Sci. 68, 51-58.

Whysner, J ., Montandon, F., McClain, R.M., Downing, J ., Verna, L.K., Steward 3”, RE.
and Williams, GM. (1998). Absence of DNA adduct formation by phenobarbital,
polychlorinated biphenyls, and chlordane in mouse liver using the 2P-postlabeling assay.
Toxicol. App. Pharmacol. 148, 14-23.

Whysner, J ., Ross, RM. and Williams, GM. (1996). Phenobarbital mechanistic data and
risk assessment: Enzyme induction, enhanced cell proliferation, and tumor promotion.
Pharmacol. Ther. 71, 153-191.

Yamamoto, Y., Moore, R., Goldsworth, T.L., Negishi, M., and Maronpot, RR. (2004).
The orphan nuclear receptor constitutive active/androstane receptor is essential for liver

tumor promotion by phenobarbital in mice. Cancer Res. 64, 7197-7200.

Ziesel, SH. (1996). Choline: A nutrient that is involved in the regulation of cell
proliferation, cell death and transformation. Adv. Exp. Med. Biol. 399, 131-141.

149

CHAPTER 2 APPENDIX

This appendix contains additional research results that are complementary to the body of
data presented in Chapter 2. The materials and methods needed to extend the analysis
and detection of altered methylation in response to promotion with PB in the promoter
region of the Ha—ras gene is provided. Results are presented along with a brief discussion
of the findings and relevant references.

150

MATERIALS AND METHODS

Gene-Specific Methylation Analysis: Methylation Sensitive Restriction Digestion of
Ha-ras

Restriction Digests

DNA was restricted with Smal or Apal endonuclease (Invitrogen), methylation sensitive
enzymes which recognize the CCCGGG and GGGCCC sequence, respectively. Smal
restricts DNA only if the cytosine immediately 5’ to guanine is not methylated. Apal will
not restrict the DNA if the cytosine immediately 3’ to guanine is methylated. The target
region will be amplified with subsequent PCR only if these target sites are methylated.
Each reaction was composed of lug DNA, 3 units Smal or Apal enzyme, 1X React4
Buffer (Invitrogen), and GDW to a final volume of lOul. Negative control digests (i.e.
Smal/Apal was omitted) for each animal were also prepared. Reactions included, lug
DNA, 1X React4 Buffer and GDW to a final volume of 10ul. All samples were

incubated at 370C for 16hours.

PCR Amplification

PCR was carried out on digested (i.e. incubated with Smal/Apal enzyme (+SmaI/Apal))
or undigested (i.e. incubated without Smal/Apal enzyme (-SmaI/Apal)) DNA. Forward
(5’ CAG GGT GGA GGC TCT GTA GT 3’) and reverse (5’ GAG AGG AGC AAG
GAA GCA CC 3’) primers (Okoji et al., 2002) amplify the ——325 to +200 region spanning
the transcriptional start site of the Ha-ras gene which contains two Smal restriction sites
and 1 Apal restriction site. Reactions consisted of IX Failsafe Buffer H (Epicentre

Technologies), 2uM each primer, 0.43ug digested or undigested DNA, and GDW to a

151

final volume of 25ul. Cycling conditions were: 80°C for 5min, 940C for 2min, 24 cycles
of 96°C for 1min, 65°C for 1min, and 720C for 2min, followed by 1 time delay cycle of
72°C for 5min. PCR products were electrophoresed (3% agarose gel) and visualized by
ethidium bromide staining. A Polaroid picture of each gel was taken, scanned and the
image was analyzed using the NIH image program. The number of PCR cycles (24) was
chosen following a pilot study in which 21, 23, and 25 cycles were tested. The chosen
number of cycles was estimated to maximize the difference in PCR product band
intensities between the undigested control DNA and the digested DNA. Because a
heterogeneous mixture of cells possibly expressing varying degrees of methylation were
sampled, increases and decreases in band intensity compared to control is proportional to

increases and decreases in methylation of the target restriction sites.

Quantification of Band Intensity

The relative intensity of a PCR product band corresponds to the relative starting
concentration of methylated DNA. Each target PCR product band was outlined and
measured for pixel number and intensity by using the NIH image program (http:/l
rsbi.info.nih.gov.). The number of pixels defined the size of the outlined region. The
same sized region was used to measure both the PCR product band and the lane
background. Total pixel intensity units (TPI) were calculated by multiplying the number
of pixels by the mean intensity units within the outlined region. This was done separately
for both the background and PCR product band. The TPI units of the background was
subtracted from the TPI units of the PCR product band to give a normalized TPI for the

outlined region within a lane.

152

Calculation of Ratios and 95 % Confidence Interval (CI)

TPI units were calculated for PCR product bands resulting from amplification of either
digested (+SmaI/Apal) or undigested (-SmaI/Apal) DNA. For each animal, a ratio was
calculated to determine the amount of PCR product formed from digested DNA versus
undigested DNA. For this, a ratio of the TPI units from +SmaI/Apal DNA to TPI units
from —SmaI/ApaI DNA was calculated. The mean (n=5) ratio with standard deviation
and 95% CI (a=0.05) was calculated for the control animals only. Treated animal ratios
were compared individually to the control 95% CI. If a ratio from a treated animal fell
above the upper limit of the C1, the internal cytosine within the SmaI/ApaI sites was
considered to be hypermethylated. If a ratio from a treated animal fell below the lower
limit of the C1, the internal cytosine within the Smal/Apal sites was considered to be

hypomethylated.

RESULTS

Due to the fact that the upstream 5’ promoter region of Ha-ras exhibited
hypomethylation at both 2 and 4 weeks only in the B6C3F1 mouse, we wanted to
investigate the methylation status of CpG sites close to the transcriptional start site. This
was achieved via methylation sensitive restriction digestion with Smal or Apal
endonuclease. Appendix Figure 1a outlines the target 525bp region which spanned the
transcriptional start site of the Ha-ras gene and contained two Smal CCCGGG
recognition sites and 1 Apal GGGCCC recognition site. In B6C3F1 mice, the percent of

animals exhibiting hypomethylation at the two Smal sites progressed from 0% to 40%

153

while hyperrnethylation was seen to decreased from 33% to 20% with 2 and 4 wks of PB
promotion, respectively (Appendix Figure 1b). The incidence of hypomethylation in a
group of 6 C57BL/6 mice remained largely the same at 2 and 4wks and hyperrnethylation
increased slightly with 4wks PB (Appendix Figure lb). The Apal recognition site is
located very near (-10nt) to the transcriptional start site (Appendix Figure 1a).
Methylation at this site was generally maintained in the B6C3F1 mice from 2 to 4wks.
Low incidences of hypo- and hyper-methylation were observed. Interestingly, the
C57BL/6 mice were considerably hypomethylated at 2wks, but by 4wks, the incidence of

hyperrnethylation increased to 100% (Appendix Figure 1c).

154

 

Appendix Figure 1 Analysis of Altered Methylation Near to the Transcriptional
Start Site of Ha-ras in B6C3F1 and C57BL/6 Mice at 2 and 4wks. A 325bp region
spanning the transcriptional start site of Ha-ras was amplified following digestion with
either Small or Apal restriction endonuclease. Recognition sites for each enzyme are
diagrammed (a). An increase in the incidence of hypomethylation at the Smal
recognition sites was observed fom 2 to 4wk in B6C3F1 but not C57BL/6 mouse liver
(b). By 4wks, 100% of C57BL/6 mice exhibited hyperrnethylation of the Apal site while
the status of methylation in B6C3F1 mouse liver DNA was unchanged (c). For each set
of 4 bars in parts (a) and (b), the first two bars represent the 2 and 4wk time points
respectively for B6C3F1 mice and the second two bars represent the 2 and 4 wk time
points for C57BL/6 mice. Analysis was performed on DNA from 6 control and 6 treated
animals at each time point

155

 

% of Total Animals

% of Total Animals

-255 -10 fl. +124 +200

 

- CCCGGG _ GGGECC I C CGGG

 

 

 

-325

0
Sma | Recognition Site CCCGGG
Apal Recognition Site GGGCCC

 

 

 

8

 

 

8

 

8

8

 

 

 

 

 

 

% of Total Animals % of Total Animals % of Total Animals
Exhibiting HypoM Exhibiting HyperM Exhibiting No Change

 

 

 

 

 

 

 

 

 

% of Total Animals % of Total Animals % of Total Animals
Exhibiting HypoM Exhibiting HyperM Exhibiting No Change

156

DISCUSSION

The results presented here are an extension of data obtained via bisulfite
sequencing of a 283bp region containing 22 CpG sites 950nt upstream of the Ha-ras
transcriptional start site. Hypomethylation exhibited by only the B6C3F1 mouse at one
possibly critical site (-1l74nt) was observed. This site was consistently hypomethylated
at 2 and 4wks in response to PB. Because this site was relatively far from the
transcriptional start site, the ability of changes in methylation to influence transcription of
the gene needed to be put in better context. Therefore, methylation sensitive restriction
digestion afforded the ability to detect differences in the methylation status of DNA at
three additional sites in control and treated animals. Hypomethylation of Smal sites were
observed only in the B6C3F1 mice which is consistent with the sequencing results.
Decreases in particular CpG sites or the density of methylation within a region can affect
the binding of proteins necessary for transcriptional repression (Ballestar and Wolffe,
2001). Digestion with Apal revealed hypermethylation of the internal cytosine of the
GGGCCC recognition sequence. Methylation of CpC sites is not a common occurrence
(Dodge et al., 2002). Therefore, the sequence context of this recognition site must be
considered. The 5 base flanking sequences of this site are as follows: 5’ m
GGGCCC GGC—CA 3’. Methylation of cytosines which are not directly 5’ to guanine but
appear in a string of cytosines, CmepG, has not been studied. Again, it has been
proposed that the density of methylation is just as important as single sites of methylation
which could be one possibility for an increase in the methylation status of this cytosine.
In Arabidopsis, CprG methylation plays a role in gene silencing and is mediated by

histone H3 lysine 9 methylation through interaction of the DNA methyltransferase gene

157

with methylated chromatin (Jackson et al., 2003). Therefore, both the increased

incidence of hypomethylation observed at the Smal only in the sensitive mice coupled to
the complete hyperrnethylation of the non-CpG site in the C57BL/6 mice supports a role
for activated Ha-ras in the facilitation of tumorigenesis and the underlying susceptibility

of the B6C3F1 mice.

REFERENCES

Ballestar, E. and Wolffe, A.P. (2001). Methyl CpG binding proteins. Targeting specific
gene repression. Eur. J. Biochem. 268, 1-6.

Dodge, J .E et al. (2002). De novo methylation of MMLV provirus in embryonic stem
cells: CpG versus non-CpG methylation. Gene 289, 41-48.

Jackson, J .P., Lindroth, A.M., Cao, X., and Jacobsen, SE. (2002). Control of CprG
DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416,
556-560.

158

CHAPTER 3

ALTERED METHYLATION IN GENE-SPECIFIC AND GC-RICH REGIONS OF
DNA IS PROGRESSIVE AND NON-RANDOM DURING PROMOTION OF SKIN
TUMORIGENESIS

This chapter represents a manuscript that will be submitted to Toxicological Sciences in

January, 2006. Authors include: Bachman, Ammie N., Curtin, Geoffrey M., Doolittle,
David J. and Goodman, Jay I.

159

ABSTRACT

Altered DNA methylation, an epigenetic mechanism, likely contributes to tumorigenesis,
with an inverse relationship existing between methylation in a promoter region and
transcription. Using the SENCAR 2-stage mouse skin tumorigenesis model, altered
methylation was characterized in precancerous tissue and in tumor tissue. Mouse skin
was initiated with 7, lZ-dimethylbenz[a] anthracene (DMBA) and promoted 3X/wk with
3, 9, 18, or 27mg cigarette smoke condensate (CSC) for 4, 8, or 29 wks; tumors were
collected at 29wks. In addition, reversibility of changes in methylation were assessed
following cessation of the promoting stimulus. DNA was isolated, and GC-rich
methylation was assessed quantitatively via methylation sensitive restriction digestion,
arbitrarily primed PCR, and electrophoretic separation of PCR products. Analysis
focused on regions of altered methylation (RAMs) which persisted from 4 to 8wks and
from 8wks to tumor tissue. Persistent RAMs (i.e. seen in precancerous tissue and carried
forward to tumors) are likely to play a key role in tumorigenesis. Twenty-two CpG sites
in an upstream region of the Ha-ras promoter were unmethylated in control skin, 27 mg
CSC and tumor tissue. At 2 CpG sites closer to the transcriptional start site the incidence
of hypomethylation increased with dose of CSC. Hypomethylation was detected in all
tumor samples. Expression of Ha-ras increased with 18 and 27mg CSC promotion, and
more so in tumor tissue. These data support our hypothesis that tumor promotion
involves an instability of the epigenome, providing an environment where changes in the
methylation status of specific regions of the genome accumulate progressively and

contribute to the clonal expansion of initiated cells that leads to tumor formation.

160

INTRODUCTION

Cancer is characterized by six fundamental changes in cell physiology including
self-sufficiency in growth signals, insensitivity to growth-inhibitory signals, evasion of
apoptosis, limitless replicative potential, sustained angiogenesis, and the ability to invade
and metastasize (Hanahan and Weinberg, 2000). These characteristics are the result of a
step-wise clonal expansion of cell populations bearing accumulated heritable genetic
changes which can involve the mis-regulation of critical cell cycle proteins, transcription
factors and signal transduction proteins.

A central paradigm in explaining events leading to tumorigenesis involves a
multistage and multi-step (e. g., multi-mechanism) model for the development of
precancerous lesions and their evolution into frank carcinomas. The stages defined by
this model are initiation, promotion and progression (Dragan et al., 1993). During
initiation, a heritable change occurs in the genome. The nature of this change can be
rooted in direct mutation of the DNA sequence or it could occur by an epigenetic
modification'(Goodman and Watson, 2002). Exceedingly aberrant subclone populations
arise from the progressive clonal expansion of initiated cells during promotion.
Promoting agents likely foster the growth of initiated cells by enhancing cell proliferation
or by inhibiting apoptosis (Schulte-Hermann et al., 1990). Reversibility is a key
characteristic of promotion (Dragan et al., 1993). If the promoting stimulus is
withdrawn, the altered cells possessing advantageous growth characteristics stop
proliferating and altered foci may “remodel”. Subsequent to the iterative nature of this
process is progression in which cells clonally expand, even in the absence of the

promoting stimulus, and typically exhibit marked changes in ploidy (Dragan et al., 1993).

161

The progressive accumulation of heritable changes is fundamental to the process
of carcinogenesis. A highly effective model to study this is the SENCAR 2-stage mouse
skin tumorigenesis model which demarcates the stages of initiation and promotion.
SENCAR mice are particularly sensitive to carcinogenesis induced by the combination of
7, 12-dimethylbenz[a]anthracene (DMBA) and 12-O-tetradecanoylphorbol 13-acetate
(TPA), when utilized as initiator and promoter; moreover, this stock of mice consistently
exhibits increased sensitivity for the induction of skin tumors when compared to other
available strains (Hennings et al., 1997; Coghlan et al, 2000). With this model the
contribution of both mutations and, importantly, epigenetic mechanisms can be
evaluated. Epigenetics is defined broadly as processes that establish heritable states of
gene expression without altering the DNA sequence. Specifically, altered patterns of
DNA methylation, (i.e. 5-methylcytosine content of DNA), have been shown to occur
during the promotion stage of skin tumorigenesis. Previous studies in our lab showed
that cigarette smoke condensate, an effective promoting agent in the 2-stage model,
induced reversible dose and time-dependent changes in methylation in GC-rich regions of
DNA as well as global decreases in methylation in tumor tissue (Watson et al., 2003).

GC-rich regions of DNA are frequently associated with the promoter regions of
genes, and the methylation status of promoter regions can be linked to the regulation of
gene expression. Specifically, increased methylation in a promoter region might decrease
gene expression, while a decrease in promoter methylation possesses the potential for up
regulating gene expression (Jones and Laird, 1999). The silencing of tumor suppressor
genes via an increase in promoter methylation has been demonstrated for genes such as

06-methylguaunine-DNA methyltransferase, cyclin-dependent kinase inhibitor 2B, and

162

RASSF 1A (Watson et al., 2004; Jones and Baylin, 2002). In addition HoxAS which
upregulates p53 expression was shown to be hypermethylated following promotion with
27 and 36mg CSC for 9wks which was associated with a decrease in expression (Watson
et al., 2004). Aberrant patterns of methylation, specifically hypomethylation, may also
facilitate the activation of oncogenes. Ha-ras, a classic oncogene, is reproducibly
activated in mouse epidermal tumors (Balmain and Pragnell, 1983; Balmain, et al.,
1984). This has mainly been attributed to early mutation induced by initiating agents
such as DMBA and dibenzo[a,l]pyrene (DB[0t,l]P). However, one study suggests that
mutations in codon 61 of Ha-ras induced by DB[a,l]P simply results in a transient
proliferation of cells. Over time only a small subpopulation of these cells persist whereas
the majority are lost (Khan, et al., 2005). It is possible that continued upregulation of Ha-
ras may be governed by changes in DNA methylation. In support of this assertion,
decreased methylation at one XhoI site within the vicinity of the Ha—ras gene exhibited
hypomethylation as compared to normal epidermis in some papillomas and carcinomas
(Ramsden et al., 1985). Based on these intriguing but limited data, further investigation
of Ha-ras promoter methylation is necessary.

We hypothesize that progressive, non-random changes in DNA methylation
contribute to tumorigenesis. SENCAR mice were initiated with DMBA and promoted
with increasing doses of CSC for 4 or 8wks. The GC-rich methylation patterns were
analyzed for dose- and time-relationships as well as reversibility in precancerous skin
tissue via a sensitive and quantitative arbitrarily primed PCR and capillary

electrophoretic approach. Regions of altered DNA methylation which persist from

163

precancerous to tumor tissue are identified, and changes in the methylation status of the

promoter region of Ha-ras are associated with changes in gene expression.

MATERIALS AND METHODS

Animals

Female SENCAR mice (ages 7-10 weeks) were obtained from the National Cancer
Institute, Frederick Cancer Research and Development Center (Frederick, Maryland).
Animals were quarantined and allowed to acclimate for a minimum of 10 days prior to
being randomly assigned to treatment groups based on body weight. All groups were
compared by ANOVA and least significant difference criteria and were demonstrated not
to be significantly different at a 5%, two-tailed risk level to ensure groups of similar
mean body weight. Animals were housed and cared for at RJ Reynold’s facilities and in
accordance with the Institute of Laboratory Animals Resources (ILAR), Commission of
Life Sciences, National Research Council document entitled, Guide for the Care and Use
of Laboratory Animals. Mice, 6-7 animals per group, were initiated with a single, topical
application of 75 ug DMBA or acetone (vehicle control), followed by thrice-weekly
applications of 27, 18, 9, or 3 mg CSC or acetone (vehicle control) promotion for 4, 8 or
29 wks, and sacrificed immediately afterwards. The recovery groups were treated with
27, 18, 9, or 3mg CSC for 8wks and allowed an 8 wk recovery period prior to sacrifice.
Two additional recovery groups for mice treated with 27mg for 4 wks were allowed a 4
or 8 wk recovery period prior to sacrifice. Mice were euthanized with 70% C02, and skin
tissue was collected from the chemical application site. Skin masses that arose during the

29wk post-initiation period were excised and identified histologically for tumor type,

164

number, and factors associated with whether the tumor was benign or malignant. All
collected samples were snap-frozen at —800C and kept until analyzed.

DNA and RNA Isolation

In order to isolate DNA, frozen skin tissue or tumor tissue was pulverized using a mortar
and pestle and then allowed to thaw following the addition of TR] Reagent (Sigma). A
dounce homogenizer was used to thoroughly homogenize the sample. RNA was isolated
according to the manufacturer’s protocol. DNA isolation was carried out according to an
alternative protocol obtained from the manufacturer. DNA is extracted with Back
Extraction Buffer in lieu of ethanol precipitation.

Preparation of CSC

Cigarette smoke condensate (CSC) was prepared as described previously (Watson et al.,

2003), and doses of 27, 18, 9, and 3mg were applied to the animals 3 times/wk.

A detailed description of the methodology associated with the following can be found in
the Materials and Methods Section of Chapter 2.
Reversibility of altered methylation: Calculations
Assumptions of the AP-PCR Data Analysis
Common and Unique Regions of Altered Methylation (RAMs)
Evaluating Total Regions of Altered Methylation (RAMs)
Percent Dissimilarity Calculations
Gene-Specific Methylation Analysis:
Sequencing of 5’ Promoter Region of Ha-ras

Smal Methylation Sensitive Restriction Digestion of Ha-ras

165

Expression of Ha-ras

Reverse Transcription of RNA

RNA samples were treated with DNaseI (Invitrogen) to purify the RNA from
contaminating DNA remaining after isolation. Each reaction contained 2ug RNA, 1X
DNaseI reaction buffer, 2 units DNaseI, and DEPC-treated GDW to a final volume of
20ul. Samples were incubated at room temp. for 15min followed by addition of MgClz to
a final concentration of 2.27mM. RNA was heated to 65°C for 10min to inactivate the
DNaseI enzyme. The TaqMan Reverse Transcription Kit (Applied Biosystems; Foster
City, CA) was used to reverse transcribed the DNaseI treated RNA. Each reverse
transcription reaction contained, 1X Reverse Transcription Reaction Buffer, 5.5mM
MgC12, 200uM of each dNTP, 2.5uM random hexarner, 20 units RNase Inhibitor, 62.5
units Multiscribe Reverse Transcriptase and DEPC-treated GDW to a final volume of
50ul. The reactions are incubated at 25°C for 10min, 42°C for 1hr, and 95°C for 5 min.
All samples were stored at 4°C until needed.

Real Time PCR

A custom TaqMan assay including primers (For 5’ TGG TGG GCA ACA AGT GTG
A3’ and Rev 5’ GGC CTG CCG AGA CTC A 3’) and probe (5’ FAM CTG GCT GCT
CGC ACT GT 3’) specific for Exon 3 of the Ha-ras gene was purchased from Applied
Biosystems. The assay was designed based on sequence information obtained from
Brown et al., 1988 and Neades et al., 1991. In addition an Applied Biosystems custom
TaqMan assay including primers (For 5’ CTA CTA CCG ATT GGA TGG TTT AGT GA
3’ and Rev 5’ GTC AAG TTC GAC CGT CTT CTC A 3’ and probe (FAM 5’ CCG TGG

GCC GAC CC3’) was used for the control gene, 18S rRNA (Accession X00686).

166

Triplicate reactions for both the gene of interest (Ha-ras) and the control gene (1 SS) were
prepared per sample. Standards were also prepared for 188 and Ha-ras and ranged from
5 x 101 copies/ul to 5 x 107 copies/ul. Each Ha-ras reaction contained 1 x Custom Assay
Mix (Applied Biosystems; Foster City, CA), 1X TaqMan Universal PCR MasterMix,
llul cDNA and GDW to a final volume of 25ul. Each 18S reaction contained 1 x
Custom Assay Mix (Applied Biosystems; Foster City, CA), 1X TaqMan Universal PCR
MasterMix, llul cDNA and GDW to a final volume of 25ul. Reactions for each standard
contained 1 x Custom Assay Mix (Applied Biosystems; Foster City, CA), 1X TaqMan
Universal PCR MasterMix, 2ul standard and GDW to a final volume of 25ul. Cycling
conditions were as follows: 50°C for 2 min, 95°C for 10min, and 40 cycles of 95°C for
15sec. and 600C for l min. The absolute standard curve method for quantifying fold

change over control was employed.

167

RESULTS

Analysis of GC-rich regions of DNA allowed for a genome-wide snapshot of
altered methylation including hypomethylations, hypermethylations and new
methylations in response to treatment. Dose-dependent changes in methylation were
discerned following promotion of mouse skin with 3, 9, 18, or 27mg CSC for 8wks. In
addition, tumor tissue was evaluated for aberrant patterns of methylations. During
promotion, a treatment-related increase in the total number of RAMs was observed
whereby the RAMs induced by 3mg and 9mg CSC were approximately half of the total
RAMs induced by 18 and 27mg CSC (Table 1). Regions of increased methylation,
including hypermethylations and new methylations, were predominant and clearly
increased with dose (Figure 1). In contrast, regions of hypomethylation did not exhibit a
dose-response relationship. Tumor tissue was somewhat distinct from precancerous
tissue in that the incidence of hypermethylations and new methylations decreased while
total regions of hypomethylation increased and comprised over half of the total
alterations detected (Figure 1).

A comparative evaluation of the total changes in altered methylation between
treatment groups allowed for the identification of developing patterns and trends which
developed with increasing doses of CSC. However, in order to clearly and simply
demonstrate the extent to which RAMs that developed during promotion and in tumor
tissue differed from controls, a percent dissimilarity was calculated for each separate dose
and for tumor tissue. Methylation patterns arising from 3 and 9mg CSC promotion were
49% and 48% dissimilar to methylation patterns of controls (Table 2). With 18mg and

27mg CSC promotion, dissimilarity increased to 70% and 66% (Table 2). This illustrates

168

Table 1. Summary of GC-Rich Regions of Altered Methylation: Comparison of
DMBA Initiated. CSC 8wk Promotion and Tumor Tissue to Cogtrol

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Regions of Regions of Regions of
Treatment Digest , a , b New TOTAL
Hypomethylation Hypermethylation Me thyia tionc
3mg CSC Hpall 5 O 3 8
Mspl 9 0 3 12
Totald 14 0 6 20
9mg CSC Hpall 2 6 6 14
Mspl 0 2 5 7
Totald 2 8 1 1 21
18mg CSC Hpall 6 10 19 35
Mspl l 0 2 3
Totald 7 10 21 38
27mg CSC Hpall 2 0 l 3
Mspl o 10 27 37
Totald 2 10 28 40
Tumor
Tissue Hpall 19 0 3 22
Mspl 3 1 11 15
Totald 22 1 14 37

 

" Hypomethylated RAMs include both statistically significant (Student’s t-test, p<0.05)
decreases and 100% decreases.
b Hypermethylated RAMs are only those increases which are statistically significant
(Student’s t-test, p<0.05).
‘ New methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer

annealing which was not formed under control conditions.
d Total RAMs including hypomethylations, hypermethylations, and new methylations for

the RsaI/Mspl and RsaI/HpaII digests combined are reported for each treatment.

169

 

 

h
C

DMBA/CSC 38 Sites of Increased
Methylation
I Sites of Decreased

Sites of Altered Methylation
..x —\ N N (to (A!
O 01 O 01 0 0|

01

 

DMBA/3mg CSC DMBA/9mg CSC DMBA/18mg CSC DMBA/27mg CSC Tumor Tissue

Figgre 1 Increases (Hypermethylations and New Methylations) in GC-Rich
Methylation are Dose-Dependent Regions of altered methylation are categorized as

increased methylation and decreased methylation. Increases in methylation exhibit a
clear dose-response relationship. Tumor tissue is shown for comparison. A table tallying
the regions of altered methylation for each dose and tumor tissue is shown as an inset in
the chart. Regions of altered methylation exhibiting increased methylation include all
hypermethylations and new methylations. Hypermethylations are increases that are
significantly different from control values (Student’s t-test, p<0.05). Regions of altered
methylation exhibiting decreased methylation include all partial hypomethylations and
complete hypomethylations (100% decreases from control). Partial hypomethylations are
decreases that are significantly different from control values (Student’s t-test, p<0.05).

170

Table 2. Measure of the Percent Dissimilarig of SWL 3, 9, 18, 27mg CSC or Tumor

to DMBA/Acetone Control

 

 

 

 

 

 

Total Re ions of Altered - - - -
T...N.......f..g....a Mjhymb Siii’lii’éiflié'é'é‘ilfi.
3mg CSC 39 20 20/39 51%
9mg CSC 44 21 21/44 48°/o
18mg CSC 54 38 38/54 70°/o
27mg CSC 61 40 40/61 66%
Tumor 47 37 37/47 79°/o

 

 

 

 

 

 

 

“ The total number of regions includes every PCR product size reporting between control
and treated groups. Each PCR product represents a region of the genome.

b Total Regions of Altered Methylation represents the number of regions exhibiting
complete hypomethylation (100%), partial hypomethylation, hyperrnethylation or new
methylation.

171

that patterns of methylation become more abnormal as the dose of CSC increases from 9
to 18mg CSC. As expected, the status of methylation in tumor tissue was the most
dissimilar (79%) to control (Table 2).

Dose-dependent changes in methylation were consistent with tumor incidence
reported for each dose of CSC. At 29wk, 297 total tumors arose due to 27mg CSC
promotion (Figure 2). Just under 200 total tumors were induced by 18mg CSC
promotion and 79 total tumors were seen with 9mg CSC promotion; 3mg tumor
promotion did not increase tumor incidence above that of control (Figure 2). Therefore,
tumor number could reflect, in part, the extent of altered patterns of methylation and the
increasing dissimilarity of patterns as the dose of CSC is increased.

With promoter stimulation, an accumulation of changes in methylation could
occur over time. Time-dependent changes in methylation were tracked following
promotion of mouse skin with 27mg CSC for 4wks and 8wks. Total RAMs increased
from 21 at 4 weeks to 40 at 8 weeks where most changes were attributable to
hypermethylations and new methylations (Table 3). This supports the notion that the
number of changes in methylation increase and accumulate with time.

A unique feature of the arbitrarily primed PCR, capillary electrophoresis approach
described is that it allows methylation changes within particular regions of the genome to
be directly compared over time. Changes in methylation induced by 4wk 27mg CSC
promotion were compared to those induced by 8wk, 27mg CSC promotion. Common
RAMs and unique RAMs were identified (Figure 3a). Common RAMs exhibiting
equivalent changes in methylation persisted from 4wk to 8wk. The persistent, “carry

forward,” RAMs included 1 hypomethylation, 3 hypermethylations, and 10 new

172

350

 

 

 

'3' DMBA/acetone .
300 1 "E'i’DMBA/Bmg csc . 2 _
E +DMBA/9mg csc , i
g 250 ‘ e DMBA/18mg CSC ‘
[-1 -¢—DMBA/27mg csc
_
5 200 4 ‘
G '. C 432‘
E‘ ("x7
‘H ‘ ,
c 150 4 . £7 , -..
In ,.
0 ‘ . ,4
E 100 ~
= a
Z . , _ ’ 9
. We .
sol . fl ,
Ix>Ig
0 r -‘ '- '- - - - -‘ 4....“ mnmmmr‘ r. 4.4—1 ’4 aw .
1 4 7 10 13 16 19 22 25 28
Week of Study

Figgre 2 Tumor Incidence Increases with Dose of CSC Tumor incidence for
animals initiated with DMBA and promoted thrice weekly with 3, 9, 18, or 27mg CSC
for 29wks is shown. Total number of tumors from 40 animals in each dosing group are

expressed over time.

173

 

Table 3. Summary of GC-Rich Regions of Altered Methylation: Comparison of
DMBA Initiated, CSC 4 and 8wk Promotion to Control

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R . . Regions of
. egions of Regions of
Treatment Digest , a , b New TOTAL
Hypomethylation Hypermethylation Me thyia tionc
4wk
Promotion
27mg CSC Hpall l 2 2 5
Mspl 0 3 l3 l6
Totala 1 5 15 21
8wk
Promotion
27mg CSC Hpall 2 0 1 3
Mspl 0 10 27 37
Totald 2 10 28 4o

 

" Hypomethylated RAMs include both statistically significant (Student’s t-test, p<0.05)
decreases and 100% decreases.

 

b Hypermethylated RAMs are only those increases which are statistically significant
(Student’s t-test, p<0.05).

c New methylations indicate the formation of a PCR product following treatment due to a
gain of methylation either at the site of primer annealing or between sites of primer
annealing which was not formed under control conditions.

d Total RAMs including hypomethylations, hypermethylations, and new methylations for
the RsaI/MspI and RsaI/HpaII digests combined are reported for each treatment

174

A.

 

 

DMBA/27mg CSC
Promoted 4wk

 

DMBA/27mg csc
Promoted 8wl<

 

 

 

 

 

\

1 Common RAMs it

 

 

Methylation Change
is Equivalent

/ ’

 

 

Persistent Changes

1. Methylation change is opposite to CSC-

30M PARE _

/\

 

\

. Methylation Change is
Statistically Different

induced methylation change
2. TUMOR changes to a greater extent than
CSC-Induced methylation change

 

 

 

 

 

 

B.
4wk 27mg CSC ”awn. 8wk 27mg CSC
Precaneerous Skin Changes Pmeaneerous Skin
2 1 2 1 1
5 3 7
E 866 8:5 99 0030
o 0
cos 15’“ 8810000018
E o o s E o o o o o o
g °o°o°s g 8689 0000000
8 o s s o o o o

 

 

 

 

 

 

 

 

TUMOR

 

 

RAM. .114... to ’

 

 

 

 

 

 

 

TUMOR .
)# Total Unique
Methylation
: Changes In
TUMOR
Persisting
Chang“ TUMOR
2 2 e o e e 2“
§ 9 ° ’ fefiegaeo
939, e 9
E . 6 o E 0
1 0
mg °
leks 6' . O 3
O O
E 8G0 0..
G G 0

 

 

 

Fi ure 3 Pro ressive Chan es in Meth lation: Chan es Which Persist Frfl 4 to 8wk and
from 8wks to Tumor A flow chart illustrates the steps necessary to determine progressive
changes in methylation (a). Regions of altered methylation (RAMs) induced by 4wk, 27mg CSC
promotion were compared to those resulting from 8wk, 27mg CSC promotion. Common regions
of altered methylation in which the magnitudes of change were equivalent (2-way ANOVA,
p<0.05) were considered persistent changes. RAMs unique to 8wk, 27mg CSC included common
RAMs in which the magnitude of change was different or the RAMs were only observed with
8wk, 27mg CSC. Similarly, RAMs induced by 8wk, 27mg CSC promotion were compared to
those in tumor. Common regions of altered methylation in which the magnitudes of change were
equivalent (2-way ANOVA, p<0.05) were considered persistent changes. RAMs unique to tumor
included common RAMs in which the magnitude of change was different or the RAMs were only
observed in tumor tissue. Regions of altered methylation in which the magnitude of change was
different included common regions in which the changes in methylation were opposite in
direction and common regions in which the change was in the same direction, but the magnitudes
of change were statistically different (2-way ANOVA, p<0.05). Persistent and unique RAMs
induced by 4wk, 27mg CSC (6), 8wk, 27mg CSC (o), and tumor (hexagons) are represented (b).
Hypomethylations (HYPOM), hypermethylations (HYPERM), and new methylations (NEWM)
are segregated. Total unique changes (minus any persistent changes) are tallied and reported for
each category of methylation change.

 

175

methylations (Figure 3b). A further comparison was made between 8wk, 27mg CSC-
induced methylation changes and those identified in tumor tissue (29wk). We were able
to categorize the 9 persisting changes into two groups 1) those that persisted from 8 to
29wks and 2) those that originated with 4wk promotion and persisted through the 8wk
time point to the tumor tissue (Figure 3b). Of the 9 persisting changes 1 of the 2
hypomethylations, and 4 of the 6 new methylations originated from the 4wk, 27mg CSC
promotion (Figure 3b). This step-wise accumulation of RAMs over time as precancerous
skin is promoted with 27mg CSC and progresses to tumor tissue likely represents critical
changes in methylation.

RAMs exhibited by the tumor tissue were a compilation analysis of 9 tumors; 3
tumors each arose from the 9, 18, and 27mg CSC promotion. Common and equivalent
RAMs (persisting changes) were identified by separately comparing 8wk, 9, 18, and
27mg CSC-induced changes to tumor tissue. Persisting changes in methylation
originating from 9, 18, and/or 27mg CSC are depicted in Figure 4b along with the unique
changes exhibited by the tumor tissue. As expected, the highest number of critical
changes in methylation which carried through to tumor was seen with 27mg CSC. Of the
13 total persistent changes in tumor, 4 (1 hypomethylation, 1 hypermethylation, and 2
new methylations) were solely attributable to the 27mg CSC promotion and 3
hypomethylations were solely attributable to the 18mg CSC promotion. Notably, 1
hypomethylation and 1 new methylation induced by all three promoting doses persisted
to the tumors. Each dose of CSC elicited changes in methylation that persisted and could

contribute to the altered pattern of methylation observed in tumor tissue (Figure 4a and

b).

176

 

Figure 4 Progressive Changes in Methylation: Changes Induced by 8wk, 9, 18,
27mg CSC and Persi_st to Tumor A flow chart illustrates the steps necessary to

determine progressive changes in methylation (a). Regions of altered methylation
(RAMs) induced by 8wk, 9, 18 and 27mg CSC promotion were compared to those
identified in tumor tissue. Common regions of altered methylation in which the
magnitudes of change were equivalent (2-way ANOVA, p<0.05) were considered
persistent changes. RAMs unique to tumor included common RAMs in which the
magnitude of change was different or the RAMs were only observed in tumor tissue.
Regions of altered methylation in which the magnitude of change was different included
common regions in which the changes in methylation were opposite in direction and
common regions in which the change was in the same direction, but the magnitudes of
change were statistically different (2-way ANOVA, p<0.05). Persistent and unique
RAMs induced by 8wk, 9,18, or 27mg CSC (G) and tumor (hexagons) are represented
(b). A total of 6 RAMs each persisted from 9 and 18mg CSC to tumor while 9 RAMs
persisted from 27mg CSC to tumor. Hypomethylations (HYPOM), hypermethylations
(HYPERM), and new methylations (N EWM) are segregated for each treatment. Total
unique changes (minus any persistent changes) are tallied and reported for each category
of methylation change. (*) Represents 3 identical RAMs which persisted from both 9
and 27mg CSC promotion

177

NEWM HYPERM-IYPOM

NEWM HYPERM-IYPOM

 

 

 

 

 

 

 

 

 

 

 

 

Methylation Change is

 

 

/

 

 

 

DMBA/9mg csc COMPARE
Promoted 8wk \
DMBA/18mg csc - Tom's.
Promoted 8wk ‘ ' 29wk,
DMBA/27mg csc /
Promoted 8wk / \
Common RAMs RAMs Unique to
" TUMOR

 

 

\

}

 

Methylation Change is
Statistically Different

 

 

 

1. Methylation change is opposite to
CSC-induced methylation change

2. TUMOR changes to a greater extent
than CSC-induced methylation change

 

 

Total Unique
Methylation
Changes in TUMOR

 

 

 

 

NEWM HYPERM HYPOM ii

 

 

 

 

 

 

 

 

 

Equivalent
Persistent Changes
PR§CANC§ROUS SKIN
GO 2
O O 8
96938.,
99898 1,
GOGO
O O
9mg CSC
sass 7
O O O 10
0988088
21
988880888
89990908
9 O O O
18mg CSC
2
GO .
O O O 10
8809999
28
:OQOQOGOQO
Q GOQOGOGOGO
9 o s s s s
27mg csc

 

 

 

Persistent Changes ‘
9mg 18mg 27mg Unique
csc Tcsc . csc cm»
i 15
2 GOG . 9 000000
E N 000000
I Om o o o
5 ............................................................................... 0
Lu
:3. O
6
“’9
g 0 * O * 00°
2 GO- O 00
TUMOR

*Represents 3 identical RAM (PCR products) which persisted from both 9

and 27mg CSC promotion

" Represents an identical RAM (PCR product) which persisted from from 9

and 18mg CSC

*** Represents an identical RAM (PCR product which persisted from 9. l8,

and 27mg CSC promotions

178

Reversibility, as an operational definition, is characteristic of tumor promotion as
continued exposure to the promoting agent is necessary for progressive clonal expansion
of initiated cells. In light of this, changes in methylation were assessed following a 4 or
8wk recovery period. Specifically, RAMs previously induced by 4 or 8wk 27mg C SC
promotion were re-analyzed following the recovery period. Reversal of the methylation
changes induced by 4wk, 27mg CSC was more complete as the duration Of recovery was
lengthened. Following 4wk recovery, 2 hypermethylations and 7 new methylations had
reversed (Figure 5a). With 8wk recovery, an additional 5 RAMs had reversed (Figure 5a
and b). In comparison, 8 wk promotion with 27mg CSC followed by 8wk recovery
resulted in the reversal of 1 of 2 hypomethylations, 10 Of 10 hypermethylations, and 24 of
28 new methylations (Figure 5b). Importantly, all changes, with the exception Of 1 new
methylation, which persisted from the 4wk promotion, were recoverable which clearly
demonstrates that changes in methylation which accumulate in response to the promoting
stimuli are largely reversible.

GC-rich regions of the genome are frequently found in the promoter regions Of
genes, and changes in methylation can be associated with changes in gene expression.
Therefore, in light of the number Of CSC and tumor-induced RAMs within GC-rich
regions, a gene-specific approach was used to assess CSC-induced changes in
methylation within the promoter region of Ha-ras. Upregulation Of Ha—ras, an oncogene,
via hypomethylation of its promoter region could lead to increased expression which
might contribute tO tumorigenesis. We first analyzed a 283bp region of the Ha-ras
promoter containing 22CpG dinucleotides which was 950m upstream of the

transcriptional start site (Figure 6a). All 22 CpG sites were umnethylated in control skin

179

 

F'ggre 5 Progressive Changes in Methylation: Reversible Methylation Changes
Following 4 and 8wk Recoveg; Hypomethylations (HYPOM), hypermethylations
(HYPERM), and new methylations (N EWM) are represented for 4wk, 27mg CSC and
8wk, 27mg CSC promotion. Following 4 and 8wk recovery periods, increasing numbers
Of changes in methylation reversed (a). RAMs induced by 4wk, 27mg CSC (O) which
reversed during each recovery period are boxed. Promotion with 4 and 8wk, 27mg CSC
followed by 8wk recovery identifies numerous reversible changes in methylation (b).
RAMs induced by 4wk, 27mg CSC (O) or 8wk, 27mg CSC (0) which reversed during
the recovery period are boxed. Total unique changes (minus any persistent changes) are
tallied and reported for each category Of methylation change. (*) One new methylation
induced by 4wk, 27mg CSC which persisted through to tumor was not reversible
following either 4 or 8wk recovery periods.

180

A.

HYPERM HYPOM

NEWM

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PRECANCEROUS SKIN
1
O
9 s
O
9e [:>
4wk
GOGOGO 15
888868
O O O
4wk 27mg CSC
4wk 27mg CSC
8wk RECOVERY
2 ‘ 0
83
;>.
m
a If 0
LL]
O4
E S
a “’8 7
O Q 8
i 9 9 Se
O @-

 

 

HYPERM HYPOM

NEWM

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4wk RECOVERY 8wk RECOVERY
O 1 5 [g 0
E
9 3 0
So i e
Didi. E S
O "
Ode ’ 3 5
OS 1 ,
8wk 27mg CSC
8wk RECOVERY
Persistent
Changes
E 0% 1
s.
E
5 0 [<3 0
E :0]
a e o
o
s a 1re re 0 3
E e re [o o
g 8 ‘ o [O] 0
O Q [0] 0
9* Q Q 0 @

 

 

 

 

 

 

 

PRECANCEROUS SKIN

181

 

DMBA] Acetone #1 0 mm
DMBA! Acetone #2 0 mm
DMBA/ Acetone #3 _o_1xm__m__o__c_c__o_crxn_n__c_o_am_o__
DMBA/ Acetone #4 __c__am_m__n__o_o__c_am_o_c_n_cxm_c__
DMBA/ Acetone #5 ..Lm_m__c__c_o_o_m_o_c_o_crm_o__
DMBA! Acetone #6 _c_cxm_m__n_o_o_o_rxm_c_n_n_mc_

DMBA! 27mg CSC #1 _o__arn_m_n_c_o__n_mn__o__o_o_rxm_o___i
DMBA] 27mg CSC #2_o_rxm_m_o_c_o_o_axn_c_n_n_mn_o_
DMBA! 27mg CSC #3__o_crm_m_o_c_n__c_am_o_o_o_am_o_
DMBA/ 27mg CSC #4_c_mn__m_o_c_o_c_cxm_c_o_o_am_o._
DMBA/ 27mg CSC #5_o_am_m_o__o_c___o_am_o_o_o_m_o._
DMBA/27mgCSC#6_L mm on o c c o am_o_o_o_mn_o_

TUMOR #1 _o__cxm_m__.o_c_c_o_am_c_o_o_axn_o___
TUMOR#2 0 mm m 0 410 Q can r) (1an

TUMOR#3 _c_rrm_cn__o_c_c_c_txm_a_c_c_am_o_
TUMOR#4 _n_crm_m_o_o_o_o_m1_o_o_o_txm_a_
TUMOR#5 _o_crm__m__o_o_o_o_m_o__o_o_rxm_o_
TUMOR#6 _o_am_m_o__c_c_o_cxm__o_o_o_rrm_o__
TUMOR#7 _o__am_rxL__o_c_o__o_crm__o__o_o_am_o_
TUMOR#8 _n_rxm___m_o_c_o__o_m_c__o_.c_am_o_
TUMOR#9 _o_1xm_m__o_o_o_c_crm_o_o_o_am.o__

Figgre 6 Methylation Status of the Promoter Region of Ha-ras. A diagram of the
Ha-ras promoter indicating location of PCR primers and CpG sites (gray lollipops) in

relation to the transcriptional start site is presented (a). Bisulfite sequencing analysis
within the region spanning 283bp revealed that all 22 CpG sites were unmethylated (open
circles) in control, 27mg CSC promoted skin and tumor tissue except for 3 samples
which exhibited methylation (closed circles) at 1 or 2 CpG sites (b).

182

 

tissue, 27mg CSC promoted precancerous skin tissue and tumor tissue (Figure 6b).
Analysis of changes in methylation occurring closer to the transcriptional start site was
similarly conducted. This was achieved via methylation sensitive restriction digestion
with Smal endonuclease. Figure 7a outlines the target 525bp region which spanned the
transcriptional start site of the Ha-ras gene and contained two Smal CCCGGG
recognition sites. The percent of animals exhibiting hypomethylation at these two sites
progressed from 43% to 57% to 85% as the promoting dose of CSC was increased from 3
to 9 to 18mg CSC (Figure 7b). Promotion with 27mg CSC resulted in hypomethylation
being observed in 67% of the total animals (Figure 7b). Hypomethylation of the region
of interest in the Ha-ras promoter was discerned in all 9 of the tumor tissues (Figure 7b).

Due to the fact that the 5’ promoter region of Ha-ras was determined to be
unmethylated and two CpG sites within Smal recognition sequences were
hypomethylated in a generalized dose-dependent manner, changes in gene expression
were assessed. Promoting doses of 18 and 27mg CSC resulted in statistically significant
(10 and 6 fold) increases in gene expression over control (Figure 8) which can be
associated with the higher incidence of hypomethylation at the CpG sites analyzed.
Although 3 and 9mg CSC also seemed to increase gene expression, the fold changes did
not reach statistical significance and this might indicate that a threshold incidence of
hypomethylation required for upregulation of Ha-ras. Strikingly, in tumor tissue the
expression of Ha—ras increased by 104 fold (Figure 8). Hypomethylation of CpG sites
close to the transcriptional start site coupled with the absence of methylation in a distant
5’ promoter region could contribute to the dramatic upregulation of Ha-ras in tumor

tissue.

183

 

A- -255

- C GGG —- C CGGG

ER
-325

_ _ +124 +200

7

 

 

 

0
Sma l Recognition Site CCCGGG

w
0

.a
8

 

§

 

co
0

 

 

 

A
O

% of Total Animals Exhibiting Hypomethylation
8 8

 

 

 

 

O

 

3mg 9mg 18mg 27mg 1-
CSC CSC CSC CSC umor

Figure 7 Smal Restriction Digest Analysis of Ha-ras A schematic illustrates the
targeted region spanning 325bp around the transcriptional start site of Ha—ras which

contains 2 Smal recognition sequences CCCGGG (a). Smal will not cut its recognition
sequence if the internal cytosine is methylated. The amount of PCR product generated is
representative of the level of methylation at the internal cytosine. In general, the
incidence of hypomethylation of the two CpG sites increased with dose of CSC (b).
Hypomethylation was detected in 100% of the tumor tissues analyzed.

184

 

 

* Statistically Significant
Increase in Expression
. over Control ..................................................... , ............ V _

 

Fold Change Over Control

 

 

 

Figure 8 Expression of Ha-ras Changes in the gene expression of Ha—ras as detected
by real-time PCR are expressed as fold change over control. A statistically significant
increase (Student’s t-test, p<0.05) over control was observed in response to 18 and 27mg
CSC promotion. Tumor tissue exhibited a very large and highly significant (p<0.001)
induction of Ha-ras expression.

185

DISCUSSION

The arbitrarily primed PCR and capillary electrophoresis method described here is
a novel approach to evaluating the methylation status of GC-rich regions of DNA, and it
allows for the simultaneous identification of three possible types of methylation changes:
hypomethylations, hypermethylations, and new methylations. As demonstrated, dose
and time dependent relationships, as well as aspects of methylation reversibility are
clearly defined. The scope of this method, when employing the methylation-sensitive
isoschizomers Hpall and Mspl, relies on changes in methylation within their CCGG
recognition site. Approximately 35.25% of these CCGG sequences lie in transposable
elements, while 64.19% are located in gene coding and promoter regions of the mouse
genome (F azarri and Greally, 2004). By targeting CCGG sites, we are sampling ~7.45%
of all CpG dinucleotides estimated for the mouse genome of which 4.37% are found
within gene coding and promoter regions (Frazarri, and Greally, 2004). Importantly, the
methylation status of CpCpG sites are concurrently evaluated which expands the capacity
and power of our technique to detect overall altered methylation.

Comparatively, common alternative genome-wide methylation status assays (i.e.
restriction landmark genomic scanning (RLGS) and amplification of intermethylated sites
(AIMS)) which rely on methylation sensitive restriction with enzymes such as NotI and
Smal, further limit the proportion of CpG dinucleotides that can be sampled (Costello et
al., 2002; Frigola et al, 2002). RLGS targets GCGGCCGC sites via NotI and drastically
reduces the proportion of CpG dinucleotides evaluated in CpG islands in the mouse
genome to 0.03% (F razzari and Greally, 2004). Approximately 14% of all Hpall and

Mspl recognition sequences are located within CpG islands. Therefore, our method is

186

less restricted allowing for a more comprehensive approach to addressing treatment (i.e.
CSC) related disruption of DNA methylation during the promotion stage of skin
tumorigenesis.

Promoting agents have been shown to act via a threshold exhibiting dose-response
with regard to changes in methylation (Watson et al., 2003). Consistent with this dose-
response characteristic, CSC was shown to induce a clear dose—dependent increase in
hypermethylations and new methylations. However, a fundamental shift in the nature of
the observed changes in methylation over time was evident. Hypermethylations and new
methylations were the predominant alteration detected in precancerous tissue at early and
intermediate time points while tumor tissue (29wk) was predominantly hypomethylated.
This key feature was reported previously in that global hypomethylation was seen to be
specific to tumor tissue and GC-rich regions of hypomethylation were infrequently
detected in precancerous skin (Watson et al., 2003). In a separate study, global loss of 5-
methylcytosine content was reported to progress in two steps. The first occurred during
the early stages of benign tumor growth and further loss of S-methylcytosine content was
seen dming the transition from an epithelial phenotype to a highly metastatic
dedifferentiated/spindle morphology (Fraga et a1 ., 2004). Given that distinct methylation
profiles are consistently identified for precancerous and tumor tissues, defining the
intermediary steps is critical. In light of the fact that carcinogenesis involves a
progressive clonal expansion of cells bearing heritable alterations of their genomes, the
key question, addressed in this report, is “Do the changes in methylation at early stages of

promotion persist and accumulate over time to facilitate tumorigenesis?”

187

 

The promotion stage of tumorigenesis involves the step-wise accumulation of
heritable changes which are critical for the selection and clonal expansion of initiated
cells (Dragan etal., 1993). A number of changes in DNA methylation persisted from 4
to 8wks with the highest promoting dose of CSC and from 8wks to the tumors.
Importantly, a few changes originating with the 4wk promotion were identified at 8wks
and, also, persisted to the tumors. Those altered regions of methylation which persisted
during promotion are likely to be critical epigenetic changes that contributed to the clonal
expansion of subsets of initiated cells. As this effect was observed with the doses that
caused tumors (i.e., 9, 18 and 27 mg/CSC), a relationship appears to exist between the
tumor incidence elicited with each dose of CSC and the number and type of persisting
changes in methylation. This supports the view that altered DNA methylation is a
causative factor in tumor formation. The notion that altered DNA methylation is a cause
and not an effect was previously tested in mice with decreased Dnmtl expression and
substantial genome-wide DNA hypomethylation in all tissues. These mice developed
aggressive T cell lymphomas which were linked to the development of chromosomal
instability and the inappropriate activation of oncogenes (Gaudet et a1 ., 2003). Although,
at this time, cause and effect can not be established unequivocally, these are fundamental
observations illustrating that changes in DNA methylation occur during the promotion
stage of tumorigenesis and precede the formation of tumors.

Reversibility is a key aspect of the promotion stage of tumorigenesis (Dragan et
al., 1993) and the potential exists for reversing altered methylation of DNA; therefore,
altered methylation may be a mechanism underlying promotion (Goodman and Watson,

2002). The balance of methylation-demethylation reactions is regulated by numerous

188

enzymes including maintenance methylases, de novo methylases and demethylases.
Therefore, changes in DNA methylation can “reverse” by a number of mechanisms
(Goodman and Watson, 2002). Active demethylation can involve removing the methyl
group from cytosine which could effectively restore a hypermethylated state to control
levels (Ramchandani et al., 1999). Erroneous maintenance methylation following DNA
replication could lead to both loss of methylation in hypermethylated regions and gain of
methylation in hypomethylated regions. Finally, proliferation of “normal” cells and
apoptosis of cells exhibiting abnormal methylation profiles could both contribute to a
“reversal” of altered methylation. Within 8wk following cessation of the 4 and 8wk
promoting stimuli, regions exhibiting altered methylation were seen to “reverse”. With
the exception of one new methylation, all persistent changes were recoverable. This is a
clear demonstration that progressive, critical changes in methylation are reversible, a
hallmark of tumor promotion.

The altered methylation observed in GC-rich regions could lead to aberrant gene
expression. Increased methylation in the promoter regions of p16, and MGMT and E-
cadherin, among others were observed during mouse skin promotion (Watson et a1 .,
2004; Fraga et al., 2004). As demonstrated by Estellar et al., 2004, unique profiles of
promoter hypermethylation can be created and used to characterize the disruption of
critical pathways in tumorigenesis. Hypermethylation associated gene silencing is
extremely important when considering the mechanisms involved in tumor formation.
However, both hypennethylation and hypomethylation are occurring simultaneously in
GC-rich regions, illustrating the point that a variety of alterations in methylation may

play a role in carcinogenesis (Counts and Goodman, 1995).

189

Activation of Ha-ras is a common feature of papillomas (Balmain and Pragnell,
1983; Balmain, et al., 1984). A novel aspect of the current research is the finding that
tumors arising from CSC-promoted skin exhibited an increased incidence of
hypomethylation of the promoter region of Ha-ras. Decreased methylation, in addition to
significant increased expression of the oncogene, suggests altered methylation during
promotion, in addition to the possibility of mutation in response to DMBA-initiation,
might contribute directly to the activation of Ha-ras which likely plays role in
tumorigenesis. At first glance, the 100 fold increase in expression of Ha—ras that was
observed in the papillomas might appear extreme. However, in a related study,
expression of Ha-ras in normal skin was virtually undetectable by western blot analysis
(Rodriguez-Puebla et al., 1999). Therefore, given that basal levels of Ha-ras expression
in normal skin are extremely low, the 100 fold increase observed in tumor tissue is not
unrealistic, and it may play a fundamental role in facilitating tumorigenesis.

The aberrant activation of Ha—ras has been implicated in facilitating numerous
aspects of a malignant phenotype (i.e. proliferation, invasion and metastasis) (Giehl,
2005). Over-expression of Ha-ras might lead to a cascade of protein kinases resulting in
the phosphorylation of Jun, which can then upregulate the activity of DNA
methyltransferase thereby increasing the methylation capacity of the cell and resulting in
aberrant methylation patterns (MacLoed et al., 1995). A consequence of this upregulated
methylation capacity includes the possibility of silencing tumor suppressor genes through
promoter hypermethylation. It was demonstrated that an oncogene, v-src, induced
overexpression of Dnmtl which led to down regulation of a candidate tumor suppressor

gene, tsg, through promoter hyperrnethylation (Sung et al., 2004). Therefore,

190

 

upregulation of ras in response to promoter hypomethylation could facilitate further
aberrant patterns of methylation, which can facilitate tumorigenesis including
hyperrnethylation of tumor suppressor genes, through indirect activation of DNA
methyltransferases.

Genetic instability is a basic feature of carcinogenesis. Mutations in genes that
normally function to maintain genetic stability might cause the formation of a mutator
phenotype (Loeb, 2001), DNA hypomethylation may lead to both elevated mutation rates
(Chen et al., 1998) and chromosomal instability (Eden et al., 2003) in addition to a CpG
island methylator phenotype (Abe et al., 2005). We have discerned widespread altered
DNA methylation on all three levels, i.e., hypomethylation, hypermethylations plus new
regions of methylation which occur in a progressive fashion during tumorigenesis. Thus,
it is now appropriate to talk about instability of the epigenome as a fundamental
component of the genetic instability that provides an environment which fosters the
aberrant gene expression involved in the transformation of a normal cell into a frank
carcinoma.

ACKNOWLEDGEMENT

Support from R]. Reynolds Tobacco Co. is acknowledged gratefully.

191

 

REFERENCES

Abe, M., Ohira, M., Kaneda, A., Yagi, Y., Yamamoto, S., Kitano, Y., Takato, T.,
Nakagawara, A., and Ushijima, T. (2005). Cancer Res. 65, 828-834.

Balmain, A., Ramsden, M., Bowden, G.T., and Smith, J. (1984). Activation of the
mouse cellular Harvey-ras gene in chemically induced benign skin papillomas. Nature
307, 658-660.

Balmain, A. and Pragnell, LB. (1983). Mouse skin carcinomas induced in vivo by
chemical carcinogens have a transforming Harvey-ras oncogene. Nature 303, 72-74.

Brown, K., et al. (1988). Isolation and characterization of the 5’ flanking region of the
mouse c-Ha-ras gene. Mol. Carcinogenesis 1, 161-170.

Chen, R.Z., Pettersson, U., Beard, C., Jackson-Grusby, L., and Jaenisch, R. (1998). DNA
hypomethylation leads to elevated mutation rates. Nature 395, 89-93.

Edean, A., Gaudet, F ., Waghmare, A., and Jaenisch, R. (2003). Chromosomal instability
and tumors promoted by DNA hypomethylation. Science 300, 455-456.

Coghlan, L.G. et al., (2000). Development and initial charaCterization of several new
inbred strains of SENCAR mice for studies of multistage skin carcinogenesis.
Carcinogenesis 21, 641-646.

Costello, J .F., Smiraglia, DJ. and Plass, C. (2002). Restriction landmark genome
scanning. Methods 27, 144-149.

Counts, .1 .L. and Goodman, J .I. (1995). Alterations in DNA methylation may play a
variety of roles in carcinogenesis. Cell 83, 13-15.

Dragan Y.P., et al., (1993). The initiation-promotion-progression model of rat
hapatocarcinogenesis. Proc. Soc. Exp. Biol. Med. 202, 16-24.

Estellar, M., Corn, P.G., Baylin, SB, and Herman, J .G. (2001). A gene
hyperrnethylation profile of human cancer. Cancer Res. 61, 3225-3229.

Fraga, M.F. et al. (2004). A mouse skin multistage carcinogenesis model reflects the
aberrant DNA methylation patterns of human tumors. Cancer Res. 64, 5527-5534.

Frazzari, M.J., and Greally, J .M. (2004). Epigenomics: Beyond CpG islands. Nature
Rev. Genet. 5, 446-455.

Frigola, J ., Ribas, M. Risques, RA. and Peinado, MA. (2002). Methylome profiling of

cancer cells by amplification of intermethylated sites (AIMS). Nucleic Acids Res. 30,
e28.

192

 

Gaudet, F., et al. (2003). Induction of tumors in mice by genomic hypomethylation.
Science. 300, 489-492.

Giehl, K. (2005). Oncogenic ras in tumour progression and metastasis. Biol. Chem.
386, 193-205.

Goodman, J .I. and Watson, RE. (2002). Altered DNA methylation: A secondary
mechanism involved in carcinogenesis. Annu. Rev. Pharmacol. T oxicol. 42, 501-525.

Gonzalgo, M.L., Liang, G., Spruck, CH. 111, Zingg, J-M., Rideout, W. M. 111, and Jones,
RA. (1997). Identification and characterization of differentially methylated regions of

genomic DNA by methylation-sensitive arbitrarily primed PCR. Cancer Res. 57, 594-
599.

Hanahan, D. and Weinberg, R.A., (2000). The Hallmarks of Cancer. Cell 100, 57-70.

Hennings, H., Lowry, D.T., Yuspa, S.H., Mock, B., and Potter, M. (1997). New strains
of inbred SENCAR mice with increased susceptibility to induction of papillomas and
squamous cell carcinomas in skin. Mol. Carcinogenesis 20, 143-150.

Jones, PA. and Baylin, SB. (2002). The fundamental role of epigenetic events in
cancer. Nature Rev. 3, 415-428.

Jones, P.A., and Laird, P.W. (1999). Cancer epigenetics comes of age Nature Genet.
21, 163-167.

Khan, et al. (2005). Harvey-ras gene expression and epidermal cell proliferation in
dibenzo[a,l]pyrene-treated early preneoplastic SENCAR mouse skin. J. Invest.
Dermatol. 125, 567-574.

Loeb, L. A. (2001). A mutator phenotype in cancer. Cancer Res. 61, 3230-3239.

MacLeod, A.R., Rouleau, J ., and Szyf, M. (1995). Regulation of DNA methylation by
the ras signalling pathway. J. Biol. Chem. 270, 11327-11337.

Neades, R., et al. (1991). Transient expression of the cloned mouse c-Ha—ras 5’
upstream region in transfected primary SENCAR mouse keratinocytes demonstrates its
power as a promoter element. Mol. Carcinogenesis 4, 369-375.

Okoji, R.S., et al. (2002). Sodium arsenite administration via drinking water increases
genome-wide and Ha-ras DNA hypomethylation in methyl-deficient C57BL/6J mice.
Carcinogenesis 23, 777-785.

Pitot, HQ and Dragan, Y.P. (1994). The multistage nature of chemically induced
hepatocarcinogenesis in the rat. Drug Metab. Rev. 26, 209-220.

193

 

Ramchandani, S. Battacharya, S.K., Cervoni, N. and Szyf, M. (1999). DNA methylation
is a reversible biological signal. Proc. Natl. Acad. Sci. USA 96, 6107-6112.

Ramsden, M., Cole, G., Smith, J ., and Balmain, A. (1985). Differential methylation of
the c-Ha-ras gene in normal mouse cells and during skin tumour progression. EMBO J.
4, 1449-1454.

Rodriguez-Puebla, M.L., LaCava, M., Bolontrade, M.F., Russell, J. and Conti, C].
(1999). Increased expression of mutated Ha-ras during premalignant progression in
SENCAR mouse skin. Mol. Carcinogenesis 26, 150-156.

Schulte-Hermann, R., Timmermann-Trosiener, I., Barthel, G., and Bursch, W. (1990).
DNA synthesis, apoptosis, and phenotypic expression as determinants of growth of
altered foci in rat liver during phenobarbital promotion. Cancer Res. 50, 5127-5135.

Sung, J. et al., (2004). Oncogene regulation of tumor suppressor genes in tumorigenesis.
Carcinogenesis 26, 487-494.

Watson, R.E., Curtin, G.M., Doolittle, D.J., and Goodman, J .I. (2003). Progressive
alterations in global and GC-rich DNA methylation during tumorigenesis. Toxicol. Sci.
75, 289-299.

Watson, R.E., Curtin, G.M., Hellmann, G.M., Doolittle, D.J., and Goodman, J .I. (2004).

Increased DNA methylation in the HoxAS promoter region correlates with decreased
expression of the gene during tumor promotion. Mol. Carcinogenesis 41, 54-66.

194

 

CHAPTER 3 APPENDIX
This appendix contains additional research results that are complementary to the body of
data presented in Chapter 3. The materials and methods needed to extend the analysis of
reversibility of altered methylation in GC-rich regions and to test the status of
methylation of LINE-1 sequences in precancerous and tumor tissue is described. In
addition experimental methods for the evaluation of expression of LIN E-l elements is
described. Results are presented along with a brief discussion of the findings and

relevant references.

195

 

MATERIALS AND METHODS

Reversibility of altered methylation: Calculations

Determining the reversibility of changes in methylation is discussed in Chapter 2, Pgs.
102-104. The following is an extension of the calculations.

PCR product sizes corresponding to statistically significant changes in methylation that
are exclusive to the recovery groups are considered unique because they arose during the
recovery period. These hypomethylations, hypermethylations, and new methylations are
accounted for separately and designated as new RAM following recovery.

Gene-Specific Methylation Analysis: Combined Bisulfite Restriction Analysis of LINE-1
Bisulfite Conversion and PCR Amplification

Bisulfite conversion of DNA effectively deaminates all un-methylated cytosines to uracil
leaving methylated cytosines unaffected. 2ug DNA was bisulfite converted using the EZ
DNA Methylation Kit (Zymo Research, Orange, CA). PCR is performed with bisulfite
converted DNA which allows for the replacement of uracil with thymine and 5-
methylcytosine with cytosine. Consequently, only cytosines that were originally
methylated remain in the DNA sequence. PCR is carried out using primers specific for
bisulfite converted DNA and containing no CpG sites. The two LINE primers
(Accession M13002), 5’ AAT TTT TGT TAG GAG TIT GGT T 3’ and 5’ ATT TTT
AAA TCT AAA TCT AAA TTT TC 3’ were used to amplify the -205nt to +132nt
(337bp) region of the LINE-l element relative to the transcriptional start site. Each PCR
reaction contained 2.0ul bisulfite converted DNA, 1X FailsafeTM Buffer E (Epicentre®;
Madison, WI) 2.5uM each primer, 1.5 units Taq Polymerase (Invitrogenm) and GDW to

a final volume of 25ul. Taq Polymerase was added following incubation at 94°C for 3

196

min Cycling conditions were: 35 cycles of 940C for 45 s, 54°C for 45 s, and 720C for 1
min, followed by 1 time delay cycle of 72°C for 3min and a 4°C soak. Amplification of

the target region was verified by gel electrophoresis on a 3% agarose gel.

Restriction Digestion with Taq] or thI restriction endonucleases

Restriction digestion with Taql or thI was carried out on the PCR products. Taql
digests contain 1x NEB Buffer 3 (NewEngland Biolabs), 1x BSA (NewEngland Biolabs),
3.0ul PCR product 20units Taql and GDW to a final volume of 10.0ul. Digests were
incubated overnight at 65°C. thI digests contain 1x NEB Buffer 4, 3.0ul PCR product,
5units thI and GDW to a final volume of lOul. Digests were incubated over night at

370C. 2.5u1 of each digest was electrophoresed via a 3% agarose gel.

Quantification of Band Intensity with the thI Digest

The relative intensity of a PCR product band corresponds to the relative starting
concentration of methylated DNA. Four regions of each lane were outlined and
measured for pixel number and intensity with NIH image analysis. 1. the 337bp PCR
product band 2. Lane background just below the 337bp PCR product 3. Restriction
Digestion Fragment (218bp) 4. Lane background just below the restriction digestion
fragment. The number of pixels defined the size of the outlined region. The same sized
region was used to measure the PCR product band, restriction digestion fragment and the
lane backgrounds. Total pixel intensity units (TPI) were calculated by multiplying the
number of pixels by the mean intensity units within the outlined region. This was done

separately for the background measurements, the PCR product band, and the 218bp

197

 

fragment. The TPI units of the respective background were subtracted from the TPI units
of either the PCR product band or the 218bp fragment band to give a normalized TPI for
the outlined region within a lane. The ratio of the normalized 218bp band over the
normalized 33 7bp band was calculated and multiplied by 100%. This percentage
represents the fraction of intensity the 218bp band is of the 337bp band. Statistical

significance was determined by Student’s t-test, p<0.05.

Expression of LINE-1

Reverse Transcription of RNA

RNA samples were treated with DNaseI (Invitrogen) to purify the RNA from
contaminating DNA remaining after isolation. Each reaction contained 2ug RNA, 1X
DNaseI reaction buffer, 2 units DNaseI, and DEPC-treated GDW to a final volume of
20u1. Samples were incubated at room temp. for 15min followed by addition of MgClz to
a final concentration of 2.27mM. RNA was heated to 65°C for 10min to inactivate the
DNaseI enzyme. The TaqMan Reverse Transcription Kit (Applied Biosystems; Foster
City, CA) was used to reverse transcribed the DNaseI treated RNA. Each reverse
transcription reaction contained, 1X Reverse Transcription Reaction Buffer, 5.5mM
MgC12, 200uM of each dNTP, 2.5uM random hexarner, 20 units RNase Inhibitor, 62.5
units Multiscribe Reverse Transcriptase and DEPC-treated GDW to a final volume of
50ul. The reactions are incubated at 25°C for 10min, 42°C for 1hr, and 95°C for 5 min.

All samples were stored at 4°C until needed.

198

Real Time PCR

A custom TaqMan assay including primers (For 5’ GGT CAA ATC TAA GTG GAT
CAA GGA ACT 3’ and Rev 5’ GCT TTT CCC CAC TTT CTC CTC TAT 3’) and probe
(5’ F AM CAG AGA CAC TGA AAC TT 3’) specific for ORF 2 of the LINE-1 element
(Accession M13002) was purchased from Applied Biosystems. In addition an Applied
Biosystems custom TaqMan assay including primers (For 5’ CTA CTA CCG ATT GGA
TGG TTT AGT GA 3’ and Rev 5’ GTC AAG TTC GAC CGT CTT CTC A 3’ and probe
(FAM 5’ CCG TGG GCC GAC CC3’) was used for the control gene, 18S rRNA
(Accession X00686). Triplicate reactions for both the gene of interest (LINE-1) and the
control gene (188) were prepared per sample. Standards were also prepared for 18S and
Ha-ras and ranged from 1 x 102 copies to l x 108 copies. Each reaction contained 1 x
Custom Assay Mix (Applied Biosystems; Foster City, CA), llul cDNA or 2ul standard,
and GDW to a final volume of 25ul. Cycling conditions were as follows: 50°C for 2
min, 95°C for 10min, and 40 cycles of 950C for lSsec. and 600C for 1 min. The absolute

standard curve method for quantifying fold change over control was employed.

199

 

RESULTS

Tumor promotion involves a basic step-wise accumulation of heritable changes.
If the promoting stimulus is withdrawn, accumulated changes in methylation could
reverse. Reversal could also result from the apoptosis of the initiated cells. Therefore,
the reversibility of regions exhibiting altered methylation in response to 4 and 8wk
promotion was measured. Specifically, RAMs previously induced by 4 or 8wk 27mg
CSC promotion were re-analyzed following the recovery period. Reversal of the
methylation changes induced by 4wk, 27mg CSC was more complete as the duration of
recovery was lengthened. Following 4wk recovery, 2 hypermethylations and 7 new
methylations had reversed (Figure 1). With 8wk recovery, an additional 4 RAMs had
reversed (Figure 1). In comparison, 8 wk promotion with 27mg CSC followed by 8wk
recovery resulted in the reversal of 1 of 2 hypomethylations, 10 of 10 hypermethylations,
and 24 of 28 new methylations (Figure 2). Importantly, all changes, with the exception
of 1 new methylation, which persisted from the 4wk promotion, were recoverable which
clearly demonstrates that changes in methylation which accumulate in response to the
promoting stimuli are largely reversible.

In addition to assessing which RAMs which did or did not reverse, unique
methylation changes occurring during recovery were analyzed. The number of
hypomethylations after 8wks induced by 3, 9, l8, and 27mg CSC were relatively few in
number; however, a notable increase in hypomethylations was evident during the 8wk
recovery period for all dosing groups except 9mg CSC (Figure 2). Unique
hypermethylations only increased with recovery following 3 and 9mg CSC but not 18

and 27mg CSC (Figure 2). This same pattern was seen with the new methylations.

200

 

 

 

01

|

15

1°
7
6 5:25:52 ‘ 6 6
- 5 5
. 3
2 Eiféaizis
R\‘ \\\ \\\‘ , . ~- i755"? \ ;

man remcsc wetness WMCSC
_ Sites of Hypomethylation (P8IN8) I Sites of Hypermethylation (P8IN8) I. Sites of New Methylation (P8IN8)
{Q Hypomethylated Sites which DE Not & Hypermethylated Sites which QM 5V New Methylation Sites which Did Not

 

 

 

d
01

12
10
g

 

 

 

.3
o

 

      
 

 

 

Regions of Altered-Methylation

 

 

 

 

 

 

 

O

 

MFollowing 8wk Recovery MFOIWNQ M Recovery BMWWHQ 3“ Recovery
fl New Sites of Hypomethylation % M Sites of Hypomethylation fl Nu Sites of New Methylation
Following Recovery Following Recovery Following Recovery

A endix lFi re 1 Reversible Meth lation Chan es Followin 8wk Promotion
Hypomethylations, hypermethylations, and new methylations are represented for 8wk, 3,
9, 18, and 27mg CSC promotion. Of those changes induced by 8wk promotion, the
number of regions of altered methylation which did not reverse following an 8wk
recovery period are presented in addition to unique changes in methylation identified
following recovery. Hypomethylated RAM include both partial hypomethylations
(statistically significant (p<0.05) decrease as compared to control) and complete (100%)
hypomethylations. Hypermethylated RAMs are only those increases which are
statistically significant (Student’s t-test, p<0.05). New methylations indicate the
formation of a PCR product following treatment due to a gain of methylation either at the
site of primer annealing or between sites of primer annealing.

 

201

 

 

 

 

 

1O

 

 

 

 

 

  

 

\

 

 

Regions of Altered Methylation
T
l
l

 

 

 

 

\

. § .

o. m R °

DMBA/27mg CSC: 4wk Promotion DMBA/27mg CSC: 4wk Promotion
Recovery Period: 4 Weeks Recovery Period: 8 Weeks

..Sites of Hypomethylation (Pat/N4) I Sites of Hypermethylation (P4IN4) I Sites of New Methylation (P4IN4)
'TCHypomethylated Sites which gram \V Hypemethylated Sites which Quint E New Methylation Sites which Qismm

MFollowing Recovery BeveraFoiioMno Recovery EmmWIoMno Recovery
mum Sitesonypomethylation a msnesornypennemyiation ables: SitesofNew Methylation

Following Recovery Following Recovery Following Recovery

A endix Fi re 2 Reversible Meth lation Chan es Followin 4 and 8wk
Promotion with 27mg CSC Hypomethylations, hypermethylations, and new
methylations are represented for 4wk, 27mg CSC promotion. Of those changes induced
by 4wk promotion, the number of regions of altered methylation which did not reverse
following a 4 or 8wk recovery period are presented in addition to unique changes in
methylation identified following recovery. Hypomethylated RAM include both partial
hypomethylations (statistically significant (p<0.05) decrease as compared to control and
complete (100%) hypomethylations. Hypermethylated RAMs are only those increases
which are statistically significant (Student’s t-test, p<0.05). New methylations indicate
the formation of a PCR product following treatment due to a gain of methylation either at
the site of primer annealing or between sites of primer annealing.

 

202

 

Twelve and 15 unique new methylations were observed during recovery following 3 and
9mg CSC respectively whereas only 2 were identified post 27mg CSC (Figure 2).
Similar to this, an increase in unique hypomethylations from 2 to 16 was observed as the
duration of recovery was extended following 4wk promotion with 27mg CSC (Figure 1).
Unique hypermethylations increased by 1 and unique new methylations decreased from 8
following 4wk to 6 after 8wks recovery (Figure 1).

GC-rich regions of the genome are frequently found in the promoter regions of
genes, and changes in methylation can be associated with changes in gene expression. In
that regard, the disregulation of LIN E-elements can be associated with genomic
instability. The methylation status of two CpG sites close to the transcriptional start site
for the first ORF were measured via COBRA analysis with Taql and thI enzymes
(Figure 3a and b). The methylation status of the CpG dinucleotide contained within the
Taql recognition site was similar in control, 27mg CSC promoted skin (Figure 4a) and
tumor samples (Figure 4b). This CpG dinucleotide was partially (~50%) unmethylated as
demonstrated by the visualization of both the 337bp PCR product and the 315bp
digestion fragment which occurred with approximately equal intensities (Figure 4a and
b). The thI CpG dinucleotide located within 3bp of the transcriptional start site was
methylated in both control and 27mg CSC promoted skin samples (Figure 5a). However,
the intensity of the 218bp digestion fragment significantly increased in tumor samples;
this indicates that the site was partially hypomethylated (Figure 5b). Expression of the
LINE-1 element showed a slight, but statistically significant, increase over control with

27mg CSC promotion while this was not seen in tumor (Figure 6).

203

 

 

 

 

ORF1
3- .9 a
3’ e E a
‘3 17» g 'as
E E t: R
U u U H
B m 9 i“
(U
E B. 53 E
(U (0
e s. 1% s. g
S 5 2 £ '0
C a) c a) to
D 2 3 2 -'
l 7:
‘ ' 400bp
300bp
200bp
- 100bp

 

 

 

 

 

 

 

Appendix Figpre 3 Tapgeted Region of LINE-1. The LINE-l promoter and part of
the first ORF (open reading frame) indicating location of PCR primers and CpG sites

(gray lollipops) in relation to the transcriptional start site is presented (a). Open lollipops
represent CpG sites which are not analyzed using the COBRA technique. Expected
fragment sizes based on the methylation status of either the Taql or th1 sites is
diagrammed (b). DNA is first bisulfite converted which all unmethylated cytosines are
deaminated to uracil. Methylated cytosines remain methylated. PCR amplification of the
bisulfite converted DNA effectively replaces all uracils with thymine. Taql recognizes
and restrict TCGA sites. Therefore, Taql will only restrict the PCR product if the
cytosine was originally methylated. This results in 22bp and 315bp digestion fragments.
thI recognizes and restricts GGTGA sites. Therefore, thI will only restrict the PCR
product if the cytosine within the GGCGA site was originally unmethylated. Through
bisulfite conversion the site becomes GGTGA and is restricted by th1 resulting in 119
and 218bp fragment sizes.

204

 

337bp
315bp

 

l i IIIIIIIIIII
T1
T2

..

 

Appendix Figure 4 Methylation Status of the Tag] Site Within a Targeted Region
of LINE-l Methylation status of a Taql TCGA site was analyzed following bisulfite

conversion of the DNA and PCR amplification of a 337bp region containing the
recognition sequence near to the transcriptional start site of LINE-1. Control (C) samples
(n=7)were compared to 8wk, 27mg CSC (CSC) promoted samples (n=6) (a), or tumor (T)
samples (n=9) (b). An undigested control (UD) PCR product was run on each gel for
comparison. Visualization of the 337bp and 315bp digestion fragments with equal
intensities indicates that the site is 50% methylated and 50% unmethylated in all samples
tested.

205

1234567

restraints
00000

C1
C

Awm

r 1 rum-

3
1 2 3 4 5 6 7 8 9 1011 12
!8388388
§ mum
: 337bp
:_ 218bp
119bp

 

Appendix Figure 5 Methylation Status of the thI Site Within a Targeted Region
of LINE-l Methylation status of an thI GGTGA site was analyzed following bisulfite

conversion of the DNA and PCR amplification of a 337bp region containing the
recognition sequence near to the transcriptional start site of LINE-1. Control (C) samples
(n=7)were compared to 8wk, 27mg CSC (CSC) promoted samples (n=6) (a), or tumor (T)
samples (n=9) (b). An undigested control (UD) PCR product was run on each gel for
comparison. Visualization of the 218bp and 119bp digestion fragments indicates that the
site is partially hypomethylated. The average pixel intensity of the 7 control 218bp
digestion fragments within Box A was calculated to be 5% of the average pixel intensity
of the 7 control 337bp PCR products. The average pixel intensity of the 6 8wk, 27mg
CSC 218bp digestion fragments within Box B was calculated to be 5% of the average
pixel intensity of the 6 8wk, 27mg CSC 337bp PCR products. The average pixel
intensity of the 9 tumor 218bp digestion fragments within Box C was calculated to be
19% of the average pixel intensity of the 9 tumor 337bp PCR products indicating a
statistically significant decrease in methylation. Statistical significance was determined
by Student’s t-test, p<0.05.

206

 

 

 

 

 

e. we
L we
4 M.W..
mm
‘ ---/-.W .mA
6 .
mm.
xm

      

...WW////////// /////M
III-
\\\\
E.-

L '-

.//////

m. w. m w. m m m. m. m
111111111

ntrol was observed in response to

was not increased in tumor tissue.

 

   

ssion of LINE-l Changes in the gene e
al-time PCR are expressed as fold change over co

ase (Student’s t-test, p<0.05) over co

 

 

 

 

as detected by re
27mg CSC promotion. LINE-1 expression

significant incre

.2250 .26.. ea: :26 so".

207

DISCUSSION

Changes in methylation were observed during recovery suggesting that altered
methylation extends past the promotion stage. However there could be numerous causes
of these changes. In general, new methylations and hypermethylations were only
observed following withdrawal of 3 and 9mg CSC or 27mg CSC (4wk). If de novo
methylases were down regulated during promotion, withdrawal of the stimulus could
result in an overcompensation or upregulation of methylase activity leading to transient
hypermethylations or new methylations. Another possibility would be that upregulation
of active demethylation in response to the hyperrnethylation during promotion could
cause transient hypomethylation of unintended regions of the genome. In addition, the
exact length of time needed to fully restore normal patterns of methylation is unknown.
Numerous time points following cessation of the promoting stimuli might be needed to
estimate full recovery. Hypomethylation is a prominent alteration observed during
recovery. This hypomethylation could be linked to increased apoptosis of abnormal cells.
The DNA methyltransferase inhibitor 5-AZA which results in hypomethylation of DNA
has been shown to induce apoptosis in colorectal cancer cells via induction of 15-
lipoxygenase-l (15-Lox-l) (His et al., 2005). lS-Lox-l has been linked to the
modulation of apoptosis and the differentiation of colorectal carcinoma cells (Ikawa et
al., 1999). Therefore, the interesting increase in hypomethylations during the recovery
period could lead to the upregulation of a number of proteins involved in apoptosis
contributing to the restoration of normal cell populations.

Analysis of the methylation status of 2 CpG sites around the transcriptional start

site of the LIN E-l element revealed slight hypomethylation of one of the two sites only

208

 

in the tumor samples. Hypomethylation of repetitive sequences is a commonly reported
feature of various cancers. Various gastric cancer cell lines exhibited LIN E-l
hypomethylation in conjunction with global hypomethylation. For example, repetitive
element hypomethylation was not seen in gastric cancer cell lines without global
hypomethylation (Kaneda et al., 2004). These results are consistent with global
hypomethylation exhibited by skin tumors (Waston et al., 2003) and evidence for
hypomethylation in a region very near to the transcriptional start site of the LIN E-l
elements. Prostate cancer also bears some similarity to observations made with
precancerous and cancerous skin samples. Coordinate hyperrnethylation of several genes
(e. g. APC, GSTPl , RARB2, and RASSF 1A) may occur early in prostate carcinoma
followed by LIN E-l hypomethylation as was observed in humans. This hypomethylation
has been associated with the progression of prostate cancer (Florl et al., 2004).
Hypomethylation of LINE-l elements has been postulated to lead to increases in
expression which would contribute to genetic instability. This has been demonstrated in
human urothelial carcinomas. Decreased methylation of LINE-1 elements was
demonstrated in conjunction with increases in RNA levels suggesting increased
expression (Florl et al., 1999). Our data did not show an increase in LIN E-l expression
in tumor tissue, however, a slight increase in expression was noted with 27mg CSC
promotion for 8wk. This indicates that the upregulation of LINE during promotion might
be controlled by regions of the promoter which were not analyzed. Therefore, genomic
instability created by the upregulation of LINE during promotion could contribute to

tumor formation, even though LINE expression is not sustained in the skin tumors.

209

 

REFERENCES

F lorl, A.R., Lower, R., Schmitz-Drager, B.J., and Schulz, WA. (1999). DNA
methylation and expression of LINE-1 and HERV-K provirus sequences in urothelial and
renal cell carcinomas. British J. Cancer 80, 1312-1321.

Florl, A.R., Steinhoff, C., Muller, M., Seifert, H-H., Hader, C., Engers, R., Ackermann,
R. and Schulz, WA. (2004). Coordinate hyperrnethylation at specific genes in prostate
carcinoma precedes LINE-1 hypomethylation. British J. Cancer 91, 985-994.

His, L.C., Xi, X., Wu, Y., and Lippman, SM. (2005). The methyltransferase inhibitor 5-
aza-2-deoxycytidine induces apoptosis via induction of 15-lipoxygenase-l in colorectal
cancer cells. Mol. Cancer Ther. 4, 1740-1746.

Ikawa, H., Kamitani, H., Calvo, B.F., Foley, J .F ., and Eling, TE. (1999). Expression of
15-Lipoxygenase-l in human colorectal cancer. Cancer Res. 59, 360-366.

Kaneda, A. Tsukamoto, T., Takamura-Enya, T., Watanabe, N., Kaminishi, M., Sugimura,
T., Tatematsu, M. and Ushijima, T. (2004). Frequent hypomethylation in multiple
promoter CpG islands is associated with global hypomethylation, but not with frequent
promoter hyperrnethylation. Cancer Sci. 95, 58-64.

Watson, R.E., Curtin, G.M., Doolittle, D.J., and Goodman, J .I. (2003). Progressive
alterations in global and GC-rich DNA methylation during tumorigenesis. Toxicol. Sci.
75, 289-299.

210

 

 

SUMMARY

In Chapter 1, the ability of diethanolamine (DEA) to alter patterns of methylation
in GC-rich regions in B6C3F1 mouse hepatocytes was tested. DEA is hypothesized to
upset choline homeostasis by inducing a choline deficient state. This choline deficiency
(CD) leads to methyl deficiency which could disrupt l-carbon metabolism and decrease
the capacity of a cell to maintain patterns of methylation resulting in the facilitation of
liver tumorigenesis. A choline devoid, methionine deficient diet has previously been
shown to lead to global hypomethylation of DNA in the livers of B6C3F1 mice (Counts
et al., 1996). The fundamental observation that DEA induces a large degree of
hypomethylation in hepatocytes is significant given that decreases in methylation have
been associated with the activation of oncogenes and transposable elements. The large
extent of hypomethylation combined with the fact that DEA- and CD-induced patterns of
altered methylation were 72% similar is suggestive of common targets and pathways
shared between them. This provides some basis to support the notion that DEA indirectly
depletes the pool of methyl groups needed for methylation of cytosine by inhibiting
choline uptake into cells. PB was also demonstrated to alter methylation with high
similarity (70%) to CD. Therefore, I concluded that altered methylation, specifically
hypomethylation, likely is a non-genotoxic mode of action underlying the abilities of
DEA, PB and CD to promote liver tumorigenesis.

In Chapter 2, mice with variable susceptibilities to liver tumorigenesis were
assessed in terms of their ability to maintain patterns of methylation during promotion
with PB. Additionally, I tested the hypothesis that a progressive accumulation of non-

random changes in methylation are involved in tumorigenesis. Analysis of altered

211

 

methylation in GC-rich regions was key to addressing this multi-part hypothesis. First,
specific regions of altered methylation (RAMs) were seen to carry forward from the 2wk
to 4wk time point, a key component of the promotion stage of tumorigenesis. The number
of accumulated changes was directly related to the relative sensitivities of the B6C3F1
and C57BL/6 mice. Secondly, by comparing specific RAMs in the relatively resistant
C57BL/6 mice to those in the relatively sensitive mice in addition to liver and kidney
tissue I demonstrated that PB-induced patterns of altered methylation are highly unique to
B6C3F1 mouse liver. Finally, gene-specific analysis in the B6C3F1 mice revealed a non-
random pattern of altered methylation in which Ha—ras hypomethylation was associated
with selective increases in gene expression and the heavily methylated LIN E-l elements
were unaffected by PB. This provided the first direct experimental evidence for
progressive, non-random changes in methylation as an important component to
tumorigenesis. My hypothesis also reinforces the conclusions of previous researchers in
this field (Counts et al., 1996; Watson and Goodman, 2002). The data suggest that altered
DNA methylation is an epigenetic mechanism underlying the ability of PB to cause liver
tumorigenesis. Therefore instability of the epi genome likely contributes to susceptibility
to tumorigenesis.

Use of the SENCAR 2-stage mouse initiation/promotion model of skin
tumorigenesis in Chapter 3 was suitable for analyzing altered methylation in GC-rich and
gene-specific regions in response to promotion. This chapter is highly complementary to
chapter 2 in that the notion of progressive changes in methylation could be tested in a
distinct model system. Patterns of DNA methylation in tumor tissue provided a highly

valuable comparison to pre-cancerous tissue. Similar to Chapter 2, RAMs carried

212

 

forward with time and, importantly, were identified in tumor tissue. With 4, 8 and 29wk
(tumor) time points, I was able to depict a more concrete example of progressive changes
in methylation. I also established a dose-response relationship for increases in
methylation and demonstrated reversibility. The observed upregulation of LINE during
promotion may be controlled by regions of the promoter which were not analyzed for
altered methylation. However, the genomic instability created by the upregulation of
LINE during promotion could contribute to tumor formation, even though LINE
expression is not sustained in the skin tumors. Based on decreased methylation in
addition to significant increased expression of Ha-ras, I proposed a role for altered
methylation during promotion which might contribute directly to the activation of Ha-ras
during skin tumorigenesis.

The specific aims, as previously presented, are individually addressed within the

context of the experimental observations.

213

 

Specific Aims
Each of the three chapters in this dissertation focused on testipg the outlined hypotheses

and objectives. Each specific aim has been eggerimentallv examined and is separately

 

addressed.
A) Examination of DEA-induced altered methylation in GC-rich regions as a
contributor to the development of liver tumors in B6C3F1 mice.

1) To assess global and GC rich methylation alterations in response to

DEA, CD, and PB in B6C3F1 mice.
Global methylation status was unchanged in response to DEA, CD, and PB strengthening
the need to specifically examine GC-rich regions for a more detailed picture of overall
altered methylation. GC-rich methylation was assessed via methylation sensitive
restriction digestion, arbitrarily primed PCR, and capillary electrophoretic separation of
PC R products. Treatment with DEA, CD and PB resulted in altered patterns of
methylation, predominantly hypomethylations, suggesting this is an epigenetic
mechanism which contributes to the facilitation of mouse liver tumorigenesis.

2) Compare changes in methylation in GC-rich regions following DEA and PB

treatment to those observed after choline deficiency.
Regions of altered methylation were highly similar between DEA and CD and between
PB and CD. This suggests that all three treatments share common pathways or targets.
The high similarity between DEA- and CD-induced RAMs supports a mechanism
where by DEA disrupts choline homeostasis to induce a methyl deficient state which leads

to altered patterns of methylation.

214

B) Effects of phenobarbital (PB) on gene specific and GC rich regional methylation
status of hepatic DNA in tumor prone and tumor—resistant mice.
1) To determine if cancer susceptibility in mice is related to differences in the
ability to maintain normal patterns of methylation in response to PB
Patterns of methylation in the B6C3F1 tumor prone mice were less stable during
promotion with PB in relation to the C5 7BL/6 mice indicating a decreased ability to
maintain patterns of methylation. This was supported by the observation that the total
number of RAMs at 4wks in B6C3F1 were more numerous than C5 7BL/6 at 4wks. Key to
answering this specific aim was the observation that, a much higher percentage of RA Ms
carried forward fiom 2wk to 4wks in B6C3F1 mice.
2) Characterize progressive changes in methylation by specifically identifying
regions of altered methylation that carry forward with time in the B6C3F1 and
C57BL/6 mice.
The arbitrarily primed PCR method is unique in that specific RAMs can be tracked from
one time point to the next. Therefore, I was able to collect experimental evidence for
progressive changes in methylation which carry forward with time during promotion.
The tumor prone-B6C3F1 mice showed a relatively large percentage of changes which
carried forward in relation to the C5 7BL/6.
3) Identify changes in methylation which are unique to B6C3F1 liver when
compared to altered DNA methylation observed in C57BL/6 liver or B6C3F1
kidney.
In the B6C3F1 mice, 5 7 unique (as compared to the C5 7BL/6) RAMs, predominantly

hypomethylation, were observed in liver after 2 wk of treatment with PB and this

215

 

increased to 86 at 4wk When comparing B6C3F1 RAMs in liver and kidney only 5
RAMs were in common and equivalent. Therefore, 81 total RAMs were unique to
B6C3F1 mouse liver.

4) To determine the effects of PB on the methylation status of the promoter

regions of Ha—ras and LINE-l elements and subsequently analyze changes in their

gene expression.
One particular CpG site 1137nt upstream of the transcriptional start site showed PB-
induced hypomethylation only in the B6C3F1 mice. In addition, the decreases were
observed close to the transcriptional start site. The observed hypomethylation, although
limited, could be linked to the observed selective increases in gene expression. As
expected LINE-I elements were heavily methylated and gene expression was unchanged
in response to PB.

5) To test the reversibility of PB-induced altered methylation in GC-Rich

regions and gene-specific promoter regions (i.e. Ha-ras and LINE-1 elements).
Reversibility was demonstrated for numerous RAMs in GC-rich regions, however, low
reversibility of hypomethylations was a critical observation that was characteristic of
both B6C 3F 1 and C5 7BL/6 mice. One site in the promoter region of Ha-ras was
hypomethylated only in the B6C 3F I mice in response to PB. Upon cessation of the
promoting stimuli this site was largely reversible. Methylation of this site in C5 7BL/6

was stable during promotion and recovery periods.

216

 

C) Characterization of GC-rich and gene specific methylation changes in tumor and
precancerous skin tissue during the promotion stage of the 2-stage,
initiation/promotion SENCAR mouse model.
1) Evaluate the methylation status of GC-rich regions following 8wk promotion
with increasing doses (3, 9, 18, 27mg) of CSC.
Changes in the methylation status of GC—rich regions in the SENCAR mouse model were
treatment related Specifically, the number of increases (hypermethylations and new
methylations) increased in a dose-dependent manner. This is consistent with previous
findings indicating a role for hyperrnethylation in the silencing of tumor suppressor genes
during promotion with CSC in the SENCAR mouse model (Watson et al., 2003 and 2004).
In addition, dose-dependent changes in methylation were consistent with tumor incidence
reported for each dose of CSC.
2) Evaluate the methylation status of GC-rich regions following 4wk and 8wk
promotion with 27mg of CSC.
The total number of changes in the methylation status of GC—rich regions in response to
27mg CSC promotion for 4 and 8wks increased with time.
3) Evaluate the methylation status of GC-rich regions in tumor tissue (29wk) and
compare to precancerous tissue.
Hypomethylated RAMs were predominant in tumor tissue, in contrast to the clear
increase in hypermethylations and new methylations in precancerous tissue. This finding
is consistent with the previous observation in which global hypomethylation was only
seen in tumor tissue (Watson et al., 2003). The total number of changes was comparable

to the 18mg and 27mg CSC promoted precancerous tissue.

217

 

4) Characterize progressive changes in methylation by specifically identifying
regions of altered methylation that carry forward with time from 4wk to 8wk and
from 8wk to tumor tissue (29wk)
GC—rich RAMs induced by CSC were seen to carry forward from 4 to 8wks and from
8wks to tumor in a manner that was consistent with the multistage carcinogenesis model
for a progressive accumulation of changes during promotion. Importantly, RAMs
originating with the 4wk promotion were identified in the tumor tissue.
5) Assess the reversibility of changes in methylation in GC-rich regions upon
cessation of CSC application.
Reversibility was demonstrated for the large majority of RAMs in GC-rich regions
following the 4wk promotion-4 or 8wk recovery and 8wk promotion—8wk recovery dosing
regimens. In addition, GC—rich regions of altered methylation were also observed
following the recovery periods and consisted mainly of hypomethylations, an important
consideration for discerning the mechanisms during recovery.
6) Relate changes in the methylation status of the promoter regions of Ha-ras
and LlNE-l elements to changes in gene expression in both precancerous and
tumor tissue.
Promotion with CSC for 8wk resulted in a generalized dose-dependent increase in the
incidence of hypomethylation at 2 CpG sites surrounding the transcriptional start site.
The incidence of hypomethylation was 100% in tumor tissue. Expression of Ha-ras
increased significantly with 18mg and 27mg CSC and even more so in tumor tissue,
supporting a role for decreased methylation in activating Ha-ras in skin tumorigenesis.

This activation of Ha-ras with 18 and 27mg CSC fit nicely and likely is associated with

218

 

the observed increase in tumor incidence at these two doses. Hypomethylation of one
CpG site in the promoter region of LINE-1 elements was seen only in tumor tissue.
However, gene expression was mildly increased in response to 27mg CSC but not in
tumor tissue suggesting my characterization of altered methylation in the promoter

region was not sufficient to link changes in methylation to changes in gene expression.

219

Support for Hypotheses

Myoverallflpothesis states that susceptibility to carcinogenesis is related
inversely to the capacity to maintain normpl DNA methylation patterns. Research,
presented in Chapters 1-3 was directly related to this hypothesis and largely focused on
demonstrating that changes in methylation are progressive and non-random during the
promotion stage of tumorigenesis in addition to connecting changes in the methylation
status of promoter regions to changes in gene expression. These characteristics do not
prove cause and effect but do strengthen the importance of altered methylation as a
contributor to the development and progression of tumors as opposed to being a
consequence of this process. My findings from three distinct model systems were
complementary and consistent indicating the power and significance of this work.

In Chapter 1, B6C3F1 mouse hepatocytes were treated in culture with DEA, PB,
and CD resulting in highly similar patterns of altered methylation. The characteristic
disruption of methylation patterns in the B6C3F1 mouse in vivo was extended in vitro
and illustrates that instability of the epigenome is a key feature of susceptibility to
tumorigenesis. The ability to induce highly reproducible changes in methylation with
three different treatments indicates that changes in methylation are not entirely random.
Instead, the observed changes are likely the end result of a mis-directed chain of factors
responsible for maintaining homeostasis. Chapter 2 was centered on experimentally
demonstrating that the tumor prone and tumor resistant mice are fundamentally different
in their abilities to maintain patterns of DNA methylation. Compelling experimental
evidence was provided to show that specific regions of altered methylation in response to

PB, carried forward with time, only in the B6C3F1 mice. Unique regions of altered

220

 

methylation as compared to C57BL/6 mice and B6C3F1 kidney were identified as likely
critical to the development of tumors. In the context of the multistage model of
tumorigenesis, DNA methylation in B6C3F1 mice is less stable and alterations are more
apt to accumulate with continued promoting stimuli. With Chapter 3, changes in
methylation occurring in precancerous tissue were directly compared to changes
occurring in tumor tissue. This was a critical link to connecting destabilized patterns of
methylation and, hence susceptibility, to tumor formation. SENCAR mice exhibited a
failure to maintain normal methylation which likely resulted in the observed step-wise
accumulation of changes during promotion. In addition, dose-dependent changes in
methylation were consistent with tumor incidence reported for each dose of CSC.

These critical changes were observed in tumor tissue which connects early
(precancerous) changes in methylation to the neoplastic state. The complementarities of
Chapters 2 and 3 shows that progressive changes in methylation are possibly a universal
feature of tumorigenesis. Also key to supporting the hypothesis was to demonstrate that
changes in methylation could lead to altered gene expression. Hypomethylation of the
Ha—ras oncogene was identified to varying degrees in both Chapters 2 and 3. This
hypomethylation was shown in conjunction with increases in gene expression suggesting
that up-regulation of oncogenes by way of decreased methylation could facilitate

tumorigenesis.

221

DISCUSSION
In my opinion, the following 4 points are the most important and significant aspects of

this research to the field of DNA methylation and carcinogenesis.

1. Development of AP-PCR/Clflas proven to be a consistent and reproducible method
for detecting changes in methylation in GC-rich regions regardless of organ type or
model system.

Methylation sensitive restriction digestion, arbitrarily primed PCR, and capillary
electrophoretic separation of PCR products is a novel approach to measuring changes in
methylation. One very unique feature of the method is the ability to simultaneously
measure treatment-related increases, decreases and new methylations in a high
throughput highly reproducible manner. The fundamental method originated with
Gonzalgo, et al., 1997 and was further developed by Watson and Goodman, 2002b. By
targeting CCGG sites, the recognition site of both Mspl and Hpall, I was able to gauge
the extent of altered methylation in ~7.45% of all CpG dinucleotides estimated for the
mouse genome (Frazzari and Greally, 2004). This portion of the genome includes gene
coding and promoter regions, CpG islands, and transposable elements. A parallel study
performed with BfaI/BssHII more directly targeted CpG islands and expanded the scope
of the method. Additionally, assessment of both CpG and CpCpG methylation via Hpall
and Mspl enzymes respectively was instructive for examining the contribution of CprG
methylation. Comparatively, other commonly employed genome-wide methods focus
only on altered methylation within CpG islands (e.g. RLGS and AIMS). Alternatively, a

gene by gene approach can be applied in which single CpG sites or limited regions are

222

 

examined. Therefore, this method is less restricted allowing for a more comprehensive
approach to addressing treatment related disruption of DNA methylation.

With the ability to more accurately separate the PCR products and quantify the
amount of PCR product amplified, the level and depth of analysis was overwhelmingly
increased. This allowed for very specific comparisons over time, between mice with
different genetic backgrounds and between target and non-target tissues. This very

feature is the sole basis of the majority of my experimental findings.

2. Treatment induced changes in methplation are not purelyrandom. but show defined
and reproducible patterns of disruption that accrue with time.

Changes in methylation on a genome-wide scale during promotion could be
perceived, in general, as a result of random “hits” throughout the genome which, by
chance, cause the activation of oncogenes and transposable elements or the suppression
of tumor suppressor genes to facilitate tumorigenesis. This is supported by seemingly
indiscriminate patterns of altered methylation and the varying extent to which particular
sites are altered from one model system to the next. In addition, global genome wide
hypomethylation is a common feature of tumor tissue, but the exact cause of each one of
those hyomethylation events might be impossible to gauge, at least by today’s
technology, leading to the conjecture that the majority are random events.

From another perspective supported by my research, and one that I am more
inclined to believe, “random” patterns of altered methylation are instead “complex”.
This implies a multi-factorial basis which may be complicated but can not be considered

stochastic. Support for this is first seen in chapter 1, in which patterns of altered

223

 

methylation in response to DEA, CD, and PB share a ~70% similarity. Highly similar
patterns of altered methylation indicate that some of the same pathways are disrupted,
mis-directed, or mis-regulated to result in the highly consistent altered patterns of
methylation. This is hypothesized for DEA and CD. DEA induces a choline deficient-
like state, therefore, methyl deficiency is the end result of both treatments. Two separate
treatments which upset homeostasis in the same way lead to the same changes in
methylation in nearly identical regions of the genome.

Chapter 2 also demonstrated and supported the idea of non-random patterns of
methylation. Highly methylated LIN E-l elements, a very large portion of the genome
(~3 3%), were unaffected by PB promotion however, changes in methylation were seen in
two separate regions of the promoter of Ha-ras. Therefore, multiple key factors involved
in maintaining stable states of methylation were likely misdirected and thus, targeted to
specific genes (e. g. Ha-ras). The idea of susceptible or targeted regions of the genome
has been suggested previously in terms of methylation-prone and methylation resistant
CpG islands (Feltus et al., 2003). The hypothesis states that CpG islands differ in their
inherent susceptibility to aberrant methylation and assumes that there are cis-acting
factors which distinguish methylation-prone and methylation-resistant CpG islands. The
data showed that CpG islands differ in their intrinsic susceptibility to de novo
methylation, and suggested that the propensity for a CpG island to become aberrantly
methylated can be predicted based on its sequence context (F eltus et al., 2003).
Therefore, the characteristics of the genome as well as the specific pathways and factors

regulating patterns of methylation result in the complex, but not random alterations.

224

3. Hypomethylation, occurring simultaneously with hypermethylationiis a predominppt

and highly sigpificant contributor to the development and proggession of tumors.

 

Hypermethylation has received strong focus with good reason as a dominant force
in the silencing of tumor suppressor genes. Methylation related gene silencing has been
demonstrated for genes involved in apoptosis, angiogenesis, differentiation, DNA repair,
metastasis and invasion, drug resistance, and signal transduction, among others (Costello
and Plass, 2001). Research by Waston et al., 2003 and 2004 outlined and demonstrated
the importance of hyperrnethylation events in the progression of skin tumorigenesis with
the finding that HOXAS, p16, and MGMT are all hypermethylated in response to CSC
promotion. These hyperrnethylation events are critical to the development of tumors
however, this approach is very much one sided.

Consistently, both increases and decreases in methylation are observed on a
genome-wide scale in precancerous tissue. Therefore, in light of previous work, I
concentrated more on the contribution of hypomethylation events to tumorigenesis. In
chapters 1 and 2, the predominant alteration induced by DEA, CD and PB was
hypomethylation within GC-rich regions. In chapter 3 a fairly dramatic increase in the
number of hypomethylated regions was seen in tumor tissue as compared to precancerous
tissue. This indicates that hypomethylation potentially plays a variety of roles in the
development of tumors and maintenance of the neoplastic phenotype. As previously
discussed, I have shown decreased methylation in the promoter region of Ha-ras and
selective increases in expression in two separate model systems by examining both
precancerous and tumor tissue. I believe, the true significance of this is really gained by

coupling both the hypomethylation events of the current research with the previously

225

 

identified hyperrnethylation. One very intriguing concept based on the complementary
actions of activation and silencing events is the possibility that upregulation of an
oncogene can lead to down regulation of a tumor suppressor gene through promoter
hyperrnethylation (Sung et al., 2004). This brings together the importance of considering
simultaneous changes as a cumulative series of epigenetic alterations leading to

deregulated cell growth.

4. As outlined by the classic multi-stgge model ofiarcinogenesis, the progpessive
accumulation of critical changes in DNA methylation, both increases and decreases,
appears to be a universg feature of tumor promotion.

As presented by the working model of multi-stage carcinogenesis, the promotion
stage of tumorigenesis involves the step-wise accumulation of heritable changes which
are critical for the selection and clonal expansion of initiated cells (Dragan et al., 1993).
Chapters 2 and 3 specifically addressed this issue and demonstrated that regions of
altered methylation seen at early time points were also identified at later precancerous
and cancerous time points. The nature of the initiated cells could be one or more of the
following three types of cells, 1) stem cells, 2) early stem cell progenitor cells or 3)
dedifferentiated cells (Bj erkvig etal., 2005). In two distinct experimental models with
two separate promoting agents, I have observed that regions of altered methylation carry
forward with time. Our method is based on detecting the a_ve_rpge change in methylation
within a region. This implies that the majority of DNA within the whole liver or skin
tissue across numerous animals had to be exhibiting either an increase or decrease within

the same region. A small minority of cells would not be able to change the statistical

226

 

significance in comparison to control levels. Based on that concept I was likely
measuring changes acquired by both mature hepatocytes and stem cells. This opens up
the possibility that altered methylation, a heritable genomic change, accumulating in a
progressive manner in the differentiated cells of the liver or stem cells could contribute to
their evolution into the cancer stem cell-like state.

In either case, the ability to experimentally measure changes in methylation
within fairly well defined regions of the genome that carry forward with time is
significant. In chapter 2, this was demonstrated at very early time points (2 and 4wks) in
relation to the appearance of foci and tumors in B6C3F1 mouse liver. Of particular
interest in this model was the observation that a large proportion of the regions of altered
methylation in B6C3F1 mouse liver carried forward in comparison to C57BL/6. This is a
highly significant observation and shows that B6C3F1 mice accumulate changes much
quicker and earlier than C57BL/6 which is in direct agreement with their relative
sensitivities to tumor formation. Complementary to this observation was the fact that 81
unique regions of altered methylation were induced by PB in B6C3F1 mouse liver.
Again in chapter 3, the same effect was seen in SENCAR mice at 4 and 8wk time points.
Here, tumor tissue contained many of the changes generated by both 4 and 8wks of
promotion. This is substantial evidence to support not only the heritability of altered
patterns of methylation but the persistence and contribution of those changes to the tumor
phenotype.

The SENCAR mouse model, was characterized by initiation with DMBA and
promotion with CSC. B6C3F1 and C57BL/6 mice were promoted with PB and had no

prior chemical initiation event. Although these two models are very distinct, I was able

227

to consistently show that changes in methylation were progressive. Importantly, this
suggests that the accumulation of critical changes in DNA methylation, both increases

and decreases, appears to be a universal feature of tumor promotion.

Conclusions

The multi-step (i.e. multi-mechanism) and multi-stage process of carcinogenesis
is considered the operational framework for experimentally testing and explaining events
leading to cell proliferation and tumor formation. Preserving normal patterns of DNA
methylation is an essential part of maintaining homeostasis within a cell. The majority of
methylation occurs at cytosines which are 5’ to guanine and these CpG dinucleotides are
located in gene coding and promoter regions (Costello and Plass, 2001) as well as in
promoter-like regions of transposable elements (Liang et al., 2002) Therefore, altered
methylation in the form of hypomethylation or hyperrnethylation in relation to control
could contribute to the initiation and progression of tumorigenesis through the aberrant
regulation of gene expression and the creation of genomic instability (Jones and Baylin,
2002)

Both a gene by gene approach to measuring changes in DNA methylation and a
genome-wide approach are necessary for extending the understanding of altered patterns
of methylation in the process of tumorigenesis, and its role in defining the susceptibility
of an organism to form tumors. Key to my research was the development of a high-
throughput and quantitative method for measuring changes in GC-rich regions of the
genome. This method provided a means for comparing treatment induced regions of

altered methylation over time, between mice with different genetic backgrounds, and

228

between target and non-target tissues. I was then able to combine this analysis with a
targeted and specific approach to examining the promoter regions of the Ha-ras oncogene
and LINE-1 elements. This was a powerful strategy which served to address my overall
hypothesis from multiple angles.

I have characterized patterns of methylation in B6C3F1 mouse hepatocytes in
culture in response to treatment with DEA, CD, or PB, all of which result in an increase
in the incidence and multiplicity of liver tumors. This established that altered
methylation is a fundamental consequence of all three treatments and patterns of
methylation are highly similar among them supporting a non-random disruption of the
genome. In addition, the simultaneous analysis of the Ha—ras oncogene and transposable
elements showed that the characteristics of the genome as well as the specific pathways
and factors regulating patterns of methylation likely result in the complex, but not
random alterations.

I have demonstrated using two separate model systems that changes in
methylation accumulate in a progressive manner with time. As seen with B6C3F1 and
C 57BL/6 mice, the extent and frequency with which changes in methylation accrue
seems to strongly relate to their relative susceptibilities to liver tumorigenesis.
Susceptibility may also directly relate to the number and identity of the unique regions of
altered methylation observed in the B6C3F1 mice. In addition, with the SENCAR mouse
skin initiation/promotion model, regions of altered methylation generated by promotion
with CSC for 4 and 8wks of promotion carried forward to tumor tissue. This is
substantial evidence to support not only the heritability of altered patterns of methylation

but the persistence and contribution of those changes to the neoplastic state. In addition,

229

the large majority of the observed changes were reversible, consistent with the major
hallmark of tumor promotion.

My research represents a logical extension of previous work by Counts et al.,
1996, Watson et al., 2002b, and Watson et al., 2003 and 2004. Importantly, I have
presented novel and detailed data to support altered patterns of DNA methylation as a
contributor to tumorigenesis. Overall, my data illustrate that instability of the epigenome

is a key feature of susceptibility to tumorigenesis.

REFERENCES FOR INTRODUCTION, SUMMARY, AND DISCUSSION
SECTIONS ARE LISTED ON PAGES 231-241.

230

 

REFERENCES

Antequera, F. (2003). Structure, function, and evolution of CpG island promoters. Cell
Mol. Life Sci. 60, 1647-1658.

Bachoo, R.M. et al., (2002). Epidermal growth factor receptor and Ink4a/Arf:
convergent mechanisms governing terminal differentiation and transformation along the
neural stem cell to astrocyte axis. Cancer Cell 1, 269-277.

Ballestar, E. and Esteller, M. (2002). The impact of chromatin in human cancer: linking
DNA methylation to gene silencing. Carcinogenesis 23, 1103-1109.

Balmain, A., Ramsden, M., Bowden, G.T., and Smith, J. (1984). Activation of the
mouse cellular Harvey-ras gene in chemically benign skin papillomas. Nature 307, 65 8-
660.

Banelli, B. et al., (2005). Distinct CpG methylation profiles characterize different
clinical groups of neuroblastic tumors. Oncogene 24, 5619-5628.

Becker, F .F. (1982). Morphological classification of mouse liver tumors based on
biological characteristics. Cancer Res. 42, 3918-3923.

Bestor, T.H. (2002). The DNA methyltransferases of mammals. Hum. Mol. Genet. 9,
2395-2402.

Bhattacharya, S.K., et al. (1999). A mammalian protein with specific demethylase
activity for meG DNA. Nature 397, 579-583.

Bird, A. (2002). DNA methylation patterns and epigenetic memory. Genes and Dev.
16, 6-21.

Bjerkvig, R., Tysnes, B.B., Aboody, D.S., Najbauer, J. and Terzis, A.J.A. (2005). The
origin of the cancer stem cell: current controversies and new insights. Nat. Rev. Cancer
5, 899-904.

Bottiglieri, T., Hyland, K., Reyonlds, EH. (1994). The clinical potential of ademetionine
(S-adenosylmethionine) in neurological disorders. Drugs 48, 137-152.

Boutwell, R.K. Some biological aspects of skin carcinogenesis. In: Homburger F (ed),
Progress in experimental tumor research. S. Karger, New York, 1964, pp 207-250.

Britten, R.J. (1997). Mobile elements inserted in the distant past have taken on
important functions. Gene 205, 177-182.

231

Bruniquel, D. and Schwartz, RH. (2003). Selective, stable demethylation of the

interleukin-2 gene enhances transcription by an active process. Nat. Immunol. 4, 23 5-
240.

Bursch, W., Chabicovsky, M., Wastl, U., Grasl-Kraupp, B., Bukowska, K., Taper, H.,
and Schulte-Hermann, R. (2005). Apoptosis in stages of mouse hepatocarcinogenesis:

failure to counterbalance cell proliferation and to account for strain differences in tumor
susceptibility. Toxicol. Sci. 85, 515-529.

Bursch, W., Wastl, U., Hufnagl, K., and Schulte-Hermann, R. (2005). No increase of
apoptosis in regressing mouse liver after withdrawal of growth stimuli or food restriction.
Toxicol. Sci. 85, 507-514.

Byeon, I-J. L. et al., (2004). Tumor suppressor p16‘NK4A: Determination of solution
structure and analyses of its interaction with cyclin-dependent kinase 4. Mol Cell 1, 421-
431.

Camell, AN. and Goodman, J.I. (2003). The long (LINES) and the short (SINEs) of it:
Altered methylation as a precursor to toxicity. Toxicol. Sci. 75, 229-235.

Castro, R. et al. (2003). Increased homocysteine and S-adenosylhomocysteine
concentrations and DNA hypomethylation in vascular disease. Clin. Chem. 49, 1292-
1296.

Chaumeil, J ., Okarnoto, I., and Heard, E. (2004). X-chromosome inactivation in mouse
embryonic stem cells: analysis of histone modifications and transcriptional activity using
immunofluorescence and FISH. Methods Enzymol. 376, 405-419.

Choi, S-W., and Mason, J .B. (2002). Folate Status: Effects on pathways of colorectal
carcinogenesis. J. Nutr. 132, 24138-24188.

Chong, J-M., et al., (2003). Global and non-random CpG island methylation in gastric
carcinoma associated with Epstein-Barr Virus. Cancer Sci. 94, 76-80.

Choumenkovitch, S.F., Selhub, J ., Bagley, P.J., Maeda, N., Nadeau, M.R., Smith, DE,
and Choi, S-W. (2002). In the cystathionine B-synthase knockout mouse, elevations in
total plasma homocysteine increase tissue s-adenosylhomocysteine, but responses of s-
adenosylmethionine and DNA methylation are tissue specific. J. Nutr. 132, 2157-2160.

Chuang, L.S., Ian, H.I., Hoh, T.W., Ng, H.H., Xu, G. and Li, BR (1997). Human DNA-

(cytosine-5)methyltransferase-PCNA complex as a target for p21WAFl. Science 277,
1996-2000.

232

Coghlan, L.G., et al. (2000). Development and initial characterization of several new
inbred strains of SENCAR mice for studies of multistage skin carcinogenesis.
Carcinogenesis 21, 641-646.

Costello, J.F., and Plass, C. (2001). Methylation matters. J. Med. Genet. 38, 285-303.

Cooper, D.N. and Krawczak, M. (1989). Cytosine methylation and the fate of CpG
dinucleotides in vertebrate genomes. Hum. Genet. 83, 181-188.

Counts, J .L., et al. (1996). Cell proliferation and global methylation status changes in
mouse liver afier phenobarbital and/or choline-devoid, methionine-deficient diet
administration. Carcinogenesis 17, 1251-1257.

Counts, J.L., McClain, R.M., and Goodman, J.I. (1997). Comparison of effect of tumor
promoter treatments on DNA methylation status and gene expression in B6C3F1 and

C57BL/6 mouse liver and in B6C3F1 mouse liver tumors. Mol. Carcinogenesis 18, 97-
106.

Dennis, K., Fan, T., Geiman, T., Yan, Q., and Muegge, K. (2001). Lsh, a member of the
SNF2 family, is required for genome-wide methylation. Genes Dev. 15, 2940-2944.

DiGiovanni, J. (1992). Multistage carcinogenesis in mouse skin. Pharmacol. Ther. 54,
63-128.

Dodge, J.E et al. (2002). De novo methylation of MMLV provirus in embryonic stem
cells: CpG versus non-CpG methylation. Gene 289, 41-48.

Dragan Y.P. et al., (1993). The initiation-promotion-progression model of rat
hapatocarcinogenesis. Proc. Soc. Exp. Biol. Med. 202, 16-24.

Drinkwater, NR, and Ginsler, JJ. (1986). Genetic control of hepatocarcinogenesis in
C57BL/6J and C3H/HeJ inbred mice. Carcinogenisis 7, 1701-1707.

Drinkwater, N.R., Hanigan, M.H., and Kemp, C]. (1989). Genetic determinants of
hepatocarcinogenesis in the B6C3F1 mouse. T oxicol. Lett. 49, 255-265.

Espino, P.S., Drobic, B., Dunn, KL, and Davie, JR. (2005). Histone modifications as a
platform for cancer therapy. J. Cell. Biochem. 94, 1088-1102.

Estellar, M., Corn, P.G., Baylin, SB, and Herman, J .G. (2001). A gene
hyperrnethylation profile of human cancer. Cancer Res. 61, 3225-3229.

Fang, J-Y., Cheng, Z-H., Chen, Y-X., Lu, R., Yang, L., Zhu, H-Y., and Lu, L-G. (2004).

Expression of Dnmtl , demethylase, MeCP2 and methylation of tumor-related genes in
human gastric cancer. W. J. Gastr. 10, 3394-3398.

233

Feinberg, A.P. (2001). Cancer epigenetics takes center stage. Proc. Natl. Acad. Sci.
USA 98, 392-394.

Feinberg, AP., and Vogelstein, B. (1983) Hypomethylation of ras oncogenes in primary
human cancers. Biochem. Biophys. Res. Commun. 111, 47-54.

Feltus, F.A., Lee, E.K., Costell, J .F ., Plass, C., and Vertino, PM. (2001). Predicting
aberrant CpG island methylation. Proc. Natl. Acad Sci. USA 100, 12253-12258.

Fuks, F ., Hurd, P.J., Deplus, R. and Kouzarides, T. (2003). The DNA methyltransferases
associate with HPl and the SUV39H1 histone methyltransferase. Nucleic Acids Res. 31,
2305—2312.

Gardiner-Garden, M. and Frommer, M. (1987). CpG islands in vertebrate genomes. J.
Mol. Bio. 196, 261-282.

Goldsworthy, TL. and Fransson-Steen, R. (2002). Quantitation of the cancer process in
C57BL/6J, B6C3F1 and C3H/HeJ mice. Toxicol. Path. 30, 97-105.

Goodman, J .I., and Watson, RE. (2002). Altered DNA methylation: A secondary
mechanism involved in carcinogenesis. Annu. Rev. Pharmacol. Toxicol. 42, 501-525.

Grassi, G. et al. Inhibitors of DNA methylation and histone deacetylation activate

cytomegalovirus promoter-controlled reporter gene expression in human gliobastoma cell
line U87. Carcinogenesis 24, 1625-1635.

Hahn, W.C., and Weinberg, R. (2002). Rules for Making Human Tumor Cells. N. Engl.
J. Med 347, 1593-1603.

Hanahan, D. and Weinberg, R.A., (2000). The Hallmarks of Cancer. Cell 100, 57-70.

Haseman, J .K., Crawford, D.D., Huff, J .E., Boorman, G.A., and McConnell, BE. (1984).
Results from 86 2-year carcinogenicity studies conducted by the National Toxicology
Program. J. Toxicol. Environ. Health 14, 621-639.

Hata, K, Okano, M., Lei, H., and Li, B. (2002). Dnmt3L co-operates with the Dnmt3
family of de novo DNA methyltransferases to establish maternal imprints in mice.
Development 129, 1983-1993.

Hennings, H. et al., (1997). New strains of inbred SENCAR mice with increased
susceptibility to induction of papillomas and squamous cell carcinomas in skin. Mol.
Carcinogenesis 20, 143-150.

Hergersberg, M. (1991). Biological aspects of cytosine methylation in eukaryotic cells.
Experientia 47, 1171-1185.

234

Herman, J .G. and Baylin, SB. (2003). Gene silencing in cancer in association with
promoter hyperrnethylation. N. Engl. J. Med. 349, 2042-2054.

Hermann, A., Gowher, H., and Jeltsch, A. (2004). Biochemistry and biology of
mammalian DNA methyltransferases. Cell. Mol. Life Sci. 61, 2571-2587.

Hiltunen, MO. and Yla-Herttuala, S. (2003). DNA methylation, smooth muscle cells,
and atherogenesis. Arterioscler. Thromb. Vasc. Biol. 23, 1750-1753.

Hochedinger, K., Yamada, Y., Beard, C., Jaenisch, R. (2005). Ectopic expression of
Oct-4 blocks progenitor-cell differentiation and causes dysplasia in epithelial tissues.
Cell 121, 465-477.

Hutt, J .A. et al. (2005). Life-span inhalation exposure to mainstream cigarette smoke
induces lung cancer in B6C3F1 mice through genetic and epigenetic pathways.
Carcinogenesis 26, l999-2009.

Inoue, S. and Oishi, M. (2005). Effects of methylation of non-CpG sequence in the

promoter region on the expression of human synaptotagmin XI (sytl 1). Gene 348, 123-
134.

Ito, S., Tsuda, M.Y., Shitake, A., Yanai and Masegi, T. (1998). Effect of phenobarbital
on hepatic gap junctional intercellular communication in rats. Toxicol. Pathol. 26, 253-
259.

Jabbari, K. and Bemardi, G. (2004). Cytosine methylation and CpG, TpG (CpA) and
TpA frequencies. Gene 333, 143-149.

Jackson, J.P., Lindroth, A.M., Cao, X., and Jacobsen, SE. (2002). Control of CprG
DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416,
556-560.

James, S.J., Pogribny, I.P., Pogribna, M., Miller, B.J., Jernigan, S. and Melnyk, S.
(2003). Mechanisms of DNA damage, DNA hypomethylation, and tumor progression in
the folate/methyl-deficient rat model of hepatocarcinogenesis. J. Nutr. 133, 37408-
37478.

Jones, PA. and Baylin, SB. (2002). The fundamental role of epigenetic events in
cancer. Nature Rev. 3, 415-428.

Jones, P.A., and Laird, P.W. (1999). Cancer epigenetics comes of age. Nature Genet.
21, 163-167.

Jutterman, R., et al. (1994) 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. USA 91, 11797-11801.

235

Kaneda, A. et al., (2004). Frequent hypomethylation in multiple promoter CpG islands is
associated with global hypomethylation, but not with frequent promoter
hypermethylation. Cancer Sci. 95, 58-64.

Kazazian, H.H., Jr., and Goodier, J.L. (2002). LINE Drive: Retrotransposition and
genome instability. Cell 110, 277-280.

Kim, G.D., Ni, J ., Kelesoglu, N., Roberts, R.J. and Pradhan, S. (2002). Co-operation and
communication between the human maintenance and de novo DNA (cytosine-5)
methyltransferases. EMBO J. 21, 4183-4195.

Klaunig, J .E., Hartnett, J .A., Ruch, R.J., Weghorst, C.M., Hampton, J.A., and Schafer,
L.D. (1990). Gap junctional intercellular communication in hepatic carcinogenesis.
Prog. Clin. Biol. Res. 340D, 165-174.

Knaak, J .B., Leung, H.-W., Stott, W.T., Busch, J., and Biski, J. (1997). Toxicology of
mono- di- and triethanolamine. Rev. Environ. Contam. T oxicol. 146, 1-86.

Knox, J .D. et al. (2000). Inhibition of DNA methyltransferase inhibits DNA replication.
J. Biol. Chem. 275, 17986-17990.

Lande-Diner, L. and Cedar, H. (2005). Silence of the genes — mechanisms of long-terrn
repression. Nat. Rev. Genet. 6, 648-654.

Lehman-McKeeman, L.D. et al., (2002). Diethanolamine induces hepatic choline
deficiency in mice. Toxicol.] Sci. 67, 38-45.

Li, B, Bestor, T.H., and Jaenisch, R. (1992). Targeted mutation of the DNA
methyltransferase gene results in embryonic lethality. Cell 69, 915-926.

Liang, G., et al. (2002). Co-operativity between DNA methyltransferases in the
maintenance methylation of repetitive elements. Mal and Cell Bio. 22, 480-491.

Lin, I.G., Han, L., Taghva, A., O’Brien, LE. and Hsieh, CL. (2002). Murine de novo
methyltransferase Dnmt3a demonstrates strand asymmetry and site preference in the
methylation of DNA in vitro. Mol. Cell. Biol. 22, 704-723.

Lund, AH. and van Lohuizen, M. (2004). Epigenetics and Cancer. Genes and Dev. 18,
2315-2335.

Manenti et al., (1994). Multiple loci affect genetic predisposition to
hepatocarcinogenesis in mice. Genomics 23, 118-124.

236

Marikawa, Y., Fujita, T.C., and Alarcon, VB. (2005). Heterogeneous DNA methylation
status of the regulatory element of the mouse Oct4 gene in adult somatic cell population.
Cloning and Stem Cells 7, 8-16.

Maronpot, R.R., et al. (1995). Mutations in the ras proto-oncogene: clues to etiology
and molecular pathogenesis of mouse liver tumors. Toxicology 101, 125-156.

Melchionne, S., Seidman, 1., Van Duuren, BL. (1986). Spontaneous tumors in
SENCAR mice. Environ. Health Perspect. 68, 135-140.

Mortusewicz, 0., Schermelleh, L., Walter, J., Cardoso, M.C., and Leonhardt, H. (2005).
Recruitment of DNA methyltransferase I to DNA repair sites. Proc. Natl. Acad. Sci. USA
102, 8905-8909.

National Toxicology Program (1999). Toxicology and carcinogenesis studies of

Diethanolamine in F 344/N and B6C3F1 mice (Dermal Studies). NTP TR 478. US.
Department of Health and Human Services, National Institutes of Health.

Newbeme, P.M., deCamagro, J .L.V., and Clark, A.J. (1982). Choline deficiency, partial
hepatectomy, and liver tumors in rats and mice. Toxicol. Path. 10, 95-106.

Niculescu, M.D., and Zeisel, SH. (2002). Diet, methyl donors and DNA methylation:
Interactions between dietary folate, methionine and choline. J. Nutr. 132, 23338-23358.

Nishigaki, M. et al., (2005). Discovery of aberrant expression of R-ras by cancer-linked
DNA hypomethylation in gastric cancer using microarrays. Cancer Res. 65, 2115-2124

Okano, M., Bell, D.W., Haber, DA, and Li, E. (1999). DNA methyltransferases
Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development.
Cell 29, 247-257.

Okoji, R.S., et al. (2002). Sodium arsenite administration via drinking water increases
genome-wide and Ha-ras DNA hypomethylation in methyl-deficient C57BL/6J mice.
Carcinogenesis 23, 777-785.

Oshimo, Y. et al., (2003). Promoter methylation of cyclin D2 gene in gastric carcinoma.
Inter. J. Oncol. 23, 1663-1670.

Ostertag, E.M., and Kazazian, H.H. Jr. (2001). Biology of mammalian L1
retrotransposons. Annu. Rev. Genet. 35, 501-538.

Parzefall, W., Kainzbauer, E., Qin, H-M., Chabicovsky, M., and Schulte-Hermann, R.
(2002). Response of isolated hepatocytes from carcinogen sensitive (C3H) and

insensitive (C57BL) mice to signals inducing replication or apoptosis. Arch. Toxicol. 76,
699-706.

237

Petronis, A. (2004). The origin of schizophrenia: Genetic thesis, epigenetic antithesis,
and resolving synthesis. Biol. Psychiatry 55, 965-970.

Petterson, CL. and Laniel, M.A. (2004). Histones and histone modifications. Curr.
Biol. 14, R546-R551.

Pfeifer, GP. and Darnmann, R. (2005). Methylation of the tumor suppressor gene
RASSFlA in human tumors. Biochem. (Moscow) 70, 699-707.

Pradhan, S. and Esteve, P.-O. (2003). Mammalian DNA (cytosine-5) methyltransferases
and their expression. Clin Immun. 109, 6-16.

Progribny, I.P., Ross, S.A., Wise, C., Pogribna, M., Jones, E.A., Tryndyak, V.P., Jill
James, S., Dragan, Y.P. and Poirier, LA. (2005). Irreversible global DNA
hypomethylation as a key step in hepatocarcinogenesis induced by dietary methyl
deficiency. Mutat. Res. (In Press).

Ramchandani, S., Bhattacharya, S.K., Cervoni, N., and Szyf, M. (1999). Methylation is
a reversible biological signal. Proc. Natl. Acad. Sci. USA 96, 6107-6112.

Ramsahoye, B.H., Biniszkiewicz, D., Lyko, F., Clark, V., Bird, A.P., and Jaenisch, R.
(2000). Non-CpG methylation is prevalent in embryonic stem cells and may be mediated
by DNA methyltransferase 3a. Cell 97, 5237-5242.

Ray, J.S., Harbison, M.L., McClain, R.M., and Goodman, J.I. (1994). Alterations in the
methylation status and expression of the raf oncogene in phenobarbital-induced and
spontaneous B6C3F1 mouse liver tumors. Mol. Carcino. 9, 155-166.

Richards, El. and Elgin, S.C.R. (2002). Epigenetic codes for heterochromatin formation
and silencing: rounding up the usual suspects. Cell 108, 489-500.

Roberston, K.D. et al., (1999). The human DNA methyltransferases (DNMTs) 1, 3a and
3b: coordinate mRNA expression in normal tissues and overexpression in tumors.
Nucleic Acids Res. 27, 2291-2298.

Roberston, K.D. (2005). DNA methylation and human disease. Nat. Rev. Genet. 6,
597-610.

Robertson, K.D. and Wolffe, A.P. (2000). DNA methylation in health and disease. Nat
Rev Genet. 1, 11-19.

Ross, SA. and Poirier, L. (2002). Proceedings of the TRANS-HHS workshop: diet
DNA methylation processes and health. J. Nutrition 132, 2329S-2332S.

Saito, Y., Kanai, Y., Sakamoto, M., Saito, H., Ishii, H., and Hirohashi, S. (2002).
Overexpression of a splice variant of DNA methyltransferase 3b, Dnmt3b4, associated

238

 

With DNA hypomethylation on pericentromeric satellite regions during human
hepatocarcinogenesis. Proc. Natl. Acad. Sci. USA 99, 10060-10065.

Shahbazian M.D., and Zoghbi, H.Y. (2002). Rett Syndrome and MeCP2: Linking
epigenetics and neuronal function. Am J Hum Genet. 71, 1259-1272.

Shi, H. et al. (2002). Expressed CpG island sequence tag microarray for dual screening

of DNA hypennethylation and gene silencing in cancer cells. Cancer Res. 62, 3214-
3220.

Shivapurkar, N. and Poirier, LA. (1983). Tissue levels of S-adenosylmethionine and S-
adenosylhomocysteine in rats fed methyl-deficient, amino acid-defined diets for one to
five weeks. Carcinogenesis 4, 1051-1057.

Singh, S.M., Murphy, B. and O’Reilly, R.L. (2003). Involvement of gene-diet/drug
interaction in DNA methylation and its contribution to complex diseases: from cancer to
schizophrenia. Clin. Genet. 64, 451-460.

Smit, A.P., and Riggs, AD. (1996). Triggers and DNA transposon fossils in the human
genome. Proc. Natl. Acad. Sci. USA. 93, 1443-1448.

Spotswood, HT. and Turner, BM. (2002). An increasingly complex code. J. Clin.
Invest. 110, 577-582.

Stern, MC. and Conti, C]. (1996). Genetic Susceptibility to tumor progression in
mouse skin carcinogenesis. Prog. Clin. Biol. Res. 395, 47-55.

Strahl, B.D., and Allis, CD. (2000). The language of covalent histone modifications.
Nature 403, 41-45.

Sung, J. et al., (2004). Oncogene regulation of tumor suppressor genes in tumorigenesis.
Carcinogenesis 26, 487-494.

Szyf, M., Pakneshan, P., and Rabbani, SA. (2004). DNA demethylation and cancer:
therapeutic implications. Cancer Lett. 211, 133-143.

Tai, M-H., Chang, C-C., Olson, L.K., and Trosko, J.E. (2005). Oct 4 expression in adult
human stem cells: evidence in support of the stem cell theory of carcinogenesis.
Carcinogenesis 26, 495-502.

Tan, N-W. and Li, B.F.L. (1990). Interaction of oligonucleotides containing 6-0-

methylguanine with human DNA (cytosine-5-)-methyltransferase. Biochem. 29, 9234-
9240.

Tao, L., Yang, S., Xie, M., Kramer, P.M., and Pereira, A. (2000). Hypomethylation and
overexpression of c-jun and c-myc protooncogenes and increased DNA methyltransferase

239

 

activity in dichloroacetic and trichloroacetic acid-promoted mouse liver tumors. Cancer
Lett. 158, 185-193.

Toyota, M., Auhja, N., Suzuki, H., Itoh, F., Che-Toyota, M., Imai, K., Baylin, SB, and
Issa, J .P. (1999). Aberrant methylation in gastric cancer associated with the CpG island
methylator phenotype. Cancer Res. 59, 5438-5442.

Trosko, J .E., and Chang, CC. (2001). Mechanism of up-regulated gap junctional

intercellular communication during chemoprevention and chemotherapy of cancer.
Mutation Res. 480, 219-229.

Tsuji-Takayama, K., Inoue, T., Ijiri, Y., Otani, T., Motoda, R., Nakamura, S., and Orita,
K. (2004). Demethylating agent, 5-azacytidine, reverses differentiation of embryonic
stem cells. Bio. Biophys. Res. Comm. 323, 86-90.

Turk, P.W., et al. (1995). DNA adduct 8-hydroxyl-2’-deoxyguanosine (8-

hydroxyguanine) affects function of human DNA methyltransferase. Carcinogenesis 16,
1253-1255.

Vachtenheim, J ., Horakova, 1., Novotna, H. (1994). Hypomethylation of CCGG sites in
the 3 region of H-ras protooncogene is frequent and is associated with H-ras allele loss in
non-small cell lung cancer. Cancer Res. 54, 1145-1148.

Vairapandi, M. (2004). Characterization of DNA demethylation in cancerous cell lines
and the regulatory role of cell cycle proteins in human DNA demethylase activity. J.
Cell. Biochem. 91, 572-583.

Van den Veyver, LB. (2002). Genetic effects of methylation diets. Annu. Rev. Nutr. 22,
255-282.

Wainfan, E., and Poirier, L. (1992). Methyl groups in carcinogenesis: effects on DNA
methylation and gene expression. Cancer Res. 52, 2071S-2077S.

Wang, K-Y. and Shen, C-K. J. (2004). DNA methyltransferase Dnmtl and mismatch
repair. Oncogene 23, 7898-7902.

Warner, K.A., Femstron, M.J., and Ruch, R.J. (2003). Inhibition of mouse hepatocytes
gap junctional intercellular communication by phenobarbital correlates with strain-
specific hepatocarcinogenesis. T oxicol. Sci. 71, 190-197.

Waterston, R.H. et al. (2002). Initial sequencing and comparative analysis of the mouse
genome. Nature 420, 520-561.

Watson, RE. and Goodman, J .I. (2002). Effects of phenobarbital on DNA methylation

in GC-rich regions of hepatic DNA from mice that exhibit different levels of
susceptibility to liver tumorigenesis. T oxicol Sci. 68, 51-8.

240

Watson, R.E., Curtin, G.M., Doolittle, D.J., and Goodman, J.I. (2003). Progressive
alterations in global and GC-rich DNA methylation during tumorigenesis. Toxicol. Sci.
75, 289-299.

Watson, R.E., Curtin, G.M., Hellmann, G.M., Doolittle, D.J., and Goodman, J .I. (2004).
Increased DNA methylation in the HoxAS promoter region correlates with decreased
expression of the gene during tumor promotion. Mol. Carcinogenesis 41, 54-66.

Whysner, J ., Ross, P.M., and Williams, GM. (1996). Phenobarbital mechanistic data

and risk assessment: Enzyme induction, enhanced cell proliferation, and tumor
promotion. Pharmacol. Ther. 71, 153-191.

Woodcock, D.M., Lawler, C.B., Linsenmeyer, M.E., Doherty, J .P., and Warren, W.D.
(1997). Asymmetric methylation in the hypermethylated CpG promoter region of the
human L1 retrotransposon. J. Biol. Chem. 272, 7810-7816.

Xie, S. et al. (1999). Cloning, expression and chromosome locations of the human
DNMT3 gene family. Gene 236, 87-95.

Xu, G.L., et al. (1999). Chromosome instability and immunodeficiency syndrome
caused be mutations in a DNA methyltransferase gene. Nature 402, 187-191.

Xu, Y., Goodyer, C.G., Deal, C., and Polychronakos, C. (1993). Functional
polymorphism in the parental imprinting of the human IGF2R gene. Biochem. Biophys.
Res. Commun. 197, 747-754.

Yoder, J .A., Walsh, CR, and Bestor, T.H. (1997). Cytosine methylation and the ecology
of intragenomic parasites. Trends Genet. 13, 335-340.

Zania, S., Lindholm, M.W. and Lund, G. (2005). Nutrition and aberrant DNA
methylation patterns in atherosclerosis: more than just hyperhomocysteinemia? J. Nutr.
135, 5-8.

Zeisel, SH. (1996). Choline. A nutrient that is involved in the regulation of cell
proliferation, cell death, and cell transformation. Adv. Exp. Med Biol. 399, 131-141.-

241

HIG

lllLllllilillillillillilliljil