University of Southern California Dissertations and Theses

Functional DNA methylation changes in normal and cancer cells

(USC Thesis Other)

Functional DNA methylation changes in normal and cancer cells

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

FUNCTIONAL DNA METHYLATION CHANGES IN NORMAL
AND CANCER CELLS

by

Fides D. Lay

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fullfillment of the
Requirement for the Degree
DOCTOR OF PHILOSOPHY
(GENETIC, MOLECULAR AND CELLULAR BIOLOGY)

DECEMBER 2014

Copyright 2014 Fides D Lay

ii

EPIGRAPH

“If you don’t have time to do it right, when will you have time to do it over?”
---- John R. Wooden

iii

DEDICATION

I dedicate this work to my parents who made the biggest sacrifice for me to be where I
am today. Everyday I remember what they taught me: hard work, perseverance and
humility.

iv

ACKNOWLEDGMENTS
I am immensely grateful of those without whom this work would not be possible:

Dr. Peter Jones, my mentor, who believes in me even when and especially when I
doubt myself. He gives me the freedom to explore science, learn how to think like a
scientist and navigate what it means to be a scientist. He fights for me and sees to it that I
have all the resources to be successful. His unwavering support keeps me going and
inspires me to be a better scientist each day.
Dr. Terry Kelly, who took me under her wings from the very first day I stepped in
to the lab as a rotation student and has become much more than just a source of technical
and emotional support over the years. I would only be so lucky to have a fraction of her
knowledge and vision.
I am also incredibly lucky and honored to have the guidance of Dr. Ben Berman.
From him, I learn to appreciate (epi)genomics and how to pay attention to details that
most people overlook and be critical of my own work. Dr. Yaping Liu, my classmate, go-
to-bioinformatician and the best collaborator I could ever hope for. There is no other
person I would trust my data with and I hope this is just the beginning of a lifelong
collaboration and friendship.
Drs. Gangning Liang and Yvonne Tsai, for overseeing the day-to-day need of the
lab so I have the space and resources to do science. My fellow graduate students in the
lab, past and present, Dr. Sheng-Fang Su, Dr. Shikhar Sharma, Yin-Wei Chang, Chris
Duymich, Ranjani Lakshminarasimhan and Alexandra Soegaard, who all made life as a
graduate student that much more interesting. Especially Yin-Wei and Ranjani, who also
had the misfortunes of having to share their desks with me, my large collection of toys

v

and bubble wraps. Postdocs and many other members of the Jones lab, past and present,
Drs. Claudia Andreu-Vieyra, Daniel de Carvalho, Jueng Soo You, Shinwu Jeong,
Yoshitomo Chihara, Jessica Charlet, Xiaojing Yang, Kamilla Mundbjerg, Minmin Liu,
Elinne Becket, Clayton Collings and Chris Duncan for their helpful advice and
continuous support throughout my time as a student. Dr. Phillippa Taberlay, whose
friendship and encouragement has spanned thousands of miles and time zones.
Dr. Peggy Farnham, whose insights and advices are always spot on. The Farnham
lab for all of their technical assistance and for reminding me that scientists are a fun
group of people. Dr. Gerry Coetzee, whose honest feedbacks as a committee member and
a faithful member of the Jones Lab Meeting I can always count on. Dr. Michael Stallcup
and the T32 CBM Training Program, for the financial support and for giving me an
additional forum where I can learn and grow as a graduate student.
My many scientific collaborators, Dr. Tim Triche, Jr, who is an amazing source of
knowledge and encouragement, and Dr. Kim Siegmund, for her helpful discussions on
how to statistically approach a large dataset. Dr. Charlie Nicolet, Selene Tyndale and
Helen Truong for their support in data generation and incredible flexibility in
accommodating my endless requests. Dr. Sia Daneshmand and Moli Chen, for their
assistance in collecting the clinical samples which resulted in my very first first-author
publication.
Finally, I would like to thank my close friends and family without whom this
journey would not be worthwhile. My parents, for letting me make mistakes. My sisters
and brother in law who know better not to ask what I do but gladly cheer for me anyway.
Howard Chen, for his unparalleled patience and love in dealing with my neurotic

vi

tendencies every step along the way. Joyce Chen, for listening to my panic rants and
indulging my stress-induced-junk-food-eating sessions among many others. Stephanie
Shih and Joanna Wung, for being my much-needed source of sanity. Eric Young, Daniel
Chang and Dustin Chang, for reminding me not to take myself so seriously.

vii

TABLE OF CONTENTS

Epigraph ii

Dedication iii

Acknowledgments iv

List of Tables x

List of Figures xi

Abstract xiv

Chapter 1: An overview of DNA methylation as a target for epigenetic therapy 1
Introduction 1
An overview of DNA methylation and the epigenetic landscape 2
Cancer epigenetics and misregulation of normal processes 6
An overview of DNMT inhibitors (DNMTis) 10
An update on the clinical use of DNMTis 14
Combining DNMTis with other epigenetic drugs in the clinic 17
Overview of thesis research 20

Chapter 2: Reprogramming of the Human Intestinal Epigenome by
Tissue Transposition 22
Introduction 22
Materials and Methods 25
Results
Tissue-specific DNA methylation as markers of sample purity 32
DNA methylation changes occur in the neobladder 36
Time-dependent de novo DNA methylation of small
intestine-specific probes 39
Time-dependent DNA demethylation of non-tissue specific probes 45
DNA methylation changes in the neobladder is a surrogate for
detecting alterations in distinct chromatin states 49
Discussion 54

Chapter 3: Developing NOMe-seq to examine the crosstalk between DNA
methylation and nucleosome occupancy 58
Introduction 58
Materials and Methods 61
Results
Determining optimal NOMe-seq conditions for accurate
footprinting of various genomic regions 67

viii

NOMe-seq detects regions exisiting in multiple chromatin
configurations 71
Measuring the strength of transcription factor binding using
NOMe-seq 73
Applying NOMe-seq genome-wide to characterize promoter
configuration in IMR90 cells 76
Genome-wide NOMe-seq detects accessibility patterns in
transcription factor binding sites 79
Discussion 82

Chapter 4: The context-dependent roles of DNA methylation in directing the
functional organization of the epigenome 85
Introduction 85
Materials and Methods 88
Results
NOMe-seq detects the depletion of nucleosomes at a subset of
genomic enhancers in methyltransferase-deficient DKO1 cells 93
The loss of DNA methylation at CGI promoters results in the re-
organization of nucleosomes 97
The loss of DNA methylation in CGI promoters results in the
acquisition of active and poised chromatin state 101
Demethylated CGI promoters uniformly gain well-phased
nucleosomes and fall into two distinct chromatin states,
distinguished by active and/or poised modification at H3K27
that recapitulate normal-like chromatin states 104
The loss of DNA methylation does not alter the chromatin
structure of non-CGI promoters 111
Long H3K9me3-marked heterochromatin domains have lower
chromatin accessibility and partially methylated DNA 115
Discussion 118

Chapter 5: The loss of DNA methylation contributes to the reversal of epigenetic
silencing in MYOD1 enhancer/promoter pair in cancer cells. 122
Introduction 122
Materials and Methods 125
Results
The MYOD1 enhancer/promoter pair acquires the multivalent
epigenetic signature in cells with reduced DNA methylation 128
Restoration of the Enhancer NDR accompanies the PRC-repressed
state in cells with reduced DNA methylation 132
Treatment of cancer cells with demethylation agents restores the
enhancer NDR, but is insufficient for transcriptional activation 135
Discussion 137

ix

Chapter 6: Mapping DNA methylation and nucleosome occupancy in primary
human tissue 140
Introduction 140
Materials and Methods 142
Results
Mapping DNA methylation and nucleosome occupancy in
uncultured human tissue 146
Comparing NOMe-seq protocol with and without fixing for fresh
frozen human tissue 150
Mapping global nucleosome occupancy and DNA methylation
pattern in fresh frozen colon tumors 153
NOMe-seq detects chromatin configurations of various promoter
types in primary colon tumors. 156
Discussion 158

Chapter 7: Discussion 160
Summary and Discussion 160
Perspective 167

References 168

Appendix: Detailed methodology of the NOMe-seq assay 191

x

LIST OF TABLES

Table 2.1: Primer list 29

Table 2.2: Neobladder sample information 30

Table 2.3: Gene ontology analyses of small intestine-specific enhancers 53

Table 3.1: NOMe-seq primer list 66

Table 5.1: Primer list 127

Table 6.1 Primer list 145

xi

LIST OF FIGURES

Figure 1.1 Epigenetic regulation in normal and cancer cells 5

Figure 1.2 Chemical structure of DNMTi 8

Figure 1.3 Metabolic pathway and mechanism of action of 5-Azanucleosides 9

Figure 2.1 Inflammatory cells-specific DNA methylation markers as surrogate
markers of neobladder sample purity 34

Figure 2.2 Quality control of neobladder samples 35

Figure 2.3 Neobladder samples show limited DNA methylation changes
within the first year of surgery 37

Figure 2.4 Neobladder samples show dramatic increase of DNA methylation
at the VTRNA2-1 locus. 38

Figure 2.5 DNA methylation patterns are cell-type specific 41

Figure 2.6 DNA methylation changes in the neobladder is a dynamic process 42

Figure 2.7 Methylation increase seen in the neobladder is not due to
contamination of inflammatory cells 43

Figure 2.8 Validation of time-dependent increased of DNA methylation in the
neobladders 44

Figure 2.9 Loss of DNA methylation in the neobladder occurs dynamically
on non-tissue specific CpG sites 47

Figure 2.10 The rate of methylation changes in the neobladder is significantly
higher than the rate of methylation changes attributed to aging 49

Figure 2.11 Characterizing functional chromatin states of small intestine 51

Figure 2.12 Alteration of the epigenetic landscape in the neobladder occurs
predominantly in the enhancers and transcribed regions. 52

Figure 3.1 Schematic of NOMe-seq assay 69

Figure 3.2 Optimization and reproducibility of M.CviPI accessibility 70

Figure 3.3 NOMe-seq detects genomic regions present in diverging
conformations 72

xii

Figure 3.4 NOMe-seq detects CTCF binding 75

Figure 3.5 NOMe-seq reveals distinct chromatin configurations at specific 78
promoter types

Figure 3.6 NOMe-seq detects chromatin configuration around transcription
factor binding sites 81

Figure 4.1 Significant loss of DNA methylation in DKO1 cells does not
dramatically increase global accessibility 95

Figure 4.2 NOMe-seq detects the depletion of nucleosomes at a subset of
genomic enhancers in methyltransferase-deficient DKO1 cells 96

Figure 4.3 HCT116 and DKO1 cells have different distribution of DNA
methylation levels at TSS 99

Figure 4.4 DNA methylation is anti-correlated with nucleosome phasing
and accessibility in CGI promoters 100

Figure 4.5 Hypomethylation of CGI promoters triggers dramatic
chromatin remodeling 103

Figure 4.6 Demethylated CGI promoters fall into two distinct chromatin
states distinguished by active and/or poised modification at
H3K27 107

Figure 4.7 Locus-specific example of chromatin changes seen in NP and
NDR promoters 108

Figure 4.8 DKO1 cells gain a normal-like chromatin landscape at CGI
promoters in the absence of DNA methylation 109

Figure 4.9 Unmethylated and accessible CGI TSSs are enriched for general
transcription factor motifs 110

Figure 4.10 DNA methylation does not negatively regulate nucleosome
organization in non-CGI promoters 113

Figure 4.11 The global loss of DNA methylation does not trigger chromatin
remodeling in non-CGI promoters 114

Figure 4.12 Long H3K9me3-marked heterochromatin domains have lower
chromatin accessibility and partially methylated DNA 117

xiii

Figure 5.1 MYOD1 is silenced in HCT116 colon cancer cells 130

Figure 5.2 Epigenetic switching at MYOD1 is reversible 131

Figure 5.3 A nucleosome depleted MYOD1 enhancer is established in cells
with reduced DNA methylation 133

Figure 5.4 A dynamic change in nucleosome occupancy occurs following
5-Aza-CdR induced demethylation 134

Figure 6.1 NOMe-seq detects chromatin accessibility of GRP78 locus of
fresh human tissue 148

Figure 6.2 NOMe-seq detects divergent chromatin structure in the VTRNA2-1
locus 149

Figure 6.3 NOMe-seq can be adapted to footprint nucleosome occupancy in
fresh-frozen colon tumor 152

Figure 6.4 Low-input genome-wide NOMe-seq can be used to detect distinct
chromatin configurations in various regulatory regions 155

Figure 6.5 Low-input NOMe-seq detects distinct chromatin configurations
of different promoter types 157

Figure 7.1 Schematic of changes in CGI promoter architecture 165

Figure 7.2 Schematic of changes in non-CGI promoter architecture 166

Figure A1 Overview of the NOMe-seq protocol 193

Figure A2 NOMe-seq primer design 201

Figure A3 NOMe-seq sequence analysis 204

Figure A4 Visualization of locus-specific and genome-wide NOMe-seq 205

xiv

ABSTRACT

In the mammalian genome, the biochemical addition of a methyl group on the
cytosine of CpG dinucleotides, or DNA methylation, is a component of the epigenetic
landscape that is most frequently associated with gene silencing. The pattern of DNA
methylation is established during development, is unique for each cell type and can be
faithfully copied or inherited during somatic cell divisions. Significant disruption of the
normal methylome has emerged as a key signature of human malignancies and thus, an
attractive target for therapy. Nevertheless, the factors driving DNA methylation changes
and the molecular implication of these changes are still poorly understood. This
dissertation aims to address the contribution of environmental cues on the alteration of
DNA methylation patterns and delineate the effects of DNA methylation changes on the
structure of the cancer epigenome.
Using an in vivo model, I hereby demonstrate that specific signals from the local
tissue environment may be required for the maintenance of a cell-type specific
methylome and that changes in the tissue environment drastically alter the epigenetic
landscape of human intestinal epithelial cells. Comprehensive investigation of the loss of
DNA methylation, using a global hypomethylation colorectal cancer cell line model and
techniques such as NOMe-seq, ChIP-seq and RNA-seq, reveal that the effects of DNA
methylation are highly variable depending on genomic context. Specifically, I show that
DNA methylation controls CpG islands (CGI), but not non-CGI, promoters at the
chromatin level such that the loss of methylation in cancer may result in the
reestablishment of a permissive and/or poised chromatin landscape, characterized by the
presence of certain histone modifications and well-phased nucleosomes. By treating

xv

cancer cells with the FDA-approved demethylating agent, 5-Aza-CdR, I also illustrate
that DNA methylation contributes to the silencing of distal regulatory regions, which
abrogates the necessary crosstalks between promoter and enhancers for gene activation.
Furthermore, I expand my work to generate the first integrated map of DNA methylation
and nucleosome positioning of colon adenocarcinomas, which is used to examine the
relationship between the epigenome and the underlying disease. Taken together, this
dissertation presents a detailed view of the human epigenome, normal and cancer, and
illustrates an often-overlooked influence of DNA methylation in shaping the chromatin
landscape.

CHAPTER 1

AN OVERVIEW OF DNA METHYLATION AS A TARGET
FOR EPIGENETIC THERAPY

This chapter is modified from a previously published review in Trends in
Pharmacological Sciences

INTRODUCTION
The epigenome consists of heritable and inter-related processes necessary for
normal cellular functions, including DNA methylation, histone variants and
modifications as well as nucleosome positioning. During mammalian development and in
differentiated somatic cells, these mechanisms play important roles in the establishment
and maintenance of cellular identity by regulating differential gene expression in all cell
types. The most-studied epigenetic mechanism, DNA methylation, is a covalent addition
of a methyl group on the fifth position of cytosines existing in the context of CpG
dinucleotides. When present in the promoter regions, DNA methylation is associated with
permanent transcriptional silencing. Disruption in the normal DNA methylation pattern is
strongly correlated with human malignancies, and as such, DNA methylation has
received much attention in the race toward developing an effective epigenetic-based
cancer therapy. However, the advance of genome-wide epigenomic studies in recent
years has revealed intricacies in the roles of DNA methylation in normal and diseased
cells and it is increasingly clear that understanding the nuanced functions of DNA
methylation requires a holistic view of the epigenetic landscape. In this chapter, I will
present an overview of the normal and cancer epigenome, factors that influence the
maintenance and dysregulation of the human DNA methylome and current strategies
1
available to target DNA methylation for cancer therapy. I will then discuss the aspects of
DNA methylation that are still unknown and outline the approaches I have undertaken to
address some of the critical questions in the field.

AN OVERVIEW OF DNA METHYLATION AND THE EPIGENETIC
LANDSCAPE
In normal mammalian cells, DNA methylation plays an indispensable role in
genomic imprinting, X-inactivation and the silencing of repetitive elements,
retrotransposon and tissue-specific genes. Most CpG dinucleotides in somatic cells are
methylated, excepting those present in the CpG islands (CGI) where more than 50% of
gene promoters are located (Figure 1.1a) (Deaton and Bird, 2011; Lister et al., 2009;
Ziller et al., 2013). The exact definition of a CGI has been debated at length though
generally, CGIs have been accepted as regions of >500bp DNA having a GC content of
>= 55%, and a ratio of observed and expected CpG>0.65 (Takai and Jones, 2002). The
current known functions of DNA methylation are based on extensive studies of CGI
promoters where the presence of DNA methylation is associated with transcriptionally
silenced genes (Deaton and Bird, 2011). Methylated CGI promoters may recruit various
methyl-binding proteins, thus promoting the compaction of the chromatin which
subsequently reduces accessibility by transcription factors and other DNA binding
proteins, thereby contributing to the silencing of said promoters. Though less definitive,
there is also growing evidence that DNA methylation directly controls gene expression of
non-CGI promoters (Han et al., 2011; Jones, 2012).
2
The establishment of unique and cell type-specific DNA methylation pattern
occurs during early development and is mediated by de novo DNA methyltransferases
DNMT3A and DNMT3B which use nucleosomal DNA as substrates (Okano et al., 1999;
Sharma et al., 2011). Embryonic stem (ES) cells are largely unmethylated with the
exception of a subset of silenced tissue-specific genes, and undergo genome-wide DNA
methylation repatterning during cellular differentiation and fate commitment in a process
often referred to as epigenetic reprogramming (Zhu et al., 2013). Once established, the
methylomes of normal differentiated cells are thought to be very stable, with later
dramatic changes associated with tumorigenesis and other disease progression (Baylin
and Jones, 2011). The faithful inheritance of DNA methylation in somatic cells is
mediated by DNMT1, which, in cooperation with UHRF1, targets hemimethylated DNA
and methylates the daughter strand during cell replication (Bostick et al., 2007). Studies
from our laboratory have subsequently demonstrated that this process also requires
DNMT3A/B (Jeong et al., 2009; Sharma et al., 2011).
It is important to note that DNA methylation works together with other epigenetic
modifications to organize the genome into a functional unit. A core histone octamer
consisting of two H2A-H2B dimers and one H3-H4 tetramer forms a nucleosome when
147 bp of DNA is wrapped around it. Covalent post-translational histone modifications,
such as acetylation, methylation, phosphorylation or sumoylation, occur in the N-terminal
tails of histone residues. Depending on the modifications, actively transcribing
euchromatin or inactive heterechromatin can be formed (Zhou et al., 2011). For example,
histone acetylation is associated with active promoters while trimethylation of lysine27
on histone H3 (H3K27me3) is a mark for polycomb-repressed promoters. The positioning
3
of nucleosomes, especially around transcription start sites (TSS), also plays an important
role in determining the accessibility of transcription factors to promoter DNA.
Altogether, DNA methylation, nucleosome positioning, as well as histone variants and
modifications, cooperate to create a dynamic epigenetic landscape that controls whether a
gene is turned on or off (Portela and Esteller, 2010; Zhou et al., 2011).
In recent years, efforts have been made to dissect the function of DNA
methylation in regions beyond the promoters. These regions, which include enhancers,
insulators and gene bodies, have been shown to have distinct DNA methylation pattern
compared to promoters and thus may point to differing regulatory processes. Enhancers,
for instance, are mostly CpG-poor and have a variable methylation status (Stadler et al.,
2011). Generally, enhancers are characterized by the presence of the H3K4me1
modification and their activity is defined by the presence of H3K27Ac modification
which is anti-correlated with DNA methylation (Heintzman et al., 2007; Rada-Iglesias et
al., 2011). Gene body methylation, unlike promoter methylation, is thought to be
positively correlated with active transcription although the specific mechanism governing
this relationship remains unclear (Jones, 1999).

4
A
B
Figure 1.1 Epigenetic Regulation in Normal and Cancer Cells. Schematics of the
epigenetic landscape in promoter regions are shown. Arrows represent transcription
start site (TSS); filled circles represent methylated CpG dinucleotides and empty
circles represent unmethylated CpG dinucleotides. (A) In normal cells, genes such as
MLH1 and CDKN2A are generally unmethylated and packaged with active modified
histone proteins (e.g. H3K4me3) as well as histone variants (e.g. H2A.Z). These
epigenetic modifications constitute an “open” chromatin structure which, with a
nucleosome depleted region (NDR), favors transcription. In other genomic regions,
such as in the repetitive elements, the CpG sites are methylated and thereby maintain a
closed chromatin structure. (B) In cancer cells,, epigenetic modifications are disrupted.
Besides cancer-specific hypomethylation (e.g. in repetitive sequences), there are two
interrelated epigenetic mechanisms to repress gene expression. Some genes
(e.g.FBXO32) can be recognized by polycomb proteins, such as EZH2 which catalyzes
H3K27 methylation, and are consequently repressed. By contrast, CpG sites within
other promoters can undergo de novo methylation by DNMT3A/B. The methylated
CpG sites attract methyl-binding proteins such as MBD, which is coupled with HDAC
proteins as well as histone methyltransferase to remove histone acetylation and
trimethylate H3K9, respectively. Additionally, nucleosomes cover the promoter region,
generating a tightly closed chromatin status to shut down gene expression.
!
5
CANCER EPIGENETICS AND MISREGULATION OF NORMAL CELLULAR
PROCESSES
Cancer has been described as a genetic disease, arising due to mutations which
disrupt the balance between oncogenes and tumor suppressor genes. It has become clear
in recent years, however, that aberrant epigenetic alterations are also involved in the
initiation and progression of cancer. Early studies measuring the global content of 5-
methylcytosine in human tumors showed that hypomethylation, which may lead to
chromosomal instability and transcriptional deregulation, was a common feature of
carcinogenesis (Eden et al., 2003; Pogribny and Beland, 2009; Wu et al., 2005).
However, most cancer epigenetic studies have been focused on focal CGI
hypermethylation in cancer, revealing that many tumor suppressor genes, cellular
functional genes, and miRNAs are silenced by promoter DNA methylation (Figure 1.1b)
(Esteller, 2007; Lujambio et al., 2008; Toyota et al., 2008).
Recent genome-wide studies have demonstrated distinct patterns of DNA
methylation in cancerous tissues compared to their normal counterparts (Baylin and
Jones, 2011; Noushmehr et al., 2010). The detailed mechanisms by which these discrete
regions undergo hyper- or hypomethylation are unclear. Early evidence suggested that
elevated DNA methyltransferase levels might trigger hypermethylation of tumor
suppressor gene promoters, causing proliferation of cancer cells (Kautiainen and Jones,
1986). In addition to this ‘selection’ model, studies have also suggested that the
establishment of aberrant epigenetic profiles in cancer is reminiscent of the epigenetic
reprogramming that occurs during development (Baylin and Jones, 2011). During cancer
initiation, the promoters of genes that are repressed by H3K27me3 in normal
6
differentiated cells become preferentially methylated and thereby set up for long-term
silencing. This ‘epigenetic switch’ could be regulated by the cooperation of polycomb
proteins and DNMTs (Gal-Yam et al., 2008; Ohm et al., 2007; Schlesinger et al., 2007).
In cooperation with DNA methylation, other epigenetic mechanisms also exhibit
abnormal regulation in cancer. For example, histone deacetylases (HDACs) are often
found to be overexpressed in various types of cancer, resulting in histone deacetylation
and a more compact chromatin structure around the TSS (Halkidou et al., 2004). In
addition, H3K4me2/3 is also selectively demethylated by the histone lysine demethylase
(LSD1), which is upregulated in cancer, making LSD1 a potential drug target (Schulte et
al., 2009). In some loci, polycomb-group (PcG) proteins associated with H3K27me work
independently of DNA methylation to aberrantly repress genes in cancer cells (Kondo et
al., 2008; Lin et al., 2007). Nucleosome occupancy is also switched from an ‘open’ to a
‘covered’ status in gene regulation elements in neoplastic cells (Lin et al., 2007).

7
A
B
Figure 1.2 Chemical Structure of DNMT inhibitors. (A) Nucleoside analogues (B)
non-nucleoside analogues.
8
Figure 1.3 Metabolic pathway and mechanism of action of 5-Azanucleosides. (A)
Human Concentrated Nucleoside Transporters 1 (hCNT1) facilitates the entry of 5-Aza-
CR and 5-Aza-CdR into cells where they are phosphorylated by uridine-cytidine kinase
and deoxycytidine kinase respectively. 5-Aza-CMP and 5-Aza-dCMP are subsequently
phosphorylated into their active triphosphate forms. Derivatives of 5-Aza-dCR can be
incorporated solely into DNA, whereas 5-Aza-CR can be incorporated into RNA as well
as DNA following the reduction of its 5-Aza-CDP form to the 5-Aza-dCDP form.
Incorporation of 5-Azanucleosides into DNA induces hypomethylation of the daughter
DNA strands whereas incorporation of 5-Aza-CR into RNA disrupts crucial cellular
process such as protein translation and causes ribosomal disassembly. (B) A working
model of 5-Azanucleosides. During DNA replication, 5-Azanucleosides are incorporated
into DNA and trap DNMTs, which are subsequently targeted for proteosomal
degradation. DNA containing azanucleosides are hemimethylated after the first round of
DNA replication and become fully demethylated after several rounds of replication.
Using hypomethylation agents, the silenced epigenetic modifications could be switched
to an active status (eg.H3K4me3 and AcH3).
A
B
9
AN OVERVIEW OF DNMT INHIBITORS
Epigenetic modifications play a crucial part in regulating normal cells, but these
processes are disrupted during tumorigenesis. The relatively reversible character of
epigenetic alterations (in contrast to genetic changes) has inspired the development of
therapeutic strategies targeting various epigenetic components. Among them, DNA
methylation and its associated enzymes have been well studied. The understanding of
their fundamental mechanism of action and correlation with other epigenetic
modifications makes them attractive drug targets.

Nucleoside analogs
5-Azacytidine (5-Aza-CR) and 5-aza-2’-deoxycitidine (5-Aza-CdR) are the two
most potent DNMT inhibitors. They have been approved by the Food and Drug
Administration (FDA) in the US for the treatment of myeloid malignancies (Figure 1.2).
They were first synthesized as cytotoxic agents. In the 1980s, these compounds were
found to have hypomethylating activity after incorporation of their derivatives into the
DNA of actively replicating tumor cells (Jones and Taylor, 1980; Taylor and Jones,
1979).
Upon transport into cells by the human concentrative nucleoside transporter 1
(hCNT1), 5-Aza-CR and 5-Aza- CdR are phosphorylated by different kinases, converting
them to their active triphosphate forms, 5-Aza-CTP and 5- Aza-dCTP, respectively
(Figure 1.3a) (Issa and Kantarjian, 2009; Rius et al., 2009). Derivative of 5-Aza-CR can
be incorporated into RNA and DNA after the reduction of 5-Aza-CDP by ribonucleotide
reductase, whereas derivative of 5-Aza-CdR incorporates into DNA after its
10
phosphorylation to 5-Aza-dCTP (Stresemann and Lyko, 2008). The incorporated 5-
azanucleoside disrupts the interaction between DNA and DNMTs through the nitrogen in
the 5’ position of the modified pyrimidine, and traps DNMTs for proteosomal
degradation (Figure 1.3b) (Ghoshal et al., 2005; Kuo et al., 2007). The depletion of
DNMTs results in the passive loss of cytosine methylation in the daughter cells after
replication. This process is also associated with reduced H3K9me3, increased H3ac and
H3K4me3 modifications around gene promoter regions, as well as the formation of a
nucleosome-deficient region (Lin et al., 2007; Nguyen et al., 2002). A recent genome-
wide study of alterations in the epigenetic landscape after 5-Aza-CR treatment further
validated this observation (Komashko and Farnham, 2010). The demethylation function
of 5-Aza-CR and 5-Aza-CdR is most evident at low drug concentration because the drugs
exhibit greater cytotoxicity, interfere with DNA synthesis, and cause DNA damage at
higher concentrations (Qin et al., 2009). The S-phase is also required for the selective and
effective incorporation of these two drugs into the DNA of rapidly proliferating cells,
thereby limiting unwanted hypomethylation in cells arrested in G0/G1 phase.
Besides inhibiting DNMTs, 5-Aza-CR incorporates into RNA and interrupts
normal cellular processes by inducing ribosomal disassembly and preventing the
translation of oncogenic proteins (Li et al., 1970; Stresemann and Lyko, 2008). The
ability of 5-Aza-CR derivatives to be incorporated into DNA and RNA increases its side
effects in vitro and in vivo because it can function in resting and dividing cells
(Hollenbach et al., 2010). However, 5-Aza-CR and 5-Aza-CdR are readily hydrolyzed in
aqueous solution and subject to deamination by cytidine deaminase. The instabilities of
these compounds inevitably present a challenge to their clinical applications.
11
To improve the stability and efficacy of 5-Azanucleosides, several other cytidine
analogs have been developed (Figure 1.2a). For example, zebularine (a cytidine analog
that lacks an amino group in the 4 position of the pyrimidine ring) can inhibit DNMTs
and cytidine deaminase after oral administration (Cheng et al., 2004; Yoo et al., 2004;
Yoo et al., 2008). Studies have shown that zebularine induces hypomethylation in breast
cancer cell lines and reactivates silenced tumor suppressor genes (Billam et al., 2010;
Flotho et al., 2009). The inefficient metabolic activation of this compound has, however,
delayed its clinical use as a single agent. As an inhibitor of cytidine deaminase, its co-
administration has increased the efficacy of 5-Aza-CdR (Lemaire et al., 2009). The
cytidine analog 5-Fluoro-20-deoxycytidine (5-F-CdR) had also been demonstrated to
have hypomethylating activity in mouse cells as well as in human breast and lung
carcinoma cells (Beumer et al., 2006; Jones and Taylor, 1980). Clinical studies further
showed that co-administration of 5-F-CdR with the cytidine deaminase inhibitor
tetrahydrouridine (THU) improved the stability of 5-FCdR (Beumer et al., 2008). The
therapeutic potential of another stable analog, dihydro-5-azacytidine (DHAC), was also
assessed for the treatment of malignant mesothelioma, but results on this compound’s
clinical efficacy have been inconsistent (Kratzke et al., 2008).
The effort to improve the stability of DNMT inhibitors includes the development
of prodrugs of the nucleoside analogs. A preclinical study showed that NPEOC-DAC (a
prodrug of 5-Aza-CdR containing a 2-(p-nitrophenyl)ethoxycarbonyl (NPEOC) group at
the 4 position of the pyrimidine ring) can be incorporated into DNA and inhibit DNMTs
after its activation by human carboxylesterase 1 in a liver cancer cell line. The NPEOC
moiety protects 5-Aza-CdR from deamination, but the compound itself is less potent
12
when administered at the same concentration as 5-Aza-CdR. Moreover, the activity of
NPEOC-DAC is dependent upon carboxylesterases, which are not expressed in all
tissues. Further studies are required to explore the use of this compound in combination
therapy (Byun et al., 2008). Alternatively, S110 (a dinucleotide containing the 5-
azacytosine ring) has also been shown to improve the efficacy of 5-Aza- CdR by
protecting it from deamination. The compound is well tolerated and can reduce the level
of DNA methylation in the CDKN2A promoter region in xenografts (Chuang et al.,
2010).

Non-nucleoside analogs
Unlike cytidine analogs, non-nucleoside DNMT inhibitors (Figure 1.2b) do not
require incorporation into DNA, and thus might exhibit less cytotoxicity. Some of the
compounds assessed for their potential to induce hypomethylation in solid tumors are
hydralazine and procainamide, the widely used vasodilator and antiarrhythmic agents,
respectively (Segura-Pacheco et al., 2006). Hydralazine has been reported to block the
activity of DNMTs by the interaction of its nitrogen atoms with the Lys-162 and Arg-240
residues of the enzyme, whereas procainamide acts similarly as a competitive inhibitor by
preferentially binding to DNMT1 (Singh et al., 2009; Song and Zhang, 2009). These
compounds, however, have limited DNA hypomethylation activity in living cells
(Chuang et al., 2005). The small molecule RG108 also shows the potential to reactivate
tumor suppressor genes in human colon cancer cells (Stresemann et al., 2006; Suzuki et
al., 2010). Recently, the lipophilic, quinoline-based compound SGI-1027 was
demonstrated to be a novel DNMT inhibitor in vitro. In RKO cells, SGI-1027 causes the
13
degradation of DNMT1 and the demethylation of the CDKN2A gene promoter as well as
reactivating silenced genes (Datta et al., 2009).
An alternative strategy to inhibit DNMT1 includes the use of short-chain
oligodeoxynucleotides and microRNAs. MG98 is a 20-bp antisense oligonucleotide that
specifically binds to the 30 UTR of human DNMT1mRNA to prevent its translation.
Despite promising results in preclinical studies, the clinical use of MG98 has not been
validated (Amato, 2007). MicroRNA miR29a, which targets DNMT3A/B directly and
DNMT1 indirectly in a similar way to MG98, can reduce global DNA methylation and
reactivate CDKN2B (Garzon et al., 2009).

AN UPDATE ON THE CLINICAL USE OF DNMT INHIBITORS
5-Azacytidine (Vidaza, Azacitidine)
In early clinical trials, 5-Aza-CR was administered at maximum tolerated doses
(MTD) to patients with osteogenic sarcoma or other cancer related diseases, but showed
unfavorable toxicity (Srinivasan et al., 1982; Velez-Garcia et al., 1977). Increasing
knowledge of the mechanism of action of 5-Aza-CR indicated that lower dosages of 5-
Aza-CR could act as demethylating agents with minimal effects on DNA synthesis (Yoo
and Jones, 2006). 5-Aza-CR was approved by the FDA for the treatment of MDS based
on the positive results from the GALGB9221 clinical trial (Silverman et al., 2002). A
reported 60% of 5-Aza-CR-treated patients exhibited various levels of response, whereas
only 5% of patients in the supportive group showed hematological improvement (HI). 5-
Aza-CR also benefited patients by delaying progression time to acute myeloid leukemia
(AML), improving quality of life, and prolonging overall survival in Refractory Anemia
14
with Excess Blasts (RAEB) or Refractory Anemia with Excess of Blasts in
Transformation (RAEB-T) subgroups (Kornblith et al., 2002). To further validate the
efficiencies of 5-Aza-CR in MDS patients, the European AZA-001 trial was conducted
for intermediate- and high-risk MDS patients. In that study, patients treated with 5-Aza-
CR showed a significant improvement in median overall survival (OS) than patients who
received conventional care regimens (CCR): 24.5 months vs 15.0 months. It was the first
time that 5-Aza-CR treatment was demonstrated to prolong OS in high-risk MDS patients
(Fenaux et al., 2009). Fenaux et al. also demonstrated that 5-Aza-CR benefited older
AML patients by prolonging the OS from 16.0 months to 24.5 months while reducing the
prevalence of side effects (Fenaux et al., 2010). Detailed analyses of AZA-001 indicated
that low-dose cytarabine (one of CCR for patients with higher-risk MDS) was less
efficient and more toxic as compared with 5-Aza-CR (Santini, 2009). Based on the
outcome of the AZA-001 trial, the National Comprehensive Cancer Network (NCCN)
recommended 5-Aza-CR as the preferred therapy for patients with high-risk MDS. In
addition to high-risk MDS, 5-Aza-CR can also be used as a potentially effective
treatment for patients with low-risk MDS (Musto et al., 2010). However, the current
FDA-approved, 7-day course of 5-Aza-CR requires weekend treatment and is
inconvenient for patients and care providers. To overcome this problem, Lyons et al.
designed three alternative regimens which avoided weekend treatment (Lyons et al.,
2009). Patients who received any one of the three regimens showed similar hematological
improvement as the previously approved 7-day 5-Aza-CR regimen, as well as a higher
transfusion independent rate. More patients who needed transfusion of red blood cells
(RBCs) at baseline became independent of RBC transfusion. Another phase-II trial
15
administered an alternative 5-day 5-Aza-CR intravenous schedule and reported a 27%
partial response (PR) + complete remission (CR) rate that was comparable with the 7-day
subcutaneous regimen (Martin et al., 2009). However, further studies are required to
define the survival benefit of these modified regimens. Although one group reported that
a limited number of treatment cycles could achieve a overall response rate (ORR) of 50%
according to new criteria set by the International Working Group, most clinical trials
indicate that prolonged exposure to 5-Aza-CR will benefit patients (Muller-Thomas et al.,
2009).

5-aza-2’-deoxycitidine (Decitabine)
5-Aza-CdR was also approved by the FDA for MDS therapy, but there is no clear
evidence indicating 5-Aza-CdR improves OS. In the USA registered trial (D-0007),
patients treated with 5-Aza-CdR had a 17% ORR, which was significantly higher than
that in the best supportive care (BSC) group (0%). When comparing the OS between the
5-Aza-CdR and control arms, a statistical improvement (14.0 vs 14.9 months) was not
observed, even though clinical benefits (e.g. independence from RBC transfusion or
elongation of median time to AML progression) were seen after 5-Aza-CdR treatment
(Kantarjian et al., 2006). A similar negative result of the OS advantage of 5-Aza-CdR for
older patients with MDS or Chronic Myelomonocytic Leukemia (CMML) had also been
reported (Lubbert et al., 2011). To increase the CR rate of patients with these diseases
taking 5-Aza-CdR in the outpatient setting, several clinical trials explored alternative
schedules. Kantarjian et al showed that a 5-day intravenous schedule with the highest
dose-intensity yielded the highest CR rate (39%) (Kantarjian et al., 2007). In a follow-up
16
study, the ADOPT trial reported an ORR of 32%, suggesting that this 5-day schedule was
as effective as the approved inpatient regimen (Steensma et al., 2009). 5-Aza-CdR is
currently being investigated in other cancer types to assess the best conditions for
administration (Schrump et al., 2006; Steensma, 2009; Stewart et al., 2009).

MG98
Several phase I/II clinical trials have been conducted to determine the tumor types
sensitive to MG98 and an appropriate working dosage. Some of the patients already
showed decreases in DNMT1 levels, but a consistent correlation of DNMT1 level and
dosage has not been observed (Klisovic et al., 2008; Winquist et al., 2006). New evidence
indicated that 7-day continuous dosing of MG98 was well tolerated for patients with
advanced solid tumors, but an objective clinical response was not reported (Plummer et
al., 2009). Most recently, a clinical trial was also conducted to determine the efficacy of
MG98 to sensitize renal cell carcinoma cells that were resistant to IFN-a therapy alone
with promising results (Amato et al., 2012).

COMBINING DNMT INHIBITORS WITH OTHER EPIGENETIC DRUGS IN
THE CLINIC
5-Aza-CR has been used in combinatinon with the HDAC inhibitor valproic acid
(VPA), a class of epigenetic drugs targeting histone deacetylases, to treat MDS and
AML. Soriano et al. conducted a phase-I/II clinical trial in which acute myeloid leukemia
patients were treated with 5-Aza-CR and VPA every day for 7 days. An ORR of 42%
indicated that this combination strategy was clinically effective (Soriano et al., 2007).
17
Another phase-II study sequentially administered 5-Aza-CR after VPA and observed a
CR+PR of 30.7%. As the plasma concentration of VPA increased, (>50 mg/mL vs <50
mg/mL), a better median survival rate was observed (18.7 vs 10 months), indicating
HDAC inhibitor have the potential to increase 5-Aza-CR efficacy (Voso et al., 2009).
Combination therapy of 5-Aza-CR + VPA has also been investigated in breast cancer,
colon cancer and other advanced cancers (Braiteh et al., 2008). Other HDAC inhibitors
have also been combined with 5-Aza-CR. For example, 46% of patients with MDS or
AML who received 5-Aza-CR supplemented with entinostat (another potential HDACi)
showed promising outcomes. Among them, 3 patients had CR, 4 patients had PR and 7
patients had hematologic improvement (Fandy et al., 2009). In a phase-I study,
researchers co-administered 5-Aza-CR and the first-generation HDAC inhibitor sodium
phenylbutyrate to patients with refractory solid tumors. The three-dose schedule
administered in that study showed mild toxicity, albeit with few benefits for the patients
(Lin et al., 2009).
Similar to 5-Aza-CR, 5-Aza-CdR has also been tested for its therapeutic efficacy
in combination with HDAC inhibitors. 5-Aza-CdR + VPA achieved a 22% objective
response in patients with leukemia although, in this case, the higher level of VPA did not
correlate with clinical activity (Garcia-Manero et al., 2006). A later study confirmed this
conclusion, demonstrating that 5-Aza-CdR itself had promising clinical activity in elderly
patients with AML. Adding 25 mg/kg/d of VPA neither enhanced the efficacy of 5-Aza-
CdR nor did it increase the re-expression of CDKN2B (Blum et al., 2007).
5-Aza-CR and 5-Aza-CdR have also been combined with other conventional
therapies. For example, a recent study combining 5-Aza-CR with the FDA-approved
18
MDS chemotherapy drug lenalidomide revealed an impressive ORR of 67% (Sekeres et
al., 2010). 5-Aza-CdR has been combined with carboplatin, imatinib, or with
gemtuzumab ozogamicin (monoclonal antibody) for the treatment of various cancer types
(Appleton et al., 2007; Chowdhury et al., 2009; Oki et al., 2007). The encouraging results
from these clinical trials could broaden the application of DNMTi to various tumor types,
including solid tumors, but further randomized clinical trials with control arms are
needed to demonstrate the feasibility of these approaches.

19
OVERVIEW OF THESIS RESEARCH
The growing use of DNA demethylating agents in the clinics, whether as single
agent approach or in combination with other drugs, makes it imperative that we
understand all aspects governing the roles and regulation of DNA methylation in order to
develop more effective therapeutic strategies. Although the tenet that DNA methylation is
involved in gene silencing remains valid, we still know surprisingly little of the
mechanisms by which DNA methylation pattern become disrupted in cancer and how
DNA methylation changes influence the overall chromatin landscape and cellular
phenotype. The development of genome-wide sequencing technologies in recent years
has allowed the cancer epigenetics field to move from focusing on individual loci to
examine the global organization of the human epigenome, and in the process, has
revealed an unprecedented view of the integrated epigenetic landscape.
My graduate dissertation employs two general strategies to study DNA
methylation changes in normal and diseased cells. First, I examine the role of the local
tissue environment on the maintenance of a normal human methylome using an in vivo
human ileal neobladder model. Exposures to external environmental cues have previously
been shown to contribute to DNA methylation alterations during early development as
well as in disease development. The reliance of these studies on large epidemiological
cohorts, however, limits our ability to determine the direct causal relationship between
the environment and the epigenome. In chapter 2, I introduce the use of an isogenic in
vivo human model to circumvent the limitation of epidemiological study in exploring the
interactions between signals from the environment and changes in global DNA
methylation pattern.
20
The remainder of my thesis focuses on how changes in DNA methylation
interacts with the other components of epigenetic mechanisms and ultimately, the
regulation of gene expression potential, a work done in close collaboration with Dr. Terry
Kelly as well as bioinformaticians Dr. Ben Berman and Yaping Liu. Chapter 3 describes
the development of NOMe-seq, a footprinting assay capable of simultaneously detecting
DNA methylation and nucleosome occupancy, for both locus-specific and genome-wide
analyses. In chapters 4 and 5, I apply the NOMe-seq assay to specifically address how the
perturbation of global DNA methylation pattern may influence the organization of a
cancer epigenome. I use the HCT116 colon cancer cell line and study the global loss of
DNA methylation using a pharmacological approach as well as a genetic derivative of
HCT116 cells, called DKO1, which lacks DNMT3B and DNMT1 activity (Rhee et al.,
2002). Based on findings in this study, I subsequently perform a pilot study to apply
NOMe-seq on primary human tissues, particularly focusing on how DNA methylation
and nucleosome positioning may be altered in primary colon tumors in comparison to the
adjacent normal tissue. This experiment will be addressed in chapter 6. Finally, in chapter
7, I briefly summarize my work and reflect on how my findings have advanced the cancer
epigenetics field.

21
CHAPTER 2
REPROGRAMMING OF THE HUMAN INTESTINAL EPIGENOME BY
SURGICAL TISSUE TRANSPOSITION

This chapter is modified from a manuscript previously published in the peer-reviewed
journal, Genome Research.

INTRODUCTION

Epigenetic mechanisms, including DNA methylation, histone modifications and
nucleosome positioning, work cooperatively to regulate differential gene expression and
act as regulators of cellular phenotype. DNA methylation, the most-studied epigenetic
mechanism, is the covalent addition of a methyl group to cytosines existing in the CpG
dinucleotide context. This epigenetic modification is heritable through somatic cell
division and has long been associated with transcriptional silencing when located at
promoters. However, the emergence of genome-wide studies suggests that the role of
DNA methylation may be dependent on its genomic-context (Bird, 2002; Jones, 2012). In
mammals, the most dramatic changes in global DNA methylation occur during
embryonic development in a process known as epigenetic reprogramming (Cantone and
Fisher, 2013). The methylome of fully differentiated cells, however, is unique for each
cell type and remarkably stable whereby aberrant alterations are often associated with
diseases such as cancer (Baylin and Jones, 2011; Halley-Stott and Gurdon, 2013).
Various local environmental factors are known to influence epigenetic
programming during mammalian development and contribute to disease susceptibility
(Feil and Fraga, 2011; Gordon et al., 2012; Jirtle and Skinner, 2007; Walker and Ho,
2012; Zhu et al., 2013). The role of local tissue environment in maintaining normal DNA
22
methylation patterns and subsequently cellular phenotype of differentiated human cells,
however, has remained elusive as studies in human subjects have mostly been limited to
in vitro models and/or epidemiological observations where the extent of environmental
exposure is often not precisely known (Cortessis et al., 2012; Feil and Fraga, 2011; Mill
and Heijmans, 2013; Rakyan et al., 2011).
Here, we introduce a novel in vivo isogenic human neobladder model to examine
the interaction between local tissue environment and the epigenome of normal,
differentiated cells. Construction of the orthotopic ileal neobladder is part of an existing
standard of care for bladder cancer patients whose bladders have been completely
removed. During the surgery, a 60-70cm segment of the patient’s small intestine or ileum
is reshaped into a bladder-like reservoir. The ureters are connected to this reservoir,
which in turn is reconnected to the urethra to allow patients to urinate normally (Freeman
et al., 1996; Hautmann et al., 2007; Stein et al., 2005). The autologous transposition of
the small intestine to form a neobladder marks the precise time point when the tissue
becomes exposed to a foreign bladder environment in which low intraluminal pH and
various urinary solutes triggers pathophysiological changes in the otherwise normal
intestinal mucosa (Aragona et al., 1998; Di Tonno et al., 2012; Gatti et al., 1999;
Philipson et al., 1987). The neobladder, however, maintains the blood supply of the
original tissue so that the predominant change of local tissue environment is in the
content of the lumen to which the intestinal cells are exposed. Subsequently, intestinal
epithelial cells become exfoliated into the urine, allowing us to non-invasively collect
them at various time-points following surgery for global DNA methylation analyses.
Using this model, we have the advantage of knowing the exact time and length of altered
23
environmental exposure, and as such, we are able to directly quantify the effect of the
local tissue environment on DNA methylation, and show a dynamic interaction in which
it directly shapes the human epigenome, particularly the enhancer regions, in a non-
diseased state.

24
MATERIALS AND METHODS
Patient Sample Collection and Ethics Statement
All tissue and urine samples were collected from patients above the age of 50
years old in accordance to institutional guidelines. For neobladder patients from whom
we were able to collect at least one neobladder urine follow-up, we also collected their
matched peripheral blood and/or small-intestine sample before surgery. Normal bladder
or urothelium sample was collected from patients undergoing radical prostatectomy for
prostate cancer and without indications of bladder cancer or related diseases.

Sample Preparation and Loci Specific Methylation Analysis
All tissue samples were examined by pathologists and processed immediately
upon collection. Fresh, uncultured small intestine tissue was cut longitudinally and
washed with ice cold PBS twice. The tissue was cleaned from excess fat and residual
fecal matter, dissected into 1-cm cubes and subsequently incubated in HBSS/30mM
EDTA solution at 37°C for 30 minutes with gentle rotation to detach the epithelial layers.
This EDTA treatment mimicked the physiological shedding of intestinal epithelial cells
in the neobladder, allowing us to collect a pure population of intestinal epithelial cells
and measure the same cell population pre- and post-surgery. Normal urothelium control
was isolated by microdissection of hematoxylin and eosin (H&E)-stained tissue. Urine
sediments from the neobladder patients were centrifuged for 10 minutes at 1,300rpm and
DNA was extracted from the cell pellet using standard phenol-choloroform method.
To detect DNA methylation level, 1 µg of genomic DNA was bisulfite converted
using the Zymo EZ Methylation kit according to the manufacturer’s instructions and used
25
for pyrosequencing, bisulfite sequencing and/or DNA methylation array. Briefly,
bisulfite-converted DNA was PCR-amplified using a biotin-labeled 3’ primer and
purified using Streptavidin-Sepharose beads. Pyrosequencing was performed using the
PSQ HS96 System and DNA methylation level was expressed for each locus as
percentage of methylated cytosines over the sum of methylated and unmethylated
cytosines. For bisulfite sequencing, bisulfite PCR fragments were cloned using the TOPO
TA cloning kit and individual colonies were screened for the insert and sequenced. All
primers used in this study are listed in Table 2.1.

Global DNA Methylation Analysis
DNA methylation of more than 450,000 CpG sites was measured using the
Infinium HumanMethylation450 BeadChip according to manufacturer’s protocol
(Bibikova et al., 2011; Sandoval et al., 2011). Details on patient samples used for the
analysis can be found in Table 2.2. Preprocessing of the array was performed using the
methylumi package available through Bioconductor and as previously described (Sean
Davis, 2012; Triche et al., 2013). We excluded samples that have a mean detection p-
value > 0.05 across all probes and performed background correction and dye-bias
equalization. Methylation level of each CpG locus is expressed in terms of beta value 0-
1.0 signifying percent methylation of 0% to 100%. At the probe level, we excluded
probes on the X and Y-chromosomes as well as probes that contain SNPs and/or
repetitive elements on the target CpG to avoid bias in our analysis. We also removed
probes containing missing data in one or more samples. After filtering, 410,808 probes
and 34 samples including 15 neobladder samples are available for downstream analyses.
26
All statistical and clustering analyses were performed on R using various Bioconductor
packages.

Description of samples used in global analysis.
To assess sample quality, we assayed 113 neobladder samples by pyrosequencing.
Of the 33 samples that passed this initial quality control, we had enough materials to
generate HM450 data for 26 samples, 3 of which were immediately excluded during
preprocessing of the array for having mean detection p-value > 0.05 across all probes. We
excluded 8 more neobladder samples because the average beta value of 2995 blood-
specifically unmethylated probes of each sample is outside of one standard deviation of
the average beta value of normal small intestine. In total, our global DNA methylation
analysis included 15 neobladder samples, 8 normal small intestines, 6 white blood cells
and 5 bladders. The samples we used in this study are listed in Table 2.2.

Analysis of variable probes
The top 5% most variable probes were determined by calculating variance of each
probe across all samples to identify probes that gain and lose methylation over time. We
used the lm and cor function on R to fit each probes in a linear regression model and
calculated their correlation with time post-surgery. Only probes with p-value<0.05 after
Benjamini-Hochberg correction were included when calculating the rate of methylation
changes. In calculating the rate of demethylation in the cluster of non-tissue specific
probes, we removed probes that showed high variation in beta value among the normal
27
tissues to increase confidence that the pattern of demethylation we observed was not due
to noises from variable methylation of the probes themselves.

Modeling the effects of sample heterogeneity
Contamination of 10%, 30% and 50% from white blood cells in the neobladder
samples was simulated to predict effects of varying levels of cellular heterogeneity. The
mock data was generated for the cluster of intestine-specific probes based on additive
projection to the beta value of normal small intestine, an accepted approach to model cell
count estimation (Houseman et al., 2012). Three-dimensional non-metric
multidimensional scaling (MDS) analyses were performed for each predicted level of
heterogeneity and the significance was expressed in terms of Kruskal stress value.

Chromatin State Discovery and Segmentation
Chromatin state calling and segmentation was performed as previously described
using publically available ChIP-seq data from the Roadmap Epigenomics Project (Ernst
and Kellis, 2012; Ernst et al., 2011). Briefly, Hidden-Markov Model (HMM) was applied
to the following histone marks for small intestine to generate a 20-state model:
H3K27Ac, H3K4me3, H3K36me3, H3K4me1, H327me3, H3K9me3 and Input (Series#
GSE16256, The Human Epigenome Atlas Release 8). We used this model to annotate
and define 10 global chromatin states. Annotations of genomic regions were obtained
through the UCSC Genome Browser.

28
Table 2.1. Primer List
Locus Sequence
Bisulfite Sequencing
VTRNA2-1
Forward
Reverse

AGTATAGAGATGGATAGATAGAA
ACCTAACAAAAAATAAAACCAC
MUC13
Forward
Reverse

TTTTGATTTATTTTGGGTTTTGAGTT
AAAAAAAAAATAATTTCCCTTCCTAA
Pyrosequencing
KLHL6
Forward
Reverse
Pyrosequencing Probe

AAGGGTTTTTGGTATTTTTTATAGATGAGTTT
AAATATCCACACACAAAATAACATCTATCAAAAA
GAAAAGGTTAAATTTGA
LAPTM5
Forward
Reverse
Pyrosequencing Probe

TGAGGAGGGTAGTTAGTAGTTTTTT
CACACTCACCACATAATAAATAACCAAAA
GTTGTTGTTTTAATGTT
LPXN
Forward
Reverse
Pyrosequencing Probe

AATTATGATTGTTTTGGGGAATTATGGGTTTAT
ACCCTTAATCTCCAATAACCTCCAAAAAA
AGAGTTTGGTGTTAAT
REG4
Forward
Reverse
Pyrosequencing Probe

TTGTAGATAAGATTTTTATGGATGGAT
AAAAAAATTACCTATTTAACAACCAAAACTCTAAAA
GATATAAAAGTTTTAGAAA
GATA4
Forward
Reverse
Pyrosequencing Probe

GTTTGGATAAAATAAAGGTTTTTGTTTTT
AAATAATTCATCCCAAAACTTTCAAAAC
ATAAAATAAAGGTTTTTGTTTTT
MX2
Forward
Reverse
Pyrosequencing Probe

TATGAAGTTAAGTAGGATAGTTT
CCACCTATATAACTCTAAAATATAA
TGTATTTGTGAGGTT
TJP2
Forward
Reverse
Pyrosequencing Probe

GGTTTTTAGATAGGATTTAAAATTTTGAG
ACTATCACCTACTTCCTTAAAACC
GTTTTTTAGGTAGT
IRAK3
Forward
Reverse
Pyrosequencing Probe

GGAGTTTTGAGTTTTGGGTTTT
GGTAATTTTTAGGTTTGGTAGG
AGGTGTAGAAGGGG

29
Table 2.2: Neobladder Sample Information

Patient
# Tissue Methylation Level(%)
Pass
450K
QC?
Global
Purity
Filter
Months
Post
Surgery Age
KLHL6 LAPTM5 LPXN
9000 SI 87.92 86.8 69.77 Yes 0 70
Neobladder 82.44 86.15 48.49 No N/A 0 71
9037 SI 87.01 95.02 58.18 Yes 0 74
Neobladder 91.78 79.39 67.33 No N/A 0 75
9188 Blood 14.63 44.94 4.92 Yes 0 67
SI 90.65 94.9 65.73 Yes 0 67
Neobladder 91.41 95.39 60.82 No N/A 0 68
9295 Blood 6.52 13.27 4.43 Yes 0 70
SI 79.14 88.55 60.47 Yes Yes 0 70
Neobladder 73.01 90.77 63.21 Yes 9 71
9307 Blood 5.49 14.15 4.39 Yes 0 77
SI 73.52 88.45 60.44 Yes 0 77
Neobladder 82.93 89.18 67.82 Yes Yes 4 78
9333 Blood 5.14 3.47 2.13 Yes 0 79
SI 77.28 85.19 53.41 Yes 0 79
Neobladder 72.18 81.44 48.73 Yes Yes 12 80
9339 Blood 6.99 2.54 4.23 Yes 0 68
SI 84.76 85.4 66.71 Yes 0 68
Neobladder 94.92 67.21 63.25 Yes Yes 12 69
9413 Blood 9.6 10.85 3.89 Yes 0 69
SI 83.56 86.47 70.01 Yes 0 69
Neobladder 93.43 75.41 64.56 Yes Yes 9 70
4489 Neobladder 1 82.17 66.72 Yes No 78 69
Neobladder 2 71.96 60.37 Yes No 84 70
5789 Neobladder 1 76.97 75.55 65.34 Yes Yes 41 73
Neobladder 2 75.36 34.76 64.81 Yes No 52.5 74
6481 Neobladder 1 93.76 76.81 59.98 Yes Yes 36 73
Neobladder 2 85.11 38.47 61.37 Yes Yes 48.5 74
6534 Neobladder 1 81.67 61.61 82.11 Yes Yes 11 80
Neobladder 2 75.86 62.62 69.44 Yes Yes 24 81
6861 Neobladder 1 81.32 93.12 N/A Yes Yes 44 62
Neobladder 2 70.54 54.56 70.39 Yes No 47 62
7065 Neobladder 1 90.09 97.31 N/A Yes Yes 26 64
Neobladder 2 79.95 70.31 N/A Yes No 32 64
7277 Neobladder 1 77.23 72.6 N/A Yes No 27 69
Neobladder 2 82.91 83.05 N/A Yes Yes 41 70
30
7464 Neobladder 1 91.56 92.09 87.82 Yes Yes 20.5 64
Neobladder 2 79.9 68.11 79.32 Yes No 25.5 65
7905 Neobladder 1 85.12 94.3 72.11 Yes Yes 9 67
Neobladder 2 72.11 61.25 68.37 Yes No 13 68
5740 Bladder 54.49 55.76 N/A Yes N/A N/A N/A
5757 Bladder 62.695 64.61 N/A Yes N/A N/A N/A
5776 Bladder 50.82 54.9 N/A Yes N/A N/A N/A
5879 Bladder 54.495 96.47 N/A Yes N/A N/A N/A
5935 Bladder 61.77 70.33 N/A Yes N/A N/A N/A

31
RESULTS
Tissue-specific DNA methylation as markers of sample purity
Unlike urine sediments from individuals with normal urothelium, the neobladder
urine specimens are highly cellular, often thick with mucus (Figure 2.1a) and may be
contaminated with white blood cells which are present in elevated numbers due to
inflammation that occurs in patients in the early stage post-surgery (Aragona et al., 1998;
Philipson et al., 1987). Mucus production and white blood cells count decrease
substantially with time, but it remains important to exclude urine samples with substantial
contaminations in order to measure the right population of cells in analyzing DNA
methylation changes in the neobladder. In our study, we used tissue-specific DNA
methylation loci as surrogate markers to confirm the identity of cell population collected
from the neobladder urine sediments (Houseman et al., 2012; Reinius et al., 2012). We
have previously generated Infinium HumanMethylation27 data for normal blood, small
intestine and bladder from which we identified inflammatory cell-specific probes which
are specifically unmethylated in white blood cells and methylated in normal small
intestine and urothelium. We chose the markers KLHL6, LAPTM5 and LPXN which have
high sensitivity and specificity to distinguish between normal white blood cells and small
intestine, and subsequently assayed DNA methylation levels of these loci in the
neobladder using pyrosequencing. Low methylation levels at these loci in the neobladder
samples indicate the enrichment of inflammatory cells in the urine sediments and these
samples were subsequently excluded from further analysis (Figure 2.1b).
For neobladder samples that were enriched for intestinal cells, we performed an
additional filter at the global level. We expanded the panel of three locus-specific
32
markers using the Infinium HumanMethylation450 (HM450) platform, from which we
identified a cluster of 2995 probes that are specifically unmethylated in blood and
methylated in the small intestine and bladder (Figure 2.2a). We calculated the average
beta value of this cluster of probes for each samples including the neobladder.
Neobladder samples whose average beta value is above one standard deviation of the
average small-intestine beta value were excluded from downstream analysis to reduce
noise and confounders from residual heterogeneity previously undetected by
pyrosequencing (Figure 2.2b).

33
Patients
* *
% Methylation
* * * * * *
Blood
Small Intestine
Neobladder
9000 9037
100
90
80
70
60
50
40
30
20
10
0
9107 9119 9165 9172 9188 9190 9217 9223 9232 9295 9307 9333 9339 9413 9421 9449 9464 9329
A
B
Figure 2.1 Inflammatory cell-specific DNA methylation markers as surrogate
markers of neobladder sample purity. (A) Urine sediment collected from a neobladder
patient is highly cellular and often thick with mucus due to the presence of mucus-
producing goblet cells. Inflammatory cells may also contaminate the urine samples due to
increased inflammation in patients immediately after surgery. Mucus production and
white blood cell count, however, decrease over time. (B) Methylation level of KLHL6, a
blood-specific gene, is measured in all matched samples using pyrosequencing. KLHL6 is
hypomethylated in all blood samples (7.57±2.45%) and highly methylated in the normal
small intestine (83.09±68.01%). Hypermethylation of this marker in the neobladder
indicates enrichment for intestinal epithelial cells, while low or intermediate methylation
indicates a significant presence of an inflammatory cell population. (*) Neobladder
samples that passed the quality-control threshold, which is a methylation level within
10% of the matched small intestine. For downstream analysis, HM450 data were
generated for neobladder samples that are enriched for intestinal cells as indicated by the
high methylation level in at least two of the three inflammatory cell markers.
34
Island
Shore
Shelf
Other
Cluster #
1
2
3
4
5
6
7
Methylation
0 --------->1
Blood
Small Intestine
Bladder
Neobladder
A
B
9295
9333
9413
9339
9307
9188
1.0
0.80
0.40
0.20
0
Beta Value
0.60
9295
9333
9413
9339
9307
9188
9000
9037
9295
9333
9413
9339
9307
9295
9333
9413
9339
9307
7905
6534
6481
7277
7065
6861
7464
6534
5789
6481
Patients ID
Blood
Neobladder
Small Intestine
Bladder
Figure 2.2 Quality Control of Neobladder Samples. (A) Secondary purity filter was
performed on global level for all neobladder samples with HM450 data. Briefly,
variance was calculated for normal blood (n=6), small intestine (n=8) and bladder (n=5)
and the top 5% most variable probes (n=20,541) were hierarchically clustered on
Euclidean distance with k-means=7 (left panel). Cluster 4 consists of probes that are
specifically unmethylated in the blood cells and methylated in the normal small
intestine and bladder (n=2995). The neobladder samples show high methylation level
for all the probes in this cluster, indicating the enrichment of intestinal epithelial cells
for each sample. (B) The average and distribution of methylation in this blood-
specifically unmethylated cluster is shown in the boxplot. The average methylation of
all the probes in each neobladder sample is within one standard deviation of the average
methylation of the small intestine (0.86±0.08). All neobladder samples with average
methylation larger than one standard deviation of the average small intestine
methylation are removed and excluded from further analysis as they may contain
significant blood cells contamination.
35
DNA methylation changes occur in the neobladder
Analysis of matched blood, small intestine and neobladder samples collected from
5 patients within the first year post-surgery show a high degree of correlation between the
neobladder with their respective matched small intestine, further confirming the
enrichment of intestinal epithelial cells in the neobladder (Figure 2.3a). We also
observed a low level increase in methylation for the majority of probes that are
differentially methylated in the neobladder one year post-surgery (Figure 2.3b). We
confirmed this change in methylation by performing bisulfite sequencing of the
individual-specific VTRNA2-1 locus on the matched samples (Figure 2.a-b). This locus
may be monoallelically methylated (Figure 2.4a) or unmethylated (Figure 2.4b) across
various tissues of different individuals, but never fully methylated (Treppendahl et al.,
2012a). The increase of methylation in this locus thus appears to be an effect of the
drastic changes in the local tissue environment of the intestinal epithelial cells in the
neobladder, rather than an artifact of sample heterogeneity.

36
0
1
2
3
4
Counts(x1000)
0 0.1 0.25 0.25 0.1
Lose Methylation Gain Methylation
Delta Beta
9295
9307
9333
9339
9413
9295
9307
9333
9339
9413
9295
9307
9333
9339
9413
9413
9295
9307
9333
9339
9413
9295
9307
9333
9339
9413
9295
9307
9333
9413
1.0
0.95
0.90
0.85
Blood Neobladder Small Intestine
Figure 2.3 Neobladder samples show limited DNA methylation changes within the
first year of surgery. A) HM450 data was generated for matched normal blood, small
intestine and neobladder samples collected from 5 patients. Blood and small intestine
samples were collected at the time of surgery and neobladder samples were collected
within the first year post-surgery. Pearson correlation was calculated for these samples
and shown as a matrix plot. A higher degree of correlation is seen within specific cell
types, blood and small intestine, than within individuals (i.e. small intestine from
different individuals are more similar to each other than to blood from the same
individual). Neobladder samples are showing high correlation with the small intestine,
further confirming that the neobladder samples are enriched for intestinal epithelial
cells. B) F-test was performed to identify differentially methylated probes between
matched normal small intestine and neobladder within 1 year post-surgery (n=12,242,
q-value<0.05). The histogram shows the distribution of methylation changes in the
neobladder where most of the probes gain methylation in minimal level (mean =0.074,
range = -0.39-0.41).
A
B
37
445bp
Small
Intestine
Blood
Neobladder
9413 9295
445bp
A
B
A
B
Figure 2.4 Neobladder samples show dramatic increase of DNA methylation at the
VTRNA2-1 Locus. (A) Bisulfite sequencing of VTRNA2-1 shows an increase of
methylation in the neobladder within the first year of surgery. VTRNA2-1, also known
as nc886 or mir886, is a non-tissue specific non-coding RNA which may be
monoallelically methylated or (D) unmethylated in different individuals. Diamond
shape (u) indicates the position of Infinium HumanMethylation450 probes.

38
Time-dependent de novo DNA methylation of small intestine specific probes
After confirming that DNA methylation changes may occur in the neobladder, we
expanded our analysis to include samples collected up to 5 years post-surgery which may
or may not be matched with normal tissue controls. Global analysis of all probes show
that samples across different individuals cluster together based on their cell type,
suggesting that variation between tissues is greater than variation between individuals
(Figure 2.5a). Although the neobladder samples cluster on the same side as the normal
small intestines, they appear to form distinct groups (Figure 2.5b) and exhibit a time-
dependent effect in which neobladder samples from earlier time points cluster closer to
the normal small intestine compared to samples from later time points.
To further characterize the methylation changes in the neobladder, we calculated
the variance across all samples and hierarchically clustered the most variable probes. We
found a significant increased of methylation in the neobladder in the cluster of probes that
were specifically unmethylated in the small intestine. Most of these altered intestine-
specific probes are located in non-CpG islands, including CpG shores and shelves,
consistent with the more dynamic and tissue-specific behavior of non-CpG island
methylation (Jones, 2012). Interestingly, a smaller subset of intestine-specifically
unmethylated probes did not show any changes and we also did not detect significant
reciprocal demethylation of probes that were specifically methylated in the small
intestine, many of which are located in CpG-islands.
We fitted each probe in the intestine-specific cluster into a linear regression model
and found 4162 probes to have a significant linear relationship with time post-surgery,
despite two individuals exhibiting minimal or no increase in DNA methylation at 16 and
39
36 months post-surgery respectively (Figure 2.6a). We further observed an increased
dissimilarity in the neobladder samples in which each neobladder sample appear to gain
its own unique DNA methylation pattern, distinct from the normal small intestine and
other neobladders (Figure 2.7a), which cannot be attributed to cellular heterogeneity in
the neobladder samples (Figure 2.7b-d). Interestingly, despite the gain of methylation, the
methylation patterns of the neobladders do not a show a shift toward normal bladders
(Figure 2.7a).
As a cluster, the gain of methylation in probes specifically unmethylated in the
intestine occurred at the average rate of 0.41% per month, or roughly 5% per year post-
surgery (Figure 2.6b). This association with time holds true for many of the probes even
when excluding normal small intestines from the linear regression model. We confirmed
the overall trend of increased methylation in the neobladder by performing
pyrosequencing of intestine specific, unmethylated loci (Figure 2.8a-b). Bisulfite
sequencing further revealed that the gain of methylation in the neobladder is not limited
to specific CpG-sites, but rather, occurred in widespread manner in broader genomic
regions (Figure 2.4, Figure 2.8c).

40
Bladder Neobladder Blood Small Intestine
A
B
Figure 2.5 DNA methylation patterns are cell-type specific. (A) Dendrogram of
hierarchical clustering performed on single Euclidean distance for normal blood,
small intestine, bladder and neobladder samples show the relationship between each
tissue type. Neobladder samples cluster closer to the small intestine than to control
bladder and blood. (B) Nonmetric multidimensional scaling (MDS) plot using all
HM450 probes for all samples shows distinct separation between normal small
intestine, blood and bladder (Kruskal’s stress = 0.0334). Each axis represents one
dimension. Although the neobladder samples cluster closer to the small intestine
control, many have diverged away, suggesting alterations in DNA methylation of
intestinal epithelial cells post-surgery.

41
Methylation
0 ------------>1
Island
Shore
Shelf
Other
0 48.5 5 20.5
Months Post Surgery
A
Blood
Small Intestine
Bladder
Neobladder
Months Post-Surgery
Average Beta Value
y = 0.004142x±0.001007+0.228762±0.022369
r = 0.7012197
p-value = 1.93x10
-4
0 10 20 30 40 50
0
0.25
0.50
0.75
1
B
Figure 2.6 DNA methylation changes in the neobladder is a dynamic process. (A) A
heatmap showing intestine-specifically unmethylated probes that have significant
(Benjamini-Hochberg adjusted p-value <0.05) linear regression value (n = 4162).
Neobladder samples are arranged from the earliest to the latest time points post-surgery
(0-48.5 months), and show increased methylation over time. (B) The rate of de novo
methylation in the neobladder was measured for individual probes and as an average of
the cluster shown in (A) and Pearson’s correlation (r) was calculated to show that
changes in DNA methylation in the neobladder is time-dependent. Time 0 refers to
normal small intestine tissues collected before being transplanted into a bladder
environment.
42
Bladder Neobladder Blood Small Intestine Mock Contamination
A
B C D
Figure 2.7 Methylation increase seen in the neobladder is not due to
contamination of inflammatory cells. (A) Nonmetric MDS plot using the cluster of
probes that gain methylation in the neobladder illustrates the distinctiveness of
neobladder samples from each other in comparison to normal tissue controls that
cluster tightly together based on their respective cell type (Kruskal’s stress=0.0098).
(B) Nonmetric MDS plot shown to include simulated data of neobladder samples
containing 10%, (C) 30% and (D) 50% of inflammatory or blood cells contamination
(green spheres). Increased level of inflammatory cells in the neobladder would cause
the samples to shift orthogonally toward the normal blood group whereas real
neobladder samples maintain their heterogeneous distribution in the three-
dimensional space (Kruskal stress = 0.012).
43
REG4
SI(0)
<12
13-24
25-48
0
10
20
30
40
50
Time Post-Surgery
%Methylation
Months Post-Surgery
% Methylation
*
*
*
Months Post-Surgery
% Methylation
MX2
SI(0)
<12 mths
13-24mths
25-48mths
0
20
40
60
80
Months Post Surgery
%Methylation
*
*
*
SI (0) <12 13-24 25-48
SI (0) <12 13-24 25-48
A B
225bp
Small
Intestine
Neobladder
C
Figure 2.8 Validation of time-dependent increased of DNA methylation in the
neobladders. (A) Pyrosequencing of REG4, an intestine specific gene, and (B) TJP2
validate the global finding that there is a significant time dependent methylation
increase of intestine-specifically unmethylated probes in the neobladder. T-test was
performed and p-value <0.05 was indicated by *. (C) Bisulfite sequencing of MUC13
locus of matched small intestine and neobladder sample Diamond shape (u) indicates
the position of Infinium HumanMethylation450 probes.
44
Time-dependent DNA demethylation of non-intestine specific probes
DNA methylation changes in the neobladder are not limited to tissue-specific
probes. We identified a subset of non-CpG island probes that were methylated across all
the normal tissues and exhibited a trend of demethylation in the neobladder. We fitted a
linear regression model for each probe within the cluster and identified 748 probes as
having significant time-dependent demethylation (Figure 2.9a). We also examined the
similarity between the normal tissue controls and the neobladders using this group of
demethylated probes and found that the methylomes of the neobladders are shifted in
heterogeneous fashion away from the normal small intestine as well as blood and bladder
controls (Figure 2.9b). Unlike the normal tissues that exhibit high similarity within each
tissue type, the neobladder exhibits higher variation in methylation between each sample
in these non-tissue specific probes.
As a cluster, loss of methylation occurs at the rate of 0.37% per month, or roughly
4.4% per year (Figure 2.9c). We validated this global trend of demethylation at locus-
specific level by pyrosequencing (Figure 2.9d). The trend of demethylation of non-
intestine specific probes is not as dramatic as the increased methylation of intestine-
specific probes; however, the rate of methylation changes in the neobladder remains
much more dramatic than the rate of methylation changes previously attributed to tissue-
specific aging (Ahuja et al., 1998; Christensen et al., 2009; Hannum et al., 2013; Heyn et
al., 2012). We measured the average rate of methylation changes in age-associated probes
in normal colon using publically available TCGA data and found that it is roughly 15-
fold lower compared to the rate of methylation changes in the neobladder (Figure 2.10).
Moreover, we found very minimal overlap between probes that are associated with aging
45
and probes that are changing in the neobladder, further suggesting that changes in the
local tissue environment, rather than aging, directly alter the methylome of normal
differentiated cells.

46
Methylation
0 ------------>1
Blood
Small Intestine
Bladder
Neobladder Island
Shore
Shelf
Other
0 48.5 5 20.5
Months Post Surgery
A
B
Bladder Neobladder
Blood Small Intestine
Months Post-Surgery
Average Beta Value
0
0.25
0.50
0.75
1
10 20 30 40 50
y = -0.003656x±0.001137+0.844896±0.025264
r = -0.5743659
p-value = 4.15x10
-3
TJP2
SI(0)
<12
13-24
25-48
0
20
40
60
80
100
Months Post-Surgery
% Methylation
*
*
*
Figure 2.9 Loss of DNA methylation in the neobladder occurs dynamically on non-
tissue specific CpG sites. (a) A heatmap showing a subset of probes (n=748) that are
methylated across all normal tissue and exhibit a significant trend of demethylation over
time in the neobladder (Benjamini-Hochberg adjusted p-value <0.05). (b) Non-metric
MDS using demethylated probes show increased variation between the neobladder
samples. (c) Linear regression analysis shows that demethylation in the neobladder is
also time dependent and occurring at the rate comparable to de novo methylation in
intestine-specific probes. (d) Demethylation of TJP2 is validated using pyrosequencing.
T-test was performed and p-value <0.05 was indicated by *.
C D
47
Years
Average Beta Value
0
0.25
0.50
0.75
1
40 60 70 80 90 50
y = 0.0034249x±0.0003689+0.0752537±0.0259873
r = 0.8398706
p-value = 4.35x10
-11
Years
Average Beta Value
0
0.25
0.50
0.75
1
40 60 70 80 90 50
y = -0.0027627x±0.0004527+0.8357352±0.0318880
r = -0.713107
p-value = 5.06x10
-7
Figure 2.10 The rate of methylation changes in the neobladder is significantly
higher than the rate of methylation changes attributed to aging. (A) The average
rate of gain or (B) loss of methylation due to aging is measured in normal colonic
epithelium for HM450 probes showing significant association with age.
A
B
48
DNA methylation changes in the neobladder is a surrogate for detecting alterations
in distinct chromatin states
DNA methylation works in concert with other epigenetic mechanisms to define
the human epigenome. To place DNA methylation changes in the context of the broader
epigenetic landscape, we used publically available ChIP-seq data of active and repressive
histone marks to characterize the chromatin configurations present in the small intestine
using the chromHMM model (Ernst et al., 2011). We generated a 20-chromatin-state
emission model and calculated the enrichment and segmentation of each chromatin state
in various genomic elements in the small intestine to further annotate each chromatin
state and subsequently collapsed the model into a 10-state model covering promoter,
enhancer, transcribed and repressed regions (Figure 2.11a-b).
We then correlated each methylation probe with the underlying chromatin states
of the small intestine and calculated the distribution of each chromatin state (Figure
2.11c). We found that in the cluster of probes that gain methylation, there was a
significant enrichment of the enhancer state where more than 50% of the probes fall in
the active and weak enhancer regions with the remaining probes in the cluster distributed
across the heterochromatic/quiescent state, transcribed, the polycomb repressed state and
to a lesser extent, the promoter regions. (Figure 2.12a). The enrichment of the enhancer
regions is specific to the small intestine as the same group of probes exhibits different
chromatin state distribution in normal bladder. These enhancer regions also have a low
methylation average in the normal small intestine (0.15±0.071 to 0.26±0.081), consistent
with what is known about enhancers having low-methylated regions or LMRs (Stadler et
al., 2011). We performed gene ontology analysis of the closest gene to these enhancers
49
using David Ontology and found that the enhancer category is enriched for genes that are
specifically expressed in the small intestine and may regulate biological functions which
include metabolism and actin cytoskeleton organization (Table 2.3).
In the cluster of probes that become demethylated, we observed a significant
enrichment of transcribed regions where more than 40% of the probes fall in this category
(Figure 2.12b). Unlike the enhancers, these regions are not specific to the intestine as the
majority of these probes also fall in the transcribed region in the normal bladder control.
Expectedly, this group of probes is significantly depleted of active promoter and
enhancer regions as DNA methylation is anti-correlated with active chromatin states
(Thurman et al., 2012). We further measured the rate of methylation change for each
chromatin state to determine whether a particular chromatin configuration drove the
average rate of change in each cluster and found that it was not the case. This suggests
that the epigenetic changes in the neobladder may occur in a tightly-controlled manner on
distinct genomic regions, leading to the loss of small intestine characteristic in the
neobladder.

50
Weak/Poised Enhancer
Weak/Poised Enhancer
Poised/Inactive Promoter
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
H3K27me3
H3K9me3
H3K36me3
H4K20me1
H3K4me1
H3K27Ac
H3K9Ac
H3K4me3
DHS
Transcription
Transcription
Transcription
Transcription
Transcription
Euchromatin
Polycomb Repressed
Active/Strong Enhancer
Active/StrongEnhancer
Active Promoter
Polycomb Repressed
Heterochromatin/Quiescent
Heterochromatin/Quiescent
Repetitive/CNV
Weak Promoter
Active Promoter
Active Promoter
State
Normalized
Probability
0 ------------------>1
A
Figure 2.11 Characterizing Functional Chromatin States of Normal Small
Intestine. (A) Chromatin states for the normal small intestine is defined based on a 20-
states chromHMM model using publically available data and underlying chromatin
states were determined for each CpG probe. Color scale measures the normalized
probability of finding a specific histone mark in a particular chromatin state, and each
state is annotated based on the levels of enrichment of each histone marks. (B)
Functional enrichment of each chromatin state in the normal small intestine based on
various genomic elements. (C) Screenshot of UCSC genome browser showing the
segmentation of chromatin state in the small intestine where a segment of ~3kb of active
enhancer state (yellow) is flanked by heterochromatic or quiescent state (brown). Nested
within the enhancer state, we found three methylation probes that are significantly
variable in the neobladder.
C
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
State
% Genome
Nuclear Lamin
CpG Island
Refseq Exon
Refseq Gene
Refseq TES
Refseq TSS
1
Refseq TSS 2kb
B
51
Data 1
Active Promoter
Weak Promoter
Poised/Inactive Promoter
Active/Strong Enhancer
Poised/Weak Enhancer
Gene Body
Polycomb Repressed
Euchromatin
Heterochromatin/Quiescent
Repetitive/CNV
0.0
0.1
0.2
0.3
0.4
0.5
Data 2
Active Promoter
Weak Promoter
Poised/Inactive Promoter
Active/Strong Enhancer
Poised/Weak Enhancer
Transcription
Polycomb Repressed
Euchromatin
Heterochromatin/Quiescent
Repetitive/CNV
0.0
0.1
0.2
0.3
0.4
0.5
***

**

***

***

***

***

Fraction Fraction
Gain Methylation
Randomized Probes
Lose Methylation
***

***

***

*

***

A
B
Figure 2.12 Alteration of the epigenetic landscape in the neobladder occurs
predominantly in the enhancers and transcribed regions. (A) Chromatin states
were collapsed into 10-distinct states as plotted in the x-axis. A cluster of 4162 probes
that gain methylation and (B) a cluster of 748 probes that lose methylation were
compared to randomized sets of equal number of loci where 1000 trials were
performed and p-value was determined using a binomial test ( *** p<2.2x10
-16
; **
p<3.3x10
-13
). The enhancer states are significantly enriched in the cluster of probes that
gain methylation whereas the cluster of probes that lose methylation predominantly
falls in the transcribed regions.
52
Table 2.3a. GO Terms of SI-Specific Enhancers
Term P-Value
Fold
Enrichment Bonferroni
GO:0030029~actin filament-based
process 7.68E-08 2.502735418 2.61E-04
GO:0009719~response to endogenous
stimulus 1.02E-07 2.092086708 3.47E-04
GO:0030036~actin cytoskeleton
organization 1.08E-07 2.541758262 3.67E-04
GO:0009725~response to hormone
stimulus 6.64E-07 2.073922375 0.002255338
GO:0008202~steroid metabolic
process 1.27E-06 2.488280675 0.00429993
GO:0048545~response to steroid
hormone stimulus 3.21E-06 2.468285563 0.01087373
GO:0043413~biopolymer
glycosylation 6.97E-06 2.804869958 0.02343667
GO:0006486~protein amino acid
glycosylation 6.97E-06 2.804869958 0.02343667
GO:0070085~glycosylation 6.97E-06 2.804869958 0.02343667
GO:0005996~monosaccharide
metabolic process 1.08E-05 2.264111245 0.036075131
GO:0009101~glycoprotein
biosynthetic process 1.16E-05 2.544975678 0.038659214

Table 2.3b. Tissue-Expression Pattern of SI-Specific Enhancers
Term P-Value Fold Enrichment Bonferroni
Small intestine 4.37E-07 2.187142721 1.37E-04
Plasma 9.05E-07 2.245512306 2.84E-04
Intestine 5.39E-05 4.378748998 0.016786804

53
DISCUSSION

The interplay between environmental cues and epigenetic changes has been
described at length in the context of cellular development and/or pathogenesis, yet not
much is known about the role of tissue environment in maintaining the epigenome of
normal somatic cells (Feil and Fraga, 2011; Zhu et al., 2013). This is partly because most
human studies have been largely limited either to epidemiological approaches where
exposure time is often unknown, or in vitro systems, which while useful, do not always
recapitulate in vivo conditions (Christensen et al., 2009; Cortessis et al., 2012; Mill and
Heijmans, 2013; Wilson and Jones, 1983; Zhu et al., 2013). Our study using the human
ileal neobladder is the first model that attempts to characterize and quantify the dynamic
interaction between local tissue environment and the epigenetics in vivo.
The epigenome of fully differentiated cells has long been thought to be stable and
less easily perturbed by environmental cues than during mammalian development (Feil
and Fraga, 2011; Halley-Stott and Gurdon, 2013). In fact, one of the hallmarks of DNA
methylation is its faithful inheritance mediated by DNA methyltransferases in somatic
cells (Jones and Liang, 2009). Our results add another layer to the currently accepted
model of somatic inheritance of DNA methylation in that the fidelity of the process
requires precise signals from the local tissue environment. In the neobladder model, the
autologous transposition of the intestinal epithelium means the cells would receive
drastically different signals from the aseptic bladder environment compared to the
intestinal tissue environment that normally includes a complex microbiome (Philipson et
al., 1987; Wullt et al., 2004). Following the tissue transposition, we observe a
widespread alteration in the methylome of the otherwise normal intestinal epithelium
which cannot be attributed to the natural aging process alone, because at the rate of 4-5%
54
per year, the changes in the neobladder are more dramatic than the cumulative changes
previously seen in aging (Ahuja et al., 1998; Hannum et al., 2013; Heyn et al., 2012). The
changes we see in the neobladder are also more dynamic and extensive compared to the
focal hypomethylation observed during differentiation of the LGR5+ intestinal stem cells
into the specialized epithelial cells (Kaaij et al., 2013). Thus, it is unlikely that the
changes we see in the neobladder are a result of over-proliferation in the intestinal stem
cells. Taken together, our study presents the first evidence that signals from the local
tissue environment may be required to ensure epigenetic maintenance in somatic cells.
The increase of DNA methylation in CpG sites located in the enhancer regions is
particularly striking because enhancer activity is known to be anti-correlated with DNA
methylation and is highly cell-type specific (Bonn et al., 2012; Heintzman et al., 2009;
Stadler et al., 2011). We have also previously shown that repressed genes may have
permissive enhancers that can initiate cell-fate reprogramming in fibroblasts (Taberlay et
al., 2011), but what we observe in the intestinal epithelium illustrates another side in the
complex regulation of cell identity. The methylation of these intestine-specific enhancers
suggests that in a bladder environment where the intestine serves as a urinary storage
instead of as a digestive organ, the intestinal epithelium gradually loses its uniquely
small-intestine epigenetic landscape presumably because the epithelium no longer
receives the necessary signals to maintain the enhancers in a permissive state. The idea
that the epigenome mediates how genomic DNA is translated into various phenotypes is
well-established, but the loss of intestinal landscape over time in the neobladder in a
“use-it-or-lose-it” manner implies that functionality may also conversely dictate how the
epigenome is shaped.
55
The requirement for the epigenome to adapt and maintain normal cellular function
is only one element in the complex interaction between DNA methylation and the local
tissue environment. One would expect a gain of urothelium-like methylome in the
neobladder to support the new function of the intestinal epithelial cells as an alternative
bladder, but we did not see such changes. Instead, we observed demethylation of non-
intestine specific probes in transcribed region that are highly methylated across all of the
normal tissues we analyzed. Methylation of transcribed or gene body regions have been
positively correlated with expression level and as such may be used as surrogate markers
for transcription level (Hellman and Chess, 2007; Jones, 1999). Demethylation of loci
that are transcribed in both normal small intestine and bladder suggests the
downregulation of non-tissue specific genes. Taken together, these changes describe a
phenomenon in which the neobladder undergoes dynamic reprogramming to adopt its
own unique epigenetic signature as a response to the physiological stress triggered by the
drastic tissue environmental changes.
It is widely-accepted that environmental agents may reprogram the epigenome to
alter gene expression and promote tumorigenesis (Walker and Ho, 2012). The
neobladder, however, shows a surprising degree of epigenetic plasticity and adaptability,
suggesting that drastic alteration of the environment may initiate in vivo reprogramming
of differentiated cells. Distinct chromatin states appear to be gaining and losing
methylation at similar rates overall, 0.41±0.1% and 0.37±0.1% per month respectively.
This comparable rate of change suggests that alterations of the methylome in normal
somatic cells may occur in a controlled process to maintain equilibrium. This is in
contrast to aberrant changes often seen in diseases such as cancer, where genome-wide
56
hypomethylation is accompanied by focal hypermethylation of CpG-islands (Baylin and
Jones, 2011; Berman et al., 2012b).
Additional studies will be needed to better understand the mechanism of somatic
maintenance as well as the specific environmental cues required in the process. For
instance, we do not yet know the sequence of events in the loss of intestinal enhancers in
the neobladder in that it is unclear whether DNA methylation drives the loss of the active
enhancer state or is a consequence of reduced function. Furthermore, we speculate that
inflammation may also play a role in epigenetic changes seen in the neobladder although
the mechanism by which inflammation mediates or alters somatic DNA methylation
maintenance is not fully understood (Maekita et al., 2006).
Overall, our results show that the neobladder is a valuable model to study the
complex cross-talk between the local tissue environment and the epigenome in vivo. This
model allows us to precisely quantify the effect of local tissue environment on the
epigenome and elucidate the susceptibility of tissue-specific enhancers to epigenetic
reprogramming. Altogether, the widespread changes in the neobladder also illustrate the
critical role that local environmental factors play in epigenetic maintenance.

57
CHAPTER 3
DEVELOPING NOME-SEQ TO EXAMINE THE CROSSTALK BETWEEN
DNA METHYLATION AND NUCLEOSOME OCCUPANCY

The following chapter is adapted from a manuscript previously published in the peer-
reviewed journal Genome Research in which I am a co-author and an invited book
chapter which I am currently preparing as a first author.

INTRODUCTION
The assembly and organization of chromatin require a coordinated effort by
epigenetic mechanisms which include DNA methylation, nucleosome positioning and
histone variants and modifications. For many years, DNA methylation has, justifiably,
received the most attention among other epigenetic mechanisms due to its well-studied
physiological role in silencing genes and strong association with cancer phenotypes when
misregulated (Baylin and Jones, 2011). The last decade, however, has seen a growing
interest in how other components of the epigenetic landscape contribute to the
modulation of gene expression (Barski et al., 2007; Bouazoune et al., 2009; Lin et al.,
2007).
Various methodologies are available to interrogate a specific component of
epigenetic mechanisms at both locus-specific and genome-wide level. For instance, DNA
methylation may be measured by affinity-based capture assay such as MeDIP-seq,
methyl-sensitive restriction enzyme digestion or by chemical conversion-based bisulfite
sequencing and microarray (Rivera and Ren, 2013). Histone modification pattern is
determined by chromatin immunoprecipitation, which when combined with next-
generation sequencing, allow for the global mapping of various chromatin states (Barski
58
et al., 2007). Nucleosome occupancy and chromatin accessibility, on the other hand, may
be determined by techniques including, but not limited to MNase-seq, DNase-seq and
FAIRE-seq, all of which have their strengths and limitations (Boyle et al., 2008; Giresi et
al., 2007; Schones et al., 2008).
It has become increasingly clear, however, that gene expression potential cannot
be determined from a single epigenetic process alone and understanding the functional
interactions between epigenetic mechanisms is critical for understanding how cellular
transcription program is regulated or misregulated during normal and disease
development (Baylin and Jones, 2011). We have developed the Nucleosome Occupancy
and Methylome sequencing (NOMe-seq) assay to directly measure the relationship
between DNA methylation and nucleosome occupancy. This assay takes advantage of an
exogenous methyltransferase M.CviPI which methylates unprotected GpC dinucleotides
to create a footprint of chromatin accessibility and generate dual nucleosome positioning
and DNA methylation information at a single DNA molecule resolution using low
number of cells as input (Kelly et al., 2012; Xu et al., 1998).
Here, we discuss strategies to optimize the assay methodology and describe in
details the application of this assay for both loci-specific and genome-wide analyses.
First, we demonstrate the unique ability for NOMe-seq to simultaneously footprint
nucleosome occupancy and DNA methylation in promoter regions, and specifically,
promoters that are present in multiple chromatin configurations such as those that are
monoallelically expressed, imprinted and/or X-inactivated. Furthermore, NOMe-seq can
be used to detect protection due to transcription factor binding and how this binding may
affect DNA methylation pattern and the global organization of surrounding nucleosomes
59
(Bell et al., 2011). Altogether, the dual epigenetic information we obtain from this assay
may provide us with a useful tool to comprehensively study the crosstalk between
epigenetic processes and ultimately, better understand their contribution to gene
regulation in normal and diseased cells.

60
MATERIALS AND METHODS
Cell Culture
IMR90 cells were maintained in DMEM supplemented with 10% fetal calf serum
and penicillin/streptomycin in accordance to ATCC recommendations.

Nucleosome Footprinting
NOMe-seq is a modified version of our methylation dependent single promoter
assay (Miranda et al., 2010b). Nuclei from IMR90 cells (ATCC) were isolated as
previously described (Miranda et al., 2010b), and were incubated with 100 or 200 units of
GpC methyltransferase (M.CviPI) and S-Adenosyl methionine (SAM) for 15 minutes at
37 °C or 200 units of GpC methyltransferase (M.CviPI) and S-Adenosyl methionine
(SAM) for 7.5 minutes at 37 °C followed by a boost with an additional 100 units
M.CviPI and SAM for 7.5 minutes. For whole genome NOMe-seq libraries were
generated from nuclei that were incubated with 200 units of GpC methyltransferase
(M.CviPI) and S-Adenosyl methionine (SAM) for 7.5 minutes at 37 °C followed by a
boost with an additional 100 units M.CviPI and SAM for 7.5 minutes. The reaction was
stopped, DNA extracted, and bisulfite converted to distinguish methylated from
unmethylated Cs. For individual regions of interest PCR was performed, using PCR
primers that do not contain any CpG or GpC dinucleotides, followed by TA cloning and
sequencing. Detailed step-by-step experimental method can be found in Appendix 1. All
primers used are listed in Table 3.1

61
Library Construction and Sequencing
For NOMe-seq, libraries were prepared from 5 ug of DNA as previously
described (Berman et al., 2012a; Kelly et al., 2010b; Lister et al., 2009). Briefly,
M.CviPI-treated DNA was fragmented into ~200bp pieces, END repaired (Epicenter),
methylated adaptors ligated (Illumina), bisulfite converted (Zymo EZ DNA methylation)
and subject to PCR. Clusters were generated following Illumina protocols and the
resulting library was sequenced on Illumina Hi-seq 2000 using the 76 bp single-end
configuration. Base calling was performed by Illumina Real Time Analysis (RTA)
software, yielding a total of 1.180 million reads that passed the Illumina quality filter.

Sequence alignment and extraction of CG and GC methylation levels
Genomic alignment and bisulfite sequence analysis was performed largely as
previously described (Berman et al., 2012a; Kelly et al., 2012; Liu et al., 2012). For
single-end IMR90 libraries, MAQ(Li et al., 2008) was used with the “-c” bisulfite mode
(as in reference (Berman et al., 2012b)). IMR90 reads were aligned to NCBI reference
genome hg19, filtering out reads with mapping quality of less than 30 which result in 678
million reads. We then removed reads starting at exactly the same genomic position as
another read (PCR “duplicate” reads), yielding a total of 156 million analyzable reads for
IMR90 (11.8 gigabases).
It is difficult or impossible to distinguish C to T SNPs in bisulfite sequencing
data, but our Illumina protocol only recovers bisulfite data from one of the two strands (G
residues complementary to cytosines are read as G whether or not the complementary
cytosine is methylated). For this “directional” bisulfite library protocol, cytosine positions
62
appear on the sequence reads as C or T depending on bisulfite conversion, whereas the
complementary G on the strand opposite the C will only be read as G (Krueger et al.,
2012). We therefore refer to two strands relative to a given cytosine position – the
“Bisulfite-C strand” (BCS) and the “genotype G strand” (GGS). The genotype G strand is
thus named because it reveals the true genotype of the position, unaffected by bisulfite
conversion. Because of the specifics of Hi-Seq paired-end sequencing, the second end of
a paired-end run is always the reverse complement the BCS sequence, and thus must be
reverse complemented before analysis to obtain the true BCS sequence.
In our analysis, we only used cytosines that had the particular trinucleotide
context in the reference genome assembly and to minimize the effect of differences
between sample genomes and the reference genome (SNPs), we only included cytosines
present in the reference genome if they included at least 3 C or T reads on the BSC
strand, and at least 90% of all reads mapped to the BSC strand were C or T. Additionally,
we only included cytosines where 90% of the reads mapped to the GGS were G (any
other base indicates a genetic variant; importantly, only the GGS strand can reveal the
C:G->T:A variants that can lead to false methylation calling). A cytosine was determined
to be in a particular XCX trinucleotide context using the same criteria, i.e. a cytosine was
considered GCH if 90% of reads were G for the preceding base and 90% of the reads
were A, C, or T (IUPAC “H” symbol includes A, C, T) for the following base. Because
of bisulfite conversion, the preceding G has to be considered in a strand-aware way –
only reads with G on the strand of interest count toward the 90%, whereas either C or T
are counted from the opposite strand. The same approach is used to determine the
following tri-nucleotides discussed in this study: HCG (H includes A, C, or T), GCG, and
63
WCG (W includes A or T).
As in (Berman et al., 2012a), we filter out the 5’ ends of reads that have apparent
bisulfite non-conversion, which is common in the Illumina protocol presumably due to
re-annealing of base pairs adjacent to the adapter sequences which are methylated and
thus have 100% base complementarity (Berman et al., 2012a; Hansen et al., 2011a). We
accomplish this by walking inward from the 5’ of the sequencing read and disregarding
any unconverted cytosine (in any sequence context) until the first converted cytosine is
encountered. From that point and all 3’ positions within the read, we include all
converted and unconverted cytosines in methylation counts.

Genomic element average profile plots
Methylation values were extracted from regions surrounding genomic landmarks
of interest (promoters, CTCF sites, etc.) and all methylation values were averaged within
moving windows of 20 bp for all plots (genomic positions without cytosines of the
correct type were not included in averages). 20bp was chosen because it is smaller than
the average distance between adjacent GCs in the genome, and clearly able to resolve
nucleosome phasing/positioning (as evidenced in CTCF alignments).
Promoter positions and chromatin marks were taken from a previous reference
(Hawkins et al., 2010)(GEO ID GSE16256). For CpG Island and non-CpG Island
promoters, we used the Takai-Jones definition (Takai and Jones, 2002). For CTCF
annotations, we used evolutionary conserved CTCF binding motifs (Xie et al., 2007) that
were bound in vivo in either HeLa cells (Kim et al., 2007) or CD4+ T-cells (Cuddapah et
al., 2009). We removed about 10% of these sites that fell within 2 kb of a known TSS.
64
Our final set contained 8,722 non-promoter CTCF sites.
Data and source code availability
NOMe-seq tracks for genomic viewers are available at http://epigenome.usc.edu
and GSE21823. All source code tools are available at
http://sourceforge.net/projects/uecgatk/.

65
Table 3.1. NOMe-seq Primer List
Locus Sequence
GRP78
Forward
Reverse

GTTGAGAATTATTTTTGGATTT
AAACTTACCACCATAAAAAATT
MLH1
Forward
Reverse

AATTTATAGAGTTGAGAAATTTGATTGG
TTACACTCCAAACAACCCTTAAAAAA
LAMB3
Forward
Reverse
ATTTTGTGAATTTTGGTTTTTGATGATTATT
CATCCAAAAATACAATCCTCCT
SNRPN
Forward
Reverse
AAAAATAGGTAGATATGTTTATTGATTTT
AATACTCCAAATCCTAAAAACTTAA
DLG3
Forward
Reverse

GTTAAGTGTGAGGTTATAGTAATTTT
AATCCCCAAAATAAATCTAAATAAAATC
CTCF
Forward
Reverse

GGTAGAGGTTTTTATTATGA
AACAAAATTCCAAACAATAAAAACA
MYOD1
Forward
Reverse

GGGTTTTTTATAAATTAGGGGATAGAGGAGTATT
CCCACCCACTATCCCCCACCCCTCCCT

66
RESULTS
Determining optimal NOMe-seq conditions for accurate footprinting of various
genomic regions.
To study the interaction between nucleosome occupancy and DNA methylation,
we adapted a previously described M-SPA assay to develop NOMe-seq, a footprinting-
based method which generates a high-resolution map of both epigenetic components
(Miranda et al., 2010a). We used M.CviPI, a commercially available bacterial
methyltransferase which is capable of methylating cytosines that are present in the GpC
dinucleotide context (Xu et al., 1998) when they are accessible and unprotected by
nucleosomes without affecting the methylation status of CpG sites to create a footprint of
nucleosome occupancy and DNA methylation following bisulfite sequencing (Figure
3.1). As GpC sites are ~5 fold more abundant in the genome compared to CpG sites and
are never endogenously methylated, we are able to profile the occupancy and positioning
of nucleosomes in high resolution while circumventing the limitation of M-SPA or other
CpG-based footprinting approach where footprinting can only be done in CpG rich
regions that are unmethylated (Gal-Yam et al., 2006; Kelly et al., 2010c; Lin et al., 2007).
We tested various enzyme concentrations to determine the appropriate reaction
condition for footprinting of various chromatin structures across the genome. Our initial
experiments were done using normal human lung fibroblast cell line, IMR90 cells, due to
the availability of other publically accessible data of the cell line (Hawkins et al., 2010).
We analyzed multiple control regions to footprint the following chromatin structures
using M.SssI treated sample as a control: unmethylated and accessible (MLH1 and
GRP78), unmethylated and inaccessible (MYOD1) and methylated and inaccessible
67
(LAMB3). Our criterion for the enzyme condition is that it must be able to map accessible
regions accurately while avoiding enzyme overtreatment that may cause known
inaccessible regions to appear accessible. Generally, the amount of enzyme required can
be region-specific as some loci are more easily footprinted than others. We found that
incubating freshly isolated nuclei with 100U or 200U for 15 minutes is sufficient to
footprint the accessibility pattern of MLH1, but not GRP78 (Figure 3.2). We then
incubated the nuclei with 200U of enzyme and substrate for 7.5 minutes followed by a
boost of half of the amount of enzyme and substrate also for 7.5 minutes. We
subsequently determined that this condition allowed us to accurately map the
accessibility of GRP78 which was known from previous works to have an open
chromatin structure (Figure 3.2) (Gal-Yam et al., 2006). This condition also sensitively
detected the occupancy of a stable nucleosome in between the two annotated TSSs of
MLH1 as well as the closed structure of polycomb repressed gene, MYOD1, and
methylated LAMB3 (Lin et al., 2007). Longer incubation time (<3 hours) does not appear
to improve the footprinting pattern in our control regions and thus for all our downstream
experiments, unless otherwise noted, we used a shorter reaction time.

68
Figure 3.1 Schematic of NOMe-seq Assay. In this assay, freshly isolated and intact
nuclei are treated with M.CviPI which methylates the cytosines within GpC
dinucleotide context that are accessible (blue circles, teal filled) and unprotected by
nucleosome (large pink-filled circle) and transcription factors (purple-filled half circle)
while retaining the endogenous DNA methylation status (black circles, white and black
filled). Following bisulfite sequencing, the methylation level of both GpC (blue circles)
and CpG sites (black circles) can be simultaneously measured at single DNA molecule
resolution and used to footprint regions with various chromatin architectures including
those present in multiple configuration such as seen in divergent chromatin allele and
linker regions.
69
Figure 3.2 Optimization and reproducibility of M.CviPI accessibility. High
resolution NOMe-seq was performed at different enzyme concentrations to footprint
various types of promoter regions. M.SssI based footprinting is included as a control. In
IMR90 cells, MLH1 and GRP78 are both active promoters and have open chromatin
structures. With 15 minutes incubation time, 200U of the enzyme is sufficient to
accurately footprint MLH1, but not GRP78. A 3-hour-incubation time with the same
amount of enzyme does not change or improve the footprinting in these two regions. By
incubating nuclei with 200U of enzyme for 7.5 minutes and a boost of 100U for an
additional 7.5 minutes of incubation, we show that we are able to accurately footprint
both MLH1 and GRP78 promoters. A repressed promoter, MYOD1’s footprinting shows
that the higher enzyme concentration does not result in aberrant accessibility. We are
also able to show the closed chromatin structure of LAMB3 using NOMe-seq, which is
impossible to do using M.SssI because LAMB3 is endogenously methylated in IMR90
cells. For M.CviPI treated nuclei, circles represent GpC dinucleotides (white,
unmethylated and inaccessible to M.CviPI; teal, methylated and accessible to M.CviPI).
Pink bars are regions ≥147bp, representing sites associated with nucleosomes. Regions
accessible to M.CviPI (teal) represent NDRs. Diagram is drawn to scale.
70
NOMe-seq detects regions existing in multiple chromatin configurations.
One of the unique strengths of NOMe-seq compared to traditional nucleosome
mapping method is its single DNA molecule resolution which can be used to map
genomic regions that are present in multiple functional configurations, termed divergent
chromatin allele or DCA (Kelly et al., 2012). Different chromatin states may exist on the
two alleles from a single cell such as in imprinted and X-inactivated genes, or in a
subpopulation of cells within a sample. As a proof of principle, we performed locus-
specific NOMe-seq to show that the maternally imprinted SNRPN and X-linked DLG3
loci consist of both open and closed chromatin architecture in a single genomic region,
each corresponding to the active and inactive allele (Figure 3.3a-b). We also profiled
ZNF597 locus which was predicted to be a novel imprinted gene based on its methylation
status (Nakabayashi et al., 2011). By sequencing individual DNA molecules of these loci,
we were able to distinguish the active alleles which have unmethylated and nucleosome
depleted conformation from the inactive alleles that are methylated and nucleosome
occupied. This result reinforces the role of DNA methylation in the permanent silencing
of promoters in normal mammalian cells (Jones, 1999). Furthermore, it establishes a
direct correlation between DNA methylation and chromatin inaccessibility, suggesting
that DNA methylation may directly control the establishment of nucleosome-occupied
regions.

71
420 bp
Pat

Mat

420 bp SNRPN SNRPN
Endogenous Methylation (CG) M.CviPI Enzyme Accessibility (GC)
Endogenous Methylation (CG) M.CviPI Enzyme Accessibility (GC)
572 bp 572 bp DLG3 DLG3
A
B
Endogenous Methylation (CG)
M.CviPI Enzyme Accessibility (GC)
ZNF597 ZNF597
C
Figure 3.3 NOMe-seq detects genomic regions present in diverging conformation.
PCR amplicons were cloned and several colonies were sequenced to visualize two
distinct chromatin configurations in (A) imprinted gene SNRPN, (B) X-inactivated
gene, DLG3 and (C) putative imprinted gene, ZNF597. For M.CviPI treated nuclei,
circles (left panel) represent GpC dinucleotides (white, unmethylated and inaccessible to
M.CviPI; teal, methylated and accessible to M.CviPI). Pink bars are regions ≥147bp,
representing sites associated with nucleosomes. Regions accessible to M.CviPI (teal)
represent NDRs. Diagram is drawn to scale
543 bp 543 bp
72
Measuring the strength of transcription factor binding using NOMe-seq.
Using NOMe-seq, we were also able to detect a protection pattern occurring due
to the binding of transcription factors or other DNA binding proteins whose recognition
sites contain at least one GpC site. One such example is CTCF, a zinc-finger protein
often described as an insulator or transcriptional repressor whose binding is highly
conserved (Cuddapah et al., 2009). The consensus sequence of CTCF often contains a
CpG site in addition to GpC sites and previous studies have shown that methylation of
the CpG residue may disrupt the protein-DNA interaction (Feldmann et al., 2013). We
performed bisulfite sequencing of a genomic region, distal from annotated TSS,
containing a known CTCF binding site and confirmed that CTCF region is largely
unmethylated. We furthermore showed that we were able to accurately footprint non-
nucleosomal protein as indicated by the discreet protection pattern of <40bp overlapping
the CTCF motif (Figure 3.4a). Additionally, we detected highly accessible linker regions,
at the length of 50-100 bp, flanking the CTCF-protected regions. These linkers are
followed by larger inaccessible regions at both 5’ and 3’ position of CTCF corresponding
to nucleosome occupancy.
We next adapted our standard NOMe-seq protocol to include a high-salt wash of
freshly isolated nuclei prior to treatment with the M.CviPI enzyme in order to measure
the strength of CTCF binding (see Methods and Appendix). We observed patches of
increased accessibility without a change in endogenous DNA methylation status in the
CTCF binding sites following a 200 mM NaCl wash step, suggesting that there was a
disruption in the binding of CTCF at non-physiological high-salt condition (Figure 3.4b).
At 400 mM salt concentration, the CTCF binding region is completely accessible in all
73
but one of the DNA molecules, due to the complete removal of protecting protein from
individual DNA molecules (Figure 3.4c). At both 200 and 400 mM, however,
nucleosome protection around the CTCF region remains mostly intact, indicating a stable
association between DNA and nucleosomal components, consistent with previous report
(Jeong et al., 2009). This data demonstrates that the NOMe-seq assay has the sensitivity
to distinguish between nucleosome and transcription factor protection and may be used to
interrogate the relationship between transcription factors, DNA methylation and
nucleosome occupancy at regions of interest.

74

10 mM
200 mM
400 mM
Endogenous DNA Methylation (CG)
Accessibility (GC)
147 bp
A
B
C
Figure 3.4 NOMe-seq detects CTCF binding. (A) High resolution NOMe-seq of
region containing the insulator CTCF, whose binding site is depicted by the red box.
The binding site is completely protected from the enzyme, whereas the surrounding
linker region, as predicted, is highly accessible. This distinct protection pattern on the
binding site is much smaller in width compared to the protection from the well-
positioned nucleosomes adjacent to the NDRs that flank the binding site. (B)
Incubating the nuclei with 200-mM and (C) 400-mM NaCl prior to enzyme treatment
can wash off binding by CTCF and remove its protection without affecting the
positioning of flanking nucleosomes and DNA methylation status.
75
Applying NOMe-seq genome-wide to characterize promoter configuration in IMR90
cells.
In order to examine the interplay between DNA methylation and nucleosome
occupancy at genome-wide scale, we coupled NOMe-seq with next-generation
sequencing (Kelly et al., 2012). We performed whole-genome bisulfite sequencing based
on previously described method and obtained 156-million unique reads (see Appendix)
(Lister et al., 2009). Methylation status of CpG and GpC sites was extracted using Bis-
SNP, excluding cytosines present in GCG trinucleotide context to eliminate ambiguity
due to endogenous and enzymatic methylation (Liu et al., 2012). We aligned NOMe-seq
reads to all TSSs and observed three distinct promoter configurations: unmethylated and
nucleosome depleted, unmethylated and nucleosome occupied, and methylated and
nucleosome occupied (Figure 3.5a). These architectures correspond to active,
repressed/poised and inactive/methylated promoters respectively. Interestingly, we saw
that active promoters also have well-positioned nucleosomes downstream of the TSS
(Figure 3a, left panel), which are absent in other promoter configuration, consistent with
previous reports that while the presence of nucleosome is anti-correlated with
transcription initiation, well-positioned nucleosomes are required to facilitate
transcriptional elongation (Lorch et al., 1987).
We further separated the TSSs into CpG islands (CGI) and non-CGI promoters
and found that generally, CGI promoters are unmethylated and nucleosome-depleted near
the TSS (+/-1kb) whereas the opposite is true for non-CGI promoters (Figure 3.5b-c, left
panel). Stratifying the promoters further based on their methylation status, we found the
average pattern of CGI promoters in IMR90 cells was driven by the unmethylated and
76
nucleosome-depleted promoters which make up the majority of CGI promoters (Figure
3.5b, middle and left panel). The small subset of methylated CGI promoters are mostly
nucleosome occupied, which suggests the antagonism between DNA methylation and
promoter accessibility (Thurman et al., 2012). Although there are more methylated than
unmethylated non-CGI promoters, the inverse relationship between DNA methylation
and accessibility also holds true for non-CGI promoters (Figure 3.5c, middle and left
panel). Furthermore, this relationship is not unique to IMR90 cells as we were also able
to reproduce this result in other cell types, including uncultured human cells (see Chapter
4 and 6) (Kelly et al., 2012).

77

CTCF
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

nonCGI TSS all
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

CGI TSS all
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

CGI TSS meC+
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

TSS K27me3
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

nonCGI TSS meC-
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

CGI TSS meC-
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

TSS K4me3
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500

nonCGI TSS meC+
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500
FIGURE 3 (Kelly et al.)

TSS meC+
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500
a b
c
d
me HCG
nuc 1-GCH
CTCF
N
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500
0 +1000
0
0.2
0.4
0.6
0.8
1
Distance from element (bp)
Fraction of reads
-1000 -500 +500
IMR90
157
248
CTCF
HCG
1-GCH
A
B
C
Figure 3.5 NOMe-seq reveals distinct chromatin configurations at specific promoter
types. (A) NOMe-seq distinguishes the three major promoter states at promoters in
IMR90 cells—active H3K4me3-marked promoters are unmethylated and contain a NDR
upstream and well-positioned nucleosomes after the TSS. TSSs are indicated on the x-
axes as 0. Repressed/poised H3K27me3-marked promoters are unmethylated and
nucleosome-occupied. Methylated promoters are nucleosome-occupied. The y-axis
indicates M.CviPI inaccessibility (1-CpG; teal) and CpG methylation level. (B) CGI
promoters are characterized by a lack of CpG methylation, an upstream NDR, and well-
positioned nucleosomes after the TSS. The majority of CGI promoters are unmethylated
(11,165) and display the same pattern, while methylated CGI promoters (781) are
nucleosome-occupied and inaccessible to M.CviPI. (C) Non-CpG island promoters are
generally characterized by CpG methylation and inaccessibility to M.CviPI, indicating
nucleosome occupancy. The few unmethylated non-CGI promoters (1397) are depleted of
nucleosomes upstream of the TSS, while the majority of non-CGI promoters (4668) are
nucleosome-occupied and inaccessible to M.CviPI. M.CviPI inaccessibility is plotted (1-
GCH) in teal and CpG methylation (CGH) in black.
78
Genome-wide NOMe-seq detects accessibility patterns in transcription factor
binding sites.
We next examined the genome-wide pattern of nucleosome occupancy and DNA
methylation in insulator regions by aligning NOMe-seq reads to the center of previously
described CTCF peaks in IMR90 cells (Cuddapah et al., 2009). Not unlike our locus
specific experiment, we observed distinct accessible regions immediately flanking a peak
of inaccessibility which corresponds to CTCF protection (Figure 3.6a, right panel).
Furthermore, we found that nucleosomes surrounding bound CTCF are spaced in well-
organized array and are strikingly anti-correlated with DNA methylation. Specifically,
DNA methylation level peaked in the accessible linker regions of CTCF, suggesting that
contrary to previously held belief, the correlation between DNA methylation and
nucleosome occupancy may vary between different regulatory elements (Chodavarapu et
al., 2010; Portela et al., 2013).
We were also able to apply NOMe-seq to measure the strength of CTCF binding
at global level. Recapitulating our loci-specific observations, we detected an increase of
accessibility in the CTCF binding sites following a 200mM-salt wash prior to M.CviPI
treatment (Figure 3.6a, middle panel). This increase is even more dramatic at 400 mM,
suggesting that the effect of high-salt wash on CTCF binding is not restricted to
individual regions (Figure 3.6a, right panel). We furthermore show that we can detect
similar change in accessibility pattern following high-salt wash in NSRF1 binding motifs
(Figure 3.6b), thus highlighting the strength of our genome-wide footprinting method.
The centers of CTCF and NSRF1 binding sites are also markedly hypomethylated
compared to the surrounding regions. Unlike CTCF, however, DNA methylation pattern
79
is not phased surrounding NSRF1 binding regions which also generally lack well-spaced
nucleosome arrays. This data illustrates that different transcription factors may have
differing roles in regulating nucleosome organization. Although this regulatory aspect is
beyond the scope of our current study, we demonstrate that NOMe-seq assay is uniquely
suited for future investigations for the role of transcription factors in the establishment of
epigenetic landscape.

80
−1000 −500 0 500 1000
0.0 0.2 0.4 0.6 0.8 1.0
CTCF_conserved.IMR90_merge_64KLYAAXX_C18FYACXX_D12D6ACXX
smooth: 0 −−−numElem: 8549 , numElemHcg: 8549
GCH
HCG
−1000 −500 0 500 1000
0.0 0.2 0.4 0.6 0.8 1.0
CTCF_conserved.ResultCount_C235LACXX_6_LIU1675A2
smooth: 0 −−−numElem: 8549 , numElemHcg: 8549
GCH
HCG
−1000 −500 0 500 1000
0.0 0.2 0.4 0.6 0.8 1.0
CTCF_conserved.ResultCount_C235LACXX_7_LIU1675A1
smooth: 0 −−−numElem: 8549 , numElemHcg: 8549
GCH
HCG
10 mM
200 mM 400 mM
Distance to Element(bp)
% Methylation
0 20 40 60 80 100
−1000 −500 0 500 1000
NRF1
smooth: 0 −numElemCenterToAlign: 19160
GCH
HCG
0 20 40 60 80 100
−1000 −500 0 500 1000
NRF1
smooth: 0 −numElemCenterToAlign: 19149
GCH
HCG
0 20 40 60 80 100
−1000 −500 0 500 1000
NRF1
smooth: 0 −numElemCenterToAlign: 19151
GCH
HCG
0
20
40
60
80
100
HCG
GCH
Figure 3.6 NOMe-seq detects chromatin configuration around transcription factor
binding sites. a) NOMe-seq demonstrates unmethylated NDRs at CTCF sites in IMR90
cells, which are marked by a peak in inaccessibility at the CTCF site itself. Well-
positioned nucleosomes flank CTCF sites, with DNA methylation peaking in between
nucleosomes. 0 indicates the middle of the CTCF binding motif. CTCF binding sites
were obtained from GSM935404. Increased accessibility at the CTCF site is seen when
nuclei are washed with high-salt buffer to disrupt CTCF binding to DNA. b) NOMe-seq
shows unmethylated NDR regions associated with NSRF1sites in IMR90 cells which
are marked by a peak of inaccessibility corresponding to NSRF1 binding site itself that
gradually disappear with increasing concentration of salt wash. Unlike CTCF, NSRF1
does not show well-phased nucleosomes.
Distance to Element(bp)
% Methylation
0
20
40
60
80
100
Distance to Element(bp)
% Methylation
0
20
40
60
80
100
Distance to Element(bp)
% Methylation
0
20
40
60
80
100
Distance to Element(bp)
% Methylation
0
20
40
60
80
100
Distance to Element(bp)
% Methylation
0
20
40
60
80
100
A
B
81
DISCUSSION
For many years, studies examining the role of nucleosome occupancy and
positioning in the regulation of transcription has been limited to lower eukaryotes such as
yeast. The last decade, however, has seen a growing interest in the epigenetic community
to map the global positioning of nucleosomes in the human genome. This is partly due to
reports elucidating the role of nucleosomes in the epigenetic silencing of tumor
suppressor genes in cancer cells and partly due to growing appreciation for how the
packaging of chromatin in nucleus contributes to gene regulation (Lin et al., 2007; Rivera
and Ren, 2013). Various methods including MNase-seq have been used to map the
pattern of nucleosome occupancy in the genome and elucidate the distinct organization of
nucleosomes across different regulatory regions (Chodavarapu et al., 2010; Schones et
al., 2008; Valouev et al., 2011). It has also become clear, however, that understanding
epigenetic gene regulation requires a comprehensive approach as to how various
epigenetic machineries influence each other and thus highlighting the need to develop
methods to integrate the analyses of epigenetic machineries.
We developed the NOMe-seq assay to simultaneously interrogate the relationship
between DNA methylation and nucleosome occupancy on a single DNA molecule
resolution. This method takes advantage of the ability of a commercially available
enzyme, M.CviPI, to access and methylate GC dinucleotides that are not bound by
nucleosome or transcription factor (Xu et al., 1998). At locus-specific level, this assay
has been used to show the pattern of nucleosome occupancy during the cell-cycle, the
presence of nucleosome depleted region (NDR) in the enhancer, the maintenance of NDR
by OCT4 and the epigenetic cross-talk between enhancers and promoters (Andreu-Vieyra
82
et al., 2011; Kelly et al., 2010c; Taberlay et al., 2011; You et al., 2011). Since the enzyme
does not target CpG sites, we are also able to retain the endogenous DNA methylation
status of CpG dinucleotides. Moreover, NOMe-seq does not require mechanical or
enzymatic fragmentation of DNA molecules, thus eliminating biases that are often
associated with preferential cutting or over-digestion of certain genomic regions. The
abundance of GpC sites in the genome may also circumvent the limitation of M.SssI-
based footprinting and allows us to map nucleosome occupancy region in CpG-poor and
endogenously methylated regions (Miranda et al., 2010a). The single molecule resolution
of the assay furthermore gives us the ability to resolve epiallele regions which may
exhibit diverging conformation in the same locus (Hawkins et al., 2010). These loci
encompass the monoallelically-expressed genes such as X-inactivated genes, imprinted
genes and autosomal, non-imprinted genes whose roles in cancer are still elusive.
Using genome-wide NOMe-seq, we can accurately footprint the global
architectures of various regulatory regions including promoters and transcription factor
binding sites. We have demonstrated that active promoters lack DNA methylation and
nucleosomes upstream of the TSSs while inactive promoters are nucleosome occupied.
We have also shown that transcription factor binding sites are generally unmethylated
and nucleosome depleted. In the case of CTCF binding, nucleosomes are well-positioned
and linker regions may be more highly methylated compared to nucleosomal DNA,
contrary to previous report (Chodavarapu et al., 2010). Distinct nucleosome organization
pattern between CTCF and NSRF1, furthermore suggests that transcription factors may
also differentially contribute to nucleosome positioning and that a dynamic regulation
83
may exist between transcription factor binding, DNA methylation and nucleosome
positioning (Feldmann et al., 2013).
Taken together, the NOMe-seq assay is a valuable approach to study the
interaction between DNA methylation and nucleosome occupancy. The simultaneous
mapping of these two epigenetic marks on the same molecule can be used to identify
combinatorial profiles present within a mixed population of cells or alleles with greater
sensitivity than the two marks alone. The epigenetic map generated by these
combinatorial epigenomic profiles will have critical implications for the profiling of
complex tissues containing multiple cell types. Furthermore, as mutations in chromatin
remodeling complexes are becoming increasingly associated with cancer (Wilson and
Roberts, 2011), locus-specific and whole genome NOMe-seq is an ideal approach to
address the effects that these mutations have, both on nucleosome positioning and DNA
methylation and can further investigate whether chromatin remodeling defects are
dependent on DNA methylation state (Kelly et al., 2012).

84
CHAPTER 4
THE CONTEXT-DEPENDENT ROLE OF DNA METHYLATION IN
DIRECTING THE FUNCTIONAL ORGANIZATION OF THE CANCER
EPIGENOME
This chapter is adapted from a manuscript to be submitted. This work was done in
collaboration with a fellow student and bioinformatician from Dr. Ben Berman’s lab,
Yaping Liu, with whom I will share a first-authorship.

Introduction
Eukaryotic genomes are controlled by the inter-related and mitotically heritable
sets of epigenetic mechanisms, consisting of DNA methylation, nucleosome positioning
and histone modifications, all of which cooperate to determine gene activation potential.
DNA methylation, the most clinically relevant epigenetic feature, is a covalent addition
of a methyl group on the cytosine of CpG dinucleotides. In mammals, DNA methylation
is critical for suppressing transcriptional activity in normal cells particularly during
imprinting, X-inactivation and silencing of retrotransposons, though recent studies have
suggested that DNA methylation may play a more subtle role in fine-tuning or
reinforcing gene silencing rather than initiating it (Jones, 2012; Rivera and Ren, 2013).
As many as 25 million CpG sites have been shown to be methylated in the human
genome, with the exception being those located in CpG island (CGI) promoters (Cohen et
al., 2011; Lister et al., 2009; Ziller et al., 2013). This methylation pattern is faithfully
copied in a cell-cycle dependent process mediated by DNA methyltransferases DNMT1
and DNMT3A/B which preferentially binds to nucleosomes (Jones and Liang, 2009;
Sharma et al., 2011).
85
The nucleosome is the primary unit of chromatin structure and consists of 147bp
of DNA wrapped around a histone octamer of H2A/B, H3 and H4. The organization of
nucleosomes, along with covalent modifications on the histone tails, is important for
maintaining a balance between compaction and accessibility of the genome by
transcription factors and other DNA binding proteins during cellular processes such as
transcription, replication and repair (Cairns, 2009; Li et al., 2007). Specifically, the
precise positioning of nucleosomes around gene promoters as well as non-coding
regulatory elements is an evolutionarily conserved mechanisms that plays a major role in
eukaryotic transcriptional regulation (Bell et al., 2011; Kelly et al., 2010c; Lorch et al.,
1987; Schones et al., 2008). Various factors such as underlying DNA sequences,
sequence-specific DNA binding factors, and ATP-dependent nucleosome remodelers
have been described as playing important roles in the positioning of nucleosomes (Bell et
al., 2011; Tillo and Hughes, 2009; Tillo et al., 2010; Valouev et al., 2011; Zhang et al.,
2011). Often overlooked and sometimes controversial, however, is the role of DNA
methylation in directing nucleosome positioning in mammalian genome (Bell et al.,
2011; Chodavarapu et al., 2010; Portela et al., 2013; Valouev et al., 2011).
Epigenetic changes, in particular aberrant DNA methylation and silencing of CGI
promoters are a common signature of cancer (Baylin and Jones, 2011). This observation
combined with the fact that more than 60% of promoters are located in CGIs has driven a
focus on CGIs as a model of study for epigenetic regulation (Deaton and Bird, 2011;
Irizarry et al., 2009; Portela et al., 2013; Tazi and Bird, 1990). The advent of the
epigenomic era has revealed that CGI promoters may not be a homogenous class and
regulatory regions outside of the CGI promoters such as the CpG island shores and non-
86
CpG island promoters and enhancers may also play a role in tumorigenesis (Aran and
Hellman, 2013; Berman et al., 2012b; Doi et al., 2009; Hovestadt et al., 2014; Irizarry et
al., 2009; Rach et al., 2011; Taberlay et al., 2014). It is becoming increasingly clear,
however, that the study of gene regulation requires an of integrated chromatin states and
a holistic understanding of how the different epigenetic machineries influence each other
(Rivera and Ren, 2013). For instance, histone marks and variants have been broadly
categorized as either active (H3K27ac), permissive (H3K4me1, H2A.Z), or repressive
(H3K27me3, H3K9me3), with their combinatorial patterns used to define distinct states
such as active vs. poised promoters and enhancers (Creyghton et al., 2010; Ernst et al.,
2011; Rada-Iglesias et al., 2011). Yet, despite the extensive study on DNA methylation
changes in cancer, we still lack an understanding as to how DNA methylation contributes
to the establishment of these integrated chromatin states (Jones, 2012).
Here, we compare a colon cancer cell line HCT116 with its almost completely
unmethylated derivative, DKO1, to evaluate the effects of DNA methylation on
nucleosome positioning and histone modifications. Our study couples NOMe-seq with
histone ChIP-seq and RNA-seq to generate integrated maps of chromatin architecture and
gene expression. Using the DKO1 model, which was genetically engineered to have a
complete depletion of DNMT3B and hypomorphic expression of DNMT1 (Rhee et al.,
2002), we profile the focal and long-range changes in chromatin structures and elucidate
how perturbations in global DNA methylation pattern may directly alter the functional
organization of the cancer epigenome and thereby, gene transcription (De Carvalho et al.,
2012; Egger et al., 2006; Rhee et al., 2002; Sharma et al., 2011).

87
Materials and Methods

Cell Culture
HCT116, obtained from ATCC, and DKO1 cells were cultured under
recommended conditions at 37°C and 5% CO
2
in McCoy’s 5A media supplemented with
10% FBS and penicillin/streptomycin. HCT116 was obtained from ATCC and DKO1
was a generous gift from Drs. Bert Vogelstein and Steve Baylin.

Genome-wide nucleosome footprinting assay
NOMe-seq was performed as previously described (Kelly et al., 2012). Briefly,
exponentially growing cells were washed with PBS, trypsinized and incubated with ice-
cold lysis buffer (10 mM Tris, pH7.4, 10 mM NaCl, 3 mM MgCl2, 0.1 mM EDTA and
0.5%NP-40) for 5 minutes on ice to isolate intact nuclei. Nuclei were washed with ice-
cold wash buffer (10 mM Tris, pH7.4, 10 mM NaCl, 3 mM MgCl2, 0.1 mM EDTA),
resuspended in ice-cold 1x GpC buffer (New England Biolabs) and treated with 200U of
M.CviPI enzyme supplemented with 1.5µl S-adenosylmethionine (SAM) for 7.5 minutes
with a boost of 100U enzyme and 0.75µl SAM for additional 7.5 minutes. Genomic DNA
was isolated by standard phenol-chloroform extraction and ethanol precipitation. WGBS
libraries were generated using 2-5 µg of DNA as previously described and sequenced on
Hiseq2000 (Berman et al., 2012b; Lister et al., 2009). Sequencing reads were mapped to
the hg19 genome and methylation levels of CpG and GpC dinucleotides were determined
using previously described pipeline (Kelly et al., 2012; Liu et al., 2012).

88
Hidden-Markov model-based approach of NDR detection
Two-state beta-binomial HMM was adapted from a previously described method
(Molaro et al., 2011) to segment regions into Methyltransferase Accessible Regions
(MARs) and Methyltransferase Protected Regions (MPRs), based on GCH methylation in
HCT116 and DKO1 cells, where training of the model was performed independently for
each biological replicate. GCH methylation and read coverage were used as inputs to the
Viterbi algorithm to determine the state of each individual GCH, and segments
containing at least 3 contiguous GCHs present in the same state were required to call the
MARs and MPRs. A one-tailed binomial test was used to calculate the significance level
of each MAR in comparison to all MPRs present in the adjacent +/-100kb region, with
only MARs having FDR-corrected p-value <0.01 considered significant. MARs having
the length >100 bp were considered as NDRs. For the analysis of Figure 1a, only NDRs
that overlapped in both biological replicates in each cell line were used.

Defining CGI and Non-CGI promoters
CGIs as defined by Takai-Jones and Gardiner-Garden-Frommer were merged and
extended by 200bp on each end. Promoters that overlapped these regions were considered
CGI promoters. To define non-CGI promoters, CGIs as defined by Takai-Jones,
Gardiner-Garden-Frommer and Irizarry were merged and extended by 500bp on each
end. Only promoters that did not overlap these regions were considered non-CGI
promoters.

89
Defining promoter methylation classes
We combined the two NOMe-seq replicates for each cell type, and filtered all
annotated Transcription Start Sites (TSSs) from the UCSC knownGenes track that had at
3 HCG sites and at least 10 reads from each cell type within the region from -300 to
+500bp around the TSS. We considered these TSS regions as unmethylated in a given
cell type if they had an average methylation level less than 5%. For methylated TSSs, we
used 60% and 25% as the lower cutoff for HCT116 and DKO1, respectively. These
different cutoffs in the two cell types were determined based on the global distribution of
methylation values in the promoter regions of each cell line. Based on these criteria, we
included 15,692 CGI promoters and 7,191 non-CGI promoters.

RNA-seq
Cells were washed with PBS and subsequently lysed in Trizol. Total RNA from
two independent cultures was purified using Direct-zol RNA MiniPrep (Zymo Research)
and libraries were constructed using the poly-A selected method of the TruSeq RNA
Sample Prep Kit (Illumina) according to manufacturer’s instructions. Sequencing reads
were mapped to the hg19 reference genome using TopHat v.1.2
(http://ccb.jhu.edu/software/tophat/index.shtml), filtering out non-uniquely mapping
reads and PCR duplicates. FPKM value was calculated using Cufflinks v.2.1.1 with the
following parameters: -F 0.3 –u –b hg19.fa. Gene annotation was obtained as a GTF file
from the UCSC Genome Browser (knownGene track). The read count for each gene was
extracted using htseq-count v0.5.4p3 (http://www-
huber.embl.de/users/anders/HTSeq/doc/overview.html). EdgeR v.3.6.0
90
(http://www.bioconductor.org/packages/release/bioc/html/edgeR.html) was used to
determine the differentially expressed genes between the two cell lines using two
biological replicates for each cell line. Only genes with FDR-corrected p-value <0.05
were considered differentially expressed.

ChIP-seq
ChIP assay was performed using 50 µg of chromatin as previously described and
according to ENCODE’s guideline (Kelly et al., 2010c; Landt et al., 2012). The following
antibodies were used: H2A.Z (Abcam, ab4174), H3K4me3 (Active Motif, 39160),
H3K4me1 (Active Motif, 39298), H3K27Ac (Active Motif, 39297), H3K27me3 (Active
Motif, 39155). Genome-wide libraries were generated from 20ng of purified ChIP and
input DNA, barcoded and sequenced for 50 single-end reads on Hiseq 2000 using
previously described protocol (Barski et al., 2007; Kelly et al., 2012). Sequencing reads
were mapped to hg19 using bwa (http://bio-bwa.sourceforge.net/), removing non-
uniquely mapping reads and PCR duplicates. All ChIP-seq reads were extended by the
sequencing library’s mean fragment size, which was estimated using the default setting of
HOMER v.4.3’s (http://homer.salk.edu/homer/chipseq/) makeTaqDirectory command.
Each dataset was normalized into a single value for each genomic position using the
Wiggler tool (https://sites.google.com/site/anshulkundaje/projects/wiggler) with default
settings and “globalmap_k20tok54” as the mappability parameter. Mean Wiggler values
were calculated in 10bp bins (Gerstein et al., 2012). To normalize variations between
biological replicates, we modified a previously described method to perform Z-score
91
transformation by subtracting the mean Wiggler value across the genome and dividing by
the standard deviation of the genome-wide Wiggler subtraction value (Xie et al., 2013).

ChromHMM
Segmentation and determination of chromatin states were calculated as previously
described (Ernst et al., 2011). Reads in two biological replicates were pooled together,
excluding non-uniquely aligned reads and PCR duplicates. ChromHMM training of 11-
states model was performed using the default setting and with the following ChIP-seq
datasets: H2A.Z, H3K4me3, H3K27ac, H3K4me1, H3K27me3, H3K36me3, H3K9me3
and Input from HCT116, DKO1 and publically available colonic mucosa (GSM621673,
GSM621672, GSM621670, GSM621671, GSM621668, GSM621669). The state
emission probability matrix and transition probability matrix were outputted and regions
were decoded and labeled using the state with the maximum posterior probability.
Long-range chromatin structure changes
The reference genome was segmented into 1Mb non-overlapping windows using
custom perl script, excluding all CGIs as defined by Takai-Jones, Gardiner-Garden-
Frommer and Irizarry. Means of the wiggler normalized signals for ChIP-seq, WCG
methylation for DNA methylation and GCH methylation for accessibility in each window
were calculated and then sorted by the accessibility level in DKO1 cells.

92
Results
NOMe-seq detects the depletion of nucleosomes at a subset of genomic enhancers in
methyltransferase-deficient DKO1 cells.
An accessible or nucleosome depleted region (NDR) is a distinct feature of active
regulatory elements (Kelly et al., 2012; Rivera and Ren, 2013). Using NOMe-seq, we
characterized the relationship between DNA methylation and chromatin accessibility by
analyzing two biological replicates of HCT116 and the severely hypomethylated DKO1
cells (Figure 4.1). To examine the effects of DNA methylation loss on discrete genomic
regions, we developed a Hidden Markov Model (HMM) approach to identify 16,245
NDRs present in one or both cell types, and hierarchically clustered them into four
distinct clusters (C1-C4) (Figure 4.2a). The vast majority of these NDRs were conserved
between cell types, but we identified a small cluster of 1,485 NDRs (C3) that were
unique to DKO1 cells. The position of these NDRs in HCT116 cells was confirmed by
examining the publically available DNAse-I hypersensitivity mapping data (data not
shown). Most of the conserved NDRs were flanked by strongly-phased nucleosomes in
both sides of the regions and were associated with unmethylated genomic regions in both
cell types. Clusters C1 and C2 were also associated with strong enrichment of the active
H3K27ac and permissive H2A.Z, H3K4me3, and H3K4me1 marks (Figure 4.2b). In
contrast, C3 NDRs exhibited weakly phased surrounding nucleosomes. These NDRs
were specific to regions that were completely methylated in HCT116, and gained
permissive and active histone marks in DKO1 that were mostly absent in the parent
HCT116 cells (Figure 4.2a-b), providing evidence that changes in DNA methylation may
alter the underlying chromatin structure. Interestingly, many of these regions were pre-
93
marked with low level H3K4me1 in HCT116, a situation reminiscent of H3K4me1
premarking of enhancers in development (Creyghton et al., 2010; Rada-Iglesias et al.,
2011).
We defined discrete combinatorial chromatin elements using chromHMM (Ernst
et al., 2011), and found that C1 and C2 consisted largely of active and weak CGI
promoters in both HCT116 and DKO1 cells, whereas C3 and C4 contained higher
fractions of distal regulatory regions (Figure 4.2c). Notably, we observed an increased
enrichment of strong enhancer state in C3 NDRs in DKO1 cells, suggesting that the
presence of DNA methylation in HCT116 cells may serve to keep these enhancers
repressed. These distal NDRs were also enriched for transcription factor motifs such as
HIF1b and AP-1, and may be responsible for the increase of expression in nearby genes
(data not shown). C4, on the other hand, primarily contained CTCF-associated weak
enhancers in HCT116 cells, where significant demethylation can be observed within
nucleosomes, but not within the highly accessible linker regions (Berman et al., 2013;
Kelly et al., 2012; Taberlay et al., 2014). Demethylation of these CTCF-associated
linkers, however, did not appear to affect the organization of surrounding nucleosomes.
Strikingly, we also observed a trend of increased poised CGI promoter state in all four
NDR clusters, particularly within cluster C1 and C2 which contained large numbers of
CGI promoters (Figure 4.2c), suggesting that rearrangement of the chromatin landscape
may occur in the absence of DNA methylation independent of accessibility changes in
order to maintain the balance between chromatin accessibility and compaction
(Komashko and Farnham, 2010). This finding prompted us to perform further
investigation of chromatin changes at promoters genome-wide.
94
DKO1
HCT116
HCT116
DKO1
CG
GC
Methylation Change (CG)
Accessibility Change (GC)
A
C
B
Figure 4.1 Significant loss of DNA methylation in DKO1 cells does not
dramatically increase global accessibility. Genome-wide NOMe-seq was
performed for HCT116 (x-axes) and DKO1 (y-axes) cells and (A) endogenous DNA
methylation and (B) accessibility levels of the two cell lines were compared.
Methylation levels of CpG and GpC sites were measured in genomic windows of
200bp. (C) Genomic windows (200bp) exhibiting change in DNA methylation (x-
axes, DKO1-HCT116) were compared to changes in accessibility (y-axes, DKO1-
HCT116). Dashed lines represent 20% change in methylation (vertical) and
accessibility (horizontal).
95
A
C
B
Figure 4.2 NOMe-seq detects the depletion of nucleosomes at a subset of genomic
enhancers in methyltransferase-deficient DKO1 cells. A) Hidden-Markov Model
(HMM) was applied to identify Methyltransferase Accessible Regions (MARs) and
MARs with length more than 100-bp were considered as nucleosome-depleted regions
(NDRs). Only NDRs that were overlapping between the two biological replicates of
HCT116 and DKO1 cells were included in this heatmap (n=16,245). NOMe-seq reads
were aligned to the center of the NDRs and plotted +/-1-kb. NDRs are hierarchically
clustered based on the accessibility within +/-250 bp of NDR centers in both cell types
and dashed horizontal lines separated the clusters. B) Enrichment level for each histone
mark was calculated in terms of z-score and the value was plotted +/-5kb from the center
of the NDRs, following the order seen in A. C) Each NDR in both cell lines was
annotated based on its chromatin state as defined by chromHMM model. and the
distribution of chromatin states within each cluster is shown as a bar graph. The total
number of NDRs contained in each cluster is indicated in parentheses.
96
The loss of DNA methylation at CGI promoters results in the re-organization of
nucleosomes
To systematically address the effects of DNA methylation loss on CGI promoters
genome-wide, we stratified all CGI promoters based on the methylation status of
HCT116 and DKO1 (Figure 4.3, see Methods). We found that the majority of CGI
promoters were unmethylated in both cell types (Figure 4.4a, UU class), consistent with
reports that CGIs were generally devoid of DNA methylation (Deaton and Bird, 2011;
Gebhard et al., 2010; Weber et al., 2007). The majority of these UU promoters had an
open architecture strongly associated with active promoters, with highly accessible
regions flanked by at least three well-phased nucleosomes in both 5’ and 3’ directions
from the TSS (Figure 4.4a, right panel) (Schones et al., 2008).
Consistent with methylation patterns observed in other cancer cell lines and
primary epithelial tumors, we also found that a significant number of CGI promoters
were methylated in wild-type HCT116 cells (Figure 4.4b-c, MM and MU classes)
(Berman et al., 2012b; Taberlay et al., 2014). DNA methylation was largely absent in
DKO1 cells and lacked the bimodal distribution seen in the parent cells (Figure 4.1a,
Figure 4.3b). However, we found that there was a small set of CGI promoters that
retained residual methylation, which we subsequently categorized as the MM class and
may contain previously reported epigenetic drivers of cancer cell survival (Figure 4.4b)
(De Carvalho et al., 2012). In contrast to the UU class, the MM class was consistently
inaccessible and lacked nucleosome phasing (Figure 4.4b, right panel). We also found a
much larger set of 3,270 CGI promoters methylated in HCT116 that became completely
demethylated in DKO1 cells (Figure 4.4c, MU class). These promoters showed a
97
dramatic reorganization of the surrounding chromatin in DKO1 into arrays of 4-5 phased
nucleosomes in both 5’ and 3’ direction, with an accompanying increase of accessibility
in a smaller subset of the promoters (Figure 4.4c). This striking observation suggests that
DNA methylation may play a direct role in inhibiting the organization of nucleosomes at
CGI promoters.

98
A
B
Figure 4.3 HCT116 and DKO1 cells have different distribution of DNA methylation
levels at TSS. A) The average methylation (CG) and accessibility (GC) levels for all
TSSs in HCT116 is shown on the y-axis of the line plot and calculated +/-2kb from the
TSS (x-axis) where 0 indicates the TSS. Methylation level in the region marked by the
tan-colored shade (-300 to +500 bp) was used to classify promoters into distinct
categories of UU, MM and MU. B) Histograms show the distribution of methylation
levels in the -300 to +500bp regions of CGI (top panel) and non-CGI (bottom panel)
promoters in HCT116 (left panel) and DKO1 cells (middle and right panel). Bimodal
distribution is observed in the promoters of HCT116 cells with non-CGI promoters
showing more highly methylated regions compared to CGIs. Tan colored shade indicates
the cutoffs of methylated (>60%) and unmethylated (<5%) promoters. Due to the
different distribution of methylation level in the promoters of DKOA1 cells, >30% and
<5% are used as cutoffs for methylated and unmethylated promoters respectively.
99
A
B
C
Figure 4.4 DNA methylation is anti-correlated with nucleosome phasing and
accessibility in CGI promoters. NOMe-seq reads were aligned to 15,692 annotated
CGI TSS and promoters were categorized based on the methylation levels in both cell
types as (A) Unmethylated in HCT116 and Unmethylated in DKO1 (UU), (B)
Methylated in HCT116 and Methylated in DKO1 (MM) and (C) Methylated in HCT116
and Unmethylated in DKO1 (MU). The number of promoters that fall in each class is
shown on the left. Heatmaps were generated for DNA methylation (left panel) and
accessibility (right panel) +/-1kb from the TSS and clustering of each row was done
based on the accessibility pattern of DKO1 cells (right most panel).

100
The loss of DNA methylation in CGI promoters results in the acquisition of active
and poised chromatin state.
To better understand the functional relevance of striking changes in nucleosome
reorganization in the absence of DNA methylation, we further analyzed the histone
modification and gene expression patterns of CGI promoters in the different methylation
classes. UU promoters were invariant between HCT116 and DKO1 cells, and on average,
had a strong enrichment of H2A.Z variant, the active mark H3K27ac, and the permissive
marks H3K4me3 and H3K4me1 (Figure 4.5a). The chromHMM states and expression
were also largely invariant, with most promoters having high expression and active
promoter status (Figure 4.5b-c, left panel). In contrast, CGI promoters that were
methylated in HCT116 (the MM and MU classes) were largely devoid of all active,
permissive, and repressive histone marks (Figure 4.5b-c, middle and right panel). In the
parent cells, these promoters corresponded to the inactive promoter status, and were
overwhelmingly associated with unexpressed genes (Figure 4.5b-c, middle and right
panel). Due to the small numbers of regions that retained methylation in DKO1 cells,
subdividing promoters cleanly into MM and MU classes was challenging (Figure 4.3b).
Nevertheless, we found that despite retaining only low levels of methylation, the MM
promoters had much less dramatic histone and expression changes in DKO1 than the MU
promoters. The MM promoters exhibited only a minimal change in active and permissive
histone marks in DKO1 cells compared to HCT116 while the MU promoters had very
strong increases of the permissive H2A.Z, H3K4me3 and H3K4me1 as well as the
repressive H3K27me3 mark (Figure 4.5a, middle and right panel). The re-acquisition of
H3K27me3, but not H3K9me3 mark, in the absence of DNA methylation was consistent
101
with earlier reports (Jin et al., 2009; Komashko and Farnham, 2010), and particularly
intriguing due to the association many of these promoters with the poised or bivalent
promoters found in embryonic stem cells (Ohm et al., 2007; Schlesinger et al., 2007;
Widschwendter et al., 2007). Another distinguishing feature of the poised promoter state
is the presence of H3K4me1 directly over the TSS as compared to active promoters such
as seen in the UU class where there is a trough of H3K4me1 over the TSSs (Figure 4.5b,
left and right panel). This spatial pattern was consistent with earlier studies (Hawkins et
al., 2010), and suggested that a significant fraction of the MU genes were gaining a
poised chromatin state in the absence of DNA methylation.

102
A
B
C
Figure 4.5 Hypomethylation of CGI promoters triggers dramatic chromatin
remodeling. (A) DNA methylation, accessibility and enrichment level of histone marks
is shown +/-3kb around the TSS as the average of all promoters in each class.
Enrichment level, expressed in terms of z-score was calculated based on normalized
experimental wiggler value compared to the input. (B) Distribution of chromatin states
for each promoter class in both cell types is shown as a bar chart. Chromatin states of
promoters are defined based on the chromHMM model. Others include chromatin states
covering various enhancers, transcribed and heterochromatic regions. (C) Transcript
level (based on FPKM) for each promoter class is shown for two biological duplicates
for both cell types.

103
Demethylated CGI promoters uniformly gain well-phased nucleosomes and fall into
two distinct chromatin states, distinguished by active and/or poised modification at
H3K27 that recapitulate normal-like chromatin states.
We next sought to characterize the chromatin changes in the MU promoter class
in more detail. We observed that the MU class consisted of two subclusters, one with
nucleosome phasing only (NP) and a second with clear nucleosome depleted regions
(NDR) in DKO1 cells (Figure 4.4c). These two subclasses exhibited distinct histone
modification patterns, with the NP class having a significantly stronger enrichment of the
repressive H3K27me3 mark and poised promoter state compared to the NDR class which
in turn had a higher average level of active H3K27ac mark (Figure 4.6a). Our result
suggests that nucleosome phasing may an unanticipated feature of polycomb-repressed
promoters and that nucleosome organization is not unique to actively-transcribing
promoters. We also observed the different H3K4me1 enrichment pattern between the two
classes of promoters with the NDR class showing more of the trough characteristic of
active promoters (Figure 4.6a). These general histone patterns are clear in individual loci
such as the CYP4X1 promoter (NP) which contains a number of phased
mononucleosomes as determined by our NOMe-seq HMM calling (Figure 4.7a, Methods)
and the ZNF214 promoter (NDR) which contains a single large NDR region (Figure
4.7b). More than half of genes associated with the NDR class including ZNF214 gained
expression in DKO1 cells, a much higher percentage than genes associated with NP
promoters which remained poised (Figure 4.6b-c). This result is consistent with previous
observations that gene reactivation following DNA demethylation requires the formation
of accessible or NDR region at the TSS (Lin et al., 2007; Yang et al., 2012).
104
The acquisition of active and poised promoter states in DKO1 cells highlights the
critical role of polycomb as an alternative repressive mechanism in the absence of DNA
methylation and may further shed light on the distinct mechanisms governing cancer
epigenetic switch (Gal-Yam et al., 2008; Jin et al., 2009). The targeting of normally
polycomb-repressed promoters for DNA methylation in cancer is well-documented, and
our results illustrate that this process may be reversed by the removal of DNA
methylation (Figure 4.8a) Remarkably, we found that both the histone and expression
patterns of the NP and NDR subclasses in DKO1 cells recapitulated patterns present in
the normal colonic mucosa (Figure 4.8a-c). Specifically, the two subclasses of promoters
were distinguished by the H3K27 state, with the NDR promoters associated with genes
that were more highly expressed in the colonic mucosa compared to the NP promoters
(Figure 4.8b-c). These NDR promoters interestingly were also enriched for binding sites
of the Sp1 transcription factor (Figure 4.9), which have been shown to protect CGI
promoters from methylation in both normal and cancer cells (Berman et al., 2012b;
Gebhard et al., 2010). These results strongly suggests that cancer-specific DNA
methylation has a role in suppressing both the active and poised CGI promoter states, and
that these normal states can be recovered with the removal of DNA methylation. We
furthermore noted that the NP class which were methylated in HCT116 cells contained
the majority of promoters poised in normal colonic mucosa whereas the NDR class
contained a small fraction of all active promoters in colonic mucosa. This supports the
idea that poised promoters may undergo a coordinated epigenetic switch from
poised/polycomb state to DNA methylated, but that methylation of active promoters
105
likely involves an additional sequence-specific binding factor to target specific CGIs
(Gal-Yam et al., 2008).

106
A
B
C
Figure 4.6 Demethylated CGI promoters fall into two distinct chromatin states,
distinguished by active and/or poised modification at H3K27. A) MU promoters
(Figure 4.4C) were separated into two clusters based on whether they gain nucleosome
phasing (NP) or nucleosome-depleted regions (NDR). Enrichment level of each histone
mark, expressed in terms of z-score is shown +/-3kb around the TSS as the average of
all promoters in clusters NP and NDR. B) Distribution of chromatin states and C)
transcript level for each promoter class in both cell types is shown as a bar chart.
107
A
B
Figure 4.7 Locus-specific example of chromatin changes seen in NP and NDR
promoters. A) IGV browser shot of promoters belonging in NP and B) NDR cluster
showing the methylation and accessibility level as well as enrichment for histone marks.
MARs indicate methyltransferase accessible regions and mono-nucleosome was called
based on the HMM model.
108
A
B
C
Figure 4.8 DKO1 cells gain a normal-like chromatin landscape at CGI promoters
in the absence of DNA methylation. A) Levels of methylation, accessibility of each
histone mark, expressed in terms of z-score is shown +/-3kb around the TSS as the
average of all promoters in each class for colonic mucosa (CM), HCT116 and DKO1
cells at MU promoters that gain nucleosome phasing (NP) and accessibility (NDR). B)
Distribution of chromatin states and C) transcript level for each promoter class in all
cell type is shown as a bar chart.
109
A
B
C
Figure 4.9. Unmethylated and accessible CGI TSSs are enriched for general
transcription factor motifs. A) Frequencies of transcription factor motifs near UU, B)
MM and C) MU promoters in CGIs were calculated using HOMER and motifs with
frequency peak > 0.03 are shown. Coordinate of the motifs were obtained from
Factorbook.
110
The loss of DNA methylation does not alter the chromatin structure of non-CGI
promoters.
Non-CGI promoters are known to have very different regulatory characteristics
compared to CGI promoters: they are expressed in a much more tissue-specific pattern,
more likely to be associated with TATA boxes and transcripts starting from a particular
base pair rather than many possible initiation sites within a “broad” region (Rach et al.,
2011), and they are almost never repressed by H3K27me3 (Mohn et al., 2008). Our
results here demonstrate another important difference between the two classes of
promoters: unlike CGI promoters, the overwhelming loss of DNA methylation in DKO1
cells did not appear to have significant effects on the nucleosome architecture or
chromatin composition of non-CGI promoters. The overwhelming majority of non-CGI
promoters (85%) corresponded to the MU state, with UU and MM promoters being much
less frequent (Figure 4.10a-c, left panel). In stark contrast to the wide-spread accessibility
seen in CGI promoters, only a very small subset of the promoters were accessible (Figure
4.10a, right panel). However, we surprisingly observed that the majority of MU
promoters in non-CGI exhibited weaker, but organized arrays of nucleosomes in HCT116
despite their high DNA methylation levels (Figure 4.10c). This phased nucleosome
pattern was maintained in both the 5’ and 3’ direction and became more defined in DKO1
cells with the apparent increase of accessibility in the linker regions. Well-positioned
nucleosomes have often been described as a feature of unmethylated and permissive
promoters, but the co-occurrence of DNA methylation and weak nucleosome phasing
suggests that the organization of nucleosomes in non-CGI promoters occurs independent
of DNA methylation status.
111
Also in contrast to CGI promoters, the loss of DNA methylation did not appear to
directly influence the histone modification patterns of non-CGI promoters. Generally, UU
promoters, specifically those exhibiting accessibility, were marked by permissive histone
modifications in both cell types (Figure 4.11a). Only a small fraction of these promoters,
however, had NDRs and were expressed (Figure 4.11b-c, left panel). Meanwhile, MM
promoters were devoid of both permissive as well as repressive histone marks and were
largely inactive (Figure 4.11a-c, middle panel). Unlike our findings at CGI promoters, the
loss of DNA methylation at non-CGI promoters did not result in dramatic shifts in
histone modification patterns, as they gained neither active nor poised histone marks, and
thus remained largely unexpressed in DKO1 cells (Figure 4.11a-c, right panel). This
observation suggests that the loss of DNA methylation alone is insufficient to remodel
and reactivate non-CGI promoters and that DNA methylation itself may not directly
modulate the chromatin landscape of non-CGI promoters.

112
A
B
C
Figure 4.10 DNA methylation does not negatively regulate nucleosome
organization in non-CGI promoters. NOMe-seq reads were aligned to 7,191
annotated non-CGI TSS and promoters were categorized based on the methylation
levels in both cell types as (A) Unmethylated in HCT116 and Unmethylated in DKO1
(UU), (B) Methylated in HCT116 and Methylated in DKO1 (MM) and (C) Methylated
in HCT116 and Unmethylated in DKO1 (MU). The number of promoters that fall in
each class is shown on the left. Heatmaps were generated for DNA methylation (left
panel) and accessibility (right panel) +/-1kb from the TSS and clustering of each row
was done based on the accessibility pattern of DKO1 cells (right most panel).
113
A
B
C
Figure 4.11 The global loss of DNA methylation does not trigger chromatin
remodeling in non-CGI promoters. (A) DNA methylation, accessibility and
enrichment level of histone marks is shown +/-3kb around the TSS as the average of all
promoters in each class. Enrichment level, expressed in terms of z-score was calculated
based on normalized experimental wiggler value compared to the input. (B) Distribution
of chromatin states for each promoter class in both cell types is shown as a bar chart.
Chromatin states of promoters are defined based on the chromHMM model. Others
include chromatin states covering various enhancers, transcribed and heterochromatic
regions. (C) Transcript level (based on FPKM) for each promoter class is shown for two
biological duplicates for both cell types.

114
Long H3K9me3-marked heterochromatin domains have lower chromatin
accessibility and partially methylated DNA.
Having established the varying roles of DNA methylation in regulating the focal
chromatin structure of enhancers and promoters, we next inspected the effects of
genome-wide loss of DNA methylation on the long-range chromatin landscape.
We calculated methylation and accessibility levels of non-overlapping 1 Mb windows
across the genome while excluding CGIs. In order to investigate global changes in
accessibility, we ranked all windows based on the average GpC accessibility signals in
DKO1 cells (Figure 4.12a). At this scale, we found that the accessibility changes were
not dramatic, and accessibility in HCT116 and DKO1 were highly correlated. In both cell
types, we also observed that a subset of windows with the lowest accessibility
surprisingly had much lower CpG methylation levels. These regions of lower
accessibility coincided with partially methylated domains (PMDs), which had previously
been linked to heterochromatic regions in cancer (Figure 4.12b) (Berman et al., 2012b;
Hansen et al., 2011b; Hon et al., 2012; Hovestadt et al., 2014; Lister et al., 2011). We
examined the long-range distribution of histone marks and indeed observed that these
windows of low accessibility were correlated with the presence of H3K9me3 in both cell
types (Figure 4.12b). Consistent with previous reports, these H3K9me3-associated PMDs
were mutually exclusive with H3K27me3 repressive domains as well as permissive
H3K4me1 domains (Gifford et al., 2013; Hon et al., 2012), but strikingly not H3K4me3
blocks (data not shown). Furthermore, it was apparent from a representative 33 Mb
region of chromosome 2 that residual DNA methylation in DKO1 cells is preferentially
retained outside of PMDs, and that the most dramatic changes in accessibility occur
115
within the boundaries of H3K9me3-enriched PMDs (Figure 4.12b), thus demonstrating
the compact and inaccessible structure of heterochromatic regions in cancer cells.

116
A
B
Figure 4.12 Long H3K9me3-marked heterochromatin domains have lower
chromatin accessibility and partially methylated DNA. a) Ranked dot plot is shown
to describe long-range chromatin changes. Each dot or data point represents the average
levels of methylation, accessibility and z-scores (y-axes) for a 1Mb genomic window
which exclude CGI (n=3,089). The data points are plotted and ordered based on the
accessibility level of DKO1 cells (*). B) IGV browser shot of a 33 mb genomic window
on chromosome 2 showing the levels of methylation, accessibility and H3K9me3
enrichment.
117
Discussion
Overwhelming evidence shows that investigating the role of epigenetics in
transcriptional regulation is key to understanding the mechanism governing the
establishment of normal mammalian phenotypes as well as diseases (Baylin and Jones,
2011). DNA methylation has been studied extensively through the lens of CpG island
(CGI) promoters whose patterns of aberrant methylation are linked to cancer (Baylin and
Jones, 2011; Deaton and Bird, 2011; Tazi and Bird, 1990). However, the remarkable
growth in the field of epigenomics, propelled by advances in high-throughput genomic
sequencing, reveals that the functions of DNA methylation may be more nuanced than
previously understood and are part of integrated epigenetic states that regulate both focal
and long range chromatin elements (Bergman and Cedar, 2013; Jones, 2012; Rivera and
Ren, 2013). It has become clear that understanding the specific role DNA methylation
plays in gene regulation thus requires a holistic examination of its interactions with other
epigenetic mechanisms, including histone modifications and nucleosome positioning.
Well-spaced arrays of nucleosomes in promoters have been shown to accompany
transcription initiation in diverse eukaryotes, but our knowledge of how DNA
methylation contributes to nucleosome organization in mammalian promoters remains
inconclusive (Berman et al., 2013; Chodavarapu et al., 2010; Kelly et al., 2012; Valouev
et al., 2011). The NOMe-seq approach, which simultaneously maps DNA methylation
and nucleosome positioning, gives us an unanticipated view of this interaction. The
relationship between DNA methylation and nucleosome organization in colon cancer
HCT116 cells highlights fundamental differences between CpG Island (CGI) and non-
CGI promoters. At CGI promoters, there is an almost complete association between an
118
unmethylated state and the presence of organized nucleosomes, while non-CGI promoters
tend to have organized nucleosomal arrays regardless of methylation state. This phasing
of nucleosomal arrays at non-CGI promoters has not been described previously, but is
consistent with an intrinsic nucleosomal affinity of G/C rich sequences, primarily
transcription factor binding sites, found within 100 bp of non-CGI transcription start sites
(Tillo et al., 2010). This intrinsic affinity for nucleosomes is thought to restrict access by
transcription factors in most cell types, as these promoters are highly tissue specific and
only utilized in specific cell types. Our ability to identify these extended nucleosomal
arrays at non-CGI promoters highlights the potential power of single base-pair resolution
NOMe-seq.
Further underscoring differences between promoters types in the mammalian
genome, we found that the loss of DNA methylation in DKO1 cells led to the
reorganization of nucleosomes at CGI promoters, but not non-CGI promoters. Upon the
loss of DNA methylation, CGI promoters that were methylated in HCT116 cells gained
nucleosome phasing as well active and poised promoter marks, whereas non-CGI
promoters do not. CGI promoters gaining nucleosome depleted regions (NDRs)
significantly gained permissive and active promoter marks while promoters that gained
only nucleosome phasing gained polycomb-repressive as well as permissive marks to
establish a poised chromatin state. Strikingly, the histone modification patterns of CGI
promoters in DKO1 cells recapitulate similar context-specific patterns observed in
normal colonic mucosa, suggesting that DNA methylation has the ability to suppress both
active and poised CGI promoters during tumorigenesis, and that this suppression can be
reversed by removing DNA methylation. The widespread methylation of polycomb-
119
regulated promoters in HCT116 cells is consistent with the global epigenetic switch by
which DNA methylation permanently silences poised promoters (Baylin and Jones, 2011;
Gal-Yam et al., 2008), through a process involving oxidative damages occurring at
poised promoters (O'Hagan et al., 2011). In contrast, the silencing of active promoters
occurs much less frequently, and likely involves additional sequence-specific binding
factors.
The presence of phased nucleosomal arrays observed at inactive non-CGI
promoters as well as poised CGI promoters in DKO1 is previously undescribed, and the
underlying mechanisms responsible for the generation of the strikingly symmetrical
arrays require further investigation. This phenomenon, along with the absence of
chromatin remodeling in non-CGI promoters following DNA methylation loss suggests
that DNA methylation may have little contribution in the organization of nucleosomes
and histone modification patterns at non-CGI promoters. The masking of nucleosome
phasing and compaction of CGI promoters by DNA methylation, on the other hand, is
presumably mediated by the recruitment of methyl-binding proteins containing CXXC
domains, which are targeted to CpG rich regions (Nan et al., 1998). Our findings
furthermore suggest that the transcriptional regulation of CGI promoters in the absence of
DNA methylation is mediated by polycomb while the repression of non-CGI promoters
in DKO1 cells is mediated by the presence of well-positioned nucleosomes around the
TSSs (Han et al., 2011). DNA methylation thus may permanently silences transcription
by modulating the poised chromatin state in CGI and by directly targeting the inherently
nucleosome occupied regions of non-CGI promoters, revealing a fundamentally different
logic in the regulation of CGI and non-CGI promoters (Cairns, 2009; Tazi and Bird,
120
1990). Indeed, the selective reestablishment of poised and active chromatin in CGI also
suggests that activation potentials of CGI promoters are facilitated by the overall poised
chromatin structure whereas activities of non-CGI promoters may depend more on other
factors such as the availability of tissue-specific transcription factors and ATP-dependent
remodeling complexes (Ramirez-Carrozzi et al., 2009).
Our data also shows the strong association between H3K9me3 domains and low
accessibility in heterochromatic regions. Interestingly, we noted that many of the
H3K9me3 domains became shorter in DKO1 cells compared to the parent cells
(unpublished observation). This result is consistent with a reversal of the epigenetic
switch from H3K27me3 in normal epithelial cells to H3K9me3 in cancer cells (Gal-Yam
et al., 2008), back to H3K27me in DKO1 cells. Previous reports have shown that
domains of H3K27me3 specifically flank the edges of H3K9me3 heterochromatic
domains and a loss of H3K27me3 in HCT116 cells thus may be consistent with an
expansion of H3K9me3 heterochromatin, along with its associated reduction in GpC
accessibility level and DNA hypomethylation.
Taken together, our study reveals the context-dependent roles of DNA
methylation in regulating focal and long-range chromatin landscape. Understanding the
mechanisms of these differences not only will contribute to better understanding as to
how DNA methylation changes influence the chromatin structure and ultimately, gene
expression of normal and cancer cells, but will also have critical implications for DNA
demethylating therapies.

121
CHAPTER 5
THE LOSS OF DNA METHYLATION CONTRIBUTES TO THE REVERSAL OF
EPIGENETIC SILENCING IN MYOD1 ENHANCER/PROMOTER PAIR IN
CANCER CELLS

INTRODUCTION

DNA methylation, histone modifications and variants, together with nucleosome
positioning collectively establish an epigenetic landscape that governs normal cellular
functions. Aberration of the epigenetic landscape, in particular the most studied DNA
methylation, deregulates biological signaling pathway and is increasingly being
recognized as a hallmark of cancer (Baylin and Jones, 2011). Previous studies have
revealed that specific type of CGI promoters are significantly predisposed to silencing by
DNA methylation in cancer cells (Gal-Yam et al., 2008; Ohm et al., 2007; Schlesinger et
al., 2007; Widschwendter et al., 2007). This process involves a change from gene
suppression by the Polycomb repressive complex (PRC) in normal cells to DNA
hypermethylation in cancer cells and is commonly termed “epigenetic switching”. Our
own study supports this phenomenon and more importantly, illustrates the reversal of the
epigenetic switch when DNA methylation is removed (Chapter 4), and thus emphasizing
the tenet that silencing by DNA methylation is reversible and can be a target for therapy
(Chapter 1) (Yang et al., 2010; Yoo and Jones, 2006).
Most cancer epigenetic studies, including our own as described in the previous
chapter, has focused to a large degree on the silencing of CGI promoters which has since
been accepted as a common signature of cancer (Deaton and Bird, 2011). In the last
122
decade however, a class of distal regulatory element, specifically enhancers, has emerged
as key players in gene regulation during mammalian development and their dysregulation
inevitably contribute to human malignancies (Heintzman et al., 2007; Hnisz et al., 2013;
Whyte et al., 2013). In normal cells, enhancer/promoter interaction may act as an
additional layer of control for the expression of genes which also include those that are
repressed by PRC in non-expressing cells or otherwise expressed in tissue-specific
manner. Interestingly and despite PRC repression, these genes may maintain regulatory
flexibility and facilitate reprogramming through their permissive enhancers (Taberlay et
al., 2011). To date however, it has been technically difficult to identify
enhancer/promoter pairs and thus, the temporal effects of enhancers on gene expression,
since they can be located at varied distances from transcription start sites (TSS) (Calo and
Wysocka, 2013; Rada-Iglesias et al., 2012). Moreover, one enhancer can also interact
with many promoters and conversely, one promoter can be regulated by multiple
enhancers (Kieffer-Kwon et al., 2013).
Here, we focus on the tissue-specific self-regulatory MYOD1 gene as a model for
understanding how enhancer/promoter pairs are epigenetically altered in cancer. MYOD1
is unique in that it has a well-characterized enhancer located ~20kB upstream of the TSS
and contains a minimal core region of 258 base pairs (bp) that is necessary for promoter
activity and reprogramming (Goldhamer et al., 1995; Taberlay et al., 2011). MYOD1 is
specifically expressed in myoblasts, but generally repressed by PRC and H3K27me3 in
normal non-muscle cells and commonly silenced by DNA methylation in cancers,
making this gene an optimal choice for investigating the epigenetic silencing of an
enhancer/promoter pair (Hiranuma et al., 2004). Not unlike promoters, we show that
123
MYOD1 enhancer can also be epigenetically silenced in cancer cells. Thus, we apply both
genetic and pharmacological approaches to remove DNA methylation in HCT116 colon
cancer cells and examine the dynamic changes in the chromatin landscape of MYOD1
regulatory regions with the goal to better understand the enhancer/promoter crosstalk that
occur during tumorigenesis.

124
MATERIALS AND METHODS
Cell Culture
Colorectal cancer cells (HCT116) were obtained from the American Type Culture
Collection (ATCC) and cultured under recommended conditions. DKO1 cells were
cultured in McCoy’s 5A medium with 10% fetal calf serum and 2%
penicillin/streptomycin solution, as described previously.

5-Aza-2’-deoxycytidine Treatments
HCT116 cells were treated with 0.3µM 5-Aza-2’-deoxycytidine (5-Aza-CdR) and
vehicle control. After 24 hours, medium was replaced and cells were cultured for 42 days
in McCoy’s 5A medium supplemented with 10% fetal calf serum and 2%
penicillin/streptomycin solution, without 5-Aza-CdR.

RNA Isolation and Quantitative PCR Analysis
RNA was isolated using Trizol reagent according to manufacturer’s instruction.
RNA was treated with DNase I, and reverse transcribed with the iScript cDNA synthesis
kit (BioRad). Amplification of cDNA was performed on the an CFX96 Real-time PCR
detection system (BioRad) using KAPA SYBR Fast qPCR Mix (Kapa Biosystems) and
the following conditions: 95ºC for 3 mins, followed by 45 cycles of 95°C for 3 seconds
then 60ºC for 30 seconds. A melt curve analysis was performed (60-95°C, rising by 1°C
every 5 seconds). All primers are listed in Table 5.1. Analyses were conducted in parallel
using human GAPDH mRNA primers for normalization. A standard curve was generated
for each primer set with serial dilutions of either a GAPDH expression plasmid or PCR
125
products to correlate the threshold (Ct) values from the amplification plots to copy
number.

NOMe-seq
Locus-specific NOMe-seq was performed as previously described (Chapter 4 and
Appendix) (Kelly et al., 2012; You et al., 2011).

ChIP-qPCR
ChIP assays were performed as previously described using 50 ug of chromatin
and the following antibodies: EZH2 (#39639, Active Motif), H2A.Z (Abcam, ab4174),
H3K4me1 (Active Motif, 39298), H3K27me3 (Active Motif, 39155). Quantitative PCR
(qPCR) for ChIP was performed as described for mRNA analyses. For each PCR, DNA
standards were included for quantitation. Immunoprecipitated DNA was calculated as a
percentage of input DNA. All primers used are listed in Table 5.1

126
Table 5.1. Primer List
Locus Sequence
Expression
MYOD1
Forward
Reverse

CGCCAGGATATGGAGCTACTC
TCGAAACACGGGTCGTCAT
GAPDH
Forward
Reverse

TGAAGGTCGGAGTCAACGG
AGAGTTAAAAGCAGCCCTGGTC
ChIP
MYOD1 P1
Forward
Reverse

GGTGTTGGAGAGGTTTGGAA
AGTCCGAGGCCAATAGGAAC
MYOD1 P2
Forward
Reverse

TCAGGCCGGACAGGAGAG
CCCGGCTGTAGATAGCAAAGTG
MYOD1 P3
Forward
Reverse
CGCCAGGATATGGAGCTACT
CGGGTCGTCATAGAAGTCGT
MYOD1 Enhancer
Forward
Reverse

CAGCCAAGTATCCTCCTCCA
AAGCTGAGCACTCTGGGAGA

127
RESULTS
The MYOD1 enhancer/promoter pair acquires the multivalent epigenetic signature
in cells with reduced DNA methylation
We first investigated the role of DNA methylation in silencing enhancers using
HCT116 colon cancer cell line and its DNMT3B/DNMT1
∆Exon2-5
genetic knockout
counterpart, DKO1 cells which have a global methylation level of ~5% (Gal-Yam et al.,
2006; Rhee et al., 2002). MYOD1 enhancer is located ~20kb upstream of the promoter
and in expressing cells, is marked by the H3K4me1 and H3K27ac signatures (Taberlay et
al., 2011). HCT116 cells, however, appear to lack the permissive marks at both promoter
and enhancer regions, indicating that it may be inactive (Figure 5.1). We confirmed that
MYOD1 was not expressed in HCT116 cells where the promoter and enhancer are highly
methylated (Figure 5.2a). Interestingly, these regions are unmethylated in DKO1 cells,
but remain inactive, suggesting that an alternative mechanism may play a role in
maintaining gene repression in the absence of DNA methylation.
We examined whether PRC was responsible for the alternate repression of
MYOD1 in DKO1 cells. While the enhancer and promoter regions were largely devoid of
EZH2 binding and H3K27me3 enrichment in HCT116 cells, these marks were
significantly more enriched at the MYOD1 promoter in DKO1 cells (Figure 5.2b), which
still generally lack H3K4me3 mark, consistent with inactive transcription (data not
shown). We also found that H2A.Z was present in the unmethylated MYOD1 promoter
and enhancer in DKO1 cells, while conspicuously absent in the methylated HCT116
cells, which is due to the known antagonism between DNA methylation and H2A.Z
(Coleman-Derr and Zilberman, 2012; Zilberman et al., 2008). H3K4me1 was also
128
detected at the MYOD1 enhancer and promoter in DKO1 cells, but not HCT116 cells,
(Figure 5.2b), consistent with a poised transcriptional state. The global removal of DNA
methylation thus illustrates that the process of epigenetic switching may be reversible and
occurs concomitant with re-establishment of a permissive enhancer epigenetic signature
(Taberlay et al., 2011).

129
Scale
chr11:
CpG Islands
DNase Clusters
RepeatMasker
Common SNPs(138)
Sequences
SNPs
Human mRNAs
Spliced ESTs
Txn Factor ChIP
Rhesus
Lamprey
10 kb
hg19
17,725,000 17,730,000 17,735,000 17,740,000
UCSC Genes (RefSeq, GenBank, CCDS, Rfam, tRNAs & Comparative Genomics)
CpG Islands (Islands < 300 Bases are Light Green)
HCT-116 H3K4me3 Histone Mod ChIP-seq Raw Sig 1 from ENCODE/UW
HCT-116 H3K4me1 Histone Modifications by ChIP-seq Signal from ENCODE/SYDH
HCT-116 H3K27ac Histone Modifications by ChIP-seq Signal from ENCODE/SYDH
H3K4Me3 Mark (Often Found Near Promoters) on 7 cell lines from ENCODE
H3K4Me1 Mark (Often Found Near Regulatory Elements) on 7 cell lines from ENCODE
H3K27Ac Mark (Often Found Near Active Regulatory Elements) on 7 cell lines from ENCODE
HCT-116 CTCF 5916 v042211.1 ChIP-seq Raw Signal Rep 2 from ENCODE/HAIB
HCT-116 Pol2 UC Davis ChIP-seq Signal from ENCODE/SYDH
Digital DNaseI Hypersensitivity Clusters in 125 cell types from ENCODE
GM12878 H2A.Z Histone Mods by ChIP-seq Signal from ENCODE/Broad
GM12878 H3K27me3 Histone Mods by ChIP-seq Signal from ENCODE/Broad
H1-hESC H2A.Z Histone Mods by ChIP-seq Signal from ENCODE/Broad
H1-hESC H3K27me3 Histone Mods by ChIP-seq Signal from ENCODE/Broad
K562 H2A.Z Histone Mods by ChIP-seq Signal from ENCODE/Broad
RefSeq Genes
K562 H3K27me3 Histone Mods by ChIP-seq Signal from ENCODE/Broad
HepG2 H2A.Z Histone Mods by ChIP-seq Signal from ENCODE/Broad
HepG2 H3K27me3 Histone Mods by ChIP-seq Signal from ENCODE/Broad
HMEC H2A.Z Histone Mods by ChIP-seq Signal from ENCODE/Broad
HMEC H3K27me3 Histone Mods by ChIP-seq Signal from ENCODE/Broad
MCF-7 H3K27me3 Histone Modifications by ChIP-seq Signal from ENCODE/SYDH
Repeating Elements by RepeatMasker
Simple Nucleotide Polymorphisms (dbSNP 138) Found in >= 1% of Samples
Publications: Sequences in Scientific Articles
Human mRNAs from GenBank
Human ESTs That Have Been Spliced
HCT-116 CTCF 5916 v042211.1 ChIP-seq Raw Signal Rep 1 from ENCODE/HAIB
Your Sequence from Blat Search
Transcription Factor ChIP-seq (161 factors) from ENCODE with Factorbook Motifs
100 vertebrates Basewise Conservation by PhyloP
Multiz Alignments of 100 Vertebrates
MYOD1
MYOD1
YourSeq
HCT16 H3K4M3 Sg 1
100 _
1 _
HCT-116 H3K4me1
20 _
3 _
HCT-116 H3K27ac
20 _
3 _
Layered H3K4Me3
150 _
0 _
Layered H3K4Me1
50 _
0 _
Layered H3K27Ac
100 _
0 _
HCT CTCF V11 2
5 _
0 _
HCT Pol2 UCD
40 _
3 _
GM12878 H2A.Z
50 _
1 _
GM12878 H3K27m3
50 _
1 _
H1-hESC H2A.Z
50 _
1 _
H1-hESC H3K27m3
50 _
1 _
K562 H2A.Z
50 _
1 _
K562 H3K27m3
50 _
1 _
HepG2 H2A.Z
50 _
1 _
HepG2 H3K27m3
50 _
1 _
HMEC H2A.Z
50 _
1 _
HMEC H3K27m3
50 _
1 _
MCF-7 H3K27me3
20 _
3 _
HCT CTCF V11 1
5 _
0 _
100 Vert. Cons
4.88 _
-4.5 _
Figure 5.1 MYOD1 is silenced in HCT116 colon cancer cells. Screenshot of UCSC
genome browser of the MYOD1 promoter and enhancer loci show the lack of
permissive of histone marks in in the non-expressing HCT116 cells. As a comparison,
some of the 7 ENCODE cell lines included show enrichment of permissive and active
histone modifications.
130
DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
0.0
0.1
0.2
0.3
0.4
An?‐EZH2 (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐EZH2 (% Input)
0.0
0.5
1.0
1.5
2.0
An?‐H3K27me3 (% Input)
0.0
1.0
2.0
3.0
An?‐H3K27me3 (% Input)
0.0
0.5
1.0
1.5
An?‐H2A.z (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐H2A.z (% Input)
0.0
0.5
1.0
1.5
2.0
2.5
An?‐K4me1 (% Input)
0.0
0.5
1.0
1.5
An?‐H3K4me1 (% Input)
Expression

Enhancer
DNA
Methyla5on

Promoter
DNA
Methyla5on

DKO1 Cells HCT116 Cells
Enhancer

P1

P2

P3

MYOD1

MIN

Legend
HCT116 HCT116

DKO1

20kB

DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
0.0
0.1
0.2
0.3
0.4
An?‐EZH2 (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐EZH2 (% Input)
0.0
0.5
1.0
1.5
2.0
An?‐H3K27me3 (% Input)
0.0
1.0
2.0
3.0
An?‐H3K27me3 (% Input)
0.0
0.5
1.0
1.5
An?‐H2A.z (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐H2A.z (% Input)
0.0
0.5
1.0
1.5
2.0
2.5
An?‐K4me1 (% Input)
0.0
0.5
1.0
1.5
An?‐H3K4me1 (% Input)
Enhancer P1 P2 P3 Enhancer P1 P2 P3
EZH2

H3K27me3

DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
DKO Cells HCT116 Cells
0.000
0.002
0.004
0.006
0
20
40
60
80
100
MYOD1 mRNA/GAPDH
DNA Methyla?on (%)
0.0
0.1
0.2
0.3
0.4
An?‐EZH2 (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐EZH2 (% Input)
0.0
0.5
1.0
1.5
2.0
An?‐H3K27me3 (% Input)
0.0
1.0
2.0
3.0
An?‐H3K27me3 (% Input)
0.0
0.5
1.0
1.5
An?‐H2A.z (% Input)
0.0
0.2
0.4
0.6
0.8
1.0
An?‐H2A.z (% Input)
0.0
0.5
1.0
1.5
2.0
2.5
An?‐K4me1 (% Input)
0.0
0.5
1.0
1.5
An?‐H3K4me1 (% Input)
Enhancer P1 P2 P3 Enhancer P1 P2 P3
H2A.Z

H3K4me1

A
B
Figure 5.2 Epigenetic switching at MYOD1 is reversible. (A)RNA was isolated,
reverse transcribed and analyzed by qPCR using primers detecting MYOD1. Data is
expressed as copies of MYOD1 relative to GAPDH expression (green bars). The mean
and SEM of 3 biological experiments is shown. The average percent DNA methylation in
HCT116 and DKO cells was determined by bisulfite genomic sequencing and is
expressed as the average of 3 CpG dinucleotides across the enhancer (black bars) and 33
CpG dinucleotides across the promoter (hatched bars). (B) ChIP assays were performed
with antibodies detecting EZH2, H3K27me3, H2A.Z and H3K4me1 on chromatin from
HCT116 (hatched bars) and DKO1 (red bars) cells. Data are presented as percent total
input. The mean and SEM of 3 biological experiments is shown.
131
Restoration of the Enhancer NDR accompanies the PRC repressed state in cells with
reduced DNA methylation
High-resolution NOMe-seq analysis shows that methylated MYOD1 promoter and
enhancer are, for the most part, occupied by nucleosomes in HCT116 cells (Figure 5.3a),
consistent with the predicted lack of DNAse hypersensitivity at inactive regulatory
regions (Lin et al., 2007; Xi et al., 2007). In contrast, MYOD1 enhancer shows a striking
region of accessibility in DKO1 cells (Figure 5.3b, left panel). This result concomitantly
validates our previous observation that most accessibility changes in DKO1 cells occur in
distal regulatory regions (Figure 4.1-Figure 4.2). An unexpected region of accessibility,
approximately covering the -2 nucleosome position near the TSS, was also detected in
DKO1 cells, suggesting that nucleosome density was lower in this region, whereas the
region immediately upstream of the TSS itself remains occupied by a nucleosome (Figure
5.2b, right panel). This state is highly similar to what we have previously observed in
normal human fibroblasts, indicating a reversal of epigenetic switching and re-
establishment of a poised chromatin state in the absence of DNA methylation in DKO1
cells. Thus far, this data suggests that a permissive epigenetic signature is lost when
MYOD1 is methylated in cancer cells, but can be restored after manipulation of DNA
methylation status.

132
Enhancer
(Nuclei
+
M.CviPI)

147bp

Promoter
(Nuclei
+
M.CviPI)

147bp

GpC
Methyla5on
(Accessibility)

DKO1
Cells
(Repressed)

Enhancer
(Endogenous
Methyla5on)
Promoter
(Endogenous
Methyla5on)

Enhancer
(Nuclei
+
M.CviPI)

147bp

Promoter
(Nuclei
+
M.CviPI)

147bp

GpC
Methyla5on
(Accessibility)

HCT116
Cells
(Silenced)

Enhancer
(Endogenous
Methyla5on)
Promoter
(Endogenous
Methyla5on)

MYOD1

MIN

A
B
Figure 5.3 A nucleosome depleted MYOD1 enhancer is established in cells with
reduced DNA methylation. A schematic diagram shows the location of the minimal
enhancer (MIN) and TSS (arrow). Nuclei were extracted from (A) HCT116 and (B)
DKO1 cells and either left untreated (black/white circles), or treated with M.CviPI GpC
methyltransferase (white/teal circles) and subjected to bisulfite conversion and cloning.
Each horizontal line represents an individual MYOD1 enhancer (left) or promoter
(right). For M.CviPI treated nuclei, circles represent GpC dinucleotides (white,
unmethylated and inaccessible to M.CviPI; teal, methylated and accessible to M.CviPI).
Pink bars are regions ≥147bp, representing sites associated with nucleosomes. Regions
accessible to M.CviPI (teal) represent NDRs. Diagram is drawn to scale.
133
Figure 5.4 A dynamic change in nucleosome occupancy occurs following 5-Aza-
CdR induced demethylation. HCT116 cells were treated with 0.3 µm 5-Aza-CdR for
24 hours and cultured for 42 days. NOMe-seq was performed at indicated time intervals
to monitor the changes in DNA methylation and nucleosome occupancy level at
MYOD1 promoter and enhancer. Each horizontal line represents an individual MYOD1
enhancer (left) or promoter (right). For M.CviPI treated nuclei, circles represent GpC
dinucleotides (white, unmethylated and inaccessible to M.CviPI; teal, methylated and
accessible to M.CviPI). Pink bars are regions ≥147bp, representing sites associated with
nucleosomes. Regions accessible to M.CviPI (teal) represent NDRs. Diagram is drawn
to scale.
134
Treatment of cancer cells with demethylation agents restores the enhancer NDR,
but is insufficient for transcriptional activation
To further investigate the dynamic reversal of epigenetic switching in the MYOD1
regulatory regions, we treated HCT116 cells with a low-dose of 5-Aza-2-deoxycitidine
(5-Aza-CdR) for 24 hours and monitored the post-treatment changes in DNA methylation
and nucleosome occupancy at the promoter and enhancer over the course of 42 days in
drug-free media. We confirmed previous report showing that MYOD1 never gain
expression following 5-Aza-CdR treatment despite losing methylation (data not shown)
(Yang et al., 2012). Here, we found that within 5 days of the drug treatment, MYOD1
enhancer has achieved complete demethylation (Figure 5.3, left panel). New NDRs are
formed concomitant with the loss of DNA methylation, suggesting the reacquisition of
permissive enhancer signature. By day 14 and 21, however, the enhancer region has
begun to regain its methylation level. Predictably, the enhancer also regains nucleosome
occupancy at these time points as nucleosome deposition has been shown to precede
DNA methylation, and indeed, a substrate for DNMTs (Sharma et al., 2011; You et al.,
2011). At day 42, the enhancer has completely regained its pre-drug treatment
nucleosome occupancy pattern, while the methylation has rebound to 70% of the initial
level, illustrating the highly dynamic response of distal regulatory regions to the
pharmacological perturbation of DNA methylation.
MYOD1 promoter is highly methylated in HCT116 colon cancer cells and in
contrast to the enhancer region, achieved maximum demethylation between day 14 and
21. Despite the reduced methylation level, the -1 position of the promoter generally is
still occupied by nucleosomes, with increased accessibility occurring mostly in the -2
135
position, reminiscent of the nucleosome status in DKO1 cells (Figure 5.2). This
nucleosome occupancy pattern is also consistent with the lack of MYOD1 reactivation
following the loss of DNA methylation. The methylation level of the promoter persists
around ~50% for the next 3 weeks when nucleosome occupancy pattern returns to its pre-
treated level. Taken together, this data also suggests that in response to pharmacological
perturbation of DNA methylation, promoter and enhancer regions may have different
dynamic rates of change in nucleosome reoocupation and rebound methylation.

136
DISCUSSION
Epigenetic silencing of CpG island promoters by DNA methylation in cancer cells
is well established, yet the status or involvement of distal regulatory regions remains
unknown. However, there is a growing evidence that disruption of enhancer function may
also drive tumorigenesis (Akhtar-Zaidi et al., 2012). The role of DNA methylation in the
establishment of this particular cancer signature, however, is still largely untapped. The
current prevailing view is that functioning enhancers have low and much more dynamic
methylation status compared to promoters (Stadler et al., 2011), and it is not an
improbable hypothesis that enhancers may also be silenced by DNA methylation.
Our previous genome-wide data has indicated that the removal DNA methylation
may restore normal-like epigenetic landscape in the promoter regions although the
detailed mechanism is unclear (Chapter 4). Using the well-characterized MYOD1 as a
model of gene-specific level study, we confirmed the global trend of reversal of
epigenetic switching we described in the previous chapter and further shed light on the
enhancer/promoter crosstalk in cancer. Previous work has shown that the enhancer of
PRC target genes permits these loci to remain permissive in somatic cells, thus retaining
epigenetic plasticity and allowing the cells to be reprogrammed (Taberlay et al., 2011).
This permissive characteristic is lost following epigenetic switching and permanent
silencing by DNA methylation and perhaps may explain why some cell types are resistant
to gene reprogramming while providing evidence that a demethylated state was important
for activity of key regulators (Boukamp et al., 1992). In DKO1 cells which has reduced
DNA methylation, we show that a poised chromatin state can be reestablished at the
137
regulatory regions of MYOD1 where the minimal enhancer gains overwhelming
accessibility and the promoter repressed by H3K27me3.
Pharmacological disruption of DNA methylation by 5-Aza-CdR further allows us
to probe the temporal and dynamic crosstalk between enhancer/promoter pair. MYOD1
has consistently been shown to lack accessibility and remain inactive following the loss
of DNA methylation (Yang et al., 2012). Here, we show that the MYOD1 enhancer is
more rapidly demethylated and depleted of nucleosomes compared to the promoter.
Reactivation of gene requires interaction between permissible promoter and enhancer and
in the case of MYOD1, the enhancer acquires its permissive state before the promoter is
sufficiently demethylated. Conversely, by the time the promoter is demethylated and
existing in the poised state, the enhancer has gradually become nucleosome occupied and
regained its DNA methylation status. This mismatched in the timing in which chromatin
remodeling takes place in the promoter and enhancer presumably inhibits the necessary
interactions for gene reactivation and thus contributes to the continuous repression of
MYOD1 in the absence of DNA methylation.
Taken together, our results suggest that activation potential may be determined by
the dynamic epigenetic profile of both regulatory elements and that DNA methylation
abrogates transcriptional competence by disrupting the crosstalk between
enhancer/promoter pairs. Removal of DNA methylation by genetic knockout or
demethylating agent may reestablish a permissive chromatin structure in both promoter
and enhancer, but is not sufficient for gene reactivation which requires much more
precise timing to allow for interactions between equally permissive regulatory elements.
This data thus illustrates an additional layer of control that may prevent non-specific gene
138
activation following 5-Aza-CdR treatment, and ultimately, can be taken into account
when designing therapeutic course which maximizes the temporal effects of the drug.

139
CHAPTER 6
MAPPING DNA METHYLATION AND NUCLEOSOME OCCUPANCY IN
PRIMARY HUMAN TISSUE

INTRODUCTION
The understanding that epigenetic processes may underlie health and disease
susceptibility has spurred large-scale collaborative projects to map the epigenomes across
normal and cancer tissues. Established in 2005, The Cancer Genome Atlas (TCGA)
which catalogues the genetic alterations across various cancer, has also elucidated the
strong correlation between somatic mutations and epigenetic changes, in particular DNA
methylation whose changes are measured globally using the Illumina Infinium
HumanMethylation27 and 450 platforms (Hinoue et al., 2012; Noushmehr et al., 2010).
Meanwhile, consortiums including ENCODE and The Roadmap Epigenomics Projects,
have spearheaded the production of reference epigenome including the mapping of
histone modifications in various primary and immortalized human cells and have become
an important resource in the study of epigenetic regulation (Beck et al., 2012; Bernstein
et al., 2010; Birney et al., 2007). To date, however, there have been no comprehensive
epigenome maps which directly compare the normal and cancer epigenomes at the single
base pair level.
One major caveat in many cell-line based epigenetic studies is the occurrence of
cell-culture specific epigenetic alterations which may not recapitulate in vivo conditions
and consequently limit our understanding of the true relationships between epigenetic
processes (Varley et al., 2013; Wilson and Jones, 1983). In our previous study using
colon cancer cell line model, we show a significant alteration occurring as a function of
140
changes in DNA methylation, specifically the establishment of normal-like active and
poised chromatin state in CGI promoters. These changes suggest that the absence of
DNA methylation may initiate a reversal of cancer-associated epigenetic switching. With
this dramatic result, it is even more pressing thus for us to determine whether our
observations truly recapitulate what goes on in primary human tissues.
Various protocols which outline the application of histone modification and DNA
methylation analyses on fresh uncultured human tissues have been published (Berman et
al., 2012b; Dahl and Collas, 2008). The occupancy and positioning of nucleosomes are
thought to be more dynamic compared to DNA methylation and thus our standard
NOMe-seq protocol requires treatment of freshly isolated nuclei to avoid experimental
artifacts. This proves challenging when it comes to primary human tissue as there is often
a lag between when the tissue is harvested from donor, examined by pathologists and
finally reaches the laboratory. Morever, we are often limited to very small amount of
samples which may not yield enough materials for genome-wide analysis. Here, we
discuss our experimental strategies to apply locus-specific and genome-wide NOMe-seq
on uncultured human tissue and circumvent the challenges associated with performing
epigenetic analyses on primary samples (Beck et al., 2012). Using human small intestine
and colon tissues, we adapted our standard NOMe-seq protocol for the treatment of fresh
and frozen human tissues and tested novel approach to generate whole-genome bisulfite
sequencing libraries using input DNA of less than 500ng for the mapping of DNA
methylation and nucleosome positioning of normal and tumor cells.

141
MATERIALS AND METHODS
Nucleosome footprinting of fresh human small intestine
Human small intestine was collected from bladder cancer patients undergoing
neobladder construction according to IRB protocol and institutional guidelines (Lay et
al., 2014). Fresh tissue samples were processed immediately upon collection and stored
on ice during transport. Briefly, the intestinal tissue was cut longitudinally, completely
removed of excess fecal matter and fatty tissue and subsequently washed with sterile ice-
cold PBS. The tissue was subsequently dissected into 1-2 cm pieces and transferred into a
conical tube containing sterile HBSS/30mM EDTA solution. The tissue suspension was
incubated for 30 minutes at 37°C with gentle rotation in order to detach the intestinal
epithelial cells from the lamina propria. Intestinal epithelial cells were further separated
from excess components of the tissue by filtering the suspension through sterile cell-
strainer with 100-µm pore size (BD). Following centrifugation, contaminating red blood
cells were lysed by incubating the cell pellet with 5 volumes of sterile 1x RBC buffer
(Sigma Aldrich) for 10 minutes at room temperature. Nuclei were then isolated by lysing
200,000 cells and immediately treated with M.CviPI for a total of 15 minutes reaction
time based on previously described conditions (see Appendix 1) (Kelly et al., 2012;
Taberlay et al., 2011). Genomic DNA was subsequently extracted, bisulfite converted,
PCR amplified and cloned into TA vector. Individual colonies were screened using the
M13 primers and colonies containing insert targeted to specific genomic regions were
sequenced using the M13 reverse primers. All primers used in this study are listed in
Table 5.1.

142
Nucleosome footprinting of frozen colon tumor
Human colon tumors were collected in accordance to institutional guidelines and
processed per TCGA standard requirements (2012; TCGA, 2012). The tissue was cut
and placed in a 2ml cryocentrifuge tube. The whole tube was submerged in isopentane at
-80°C for 60-90 seconds and stored in -80°C long-term. Detailed step-by-step processing
of the tissue for NOMe-seq can be found in Appendix 1. Briefly, colon tissue was cut into
1-3 mm pieces while still frozen and resuspended in DMEM media, using 10 mL of
media per gram of tissue in a conical tube. To cross-link, 37% formaldehyde was added
to a final concentration of 1% and the tube was gently rotated at room temperature for no
more than 15 minutes. A final concentration of cold 0.125M glycine was added to the
tube to stop the reaction and the tube was rotated for additional 5 minutes at room
temperature. After cross-linking, the tissue was washed twice with cold PBS. To prepare
the nuclei, the tissue was resupended in 1mL cold PBS per 100mg of tissue and dounced
for 20 strokes using a chilled dounce-homogenizer. After centrifugation and removal of
supernatant, the nuclei pellet was washed with ice-cold NOMe-seq wash buffer. Nuclei
were resuspended in 1X GpC buffer and sonicated for 3 cycles of 30s on and 30s off
using the Bioruptor system (Diagenode). NOMe-seq was performed as previously
described with the following modifications: the nuclei were incubated with 100U
M.CviPI and 1.5ul of SAM for 60 minutes followed by a boost of an additional 100U
M.CviPI and 0.75ul of SAM for 60 minutes twice before stopping the reaction. The
chromatin was reverse cross-linked overnight at 65°C and DNA was extracted using
standard phenol-chloroform protocol and ethanol precipitation.

143
Low-input generation of whole-genome bisulfite sequencing library
WGBS libraries were generated using the EpiGnome Methyl-Seq Kit (Epicentre).
Briefly, 50 ng of bisulfite converted DNA were denatured and DNA synthesis primer was
annealed at 95° C for 5 minutes. DNA copy was subsequently synthesized and excess
random primer was removed by exonuclease digestion. In the second round of DNA
synthesis, terminal-tagging oligo was annealed to generate di-tagged cDNA which was
subsequently purified using AMPure magnetic beads (BD) and whole-genome amplified
using high-fidelity Taq Polymerase, yielding adaptor-tagged whole genome bisulfite
sequencing libraries. Libraries were sequenced on HiSeq-2000 using the 75PE method
and raw data was processed based on previously described methods (Berman et al.,
2012b; Kelly et al., 2012; Liu et al., 2012).

144
Table 6.1. Primer List
Locus Sequence
WNT2
Forward
Reverse

GTTGAGAATTATTTTTGGATTT
AAACTTACCACCATAAAAAATT
VTRNA2-1 Amplicon 1
Forward
Reverse

GGGGAGATTTTATGGAG
ACTAAAACTTCTAAACCATAAAAAAATA
VTRNA2-1 Amplicon 2
Forward
Reverse

GGGAGGAATTGAGAGTTTTT
CTTTCTATCACACCTTCAAAATAAC
M13
Forward
Reverse

GTAAAACGACGGCCAGT
AACAGCTATGACCATG

145
RESULTS
Mapping DNA methylation and nucleosome occupancy in uncultured human tissue
We initiated a pilot experiment to apply NOMe-seq on primary human tissue by
using freshly isolated small intestine tissue. Normal small intestine was surgically
removed from patients who undergo neobladder construction (see Chapter 2), and excess
tissue was immediately processed following pathological examination. Intestinal
epithelial cells were purified based on previously described method and subsequently
lysed and treated with M.CviPI as outlined previously (Kelly et al., 2012; Lay et al.,
2014). We used our standard experimental protocol (Appendix) and were able to
accurately footprint the high accessibility pattern of the GRP78 locus for two independent
biological samples (Figure 2a-b).
As a control region, we examined the non-coding VTRNA2-1 locus whose
individual specific methylation pattern may have prognostic value in determining cell
survival (Lay et al., 2014; Romanelli et al., 2014; Treppendahl et al., 2012b). Here, we
confirm that VTRNA2-1 can be monoallelically methylated (Figure 6.2a, left panel) or
biallelically unmethylated in normal small intestine in individual specific manner (Figure
6.2b, right panel). Sample 11048 is monoallelically methylated and show distinct
chromatin configurations in that the unmethylated strands are accessible at the -1 position
whereas the methylated strands are inaccessible (Figure 6.2a), consistent with our
previous report that specific genomic loci may contain diverging chromatin structures
(Kelly et al., 2012). Sample 11253 on the other hand, is unmethylated and largely
accessible at the -1 position of the TSS (Figure 6.2b). Interestingly, both biological
samples also show a distinct region of inaccessibility within the -2 position, independent
146
of their endogenous DNA methylation status as well as the -1 accessibility level. This
protection, however, is not large enough to accommodate a nucleosome and may be due
to a DNA binding complex that is stably and consistently bound in this locus. The
epigenetic variation between individuals is a critical component in understanding disease
susceptibility which is not captured in cell line model. This experiment thus illustrates the
significance of performing single base pair resolution epigenetic analyses on uncultured
human cells and importantly demonstrates the feasibility of performing NOMe-seq assay
on primary human tissue.

147
A
B
Figure 6.1 NOMe-seq detects chromatin accessibility of GRP78 locus of fresh
human tissue. Nuclei from normal intestinal epithelial cells of patient (A) 11048
and (B) 11253 were treated with M.CviPI and bisulfite sequencing was performed
for the GRP78 locus (447bp). GRP78, consistent with previous observation is
endogenously unmethylated (black circles, white filled) and accessible (blue
circles, teal filled). Regions of inaccessibility larger than 147bp are covered by a
pink rectangle indicating nucleosome occupancy.

148
A
B
Figure 6.2 NOMe-seq detects divergent chromatin structure in the VTRNA2-1 locus.
Nuclei from normal intestinal epithelial cells of patient (A) 11048 and (B) 11253 were
treated with M.CviPI and bisulfite sequencing was performed for the VTRNA2-1 locus
which can be monoallelically methylated and unmethylated in individual-specific manner.
Two amplicons (345 and 387 bp respectively) were used in order to resolve larger
genomic regions. When endogenously unmethylated (black circles, white filled),
accessible regions (blue circles, teal filled) can be detected at the -1 position of each DNA
molecules. Methylated strands (black circles, black filled) are largely inaccessible and
nucleosome occupied (blue circles, white filled and pink bar).

149
Comparing NOMe-seq protocol with and without fixing for fresh frozen human
tissue
Next, we extended our pilot experiment to apply the assay to fresh-frozen
primary human tissue, partly because our relationship with the TCGA project through the
USC Epigenome center provided us with an invaluable access to a repository of fresh-
frozen human normal and tumor tissues which have been genotyped and epigenotyped.
Due to sample availability and our continuous interest on epigenetic changes in colorectal
cancer, we isolated nuclei from fresh-frozen colon tumor. The tissue was thawed and
processed immediately using our standard NOMe-seq protocol outlined in previous
chapters. The enzyme treatment can be used to map the accessibility pattern of GRP78
(Figure 5.3a, left panel), but show aberrant accessibility in the highly methylated WNT2
locus (Figure 5.3a, right panel), which may be due to the bursting of nuclei during tissue
thawing process, resulting in the treatment of free DNA rather than chromatin.
To circumvent this problem, we fixed the tissue in 1% formaldehyde solution
before mechanically isolating the nuclei (Appendix 1). We found that a total of 15
minutes incubation time as previously determined (Chapter 3) was not sufficient to
footprint the nucleosome occupancy pattern of fixed tissue despite using a total of 300U
of enyzme (data not shown). We thus incubated the nuclei in 100U of enzyme and 1.5 µl
of SAM and boosted the reaction with equal amount of enzyme and SAM twice for a
total reaction of 3 hours and 300U of enzyme. Using this condition, we were able to
accurately footprint the accessibility of GRP78 (Figure 3b, left panel) and the
inaccessibility of WNT2 locus (Figure 3b, right panel), demonstrating that our alternative
method may be used to obtain biologically meaningful observations. We also show that
150
the two biological samples we used for this pilot experiment exhibit distinct DNA
methylation pattern in the WNT2 locus (Figure 5.3, right panel). Interestingly, our
previous works show that WNT2 promoter is highly methylated in HCT116 colon cancer
cell line and retains methylation in the derivative cells, DKO1, although it remains
unclear whether the methylation status is cancer-specific or cell-culture specific (De
Carvalho et al., 2012). Our observations with the uncultured colon tumor samples
indicate that WNT2 methylation is more variable in tumors than in cell lines and that its
methylation status may not represent epigenetic driver of tumorigenesis which thus again
highlights the importance of mapping the epigenome in primary tissue.

151
A
B
Figure 6.3 NOMe-seq can be adapted to footprint nucleosome occupancy in fresh-
frozen colon tumor. Locus-specific NOMe-seq was performed for GRP78 and WNT2
loci using two methods of sample preparation: (A) without fixing and (B) with fixing.
Both methods detect the accessibility (blue cirlcles, teal filled) in GRP78 control region
(left panel), but aberrant accessibility is seen in endogenously methylated (black circles,
black filled) WNT2 locus when the tissue was not fixed prior to enzyme treatment.
Regions of inaccessibility larger than 147bp are covered by a pink rectangle indicating
nucleosome occupancy.
152
Mapping global nucleosome occupancy and DNA methylation pattern in fresh
frozen colon tumors
After determining that it was possible to perform NOMe-seq on fresh frozen
human tissues, we selected two adenocarcinoma specimens, E485101 and E237101,
which were collected from the right proximal colon of two female patients for genome-
wide NOMe-seq analysis. These tumors were previously genotyped and found to be wild-
type for KRAS, BRAF and TP53 (Hinoue et al., 2012). One of the limiting factors in our
effort to globally map the nucleosome occupancy and DNA methylation pattern of
uncultured human tumors is the low starting amount of tissue and subsequently, the yield
of DNA following enzyme treatment, in particular when dealing with adjacent normal
tissues. To circumvent this issue, we tested the EpiGnome Methyl-Seq Kit (Epicentre)
which required <100ng of input DNA for the generation of genome-wide NOMe-seq
libraries and performed low coverage sequencing to generate 62-63 million uniquely
mappable reads for each tumor sample (>90% mapping quality).
We then aligned both samples to distal CTCF consensus sequence and showed
that we were able to visualize the distinct chromatin structure of CTCF regions which
from previous works were known to have highly-positioned nucleosomes that are anti-
correlated with DNA methylation (Figure 5.4a) (Cuddapah et al., 2009; Kelly et al.,
2012). These patterns were conserved between the two biological samples which also had
comparable DNA methylation level in these regions. Interestingly, we found that
E485101 exhibits increased baseline accessibility compared to E237101. In both samples,
however, the difference between the peak of accessibility around the immediate CTCF
binding sites and baseline accessibility is ~20%, suggesting that the discrepancy between
153
the two samples’ accessibility may be an artifact of unequal coverage in sequencing reads
rather than real biological differences and that modification in our bioinformatics analysis
pipeline will be needed to properly normalize between experiments.
We next aligned the sequencing reads to CGI TSSs and examined the average
DNA methylation and accessibility of each tumor. We show that even in low coverage
sequencing, we are able to detect the average high accessibility and low endogenous
methylation pattern which are characteristics of most CGI promoters (Figure 5.4b).
Sample E485101 again shows increased baseline accessibility in the CGI TSS, consistent
with our observation in the CTCF regions. However, we also had enough resolution to
detect the well-positioned +1 nucleosome in both samples, indicating that our low-input
library generation method is a feasible approach to be incorporated in our experimental
pipeline.

154
Distance to Element(bp)
% Methylation
Distance to Element(bp)
% Methylation
Distance to Element(bp)
% Methylation
Distance to Element(bp)
% Methylation
A
B
Figure 6.4 Low-input genome-wide NOMe-seq can be used to detect distinct
chromatin configurations in various regulatory regions. (A) NOMe-seq reads from
two independent biological colon tumor samples, 237101(left panel) and 485101(right
panel), were aligned to CTCF sites and showed unmethylated accessible regions which
are marked by a valley of inaccessibility at the CTCF binding sites. CTCF sites are
flanked by well-phased nucleosomes which are anti-correlated with the peak of DNA
methylation. (B) NOMe-seq reads were aligned to CGI TSSs (0 on the x-axes). Both
samples show a peak of accessibility upstream of the TSS and a dip corresponding to
well-positioned +1 nucleosome downstream of the TSS.
155
NOMe-seq detects chromatin configurations of various promoter types in primary
colon tumors.
Works from our lab and others have shown that promoters can be categorized
broadly based on whether they are active, poised or inactive (Baylin and Jones, 2011;
Kelly et al., 2012; Ohm et al., 2007; Schlesinger et al., 2007). In order to characterize the
promoter types within the tumor colon, we took advantage of publically available ChIP-
seq data of normal colonic mucosa in order to annotate the genome based on the
functional chromatin states (Ernst et al., 2011). We then examined the changes in DNA
methylation and nucleosome occupancy pattern in the functional promoter states in the
both colon tumors. We show that generally active promoters remain unmethylated and
accessible (Figure 5.5, left panel), while poised (Figure 5.5, middle panel) and inactive
promoters (Figure 5.5, right panel) are unmethylated and inaccessible and methylated and
inaccessible respectively in the tumors.
Strikingly, we observed a distinct accessibility pattern within the active promoters
of the two colon tumors. E237101 showed peak accessibility immediately upstream of
the TSS (Figure 5.5a, left panel) whereas the accessibility of E485101 peaked
downstream of the TSS (Figure 5.5b, left panel). We also observed a minimal increased
of accessibility in the TSS of poised promoter state of E237101 (Figure 5.5a, middle
panel) compared to baseline accessibility which was not observed in E485101 (Figure
5.5b, middle panel). Higher depth sequencing will be needed to comprehensively
examine the differences between these samples and to determine whether different tumor
subtypes may have different patterns of nucleosome positioning and chromatin
accessibility.
156
% Methylation % Methylation
Ac5ve

Poised

Inac5ve

A
B
Figure 6.5 Low-input NOMe-seq detects distinct chromatin configurations of
different promoter types. NOMe-seq reads from two biological samples (A)237101
and (B)485101 are aligned to TSS of active (left panel), poised (middle panel) and
inactive promoters (right panel). The chromatin state was determined by training ChIP-
seq data of normal colonic mucosa into chromHMM model as described previously in
Chapter 2 and 4.

157
DISCUSSION
A major caveat in current epigenomic study is the reliance on cell line model
which while useful, may not always recapitulate in vivo conditions. Culture-specific
changes in DNA methylation have been shown and consequently, may add confounding
noise when studying epigenetic alterations occurring in cancer (Wilson and Jones, 1983).
The Cancer Genome Atlas and the Roadmap Epigenomic Project have spearheaded the
mapping of DNA methylation and histone modification marks in uncultured normal and
cancerous human tissues (Beck et al., 2012; Bernstein et al., 2010; Weinstein et al.,
2013). However, the dynamic nature of nucleosome positioning remains one of the most
of the challenging aspect in mapping this specific epigenetic process and in our attempt to
apply NOMe-seq to uncultured human tissue. Although our ideal experimental strategy
involves performing NOMe-seq immediately on freshly isolated tissue, in reality, there is
often a significant lag time between when the tissue is removed and when it reaches the
laboratory for processing due to institutional protocol which requires all specimens to be
examined by pathologists beforehand. Moreover, using fresh tissue limits us to fewer
samples that can be collected and prevents us from taking advantage of large repository
of tissue available from consortiums such as TCGA.
Here, we show that we can adapt NOMe-seq to footprint the nucleosome
occupancy of fresh-frozen primary colon tumors at locus-specific and genome-wide
level. By fixing the tissue before nuclear extraction, we chemically immobilized the
positioning of nucleosomes as well as the binding of transcription factors on the DNA,
thereby limiting the variations that may occur during M.CviPI treatment due to
contaminating free DNA. By performing the fixation step immediately upon removing
158
the frozen tissue from storage, we significantly eliminate the uncertainty that changes in
nucleosome occupancy occur during the thawing process and subsequent nuclei isolation.
Additional benefit of this alternative protocol is the increased resolution for us to detect
the protection by weakly-binding transcription factors which in our standard protocol
may be washed away. This approach is not without caveat, however, as significant
adjustment may need to be done in the bioinformatics analysis pipeline in order to
account for increased overall protection level within each library.
Our pilot experiments overall provide a proof-of-principle that epigenome-wide
mapping of uncultured normal and tumor tissue can be done. We selected colon tissues
as they are readily available at the USC Norris Comprehensive Cancer Center as well as
through TCGA. Moreover, the first complete methylome of colon cancer has been
published, showing dramatic alterations occur in the tumor when compared to the
adjacent normal, giving us confidence that our efforts will yield significant information
regarding the epigenetic changes occurring in cancer (Berman et al., 2012b). Our own
work using cell culture model suggests that significant reorganization of nucleosome
positioning occurs in response to changes in DNA methylation. Performing a
comprehensive genome-wide analysis on uncultured human tissue in a normal and
diseased state thus will be the logical and critical trajectory in order to accurately assess
how the underlying epigenome is altered in cancer.

159
CHAPTER 7

DISCUSSION

Summary and Discussion

Recent outcomes of comprehensive cancer exome sequencing efforts reveal the
accumulation of mutations in genes controlling epigenetic regulatory processes in human
malignancies (Shen and Laird, 2013; Vogelstein et al., 2013; You and Jones, 2012). One
such common mutation occurs in the body of DNMT3A gene that controls the
establishment and maintenance of DNA methylation, a critical component of the
epigenetic process (Ley et al., 2010; Walter et al., 2011; Yan et al., 2011). The C to T
mutation in this gene is associated with poor prognostic in patients with MDS and AML
and contributes to the disruption of normal DNA methylation pattern (Yan et al., 2011).
Similarly, mutations in DNMT1 and DNMT3B have also been described in colorectal
cancer and ICF syndrome and are correlated with altered DNA methylation landscape
(Jiang et al., 2005; Kanai et al., 2003).
DNA hypermethylation has long been described to mediate aberrant silencing of
tumor suppressor genes, thereby driving tumorigenesis (Baylin and Jones, 2011).
Invariably, DNA methylation signatures have emerged as a valuable marker to stratify
disease subtypes and predict clinical outcomes (Figueroa et al., 2010; Noushmehr et al.,
2010; Shen et al., 2007; Su et al., 2014). It has also been apparent that targeting DNA
methylation using DNA demethylating agents as reviewed in Chapter 1 may constitute an
effective approach toward a more sensitive cancer therapy (Jones, 2014; Kelly et al.,
2010a). Yet despite the progresses made in recent years there are still many aspects of
DNA methylation that are poorly understood, one of which is how changes in DNA
160
methylation affect the epigenome. This underlying question is the overarching theme of
my study where I use a combination of primary human tissue and cell line models as well
genome-wide and gene-centric approaches to elucidate some of the pervasive issues
regarding the functions and regulation of DNA methylation.
First, I demonstrated that changes in the local tissue environment may
dramatically alter normal DNA methylation patterns (Lay et al., 2014). This study was
done using the in vivo human ileal neobladder model where an “artificial” bladder was
constructed for radical cystectomy patients using a piece of their own ileum or small
intestine. Longitudinal analysis of the methylomes of the intestinal epithelial cells before
and after the surgery reveals a time-dependent increase of DNA methylation in the
intestine-specific enhancers and corresponding decrease of DNA methylation level in the
non-tissue specific transcribed regions at a rate that exceeds age-dependent changes by
approximately 15 fold. The change in DNA methylation correlates with the loss of the
epigenetic landscape of the intestinal epithelium and the establishment of new epigenome
in response to environmental changes. This study directly implicates local tissue
environmental cues in the shaping of the human methylomes and importantly suggests
that the heritable maintenance of the unique epigenomes requires specific signals which
are yet to be determined.
Having described the importance of DNA methylation in the maintenance of
cellular identity, the bulk of my study, done in collaboration with Dr. Terry Kelly, Dr.
Ben Berman and Yaping Liu, focused on determining the role of DNA methylation on
the structure of the epigenome. I have discussed extensively the development of the
NOMe-seq assay as an innovative experimental method to examine how DNA
161
methylation and nucleosome occupancy relate to each other with a higher precision than
techniques such as MNAse. The strength of this approach lies in its ability to
simultaneously examine nucleosome occupancy and endogenous DNA methylation
patterns from a single DNA molecule both at the locus-specific and a genome-wide level
without requiring mechanical disruption of regions of interest. The NOMe-seq assay also
has the sensitivity to detect regions associated with transcription factor binding as well as
regions present in multiple chromatin configurations (or epistates) with the caveat that
these regions must have GpC sites. Regardless, such resolution has critical implications
for the study of biology, as transcription factors have been suggested to play a role in
modulating the epigenetic landscape during development, reprogramming and disease
development (Feldmann et al., 2013).
Using NOMe-seq in parallel with ChIP-seq and RNA-seq, I examined the
interplay between different epigenetic processes and specifically analyzed how changes
in DNA methylation altered the functional organization of cancer epigenome. The
comparison between HCT116 colon cancer cell line and its hypomethylated derivative
DKO1 cells unexpectedly revealed a context-specific role of DNA methylation in the
organization of the cancer epigenome (Figure 7.1 and Figure 7.2). DNA methylation
antagonizes nucleosome phasing at CGI, but not non-CGI promoters, although careful
studies are still needed to determine whether DNA methylation acts by masking the
underlying phasing or by directly blocking the positioning of nucleosomes. Morever, the
striking contrast in the influence of DNA methylation at CGI and non-CGI promoters
underlies a fundamentally different mode of transcription regulation. CGI promoters have
a chromatin landscape that is inherently permissive for transcription and thus, they
162
require a control mechanism to ensure some genes remain repressed. This repression is
mediated by the polycomb repressive complex and fine-tuned by DNA methylation
which permanently silences gene transcription by modulating the chromatin structure.
Non-CGI promoters on the other hand, are preferentially methylated and inactive.
Consequently, unlike CGI promoters, they require chromatin regulator complexes and
tissue-specific transcription factors to establish a chromatin landscape permissive for
transcription.
I have also shown that the genome-wide removal of DNA methylation may
restore a normal-like poised chromatin state in CGI promoters (Figure 7.1). This
observation was confirmed by a more detailed analysis on the MYOD1
promoter/enhancer pair where the previously methylated promoter acquired H3K27me3
and the minimal enhancer lost its nucleosomes. Inhibition of DNA methylation by
demethylating agent further shows that the resetting of the chromatin landscape is a
dynamic process and that the promoter/enhancer crosstalk provides an additional layer of
control to increase the precision and specificity of gene activation following the drug
treatment, a critical issue that needs to be considered when designing epigenetic targeted
therapy.
Finally, I have initiated a pilot experiment to map the epigenome of uncultured
normal human tissue and primary tumors with the overall goal of assessing the effects of
the disease on the underlying epigenome in an unbiased manner. Performing epigenomic
analyses on uncultured human cells remain challenging technically and logistically
despite significant progress made in this area of research. The cellular heterogeneity of
the tissue as well as the limited material available is among the roadblocks that need to be
163
overcome. So far, I have shown the feasibility of performing NOMe-seq on fresh-frozen
colon tumors at locus-specific and genome-wide level using low-input starting material
and DNA yield. Preliminary analysis of low coverage sequencing of two biologically
independent colon adenocarcinomas indicates that different tumors may exhibit, not only
aberrant DNA methylation signatures, but also have distinct nucleosome occupancy
patterns. Further study is currently underway to comprehensively compare the epigenome
of colon tumors to their adjacent normal tissues in order to better understand how the
human epigenome is altered in cancer and ultimately how to better target the epigenome
for therapy.

164
Figure 7.1 Schematic of changes in CGI promoter architecture. In normal
differentiated cells, CGI promoters exhibit distinct chromatin structures: inactive
(methylated, inaccessible and lack phased nucleosomes), active (unmethylated and
accessible) and poised (unmethylated, inaccessible and repressed by H3K27me3, but
have phased nucleosomes). Active and in particular, poised promoters are targeted for
silencing by DNA methylation in cancer cells and the chromatin landscape becomes
more compact. Removal of DNA methylation by genetic knockout or demethylating
agent may restore the permissive epigenetic landscape of normal cells.
165
Figure 7.2 Schematic of changes in non-CGI promoter architecture. Non-CGI
promoters are largely methylated and inactive in normal cells while still exhibiting
phased nucleosomes. Active promoters are unmethylated and accessible in a tissue-
specific manner and their activation potential may require additional factors in order to
establish the permissive chromatin structure needed for transcription. In cancer cells,
epigenetic silencing may occur via DNA methylation or by nucleosome-mediated
repression. The removal of DNA methylation does not result in dramatic activation for
most promoters which remain repressed by the presence of nucleosomes.
166
Perspectives

The application of a genome-wide approach in the study of DNA methylation,
supported by microarray and sequencing-based technologies, has allowed for an
unprecedented view of the structure of the epigenome and how it is altered in cancer.
Beyond promoter regions, there is growing evidence suggesting that DNA methylation
may also influence the function of enhancers, transcription factor binding, and gene bodiy
transcription, all of which are critical for normal cellular processes (Blattler and Farnham,
2013; Jones, 1999; Jones, 2012). Enhancer/promoter crosstalk is a critical component of
transcriptional regulation and understanding this relationship requires not only careful
and detailed analyses of their physical interaction but also investigation of how the
epigenetic landscape of enhancers is established, maintained and ultimately disrupted in
cancer, both at genome-wide level and mechanistically at specific loci. Understanding the
role of DNA methylation in various regulatory regions is ultimately critical for
elucidating the molecular mechanism governing gene transcription. My own study barely
touched the surface of these burgeoning areas of research, but the genome-wide data that
has been generated and the experimental pipeline established throughout my work may
provide a wealth of resources for others to examine the relationship between DNA
methylation and the epigenetic landscape in non-promoter regions in both normal and
cancer cells.

167
REFERENCES

Ahuja, N., Li, Q., Mohan, A. L., Baylin, S. B., and Issa, J. P. (1998). Aging and DNA
methylation in colorectal mucosa and cancer. Cancer research 58, 5489-5494.
Akhtar-Zaidi, B., Cowper-Sal-lari, R., Corradin, O., Saiakhova, A., Bartels, C. F.,
Balasubramanian, D., Myeroff, L., Lutterbaugh, J., Jarrar, A., Kalady, M. F., et al.
(2012). Epigenomic enhancer profiling defines a signature of colon cancer. Science 336,
736-739.
Amato, R. J. (2007). Inhibition of DNA methylation by antisense oligonucleotide MG98
as cancer therapy. Clinical genitourinary cancer 5, 422-426.
Amato, R. J., Stephenson, J., Hotte, S., Nemunaitis, J., Belanger, K., Reid, G., and
Martell, R. E. (2012). MG98, a second-generation DNMT1 inhibitor, in the treatment of
advanced renal cell carcinoma. Cancer investigation 30, 415-421.
Andreu-Vieyra, C., Lai, J., Berman, B. P., Frenkel, B., Jia, L., Jones, P. A., and Coetzee,
G. A. (2011). Dynamic nucleosome-depleted regions at androgen receptor enhancers in
the absence of ligand in prostate cancer cells. Molecular and cellular biology 31, 4648-
4662.
Appleton, K., Mackay, H. J., Judson, I., Plumb, J. A., McCormick, C., Strathdee, G., Lee,
C., Barrett, S., Reade, S., Jadayel, D., et al. (2007). Phase I and pharmacodynamic trial of
the DNA methyltransferase inhibitor decitabine and carboplatin in solid tumors. Journal
of clinical oncology : official journal of the American Society of Clinical Oncology 25,
4603-4609.
Aragona, F., De Caro, R., Parenti, A., Artibani, W., Bassi, P., Munari, P. F., and Pagano,
F. (1998). Structural and ultrastructural changes in ileal neobladder mucosa: a 7-year
follow-up. British journal of urology 81, 55-61.
Aran, D., and Hellman, A. (2013). DNA methylation of transcriptional enhancers and
cancer predisposition. Cell 154, 11-13.
Barski, A., Cuddapah, S., Cui, K., Roh, T. Y., Schones, D. E., Wang, Z., Wei, G.,
Chepelev, I., and Zhao, K. (2007). High-resolution profiling of histone methylations in
the human genome. Cell 129, 823-837.
Baylin, S. B., and Jones, P. A. (2011). A decade of exploring the cancer epigenome -
biological and translational implications. Nature reviews Cancer 11, 726-734.
Beck, S., Bernstein, B. E., Campbell, R. M., Costello, J. F., Dhanak, D., Ecker, J. R.,
Greally, J. M., Issa, J. P., Laird, P. W., Polyak, K., et al. (2012). A blueprint for an
international cancer epigenome consortium. A report from the AACR Cancer Epigenome
Task Force. Cancer research 72, 6319-6324.
168
Bell, O., Tiwari, V. K., Thoma, N. H., and Schubeler, D. (2011). Determinants and
dynamics of genome accessibility. Nature reviews Genetics 12, 554-564.
Bergman, Y., and Cedar, H. (2013). DNA methylation dynamics in health and disease.
Nature structural & molecular biology 20, 274-281.
Berman, B. P., Liu, Y., and Kelly, T. K. (2013). DNA methylation marks inter-
nucleosome linker regions throughout the human genome. PeerJ PrePrints e27v3.
Berman, B. P., Weisenberger, D. J., Aman, J. F., Hinoue, T., Ramjan, Z., Liu, Y.,
Noushmehr, H., Lange, C. P., van Dijk, C. M., Tollenaar, R. A., et al. (2012a). Regions
of focal DNA hypermethylation and long-range hypomethylation in colorectal cancer
coincide with nuclear lamina-associated domains. Nat Genet 44, 40-46.
Berman, B. P., Weisenberger, D. J., Aman, J. F., Hinoue, T., Ramjan, Z., Liu, Y.,
Noushmehr, H., Lange, C. P., van Dijk, C. M., Tollenaar, R. A., et al. (2012b). Regions
of focal DNA hypermethylation and long-range hypomethylation in colorectal cancer
coincide with nuclear lamina-associated domains. Nature genetics 44, 40-46.
Bernstein, B. E., Stamatoyannopoulos, J. A., Costello, J. F., Ren, B., Milosavljevic, A.,
Meissner, A., Kellis, M., Marra, M. A., Beaudet, A. L., Ecker, J. R., et al. (2010). The
NIH Roadmap Epigenomics Mapping Consortium. Nature biotechnology 28, 1045-1048.
Beumer, J. H., Eiseman, J. L., Parise, R. A., Joseph, E., Holleran, J. L., Covey, J. M., and
Egorin, M. J. (2006). Pharmacokinetics, metabolism, and oral bioavailability of the DNA
methyltransferase inhibitor 5-fluoro-2'-deoxycytidine in mice. Clinical cancer research :
an official journal of the American Association for Cancer Research 12, 7483-7491.
Beumer, J. H., Parise, R. A., Newman, E. M., Doroshow, J. H., Synold, T. W., Lenz, H.
J., and Egorin, M. J. (2008). Concentrations of the DNA methyltransferase inhibitor 5-
fluoro-2'-deoxycytidine (FdCyd) and its cytotoxic metabolites in plasma of patients
treated with FdCyd and tetrahydrouridine (THU). Cancer chemotherapy and
pharmacology 62, 363-368.
Bibikova, M., Barnes, B., Tsan, C., Ho, V., Klotzle, B., Le, J. M., Delano, D., Zhang, L.,
Schroth, G. P., Gunderson, K. L., et al. (2011). High density DNA methylation array with
single CpG site resolution. Genomics 98, 288-295.
Billam, M., Sobolewski, M. D., and Davidson, N. E. (2010). Effects of a novel DNA
methyltransferase inhibitor zebularine on human breast cancer cells. Breast cancer
research and treatment 120, 581-592.
Bird, A. (2002). DNA methylation patterns and epigenetic memory. Genes &
development 16, 6-21.
Birney, E., Stamatoyannopoulos, J. A., Dutta, A., Guigo, R., Gingeras, T. R., Margulies,
E. H., Weng, Z., Snyder, M., Dermitzakis, E. T., Thurman, R. E., et al. (2007).
169
Identification and analysis of functional elements in 1% of the human genome by the
ENCODE pilot project. Nature 447, 799-816.
Blattler, A., and Farnham, P. J. (2013). Cross-talk between site-specific transcription
factors and DNA methylation states. The Journal of biological chemistry 288, 34287-
34294.
Blum, W., Klisovic, R. B., Hackanson, B., Liu, Z., Liu, S., Devine, H., Vukosavljevic, T.,
Huynh, L., Lozanski, G., Kefauver, C., et al. (2007). Phase I study of decitabine alone or
in combination with valproic acid in acute myeloid leukemia. Journal of clinical
oncology : official journal of the American Society of Clinical Oncology 25, 3884-3891.
Bonn, S., Zinzen, R. P., Girardot, C., Gustafson, E. H., Perez-Gonzalez, A., Delhomme,
N., Ghavi-Helm, Y., Wilczynski, B., Riddell, A., and Furlong, E. E. (2012). Tissue-
specific analysis of chromatin state identifies temporal signatures of enhancer activity
during embryonic development. Nature genetics 44, 148-156.
Bostick, M., Kim, J. K., Esteve, P. O., Clark, A., Pradhan, S., and Jacobsen, S. E. (2007).
UHRF1 plays a role in maintaining DNA methylation in mammalian cells. Science 317,
1760-1764.
Bouazoune, K., Miranda, T. B., Jones, P. A., and Kingston, R. E. (2009). Analysis of
individual remodeled nucleosomes reveals decreased histone-DNA contacts created by
hSWI/SNF. Nucleic acids research 37, 5279-5294.
Boukamp, P., Chen, J., Gonzales, F., Jones, P. A., and Fusenig, N. E. (1992). Progressive
stages of "transdifferentiation" from epidermal to mesenchymal phenotype induced by
MyoD1 transfection, 5-aza-2'-deoxycytidine treatment, and selection for reduced cell
attachment in the human keratinocyte line HaCaT. The Journal of cell biology 116, 1257-
1271.
Boyle, A. P., Davis, S., Shulha, H. P., Meltzer, P., Margulies, E. H., Weng, Z., Furey, T.
S., and Crawford, G. E. (2008). High-resolution mapping and characterization of open
chromatin across the genome. Cell 132, 311-322.
Braiteh, F., Soriano, A. O., Garcia-Manero, G., Hong, D., Johnson, M. M., Silva Lde, P.,
Yang, H., Alexander, S., Wolff, J., and Kurzrock, R. (2008). Phase I study of epigenetic
modulation with 5-azacytidine and valproic acid in patients with advanced cancers.
Clinical cancer research : an official journal of the American Association for Cancer
Research 14, 6296-6301.
Byun, H. M., Choi, S. H., Laird, P. W., Trinh, B., Siddiqui, M. A., Marquez, V. E., and
Yang, A. S. (2008). 2'-Deoxy-N4-[2-(4-nitrophenyl)ethoxycarbonyl]-5-azacytidine: a
novel inhibitor of DNA methyltransferase that requires activation by human
carboxylesterase 1. Cancer letters 266, 238-248.
Cairns, B. R. (2009). The logic of chromatin architecture and remodelling at promoters.
Nature 461, 193-198.
170
Calo, E., and Wysocka, J. (2013). Modification of enhancer chromatin: what, how, and
why? Molecular cell 49, 825-837.
Cantone, I., and Fisher, A. G. (2013). Epigenetic programming and reprogramming
during development. Nature structural & molecular biology 20, 282-289.
Cheng, J. C., Yoo, C. B., Weisenberger, D. J., Chuang, J., Wozniak, C., Liang, G.,
Marquez, V. E., Greer, S., Orntoft, T. F., Thykjaer, T., and Jones, P. A. (2004).
Preferential response of cancer cells to zebularine. Cancer cell 6, 151-158.
Chodavarapu, R. K., Feng, S., Bernatavichute, Y. V., Chen, P. Y., Stroud, H., Yu, Y.,
Hetzel, J. A., Kuo, F., Kim, J., Cokus, S. J., et al. (2010). Relationship between
nucleosome positioning and DNA methylation. Nature 466, 388-392.
Chowdhury, S., Seropian, S., and Marks, P. W. (2009). Decitabine combined with
fractionated gemtuzumab ozogamicin therapy in patients with relapsed or refractory acute
myeloid leukemia. American journal of hematology 84, 599-600.
Christensen, B. C., Houseman, E. A., Marsit, C. J., Zheng, S., Wrensch, M. R., Wiemels,
J. L., Nelson, H. H., Karagas, M. R., Padbury, J. F., Bueno, R., et al. (2009). Aging and
environmental exposures alter tissue-specific DNA methylation dependent upon CpG
island context. PLoS genetics 5, e1000602.
Chuang, J. C., Warner, S. L., Vollmer, D., Vankayalapati, H., Redkar, S., Bearss, D. J.,
Qiu, X., Yoo, C. B., and Jones, P. A. (2010). S110, a 5-Aza-2'-deoxycytidine-containing
dinucleotide, is an effective DNA methylation inhibitor in vivo and can reduce tumor
growth. Molecular cancer therapeutics 9, 1443-1450.
Chuang, J. C., Yoo, C. B., Kwan, J. M., Li, T. W., Liang, G., Yang, A. S., and Jones, P.
A. (2005). Comparison of biological effects of non-nucleoside DNA methylation
inhibitors versus 5-aza-2'-deoxycytidine. Molecular cancer therapeutics 4, 1515-1520.
Cohen, N. M., Kenigsberg, E., and Tanay, A. (2011). Primate CpG islands are maintained
by heterogeneous evolutionary regimes involving minimal selection. Cell 145, 773-786.
Coleman-Derr, D., and Zilberman, D. (2012). DNA methylation, H2A.Z, and the
regulation of constitutive expression. Cold Spring Harbor symposia on quantitative
biology 77, 147-154.
Cortessis, V. K., Thomas, D. C., Levine, A. J., Breton, C. V., Mack, T. M., Siegmund, K.
D., Haile, R. W., and Laird, P. W. (2012). Environmental epigenetics: prospects for
studying epigenetic mediation of exposure-response relationships. Human genetics 131,
1565-1589.
Creyghton, M. P., Cheng, A. W., Welstead, G. G., Kooistra, T., Carey, B. W., Steine, E.
J., Hanna, J., Lodato, M. A., Frampton, G. M., Sharp, P. A., et al. (2010). Histone
H3K27ac separates active from poised enhancers and predicts developmental state.
171
Proceedings of the National Academy of Sciences of the United States of America 107,
21931-21936.
Cuddapah, S., Jothi, R., Schones, D. E., Roh, T. Y., Cui, K., and Zhao, K. (2009). Global
analysis of the insulator binding protein CTCF in chromatin barrier regions reveals
demarcation of active and repressive domains. Genome research 19, 24-32.
Dahl, J. A., and Collas, P. (2008). A rapid micro chromatin immunoprecipitation assay
(microChIP). Nature protocols 3, 1032-1045.
Datta, J., Ghoshal, K., Denny, W. A., Gamage, S. A., Brooke, D. G., Phiasivongsa, P.,
Redkar, S., and Jacob, S. T. (2009). A new class of quinoline-based DNA
hypomethylating agents reactivates tumor suppressor genes by blocking DNA
methyltransferase 1 activity and inducing its degradation. Cancer research 69, 4277-4285.
De Carvalho, D. D., Sharma, S., You, J. S., Su, S. F., Taberlay, P. C., Kelly, T. K., Yang,
X., Liang, G., and Jones, P. A. (2012). DNA methylation screening identifies driver
epigenetic events of cancer cell survival. Cancer cell 21, 655-667.
Deaton, A. M., and Bird, A. (2011). CpG islands and the regulation of transcription.
Genes & development 25, 1010-1022.
Di Tonno, F., Siracusano, S., Ciciliato, S., Visalli, F., Lampropoulou, N., and Lavelli, D.
(2012). Morphological changes on the intestinal mucosa in orthotopic neobladder.
Urologia internationalis 89, 67-70.
Doi, A., Park, I. H., Wen, B., Murakami, P., Aryee, M. J., Irizarry, R., Herb, B., Ladd-
Acosta, C., Rho, J., Loewer, S., et al. (2009). Differential methylation of tissue- and
cancer-specific CpG island shores distinguishes human induced pluripotent stem cells,
embryonic stem cells and fibroblasts. Nature genetics 41, 1350-1353.
Eden, A., Gaudet, F., Waghmare, A., and Jaenisch, R. (2003). Chromosomal instability
and tumors promoted by DNA hypomethylation. Science 300, 455.
Egger, G., Jeong, S., Escobar, S. G., Cortez, C. C., Li, T. W., Saito, Y., Yoo, C. B., Jones,
P. A., and Liang, G. (2006). Identification of DNMT1 (DNA methyltransferase 1)
hypomorphs in somatic knockouts suggests an essential role for DNMT1 in cell survival.
Proceedings of the National Academy of Sciences of the United States of America 103,
14080-14085.
Ernst, J., and Kellis, M. (2012). ChromHMM: automating chromatin-state discovery and
characterization. Nature methods 9, 215-216.
Ernst, J., Kheradpour, P., Mikkelsen, T. S., Shoresh, N., Ward, L. D., Epstein, C. B.,
Zhang, X., Wang, L., Issner, R., Coyne, M., et al. (2011). Mapping and analysis of
chromatin state dynamics in nine human cell types. Nature 473, 43-49.
172
Esteller, M. (2007). Cancer epigenomics: DNA methylomes and histone-modification
maps. Nature reviews Genetics 8, 286-298.
Fandy, T. E., Herman, J. G., Kerns, P., Jiemjit, A., Sugar, E. A., Choi, S. H., Yang, A. S.,
Aucott, T., Dauses, T., Odchimar-Reissig, R., et al. (2009). Early epigenetic changes and
DNA damage do not predict clinical response in an overlapping schedule of 5-azacytidine
and entinostat in patients with myeloid malignancies. Blood 114, 2764-2773.
Feil, R., and Fraga, M. F. (2011). Epigenetics and the environment: emerging patterns
and implications. Nature reviews Genetics 13, 97-109.
Feldmann, A., Ivanek, R., Murr, R., Gaidatzis, D., Burger, L., and Schubeler, D. (2013).
Transcription factor occupancy can mediate active turnover of DNA methylation at
regulatory regions. PLoS genetics 9, e1003994.
Fenaux, P., Gattermann, N., Seymour, J. F., Hellstrom-Lindberg, E., Mufti, G. J.,
Duehrsen, U., Gore, S. D., Ramos, F., Beyne-Rauzy, O., List, A., et al. (2010). Prolonged
survival with improved tolerability in higher-risk myelodysplastic syndromes: azacitidine
compared with low dose ara-C. British journal of haematology 149, 244-249.
Fenaux, P., Mufti, G. J., Hellstrom-Lindberg, E., Santini, V., Finelli, C., Giagounidis, A.,
Schoch, R., Gattermann, N., Sanz, G., List, A., et al. (2009). Efficacy of azacitidine
compared with that of conventional care regimens in the treatment of higher-risk
myelodysplastic syndromes: a randomised, open-label, phase III study. The lancet
oncology 10, 223-232.
Figueroa, M. E., Lugthart, S., Li, Y., Erpelinck-Verschueren, C., Deng, X., Christos, P.
J., Schifano, E., Booth, J., van Putten, W., Skrabanek, L., et al. (2010). DNA methylation
signatures identify biologically distinct subtypes in acute myeloid leukemia. Cancer cell
17, 13-27.
Flotho, C., Claus, R., Batz, C., Schneider, M., Sandrock, I., Ihde, S., Plass, C., Niemeyer,
C. M., and Lubbert, M. (2009). The DNA methyltransferase inhibitors azacitidine,
decitabine and zebularine exert differential effects on cancer gene expression in acute
myeloid leukemia cells. Leukemia 23, 1019-1028.
Freeman, J. A., Tarter, T. A., Esrig, D., Stein, J. P., Elmajian, D. A., Chen, S. C.,
Groshen, S., Lieskovsky, G., and Skinner, D. G. (1996). Urethral recurrence in patients
with orthotopic ileal neobladders. The Journal of urology 156, 1615-1619.
Gal-Yam, E. N., Egger, G., Iniguez, L., Holster, H., Einarsson, S., Zhang, X., Lin, J. C.,
Liang, G., Jones, P. A., and Tanay, A. (2008). Frequent switching of Polycomb
repressive marks and DNA hypermethylation in the PC3 prostate cancer cell line.
Proceedings of the National Academy of Sciences of the United States of America 105,
12979-12984.
173
Gal-Yam, E. N., Jeong, S., Tanay, A., Egger, G., Lee, A. S., and Jones, P. A. (2006).
Constitutive nucleosome depletion and ordered factor assembly at the GRP78 promoter
revealed by single molecule footprinting. PLoS genetics 2, e160.
Garcia-Manero, G., Kantarjian, H. M., Sanchez-Gonzalez, B., Yang, H., Rosner, G.,
Verstovsek, S., Rytting, M., Wierda, W. G., Ravandi, F., Koller, C., et al. (2006). Phase
1/2 study of the combination of 5-aza-2'-deoxycytidine with valproic acid in patients with
leukemia. Blood 108, 3271-3279.
Garzon, R., Liu, S., Fabbri, M., Liu, Z., Heaphy, C. E., Callegari, E., Schwind, S., Pang,
J., Yu, J., Muthusamy, N., et al. (2009). MicroRNA-29b induces global DNA
hypomethylation and tumor suppressor gene reexpression in acute myeloid leukemia by
targeting directly DNMT3A and 3B and indirectly DNMT1. Blood 113, 6411-6418.
Gatti, R., Ferretti, S., Bucci, G., Simonazzi, M., Cortellini, P., and Orlandini, G. (1999).
Histological adaptation of orthotopic ileal neobladder mucosa: 4-year follow-up of 30
patients. European urology 36, 588-594.
Gebhard, C., Benner, C., Ehrich, M., Schwarzfischer, L., Schilling, E., Klug, M.,
Dietmaier, W., Thiede, C., Holler, E., Andreesen, R., and Rehli, M. (2010). General
transcription factor binding at CpG islands in normal cells correlates with resistance to de
novo DNA methylation in cancer cells. Cancer research 70, 1398-1407.
Gerstein, M. B., Kundaje, A., Hariharan, M., Landt, S. G., Yan, K. K., Cheng, C., Mu, X.
J., Khurana, E., Rozowsky, J., Alexander, R., et al. (2012). Architecture of the human
regulatory network derived from ENCODE data. Nature 489, 91-100.
Ghoshal, K., Datta, J., Majumder, S., Bai, S., Kutay, H., Motiwala, T., and Jacob, S. T.
(2005). 5-Aza-deoxycytidine induces selective degradation of DNA methyltransferase 1
by a proteasomal pathway that requires the KEN box, bromo-adjacent homology domain,
and nuclear localization signal. Molecular and cellular biology 25, 4727-4741.
Gifford, C. A., Ziller, M. J., Gu, H., Trapnell, C., Donaghey, J., Tsankov, A., Shalek, A.
K., Kelley, D. R., Shishkin, A. A., Issner, R., et al. (2013). Transcriptional and epigenetic
dynamics during specification of human embryonic stem cells. Cell 153, 1149-1163.
Giresi, P. G., Kim, J., McDaniell, R. M., Iyer, V. R., and Lieb, J. D. (2007). FAIRE
(Formaldehyde-Assisted Isolation of Regulatory Elements) isolates active regulatory
elements from human chromatin. Genome research 17, 877-885.
Goldhamer, D. J., Brunk, B. P., Faerman, A., King, A., Shani, M., and Emerson, C. P., Jr.
(1995). Embryonic activation of the myoD gene is regulated by a highly conserved distal
control element. Development 121, 637-649.
Gordon, L., Joo, J. E., Powell, J. E., Ollikainen, M., Novakovic, B., Li, X., Andronikos,
R., Cruickshank, M. N., Conneely, K. N., Smith, A. K., et al. (2012). Neonatal DNA
methylation profile in human twins is specified by a complex interplay between
174
intrauterine environmental and genetic factors, subject to tissue-specific influence.
Genome research 22, 1395-1406.
Halkidou, K., Gaughan, L., Cook, S., Leung, H. Y., Neal, D. E., and Robson, C. N.
(2004). Upregulation and nuclear recruitment of HDAC1 in hormone refractory prostate
cancer. The Prostate 59, 177-189.
Halley-Stott, R. P., and Gurdon, J. B. (2013). Epigenetic memory in the context of
nuclear reprogramming and cancer. Briefings in functional genomics 12, 164-173.
Han, H., Cortez, C. C., Yang, X., Nichols, P. W., Jones, P. A., and Liang, G. (2011).
DNA methylation directly silences genes with non-CpG island promoters and establishes
a nucleosome occupied promoter. Human molecular genetics 20, 4299-4310.
Hannum, G., Guinney, J., Zhao, L., Zhang, L., Hughes, G., Sadda, S., Klotzle, B.,
Bibikova, M., Fan, J. B., Gao, Y., et al. (2013). Genome-wide methylation profiles reveal
quantitative views of human aging rates. Molecular cell 49, 359-367.
Hansen, K. D., Timp, W., Bravo, H. C., Sabunciyan, S., Langmead, B., McDonald, O. G.,
Wen, B., Wu, H., Liu, Y., Diep, D., et al. (2011a). Increased methylation variation in
epigenetic domains across cancer types. Nat Genet 43, 768-775.
Hansen, K. D., Timp, W., Bravo, H. C., Sabunciyan, S., Langmead, B., McDonald, O. G.,
Wen, B., Wu, H., Liu, Y., Diep, D., et al. (2011b). Increased methylation variation in
epigenetic domains across cancer types. Nature genetics 43, 768-775.
Hautmann, R. E., Abol-Enein, H., Hafez, K., Haro, I., Mansson, W., Mills, R. D., Montie,
J. D., Sagalowsky, A. I., Stein, J. P., Stenzl, A., et al. (2007). Urinary diversion. Urology
69, 17-49.
Hawkins, R. D., Hon, G. C., Lee, L. K., Ngo, Q., Lister, R., Pelizzola, M., Edsall, L. E.,
Kuan, S., Luu, Y., Klugman, S., et al. (2010). Distinct epigenomic landscapes of
pluripotent and lineage-committed human cells. Cell stem cell 6, 479-491.
Heintzman, N. D., Hon, G. C., Hawkins, R. D., Kheradpour, P., Stark, A., Harp, L. F.,
Ye, Z., Lee, L. K., Stuart, R. K., Ching, C. W., et al. (2009). Histone modifications at
human enhancers reflect global cell-type-specific gene expression. Nature 459, 108-112.
Heintzman, N. D., Stuart, R. K., Hon, G., Fu, Y., Ching, C. W., Hawkins, R. D., Barrera,
L. O., Van Calcar, S., Qu, C., Ching, K. A., et al. (2007). Distinct and predictive
chromatin signatures of transcriptional promoters and enhancers in the human genome.
Nature genetics 39, 311-318.
Hellman, A., and Chess, A. (2007). Gene body-specific methylation on the active X
chromosome. Science 315, 1141-1143.
Heyn, H., Li, N., Ferreira, H. J., Moran, S., Pisano, D. G., Gomez, A., Diez, J., Sanchez-
Mut, J. V., Setien, F., Carmona, F. J., et al. (2012). Distinct DNA methylomes of
175
newborns and centenarians. Proceedings of the National Academy of Sciences of the
United States of America 109, 10522-10527.
Hinoue, T., Weisenberger, D. J., Lange, C. P., Shen, H., Byun, H. M., Van Den Berg, D.,
Malik, S., Pan, F., Noushmehr, H., van Dijk, C. M., et al. (2012). Genome-scale analysis
of aberrant DNA methylation in colorectal cancer. Genome research 22, 271-282.
Hiranuma, C., Kawakami, K., Oyama, K., Ota, N., Omura, K., and Watanabe, G. (2004).
Hypermethylation of the MYOD1 gene is a novel prognostic factor in patients with
colorectal cancer. International journal of molecular medicine 13, 413-417.
Hnisz, D., Abraham, B. J., Lee, T. I., Lau, A., Saint-Andre, V., Sigova, A. A., Hoke, H.
A., and Young, R. A. (2013). Super-enhancers in the control of cell identity and disease.
Cell 155, 934-947.
Hollenbach, P. W., Nguyen, A. N., Brady, H., Williams, M., Ning, Y., Richard, N.,
Krushel, L., Aukerman, S. L., Heise, C., and MacBeth, K. J. (2010). A comparison of
azacitidine and decitabine activities in acute myeloid leukemia cell lines. PloS one 5,
e9001.
Hon, G. C., Hawkins, R. D., Caballero, O. L., Lo, C., Lister, R., Pelizzola, M., Valsesia,
A., Ye, Z., Kuan, S., Edsall, L. E., et al. (2012). Global DNA hypomethylation coupled to
repressive chromatin domain formation and gene silencing in breast cancer. Genome
research 22, 246-258.
Houseman, E. A., Accomando, W. P., Koestler, D. C., Christensen, B. C., Marsit, C. J.,
Nelson, H. H., Wiencke, J. K., and Kelsey, K. T. (2012). DNA methylation arrays as
surrogate measures of cell mixture distribution. BMC bioinformatics 13, 86.
Hovestadt, V., Jones, D. T., Picelli, S., Wang, W., Kool, M., Northcott, P. A., Sultan, M.,
Stachurski, K., Ryzhova, M., Warnatz, H. J., et al. (2014). Decoding the regulatory
landscape of medulloblastoma using DNA methylation sequencing. Nature 510, 537-541.
Irizarry, R. A., Ladd-Acosta, C., Wen, B., Wu, Z., Montano, C., Onyango, P., Cui, H.,
Gabo, K., Rongione, M., Webster, M., et al. (2009). The human colon cancer methylome
shows similar hypo- and hypermethylation at conserved tissue-specific CpG island
shores. Nature genetics 41, 178-186.
Issa, J. P., and Kantarjian, H. M. (2009). Targeting DNA methylation. Clinical cancer
research : an official journal of the American Association for Cancer Research 15, 3938-
3946.
Jeong, S., Liang, G., Sharma, S., Lin, J. C., Choi, S. H., Han, H., Yoo, C. B., Egger, G.,
Yang, A. S., and Jones, P. A. (2009). Selective anchoring of DNA methyltransferases 3A
and 3B to nucleosomes containing methylated DNA. Molecular and cellular biology 29,
5366-5376.
176
Jiang, Y. L., Rigolet, M., Bourc'his, D., Nigon, F., Bokesoy, I., Fryns, J. P., Hulten, M.,
Jonveaux, P., Maraschio, P., Megarbane, A., et al. (2005). DNMT3B mutations and DNA
methylation defect define two types of ICF syndrome. Human mutation 25, 56-63.
Jin, B., Yao, B., Li, J. L., Fields, C. R., Delmas, A. L., Liu, C., and Robertson, K. D.
(2009). DNMT1 and DNMT3B modulate distinct polycomb-mediated histone
modifications in colon cancer. Cancer research 69, 7412-7421.
Jirtle, R. L., and Skinner, M. K. (2007). Environmental epigenomics and disease
susceptibility. Nature reviews Genetics 8, 253-262.
Jones, P. A. (1999). The DNA methylation paradox. Trends in genetics : TIG 15, 34-37.
Jones, P. A. (2012). Functions of DNA methylation: islands, start sites, gene bodies and
beyond. Nature reviews Genetics 13, 484-492.
Jones, P. A. (2014). At the tipping point for epigenetic therapies in cancer. The Journal of
clinical investigation 124, 14-16.
Jones, P. A., and Liang, G. (2009). Rethinking how DNA methylation patterns are
maintained. Nature reviews Genetics 10, 805-811.
Jones, P. A., and Taylor, S. M. (1980). Cellular differentiation, cytidine analogs and
DNA methylation. Cell 20, 85-93.
Kaaij, L. J., van de Wetering, M., Fang, F., Decato, B., Molaro, A., van de Werken, H. J.,
van Es, J. H., Schuijers, J., de Wit, E., de Laat, W., et al. (2013). DNA methylation
dynamics during intestinal stem cell differentiation reveals enhancers driving gene
expression in the villus. Genome biology 14, R50.
Kanai, Y., Ushijima, S., Nakanishi, Y., Sakamoto, M., and Hirohashi, S. (2003).
Mutation of the DNA methyltransferase (DNMT) 1 gene in human colorectal cancers.
Cancer letters 192, 75-82.
Kantarjian, H., Issa, J. P., Rosenfeld, C. S., Bennett, J. M., Albitar, M., DiPersio, J.,
Klimek, V., Slack, J., de Castro, C., Ravandi, F., et al. (2006). Decitabine improves
patient outcomes in myelodysplastic syndromes: results of a phase III randomized study.
Cancer 106, 1794-1803.
Kantarjian, H., Oki, Y., Garcia-Manero, G., Huang, X., O'Brien, S., Cortes, J., Faderl, S.,
Bueso-Ramos, C., Ravandi, F., Estrov, Z., et al. (2007). Results of a randomized study of
3 schedules of low-dose decitabine in higher-risk myelodysplastic syndrome and chronic
myelomonocytic leukemia. Blood 109, 52-57.
Kautiainen, T. L., and Jones, P. A. (1986). DNA methyltransferase levels in tumorigenic
and nontumorigenic cells in culture. The Journal of biological chemistry 261, 1594-1598.
177
Kelly, T. K., De Carvalho, D. D., and Jones, P. A. (2010a). Epigenetic modifications as
therapeutic targets. Nature biotechnology 28, 1069-1078.
Kelly, T. K., Liu, Y., Lay, F. D., Liang, G., Berman, B. P., and Jones, P. A. (2012).
Genome-wide mapping of nucleosome positioning and DNA methylation within
individual DNA molecules. Genome research 22, 2497-2506.
Kelly, T. K., Miranda, T. B., Liang, G., B.P, B., J.C., L., A., T., and Jones, P. A. (2010b).
H2A.Z maintenance during mitosis reveals nucleosome shifting on mitotically silenced
genes Molecular Cell 39, 901-911.
Kelly, T. K., Miranda, T. B., Liang, G., Berman, B. P., Lin, J. C., Tanay, A., and Jones,
P. A. (2010c). H2A.Z maintenance during mitosis reveals nucleosome shifting on
mitotically silenced genes. Molecular cell 39, 901-911.
Kieffer-Kwon, K. R., Tang, Z., Mathe, E., Qian, J., Sung, M. H., Li, G., Resch, W., Baek,
S., Pruett, N., Grontved, L., et al. (2013). Interactome maps of mouse gene regulatory
domains reveal basic principles of transcriptional regulation. Cell 155, 1507-1520.
Kim, T. H., Abdullaev, Z. K., Smith, A. D., Ching, K. A., Loukinov, D. I., Green, R. D.,
Zhang, M. Q., Lobanenkov, V. V., and Ren, B. (2007). Analysis of the vertebrate
insulator protein CTCF-binding sites in the human genome. Cell 128, 1231-1245.
Klisovic, R. B., Stock, W., Cataland, S., Klisovic, M. I., Liu, S., Blum, W., Green, M.,
Odenike, O., Godley, L., Burgt, J. V., et al. (2008). A phase I biological study of MG98,
an oligodeoxynucleotide antisense to DNA methyltransferase 1, in patients with high-risk
myelodysplasia and acute myeloid leukemia. Clinical cancer research : an official journal
of the American Association for Cancer Research 14, 2444-2449.
Komashko, V. M., and Farnham, P. J. (2010). 5-azacytidine treatment reorganizes
genomic histone modification patterns. Epigenetics : official journal of the DNA
Methylation Society 5.
Kondo, Y., Shen, L., Cheng, A. S., Ahmed, S., Boumber, Y., Charo, C., Yamochi, T.,
Urano, T., Furukawa, K., Kwabi-Addo, B., et al. (2008). Gene silencing in cancer by
histone H3 lysine 27 trimethylation independent of promoter DNA methylation. Nature
genetics 40, 741-750.
Kornblith, A. B., Herndon, J. E., 2nd, Silverman, L. R., Demakos, E. P., Odchimar-
Reissig, R., Holland, J. F., Powell, B. L., DeCastro, C., Ellerton, J., Larson, R. A., et al.
(2002). Impact of azacytidine on the quality of life of patients with myelodysplastic
syndrome treated in a randomized phase III trial: a Cancer and Leukemia Group B study.
Journal of clinical oncology : official journal of the American Society of Clinical
Oncology 20, 2441-2452.
Kratzke, R. A., Wang, X., Wong, L., Kratzke, M. G., Green, M. R., Vokes, E. E.,
Vogelzang, N. J., Kindler, H. L., and Kern, J. A. (2008). Response to the methylation
inhibitor dihydro-5-azacytidine in mesothelioma is not associated with methylation of
178
p16INK4a: results of cancer and leukemia group B 159904. Journal of thoracic oncology
: official publication of the International Association for the Study of Lung Cancer 3,
417-421.
Krueger, F., Kreck, B., Franke, A., and Andrews, S. R. (2012). DNA methylome analysis
using short bisulfite sequencing data. Nature methods 9, 145-151.
Kuo, H. K., Griffith, J. D., and Kreuzer, K. N. (2007). 5-Azacytidine induced
methyltransferase-DNA adducts block DNA replication in vivo. Cancer research 67,
8248-8254.
Landt, S. G., Marinov, G. K., Kundaje, A., Kheradpour, P., Pauli, F., Batzoglou, S.,
Bernstein, B. E., Bickel, P., Brown, J. B., Cayting, P., et al. (2012). ChIP-seq guidelines
and practices of the ENCODE and modENCODE consortia. Genome research 22, 1813-
1831.
Lay, F. D., Triche, T. J., Jr., Tsai, Y. C., Su, S. F., Martin, S. E., Daneshmand, S.,
Skinner, E. C., Liang, G., Chihara, Y., and Jones, P. A. (2014). Reprogramming of the
human intestinal epigenome by surgical tissue transposition. Genome research 24, 545-
553.
Lemaire, M., Momparler, L. F., Raynal, N. J., Bernstein, M. L., and Momparler, R. L.
(2009). Inhibition of cytidine deaminase by zebularine enhances the antineoplastic action
of 5-aza-2'-deoxycytidine. Cancer chemotherapy and pharmacology 63, 411-416.
Ley, T. J., Ding, L., Walter, M. J., McLellan, M. D., Lamprecht, T., Larson, D. E.,
Kandoth, C., Payton, J. E., Baty, J., Welch, J., et al. (2010). DNMT3A mutations in acute
myeloid leukemia. The New England journal of medicine 363, 2424-2433.
Li, B., Carey, M., and Workman, J. L. (2007). The role of chromatin during transcription.
Cell 128, 707-719.
Li, H., Ruan, J., and Durbin, R. (2008). Mapping short DNA sequencing reads and
calling variants using mapping quality scores. Genome research 18, 1851-1858.
Li, L. H., Olin, E. J., Buskirk, H. H., and Reineke, L. M. (1970). Cytotoxicity and mode
of action of 5-azacytidine on L1210 leukemia. Cancer research 30, 2760-2769.
Lin, J., Gilbert, J., Rudek, M. A., Zwiebel, J. A., Gore, S., Jiemjit, A., Zhao, M., Baker, S.
D., Ambinder, R. F., Herman, J. G., et al. (2009). A phase I dose-finding study of 5-
azacytidine in combination with sodium phenylbutyrate in patients with refractory solid
tumors. Clinical cancer research : an official journal of the American Association for
Cancer Research 15, 6241-6249.
Lin, J. C., Jeong, S., Liang, G., Takai, D., Fatemi, M., Tsai, Y. C., Egger, G., Gal-Yam,
E. N., and Jones, P. A. (2007). Role of nucleosomal occupancy in the epigenetic silencing
of the MLH1 CpG island. Cancer cell 12, 432-444.
179
Lister, R., Pelizzola, M., Dowen, R. H., Hawkins, R. D., Hon, G., Tonti-Filippini, J.,
Nery, J. R., Lee, L., Ye, Z., Ngo, Q. M., et al. (2009). Human DNA methylomes at base
resolution show widespread epigenomic differences. Nature 462, 315-322.
Lister, R., Pelizzola, M., Kida, Y. S., Hawkins, R. D., Nery, J. R., Hon, G., Antosiewicz-
Bourget, J., O'Malley, R., Castanon, R., Klugman, S., et al. (2011). Hotspots of aberrant
epigenomic reprogramming in human induced pluripotent stem cells. Nature 471, 68-73.
Liu, Y., Siegmund, K. D., Laird, P. W., and Berman, B. P. (2012). Bis-SNP: Combined
DNA methylation and SNP calling for Bisulfite-seq data. Genome biology 13, R61.
Lorch, Y., LaPointe, J. W., and Kornberg, R. D. (1987). Nucleosomes inhibit the
initiation of transcription but allow chain elongation with the displacement of histones.
Cell 49, 203-210.
Lubbert, M., Suciu, S., Baila, L., Ruter, B. H., Platzbecker, U., Giagounidis, A.,
Selleslag, D., Labar, B., Germing, U., Salih, H. R., et al. (2011). Low-dose decitabine
versus best supportive care in elderly patients with intermediate- or high-risk
myelodysplastic syndrome (MDS) ineligible for intensive chemotherapy: final results of
the randomized phase III study of the European Organisation for Research and Treatment
of Cancer Leukemia Group and the German MDS Study Group. Journal of clinical
oncology : official journal of the American Society of Clinical Oncology 29, 1987-1996.
Lujambio, A., Calin, G. A., Villanueva, A., Ropero, S., Sanchez-Cespedes, M., Blanco,
D., Montuenga, L. M., Rossi, S., Nicoloso, M. S., Faller, W. J., et al. (2008). A
microRNA DNA methylation signature for human cancer metastasis. Proceedings of the
National Academy of Sciences of the United States of America 105, 13556-13561.
Lyons, R. M., Cosgriff, T. M., Modi, S. S., Gersh, R. H., Hainsworth, J. D., Cohn, A. L.,
McIntyre, H. J., Fernando, I. J., Backstrom, J. T., and Beach, C. L. (2009). Hematologic
response to three alternative dosing schedules of azacitidine in patients with
myelodysplastic syndromes. Journal of clinical oncology : official journal of the
American Society of Clinical Oncology 27, 1850-1856.
Maekita, T., Nakazawa, K., Mihara, M., Nakajima, T., Yanaoka, K., Iguchi, M., Arii, K.,
Kaneda, A., Tsukamoto, T., Tatematsu, M., et al. (2006). High levels of aberrant DNA
methylation in Helicobacter pylori-infected gastric mucosae and its possible association
with gastric cancer risk. Clinical cancer research : an official journal of the American
Association for Cancer Research 12, 989-995.
Martin, M. G., Walgren, R. A., Procknow, E., Uy, G. L., Stockerl-Goldstein, K., Cashen,
A. F., Westervelt, P., Abboud, C. N., Kreisel, F., Augustin, K., et al. (2009). A phase II
study of 5-day intravenous azacitidine in patients with myelodysplastic syndromes.
American journal of hematology 84, 560-564.
Mill, J., and Heijmans, B. T. (2013). From promises to practical strategies in epigenetic
epidemiology. Nature reviews Genetics 14, 585-594.
180
Miranda, T. B., Kelly, T. K., Bouazoune, K., and Jones, P. A. (2010a). Methylation-
sensitive single-molecule analysis of chromatin structure. Current protocols in molecular
biology / edited by Frederick M Ausubel [et al] Chapter 21, Unit 21 17 21-16.
Miranda, T. B., Kelly, T. K., Bouazoune, K., and Jones, P. A. (2010b). Methylation-
sensitive single-molecule analysis of chromatin structure. Curr Protoc Mol Biol Chapter
21, Unit 21 17 21-16.
Mohn, F., Weber, M., Rebhan, M., Roloff, T. C., Richter, J., Stadler, M. B., Bibel, M.,
and Schubeler, D. (2008). Lineage-specific polycomb targets and de novo DNA
methylation define restriction and potential of neuronal progenitors. Molecular cell 30,
755-766.
Molaro, A., Hodges, E., Fang, F., Song, Q., McCombie, W. R., Hannon, G. J., and Smith,
A. D. (2011). Sperm methylation profiles reveal features of epigenetic inheritance and
evolution in primates. Cell 146, 1029-1041.
Muller-Thomas, C., Schuster, T., Peschel, C., and Gotze, K. S. (2009). A limited number
of 5-azacitidine cycles can be effective treatment in MDS. Annals of hematology 88,
213-219.
Musto, P., Maurillo, L., Spagnoli, A., Gozzini, A., Rivellini, F., Lunghi, M., Villani, O.,
Aloe-Spiriti, M. A., Venditti, A., and Santini, V. (2010). Azacitidine for the treatment of
lower risk myelodysplastic syndromes : a retrospective study of 74 patients enrolled in an
Italian named patient program. Cancer 116, 1485-1494.
Nakabayashi, K., Trujillo, A. M., Tayama, C., Camprubi, C., Yoshida, W., Lapunzina, P.,
Sanchez, A., Soejima, H., Aburatani, H., Nagae, G., et al. (2011). Methylation screening
of reciprocal genome-wide UPDs identifies novel human-specific imprinted genes.
Human molecular genetics 20, 3188-3197.
Nan, X., Ng, H. H., Johnson, C. A., Laherty, C. D., Turner, B. M., Eisenman, R. N., and
Bird, A. (1998). Transcriptional repression by the methyl-CpG-binding protein MeCP2
involves a histone deacetylase complex. Nature 393, 386-389.
Nguyen, C. T., Weisenberger, D. J., Velicescu, M., Gonzales, F. A., Lin, J. C., Liang, G.,
and Jones, P. A. (2002). Histone H3-lysine 9 methylation is associated with aberrant gene
silencing in cancer cells and is rapidly reversed by 5-aza-2'-deoxycytidine. Cancer
research 62, 6456-6461.
Noushmehr, H., Weisenberger, D. J., Diefes, K., Phillips, H. S., Pujara, K., Berman, B.
P., Pan, F., Pelloski, C. E., Sulman, E. P., Bhat, K. P., et al. (2010). Identification of a
CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer cell
17, 510-522.
O'Hagan, H. M., Wang, W., Sen, S., Destefano Shields, C., Lee, S. S., Zhang, Y. W.,
Clements, E. G., Cai, Y., Van Neste, L., Easwaran, H., et al. (2011). Oxidative damage
181
targets complexes containing DNA methyltransferases, SIRT1, and polycomb members
to promoter CpG Islands. Cancer cell 20, 606-619.
Ohm, J. E., McGarvey, K. M., Yu, X., Cheng, L., Schuebel, K. E., Cope, L., Mohammad,
H. P., Chen, W., Daniel, V. C., Yu, W., et al. (2007). A stem cell-like chromatin pattern
may predispose tumor suppressor genes to DNA hypermethylation and heritable
silencing. Nature genetics 39, 237-242.
Okano, M., Bell, D. W., Haber, D. A., and Li, E. (1999). DNA methyltransferases
Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development.
Cell 99, 247-257.
Oki, Y., Kantarjian, H. M., Gharibyan, V., Jones, D., O'Brien, S., Verstovsek, S., Cortes,
J., Morris, G. M., Garcia-Manero, G., and Issa, J. P. (2007). Phase II study of low-dose
decitabine in combination with imatinib mesylate in patients with accelerated or myeloid
blastic phase of chronic myelogenous leukemia. Cancer 109, 899-906.
Philipson, B. M., Hockenstrom, T., and Akerlund, S. (1987). Biological consequences of
exposing ileal mucosa to urine. World journal of surgery 11, 790-797.
Plummer, R., Vidal, L., Griffin, M., Lesley, M., de Bono, J., Coulthard, S., Sludden, J.,
Siu, L. L., Chen, E. X., Oza, A. M., et al. (2009). Phase I study of MG98, an
oligonucleotide antisense inhibitor of human DNA methyltransferase 1, given as a 7-day
infusion in patients with advanced solid tumors. Clinical cancer research : an official
journal of the American Association for Cancer Research 15, 3177-3183.
Pogribny, I. P., and Beland, F. A. (2009). DNA hypomethylation in the origin and
pathogenesis of human diseases. Cellular and molecular life sciences : CMLS 66, 2249-
2261.
Portela, A., and Esteller, M. (2010). Epigenetic modifications and human disease. Nature
biotechnology 28, 1057-1068.
Portela, A., Liz, J., Nogales, V., Setien, F., Villanueva, A., and Esteller, M. (2013). DNA
methylation determines nucleosome occupancy in the 5'-CpG islands of tumor suppressor
genes. Oncogene 32, 5421-5428.
Qin, T., Jelinek, J., Si, J., Shu, J., and Issa, J. P. (2009). Mechanisms of resistance to 5-
aza-2'-deoxycytidine in human cancer cell lines. Blood 113, 659-667.
Rach, E. A., Winter, D. R., Benjamin, A. M., Corcoran, D. L., Ni, T., Zhu, J., and Ohler,
U. (2011). Transcription initiation patterns indicate divergent strategies for gene
regulation at the chromatin level. PLoS genetics 7, e1001274.
Rada-Iglesias, A., Bajpai, R., Prescott, S., Brugmann, S. A., Swigut, T., and Wysocka, J.
(2012). Epigenomic annotation of enhancers predicts transcriptional regulators of human
neural crest. Cell stem cell 11, 633-648.
182
Rada-Iglesias, A., Bajpai, R., Swigut, T., Brugmann, S. A., Flynn, R. A., and Wysocka, J.
(2011). A unique chromatin signature uncovers early developmental enhancers in
humans. Nature 470, 279-283.
Rakyan, V. K., Down, T. A., Balding, D. J., and Beck, S. (2011). Epigenome-wide
association studies for common human diseases. Nature reviews Genetics 12, 529-541.
Ramirez-Carrozzi, V. R., Braas, D., Bhatt, D. M., Cheng, C. S., Hong, C., Doty, K. R.,
Black, J. C., Hoffmann, A., Carey, M., and Smale, S. T. (2009). A unifying model for the
selective regulation of inducible transcription by CpG islands and nucleosome
remodeling. Cell 138, 114-128.
Reinius, L. E., Acevedo, N., Joerink, M., Pershagen, G., Dahlen, S. E., Greco, D.,
Soderhall, C., Scheynius, A., and Kere, J. (2012). Differential DNA methylation in
purified human blood cells: implications for cell lineage and studies on disease
susceptibility. PloS one 7, e41361.
Rhee, I., Bachman, K. E., Park, B. H., Jair, K. W., Yen, R. W., Schuebel, K. E., Cui, H.,
Feinberg, A. P., Lengauer, C., Kinzler, K. W., et al. (2002). DNMT1 and DNMT3b
cooperate to silence genes in human cancer cells. Nature 416, 552-556.
Rius, M., Stresemann, C., Keller, D., Brom, M., Schirrmacher, E., Keppler, D., and Lyko,
F. (2009). Human concentrative nucleoside transporter 1-mediated uptake of 5-
azacytidine enhances DNA demethylation. Molecular cancer therapeutics 8, 225-231.
Rivera, C. M., and Ren, B. (2013). Mapping human epigenomes. Cell 155, 39-55.
Romanelli, V., Nakabayashi, K., Vizoso, M., Moran, S., Iglesias-Platas, I., Sugahara, N.,
Simon, C., Hata, K., Esteller, M., Court, F., and Monk, D. (2014). Variable maternal
methylation overlapping the nc886/vtRNA2-1 locus is locked between hypermethylated
repeats and is frequently altered in cancer. Epigenetics : official journal of the DNA
Methylation Society 9.
Sandoval, J., Heyn, H., Moran, S., Serra-Musach, J., Pujana, M. A., Bibikova, M., and
Esteller, M. (2011). Validation of a DNA methylation microarray for 450,000 CpG sites
in the human genome. Epigenetics : official journal of the DNA Methylation Society 6,
692-702.
Santini, V. (2009). Azacitidine in lower-risk myelodysplastic syndromes. Leukemia
research 33 Suppl 2, S22-26.
Schlesinger, Y., Straussman, R., Keshet, I., Farkash, S., Hecht, M., Zimmerman, J., Eden,
E., Yakhini, Z., Ben-Shushan, E., Reubinoff, B. E., et al. (2007). Polycomb-mediated
methylation on Lys27 of histone H3 pre-marks genes for de novo methylation in cancer.
Nature genetics 39, 232-236.
183
Schones, D. E., Cui, K., Cuddapah, S., Roh, T. Y., Barski, A., Wang, Z., Wei, G., and
Zhao, K. (2008). Dynamic regulation of nucleosome positioning in the human genome.
Cell 132, 887-898.
Schrump, D. S., Fischette, M. R., Nguyen, D. M., Zhao, M., Li, X., Kunst, T. F., Hancox,
A., Hong, J. A., Chen, G. A., Pishchik, V., et al. (2006). Phase I study of decitabine-
mediated gene expression in patients with cancers involving the lungs, esophagus, or
pleura. Clinical cancer research : an official journal of the American Association for
Cancer Research 12, 5777-5785.
Schulte, J. H., Lim, S., Schramm, A., Friedrichs, N., Koster, J., Versteeg, R., Ora, I.,
Pajtler, K., Klein-Hitpass, L., Kuhfittig-Kulle, S., et al. (2009). Lysine-specific
demethylase 1 is strongly expressed in poorly differentiated neuroblastoma: implications
for therapy. Cancer research 69, 2065-2071.
Sean Davis, P. D., Sven Bilke, Tim Triche,Jr and Moiz Bootwalla (2012). methylumi:
Handle Illumina methylation data. R package version 2.4.0. In.
Segura-Pacheco, B., Perez-Cardenas, E., Taja-Chayeb, L., Chavez-Blanco, A., Revilla-
Vazquez, A., Benitez-Bribiesca, L., and Duenas-Gonzalez, A. (2006). Global DNA
hypermethylation-associated cancer chemotherapy resistance and its reversion with the
demethylating agent hydralazine. Journal of translational medicine 4, 32.
Sekeres, M. A., List, A. F., Cuthbertson, D., Paquette, R., Ganetzky, R., Latham, D.,
Paulic, K., Afable, M., Saba, H. I., Loughran, T. P., Jr., and Maciejewski, J. P. (2010).
Phase I combination trial of lenalidomide and azacitidine in patients with higher-risk
myelodysplastic syndromes. Journal of clinical oncology : official journal of the
American Society of Clinical Oncology 28, 2253-2258.
Sharma, S., De Carvalho, D. D., Jeong, S., Jones, P. A., and Liang, G. (2011).
Nucleosomes containing methylated DNA stabilize DNA methyltransferases 3A/3B and
ensure faithful epigenetic inheritance. PLoS genetics 7, e1001286.
Shen, H., and Laird, P. W. (2013). Interplay between the cancer genome and epigenome.
Cell 153, 38-55.
Shen, L., Toyota, M., Kondo, Y., Lin, E., Zhang, L., Guo, Y., Hernandez, N. S., Chen,
X., Ahmed, S., Konishi, K., et al. (2007). Integrated genetic and epigenetic analysis
identifies three different subclasses of colon cancer. Proceedings of the National
Academy of Sciences of the United States of America 104, 18654-18659.
Silverman, L. R., Demakos, E. P., Peterson, B. L., Kornblith, A. B., Holland, J. C.,
Odchimar-Reissig, R., Stone, R. M., Nelson, D., Powell, B. L., DeCastro, C. M., et al.
(2002). Randomized controlled trial of azacitidine in patients with the myelodysplastic
syndrome: a study of the cancer and leukemia group B. Journal of clinical oncology :
official journal of the American Society of Clinical Oncology 20, 2429-2440.
184
Singh, N., Duenas-Gonzalez, A., Lyko, F., and Medina-Franco, J. L. (2009). Molecular
modeling and molecular dynamics studies of hydralazine with human DNA
methyltransferase 1. ChemMedChem 4, 792-799.
Song, Y., and Zhang, C. (2009). Hydralazine inhibits human cervical cancer cell growth
in vitro in association with APC demethylation and re-expression. Cancer chemotherapy
and pharmacology 63, 605-613.
Soriano, A. O., Yang, H., Faderl, S., Estrov, Z., Giles, F., Ravandi, F., Cortes, J., Wierda,
W. G., Ouzounian, S., Quezada, A., et al. (2007). Safety and clinical activity of the
combination of 5-azacytidine, valproic acid, and all-trans retinoic acid in acute myeloid
leukemia and myelodysplastic syndrome. Blood 110, 2302-2308.
Srinivasan, U., Reaman, G. H., Poplack, D. G., Glaubiger, D. L., and LeVine, A. S.
(1982). Phase II study of 5-azacytidine in sarcomas of bone. American journal of clinical
oncology 5, 411-415.
Stadler, M. B., Murr, R., Burger, L., Ivanek, R., Lienert, F., Scholer, A., van Nimwegen,
E., Wirbelauer, C., Oakeley, E. J., Gaidatzis, D., et al. (2011). DNA-binding factors
shape the mouse methylome at distal regulatory regions. Nature 480, 490-495.
Steensma, D. P. (2009). Decitabine treatment of patients with higher-risk myelodysplastic
syndromes. Leukemia research 33 Suppl 2, S12-17.
Steensma, D. P., Baer, M. R., Slack, J. L., Buckstein, R., Godley, L. A., Garcia-Manero,
G., Albitar, M., Larsen, J. S., Arora, S., Cullen, M. T., and Kantarjian, H. (2009).
Multicenter study of decitabine administered daily for 5 days every 4 weeks to adults
with myelodysplastic syndromes: the alternative dosing for outpatient treatment
(ADOPT) trial. Journal of clinical oncology : official journal of the American Society of
Clinical Oncology 27, 3842-3848.
Stein, J. P., Clark, P., Miranda, G., Cai, J., Groshen, S., and Skinner, D. G. (2005).
Urethral tumor recurrence following cystectomy and urinary diversion: clinical and
pathological characteristics in 768 male patients. The Journal of urology 173, 1163-1168.
Stewart, D. J., Issa, J. P., Kurzrock, R., Nunez, M. I., Jelinek, J., Hong, D., Oki, Y., Guo,
Z., Gupta, S., and Wistuba, II (2009). Decitabine effect on tumor global DNA
methylation and other parameters in a phase I trial in refractory solid tumors and
lymphomas. Clinical cancer research : an official journal of the American Association for
Cancer Research 15, 3881-3888.
Stresemann, C., Brueckner, B., Musch, T., Stopper, H., and Lyko, F. (2006). Functional
diversity of DNA methyltransferase inhibitors in human cancer cell lines. Cancer
research 66, 2794-2800.
Stresemann, C., and Lyko, F. (2008). Modes of action of the DNA methyltransferase
inhibitors azacytidine and decitabine. International journal of cancer Journal international
du cancer 123, 8-13.
185
Su, S. F., de Castro Abreu, A. L., Chihara, Y., Tsai, Y., Andreu-Vieyra, C., Daneshmand,
S., Skinner, E. C., Jones, P. A., Siegmund, K. D., and Liang, G. (2014). A panel of three
markers hyper- and hypomethylated in urine sediments accurately predicts bladder cancer
recurrence. Clinical cancer research : an official journal of the American Association for
Cancer Research 20, 1978-1989.
Suzuki, T., Tanaka, R., Hamada, S., Nakagawa, H., and Miyata, N. (2010). Design,
synthesis, inhibitory activity, and binding mode study of novel DNA methyltransferase 1
inhibitors. Bioorganic & medicinal chemistry letters 20, 1124-1127.
Taberlay, P. C., Kelly, T. K., Liu, C. C., You, J. S., De Carvalho, D. D., Miranda, T. B.,
Zhou, X. J., Liang, G., and Jones, P. A. (2011). Polycomb-repressed genes have
permissive enhancers that initiate reprogramming. Cell 147, 1283-1294.
Taberlay, P. C., Statham, A. L., Kelly, T. K., Clark, S. J., and Jones, P. A. (2014).
Reconfiguration of nucleosome depleted regions at distal regulatory elements
accompanies DNA methylation of enhancers and insulators in cancer. Genome research.
Takai, D., and Jones, P. A. (2002). Comprehensive analysis of CpG islands in human
chromosomes 21 and 22. Proceedings of the National Academy of Sciences of the United
States of America 99, 3740-3745.
Taylor, S. M., and Jones, P. A. (1979). Multiple new phenotypes induced in 10T1/2 and
3T3 cells treated with 5-azacytidine. Cell 17, 771-779.
Tazi, J., and Bird, A. (1990). Alternative chromatin structure at CpG islands. Cell 60,
909-920.
The Cancer Genome Atlas (2012). Comprehensive molecular characterization of human
colon and rectal cancer. Nature 487, 330-337.
Thurman, R. E., Rynes, E., Humbert, R., Vierstra, J., Maurano, M. T., Haugen, E.,
Sheffield, N. C., Stergachis, A. B., Wang, H., Vernot, B., et al. (2012). The accessible
chromatin landscape of the human genome. Nature 489, 75-82.
Tillo, D., and Hughes, T. R. (2009). G+C content dominates intrinsic nucleosome
occupancy. BMC bioinformatics 10, 442.
Tillo, D., Kaplan, N., Moore, I. K., Fondufe-Mittendorf, Y., Gossett, A. J., Field, Y.,
Lieb, J. D., Widom, J., Segal, E., and Hughes, T. R. (2010). High nucleosome occupancy
is encoded at human regulatory sequences. PloS one 5, e9129.
Toyota, M., Suzuki, H., Sasaki, Y., Maruyama, R., Imai, K., Shinomura, Y., and Tokino,
T. (2008). Epigenetic silencing of microRNA-34b/c and B-cell translocation gene 4 is
associated with CpG island methylation in colorectal cancer. Cancer research 68, 4123-
4132.
186
Treppendahl, M. B., Qiu, X., Sogaard, A., Yang, X., Nandrup-Bus, C., Hother, C.,
Andersen, M. K., Kjeldsen, L., Mollgaard, L., Hellstrom-Lindberg, E., et al. (2012a).
Allelic methylation levels of the noncoding VTRNA2-1 located on chromosome 5q31.1
predict outcome in AML. Blood 119, 206-216.
Treppendahl, M. B., Qiu, X., Sogaard, A., Yang, X., Nandrup-Bus, C., Hother, C.,
Andersen, M. K., Kjeldsen, L., Mollgard, L., Hellstrom-Lindberg, E., et al. (2012b).
Allelic methylation levels of the noncoding VTRNA2-1 located on chromosome 5q31.1
predict outcome in AML. Blood 119, 206-216.
Triche, T. J., Jr., Weisenberger, D. J., Van Den Berg, D., Laird, P. W., and Siegmund, K.
D. (2013). Low-level processing of Illumina Infinium DNA Methylation BeadArrays.
Nucleic acids research 41, e90.
Valouev, A., Johnson, S. M., Boyd, S. D., Smith, C. L., Fire, A. Z., and Sidow, A.
(2011). Determinants of nucleosome organization in primary human cells. Nature 474,
516-520.
Varley, K. E., Gertz, J., Bowling, K. M., Parker, S. L., Reddy, T. E., Pauli-Behn, F.,
Cross, M. K., Williams, B. A., Stamatoyannopoulos, J. A., Crawford, G. E., et al. (2013).
Dynamic DNA methylation across diverse human cell lines and tissues. Genome research
23, 555-567.
Velez-Garcia, E., Vogler, W. R., Bartolucci, A. A., and Arkun, S. N. (1977). Twice
weekly 5-azacytidine infusion in dissmeinated metastatic cancer: a phase II study. Cancer
treatment reports 61, 1675-1677.
Vogelstein, B., Papadopoulos, N., Velculescu, V. E., Zhou, S., Diaz, L. A., Jr., and
Kinzler, K. W. (2013). Cancer genome landscapes. Science 339, 1546-1558.
Voso, M. T., Santini, V., Finelli, C., Musto, P., Pogliani, E., Angelucci, E., Fioritoni, G.,
Alimena, G., Maurillo, L., Cortelezzi, A., et al. (2009). Valproic acid at therapeutic
plasma levels may increase 5-azacytidine efficacy in higher risk myelodysplastic
syndromes. Clinical cancer research : an official journal of the American Association for
Cancer Research 15, 5002-5007.
Walker, C. L., and Ho, S. M. (2012). Developmental reprogramming of cancer
susceptibility. Nature reviews Cancer 12, 479-486.
Walter, M. J., Ding, L., Shen, D., Shao, J., Grillot, M., McLellan, M., Fulton, R.,
Schmidt, H., Kalicki-Veizer, J., O'Laughlin, M., et al. (2011). Recurrent DNMT3A
mutations in patients with myelodysplastic syndromes. Leukemia 25, 1153-1158.
Weber, M., Hellmann, I., Stadler, M. B., Ramos, L., Paabo, S., Rebhan, M., and
Schubeler, D. (2007). Distribution, silencing potential and evolutionary impact of
promoter DNA methylation in the human genome. Nature genetics 39, 457-466.
187
Weinstein, J. N., Collisson, E. A., Mills, G. B., Shaw, K. R., Ozenberger, B. A., Ellrott,
K., Shmulevich, I., Sander, C., and Stuart, J. M. (2013). The Cancer Genome Atlas Pan-
Cancer analysis project. Nature genetics 45, 1113-1120.
Whyte, W. A., Orlando, D. A., Hnisz, D., Abraham, B. J., Lin, C. Y., Kagey, M. H.,
Rahl, P. B., Lee, T. I., and Young, R. A. (2013). Master transcription factors and
mediator establish super-enhancers at key cell identity genes. Cell 153, 307-319.
Widschwendter, M., Fiegl, H., Egle, D., Mueller-Holzner, E., Spizzo, G., Marth, C.,
Weisenberger, D. J., Campan, M., Young, J., Jacobs, I., and Laird, P. W. (2007).
Epigenetic stem cell signature in cancer. Nature genetics 39, 157-158.
Wilson, B. G., and Roberts, C. W. (2011). SWI/SNF nucleosome remodellers and cancer.
Nature reviews Cancer 11, 481-492.
Wilson, V. L., and Jones, P. A. (1983). DNA methylation decreases in aging but not in
immortal cells. Science 220, 1055-1057.
Winquist, E., Knox, J., Ayoub, J. P., Wood, L., Wainman, N., Reid, G. K., Pearce, L.,
Shah, A., and Eisenhauer, E. (2006). Phase II trial of DNA methyltransferase 1 inhibition
with the antisense oligonucleotide MG98 in patients with metastatic renal carcinoma: a
National Cancer Institute of Canada Clinical Trials Group investigational new drug study.
Investigational new drugs 24, 159-167.
Wu, H., Chen, Y., Liang, J., Shi, B., Wu, G., Zhang, Y., Wang, D., Li, R., Yi, X., Zhang,
H., et al. (2005). Hypomethylation-linked activation of PAX2 mediates tamoxifen-
stimulated endometrial carcinogenesis. Nature 438, 981-987.
Wullt, B., Holst, E., Steven, K., Carstensen, J., Pedersen, J., Gustafsson, E., Colleen, S.,
and Mansson, W. (2004). Microbial flora in ileal and colonic neobladders. European
urology 45, 233-239.
Xi, H., Shulha, H. P., Lin, J. M., Vales, T. R., Fu, Y., Bodine, D. M., McKay, R. D.,
Chenoweth, J. G., Tesar, P. J., Furey, T. S., et al. (2007). Identification and
characterization of cell type-specific and ubiquitous chromatin regulatory structures in
the human genome. PLoS genetics 3, e136.
Xie, W., Schultz, M. D., Lister, R., Hou, Z., Rajagopal, N., Ray, P., Whitaker, J. W.,
Tian, S., Hawkins, R. D., Leung, D., et al. (2013). Epigenomic analysis of multilineage
differentiation of human embryonic stem cells. Cell 153, 1134-1148.
Xie, X., Mikkelsen, T. S., Gnirke, A., Lindblad-Toh, K., Kellis, M., and Lander, E. S.
(2007). Systematic discovery of regulatory motifs in conserved regions of the human
genome, including thousands of CTCF insulator sites. Proceedings of the National
Academy of Sciences of the United States of America 104, 7145-7150.
188
Xu, M., Kladde, M. P., Van Etten, J. L., and Simpson, R. T. (1998). Cloning,
characterization and expression of the gene coding for a cytosine-5-DNA
methyltransferase recognizing GpC. Nucleic acids research 26, 3961-3966.
Yan, X. J., Xu, J., Gu, Z. H., Pan, C. M., Lu, G., Shen, Y., Shi, J. Y., Zhu, Y. M., Tang,
L., Zhang, X. W., et al. (2011). Exome sequencing identifies somatic mutations of DNA
methyltransferase gene DNMT3A in acute monocytic leukemia. Nature genetics 43, 309-
315.
Yang, X., Lay, F., Han, H., and Jones, P. A. (2010). Targeting DNA methylation for
epigenetic therapy. Trends in pharmacological sciences 31, 536-546.
Yang, X., Noushmehr, H., Han, H., Andreu-Vieyra, C., Liang, G., and Jones, P. A.
(2012). Gene reactivation by 5-aza-2'-deoxycytidine-induced demethylation requires
SRCAP-mediated H2A.Z insertion to establish nucleosome depleted regions. PLoS
genetics 8, e1002604.
Yoo, C. B., Cheng, J. C., and Jones, P. A. (2004). Zebularine: a new drug for epigenetic
therapy. Biochemical Society transactions 32, 910-912.
Yoo, C. B., Chuang, J. C., Byun, H. M., Egger, G., Yang, A. S., Dubeau, L., Long, T.,
Laird, P. W., Marquez, V. E., and Jones, P. A. (2008). Long-term epigenetic therapy with
oral zebularine has minimal side effects and prevents intestinal tumors in mice. Cancer
prevention research 1, 233-240.
Yoo, C. B., and Jones, P. A. (2006). Epigenetic therapy of cancer: past, present and
future. Nature reviews Drug discovery 5, 37-50.
You, J. S., and Jones, P. A. (2012). Cancer genetics and epigenetics: two sides of the
same coin? Cancer cell 22, 9-20.
You, J. S., Kelly, T. K., De Carvalho, D. D., Taberlay, P. C., Liang, G., and Jones, P. A.
(2011). OCT4 establishes and maintains nucleosome-depleted regions that provide
additional layers of epigenetic regulation of its target genes. Proceedings of the National
Academy of Sciences of the United States of America 108, 14497-14502.
Zhang, Z., Wippo, C. J., Wal, M., Ward, E., Korber, P., and Pugh, B. F. (2011). A
packing mechanism for nucleosome organization reconstituted across a eukaryotic
genome. Science 332, 977-980.
Zhou, V. W., Goren, A., and Bernstein, B. E. (2011). Charting histone modifications and
the functional organization of mammalian genomes. Nature reviews Genetics 12, 7-18.
Zhu, J., Adli, M., Zou, J. Y., Verstappen, G., Coyne, M., Zhang, X., Durham, T., Miri,
M., Deshpande, V., De Jager, P. L., et al. (2013). Genome-wide chromatin state
transitions associated with developmental and environmental cues. Cell 152, 642-654.
189
Zilberman, D., Coleman-Derr, D., Ballinger, T., and Henikoff, S. (2008). Histone H2A.Z
and DNA methylation are mutually antagonistic chromatin marks. Nature 456, 125-129.
Ziller, M. J., Gu, H., Muller, F., Donaghey, J., Tsai, L. T., Kohlbacher, O., De Jager, P.
L., Rosen, E. D., Bennett, D. A., Bernstein, B. E., et al. (2013). Charting a dynamic DNA
methylation landscape of the human genome. Nature 500, 477-481.

190
APPENDIX

DETAILED METHODOLOGY OF NUCLEOSOME OCCUPANCY AND
METHYLOME SEQUENCING (NOME-SEQ)

This chapter is adapted from an invited book chapter that will be published in the third
edition of DNA Methylation Protocols in Methods in Molecular Biology.

1. Introduction

This chapter covers procedures for performing locus-specific and genome-wide
NOMe-seq (Kelly et al., 2012; Taberlay et al., 2011; You et al., 2011). Both protocols
begin by incubating freshly isolated nuclei from cell cultures or primary tissue with the
enzyme and its substrate, S-adenosylhomocysteine (SAM) for a total of two 7.5 minutes
reaction times. Following purification of enzyme-treated genomic DNA, DNA
methylation and nucleosome occupancy pattern in specific genomic loci is determined by
cloning the products of PCR-amplified bisulfite converted DNA and sequencing
individual colony (Figure A1, left panel). For global analysis, whole-genome bisulfite
sequencing library is generated by ligating methylated adapters to the end-repaired and
A-tailed DNA fragments before bisulfite conversion (Figure A1, right panel). Following
whole-genome amplification of bisulfite DNA, gel purification is then performed in order
to remove residual and self-ligated adapters and select a size-range of templates to be
sequenced (Hawkins et al., 2010; Lister et al., 2009).
Here, we also delineate an alternative method for treating formaldehyde-fixed
chromatin isolated from primary human tissues, fresh and/or fresh-frozen. The fixing of
chromatin reduces the effects of dynamic nucleosome movement that may occur during
the mechanical isolation of nuclei and inadvertently results in the methylation of GpC
191
sites otherwise inaccessible. This approach, however, involves a much longer enzyme
incubation time which is necessary in order to ensure the efficient methylation of all
accessible GpC sites while reducing the bias that may arise from enzyme overtreatment.
Similar to our standard protocol, the isolated DNA can be subsequently used for locus-
specific and genome-wide analyses to examine the chromatin landscape of human tissues.

192
Figure A1 Overview of the NOMe-seq protocol. The assay starts by treating freshly
isolated nuclei with M.CviPI in a total reaction time of 15 minutes. M.CviPI-treated
DNA can be analyzed in loci-specific or genome-wide manner, the steps of which are
outlined in flow chart including the approximate time required for each step
193
2. Materials

2.1.Isolation of Nuclei
1. 1X PBS, sterile
2. Trypsin for detaching adherent cells
3. Trypan blue and hemocytometer
4. Lysis Buffer (10 mM Tris, pH 7.4, 10mM NaCl, 3mM MgCl2, 0.1mM EDTA
0.5% NP-40)
5. Wash Buffer (10 mM Tris, pH 7.4, 10mM NaCl, 3mM MgCl2, 0.1 mM EDTA)
6. Proteinase-inhibitor cocktail
7. Optional: High-Salt Wash Buffer (10mM Tris, pH 7.4, 3mM MgCl2, 0.3M
Sucrose, 400mM NaCl)
8. Optional: Dounce Homogenizer
9. Optional: 37% Formaldehyde to crosslink cells
10. Optional: 1.25 M glycine
11. Optional: Bioruptor (Diagenode) or other sonicator
2.2. GpC Methyltransferase reaction
1. 10X GpC Buffer
2. 32 mM S-adenosylhomocysteine (SAM)
3. 4U/ul M.CvIPI
4. 1M Sucrose
5. Nuclease-free water
6. Stop Buffer (20mM Tris, pH 7.4, 600 mM NaCl, 1% SDS, 10mM EDTA)
7. Proteinase-inhibitor cocktail
2.3. DNA Extraction
1. TE buffer
2. Proteinase K
3. 1:1 Phenol: Chloroform
4. 100% Ethanol
5. Nanodrop Spectrophotometer
6. Optional: Glycogen
7. Optional: Phase-lock gel tubes
194
8. Optional: 5M NaCl to reverse crosslink cells if using alternative method
2.4. Bisulfite conversion.
1. EZ DNA Methylation kit (Zymo) or other commercially available kit
2. Nuclease-free water
2.5. PCR
1. Target Primers
2. PCR Master Mix
3. Gel Extraction Kit
2.6. TA Cloning
1. TOPO TA Cloning Kit (Life Technologies)
2. OneShot TOP10 Competent Cells (Life Technologies)
3. LB agar plate supplemented with ampicillin
4. 20mg/ml X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside)
5. Insert Primers or M13 Primers
6. TempliPhi (GE Healthcare Life Sciences)
2.7. Locus-specific Sequence Analysis
1. BiQ Analyzer
2. Optional: Methylviewer
2.8. Genome-wide bisulfite sequencing library construction
1. End-Repair DNA kit (Epicentre)
2. Klenow, DNA Polymerase I (NEB)
3. 1mM dATP (NEB)
4. Minelute Reaction Cleanup Kit (Qiagen)
5. Rapid T4 DNA Ligase (Enzymatics)
6. 2x Rapid Ligation Buffer (Enzymatics)
7. 10 mM dNTP Mix
8. DNA Methyl Adapter or DNA Adapter Index (Illumina)
9. Covaris S2 system
10. Optional: Agencourt AMPure XP magnetic beads
11. Optional: Illumina Pair-End Sample Prep Kit
12. Optional: Truseq DNA Sample Prep Kit
195
2.9. Whole-genome amplification
1. KAPA HiFi HotStart Uracil+ ReadyMix PCR Kit (KAPA Biosystems)
2. Gel Purification Kit
3. Optional: Agencourt AMPure XP magnetic beads
4. Agencourt Bioanalyzer

3. Method
3.1 Isolation of Nuclei
1. Trypsinize exponentially growing adherent cells or collect susupension cells and
centrifuge at 250 g at 4°C for 5 minutes.
2. Remove the media and wash cells with 10 mL sterile PBS. Take care to keep cells
on ice or at 4°C at all times. Centrifuge at 250 g at 4°C for 5 minutes and remove
PBS wash.
3. Resuspend 1 million cells in 1mL ice-cold lysis buffer and let sit undisturbed on
ice for 5-10 minutes (see Note 1). We use 250,000 cells per reaction, but
recommend performing each reaction in duplicate and including no enzyme
control. Check a small aliquot of cells under the microscope using trypan blue and
hemocytometer for intact nuclei and to ensure that lysis is complete before
proceeding to the next step (Note 2).
4. Following the incubation, centrifuge for 5 minutes at 750 g in 4°C and discard
supernatant, taking care not to disturb the nuclear pellet.
5. Gently resuspend nuclei in 1 mL ice-cold wash buffer (see Note 3). Centrifuge for
5 minutes at 750 g in 4°C, discard supernatant and immediately proceed to
M.CviPI treatment.

3.2. Treating Nuclei with M.CviPI.
1. Resuspend nuclei in 1x GpC buffer to obtain a final concentration of 250,000
cells per 282 µl volume (see Note 4).
2. In a microcentrifuge tube, prepare the reaction mixture in the following order
(Note 5):

196
1M Sucrose 150 µl
10x GpC Buffer 17 µl
Nuclei (250,000) 282 µl
32 mM SAM 1.5 µl
4U/µl M.CviPI 50 µl
----------------------
Total 500 µl

We recommend using fresh GpC buffer and SAM each time. SAM in particular
is unstable and thus proper storage based on the manufacturer’s instruction is
crucial. Freeze-thawing of reagents should also be avoided.
3. Incubate the reaction for 7.5 minutes at 37°C.
4. Add a boost of enzyme and substrate in the following amount (Note 6):

32 mM SAM 1.5 µl
4U/µl M.CviPI 25 µl
----------------------
Total 26.5 µl

5. Incubate for additional 7.5 minutes at 37°C. Do not incubate for longer as over-
treatment may occur and result in methylation of inaccessible regions.
6. Stop the reaction by adding an equal volume of stop buffer (526.5 µl), which was
pre-warmed at 37°C to dissolve precipitate (Note 7).

3.1. Alternative method: Isolation of nuclei
The following alternative method is adapted for fresh or fresh frozen primary tissue
(Note 8).
1. Using clean scalpel, dissect tissue into 1-3mm pieces and resuspend them in
10mL of DMEM per gram of tissue (Note 9).
2. Add formaldehyde to a final concentration of 1% and gently rotate for 10-15
minutes at room temperature.
197
3. Stop the crosslinking by adding glycine to a final concentration of 0.125M and
rotate for 5 minutes.
4. Wash the crosslinked tissue twice with ice cold PBS supplemented with
proteinase inhibitors (PIC). Centrifuge at 300 g for 5 minutes at 4°C and remove
supernatant.
5. Resuspend tissue in 1mL of ice cold lysis buffer per 100mg of tissue and
incubate on ice for 5-10 minutes. For subsequent steps, scale up the reaction if
using more than 100mg of tissue, and separate reactions into multiple tubes if
necessary.
6. Transfer the mixture into a chilled dounce homogenizer. Using the glass pestle,
grind the tissue on ice for 10-20 strokes to release the nuclei. Depending on the
tissue type, extra strokes may be required. We recommend monitoring the lysis
process using a microscope.
7. Transfer nuclei into a 1.5ml microcentrifuge tube and centrifuge at 750 g for 10
minutes at 4°C. Remove supernatant.
8. Resuspend tissue in ice-cold wash buffer and centrifuge at 750 g for 10 minutes
at 4°C. Remove supernatant.

3.2. Alternative Method: Treating Nuclei with M.CviPI
1. Resuspend pellet in 282 µl of ice-cold 1x GpC buffer supplemented with PIC.
2. Sonicate the chromatin to a fragment 1-2kb in 4°C. We regularly perform 3 cycles
of 30 seconds on and 30 seconds off on the Bioruptor (Diagenode) at 4°C.
3. To the microcentrifuge tube, add the following reaction mixture
Chromatin 282 µl
1M Sucrose 150 µl
10x GpC Buffer 17 µl
32 mM SAM 1 µl
4U/ul M.CviPI 25 µl
----------------------+
475 µl

198
4. Incubate at 37°C for 1 hour.
5. Boost with 1 µl of SAM and incubate for additional hour.
6. Repeat step 5.
7. After a total of 3 hours incubation time, add an equal volume (477 µl) of stop
buffer (Note 10).

3.3. Purification of enzyme-treated DNA
Follow step 1 if using standard method and step 2 if using alternative method.
1. Add 200 µg/ml of Proteinase K to each reaction tube and incubate for 16 hours at
55°C to remove excess enzyme.
2. Reverse cross-link chromatin by adding 5M NaCl to a final concentration of
0.2M and incubating at 65°C for 4-6 hours. Add 200 µg/ml of Proteinase K and
incubate for 16 hours at 55°C.
3. Purify DNA using a standard phenol/chloroform extraction method followed by
ethanol precipitation or DNA purification columns. (Optional: during
phenol/chloroform extraction, phased-lock gel can be used to assist the
separation of aqueous and organic phase). Ethanol precipitation can be carried
out at -20°C for overnight or at -80°C for 1-2 hours. During ethanol precipitation,
20 µg/ml of glycogen can also be added as a carrier.
4. Resuspend DNA pellet in 20 µl of nuclease-free water or TE buffer. Quantify
DNA and store long term in -20°C.

3.4. Bisulfite conversion
1. Bisulfite conversion can be performed using a commercially available kit
according to the manufacturer’s instructions. We regularly use EZ DNA
Methylation kit from Zymo Research to convert 500ng-1ug of DNA per reaction
although other kits are also available.
2. Elute DNA in 20 µl of nuclease-free water. Bisulfite converted DNA is stable in -
20°C for up to a year.

199
3.5. PCR of bisulfite-converted DNA
1. Design primers for regions of interest that are specific for bisulfite-converted
DNA (Figure 3, Note 11).
2. Perform PCR using Taq polymerase for 40-45 cycles using 2 µl of bisulfite
converted DNA and Taq polymerase that adds 3’ –A overhangs to the end of PCR
products which is necessary for subsequent TA cloning. Make sure that the Taq
polymerase is able to tolerate the presence of Uracils in the DNA template. A
final extension time of up to 10 minutes, depending on the robustness of the
chosen Taq, should be included to ensure that the polymerase appends the end of
the PCR products with an –A overhang. The following is an example of PCR
condition using Bioline’s MyTaq Red Mix:
2x MyTaq Red Mix 12.5 µl
10 µM Forward Primer 1 µl
10 µM Reverse Primer 1 µl
Nuclease-free Water 8.5 µl
Bisulfite DNA 2 µl
----------------------
25 µl
Cycling condition:
a. 95°C 1 min
b. 95°C 30 secs
c. Ta (52°-62°C) 30 secs
d. 72°C 30 secs
Repeat b-d for 40-45 cycles
e. 72°C 5 mins
Annealing temperature (Ta) should be optimized for each primer set so that it is
specific to the bisulfite converted DNA and does not amplify genomic DNA.
3. It is recommended to always include no template and genomic DNA control to
ensure purity and efficiency (Note 12).

200
Figure A2 NOMe-seq primer design. In the sense strand, all C’s that are not part of
CpG (bold, italicized) or GpC (bold, underlined) dinucleotide to T (upper case). Then
design a forward primer that is complementary to the antisense strand. The primer (bold,
lower case) should be 20-30bp long with an annealing temperature of 50-60°C, does not
contain a CpG or GpC sites and if possible, end with converted Cs (boxed). The reverse
primer should follow the same criteria, but designed to complement the sense strand.
201
3.6. TA-Cloning
1. PCR products that are specific to target region can be used directly for TA cloning
although it is highly recommended to perform gel extraction to remove excess
primers and polymerase which might interfere with subsequent ligation step.
2. Clone freshly purified PCR products (within 24 hours) using the TOPO TA
Cloning for Sequencing kit according to manufacturer’s instruction and transform
using TOP10 competent cells (Note 13). If ligation within 24 hours is not
possible, incubate the PCR products with Taq and dATP for 10 minutes at 72°C
to add the –A overhangs prior to ligation.
3. On each LB-agar plate containing ampicillin, add 40 µl of 20mg/ml X-gal and
spread evenly.
4. Plate recovered TOP10 cells and incubate at 37°C overnight.
5. Screen for colonies with positive insert using either the insert primers or M13
primers located in the backbone of the vector. Plasmid DNA can be amplified
with Templiphi or purified with minipreps according to manufacturer’s
instruction.
6. Sequence individual clones. We recommend using M13 reverse primer for
sequencing, although primers specific to the insert can also be used (Note 14).

3.7. Sequence Analysis
Various softwares are available for bisulfite sequencing analysis, such as BiQ
Analyzer (Bock et al., 2005; Lutsik et al., 2011) and MethylViewer (Pardo et al.,
2011). The theory behind the analysis of M.CviPI-treated sequences is similar to the
analysis of regular bisulfite sequencing with the following modifications:
• GpC dinucleotides that are not occupied by tight binding transcription factors or
nucleosomes will be accessible and thus methylated by the enzyme. Following
bisulfite sequencing, these GpC sites will remain as “GC”, whereas inaccessible
and unmethylated sites will be sequenced as “GT”. Nucleosome-occupied regions
are defined as inaccessible regions of >147-bp DNA whereas protection due to
DNA binding factors generally are smaller (<50 bp).
202
• GCG sites are excluded from downstream analysis as we cannot determine
whether the methylation status of the cytosines is due to endogenous DNA
methylation or due to GpC methyltransferase activity (Figure 4).
• Depending on the alignment software used, it may be necessary to reduce the
minimal alignment requirement to ~60% to compensate for the Cs in the GC
context that have not been bisulfite converted.
• For visualization, we often generate bubble maps to represent the methylation
status of CG and GC sites which are drawn to scale (Figure 5).

203
Figure A3: NOMe-seq sequence analysis. Align sequenced clones to bisulfite-
converted sequence. When present in GpC (bold, underlined) or CpG (bold, italicized)
context, methylated C will be read as C and unmethylated C will be read as T.
Methylation status is often represented in a bubble map where methylated CpG is shown
as black circle and methylated GpC is shown as teal circle. Cs present in GCG context
(overlined) are excluded from the analysis as it cannot be determined whether the
methylation status is endogenous, or enzyme-treated.

187
204
A
B
Figure A4 Visualization of locus-specific and genome-wide NOMe-seq. (A) Bubble
map chart is shown for the Xist locus which is present in multiple chromatin
configurations, endogenously unmethylated (black circles, white filled) and accessible
(blue circles, teal filled), and endogenously methylated (black circles, black filled) and
inaccessible (blue circles, white filled). Regions of inaccessibility larger than 147bp are
covered by a pink rectangle indicating nucleosome occupancy. (B) Screenshot of IGV
browser showing the average HCG and GCH methylation status in the GAPDH locus
following whole-genome bisulfite sequencing.

205
3.8. Whole-genome bisulfite sequencing library construction
Before starting, see Note 15.
1. Sonicate 3-5 µg of enzyme-treated genomic DNA to 150bp (Note 16).
2. Ethanol precipitate sonicated DNA and resuspend in 34 µl of nuclease free water.
3. End-repair sonicated DNA using the End-It DNA End-Repair Kit (Epicenter). In a
1.5ml microcentrifuge tube, add the following reaction mixture:
DNA (3-5 µg) : 34 µl
10x End-Repair Buffer: 5 µl
2.5 mM dNTPs : 5 µl
10 mM ATP : 5 µl
END-IT enzyme mix : 1 µl
----------------------------
Total 50 µl
4. Incubate at room temperature for 45 minutes.
5. Purify DNA using Minelute kit (Qiagen) according to the manufacturer’s
instruction, and elute in 32 µl of EB buffer. Alternatively, purification step can be
done using magnetic beads (AMPure) according to the manufacturer’s instruction.
6. Add an A’ base to the 3’ end of the DNA in following reaction mixture:
DNA : 32 µl
10x Klenow Buffer : 5 µl
1mM dATP : 10 µl
Klenow (5U/µl) : 3 µl
-----------------------------
Total 50 µl
7. Incubate for 30 minutes at 37°C.
8. Purify DNA using Minelute kit (Qiagen) according to manufacturer’s instruction
and elute in 14 µl of EB buffer. Alternatively, the purification step can be done
using magnetic beads such as AMPure beads.
9. Ligate methylated adapters to the ends of the DNA fragments (Note 17). Prepare
the following mixture:
DNA : 14 µl
206
2X Ligase Buffer : 25 µl
Methylated Adapter : 10 µl
DNA Ligase : 1 µl
-----------------------------
Total 50 µl

10. Incubate for 15 minutes at room temperature.
11. Purify DNA using Minelute kit (Qiagen) according to the manufacturer’s
instruction and elute in 25 µl of EB buffer. Alternatively, the purification step can
be done using magnetic beads or by gel extraction (Note 18).
12. Quantify purified DNA (Note 19).
13. Perform bisulfite conversion (see section 3.4). Divide DNA into multiple
reactions, converting no more than 1 µg of DNA per reaction. Elute each reaction
in 10 µl of nuclease free water and pool bisulfite converted DNA together in a
microcentrifuge tube.

3.9. Whole Genome Amplification
1. Perform no more than 8 cycles of PCR using high-fidelity Taq polymerase. We
routinely used 2X KAPA-Hifi+Uracil master mix to amplify 20 µl of purified
bisulfite converted library using a final concentration of 500nM primers designed
to amplify the methylated adapters.
2. Use PCR cycle conditions as recommended by the manufacturer.
Cycling Condition (Note 20):
a. 98°C 45 secs
b. 98°C 15 secs
c. 65°C 30 secs
d. 72°C 30 secs
Repeat b-d for 5-8 cycles
e. 72°C 60 secs
207
3. Add loading buffer to a final concentration of 1X and load all PCR product and
DNA ladder to a 2% gel. Run gel at 120V for 45-60 minutes or until clear
separation for lower molecular weight products is achieved (Note 21).
4. With a clean scalpel, excise bands with fragment size of 200-500bp (Note 22).
5. Purify DNA using a gel extraction kit. Melt the gel slowly at room temperature
and proceed with the protocol as indicated by the manufacturer. Elute in no more
than 20µl of nuclease-free water and store in -20°C long term.
6. Measure the concentration and size distribution of library using Bioanalyzer.
7. Next generation sequencing is commonly done on HiSeq-2000 using 50 or 75 PE
or 100 SE reads (Note 22). Methylation status of CG and GC sites can be
extracted using Bis-SNP or other WGBS analysis tools (Liu et al., 2012).

4. Notes
1. We recommend optimizing the lysis condition for each cell type by adjusting the
incubation time and concentration of NP-40 in the lysis buffer as some cells are
more easily lysed than the others. For example, fibroblasts such as IMR90 cells
require longer incubation time in lysis buffer containing 0.5%NP-40 whereas
K562 and GM12878 cells can be lysed in <3 minutes in 0.25% NP-40. Nuclei can
also be isolated by mechanically disrupting the cells using dounce homogenizer in
the presence of NP-40. Monitor lyses by removing a few µl every minute until
adequate cell lysis is achieved.
2. For genome-wide application, due to the loses seen with bisulfite conversion, 3-
5µg of footprinted DNA is required to start the library preparation procedure. We
recommend performing multiple enzyme reactions for each experimental
condition. DNA extracted from multiple reactions can be pooled together as
technical replicates.
3. It is critical that the nuclei remain intact throughout the wash and reaction step.
Intact nuclei will appear as blue circles under the microscope when visualized
with trypan-blue. To resuspend nuclei in the wash buffer, we recommend using a
p-1000 pipette tip which has wider opening than p-200 tip in order to minimize
friction which may cause the nuclei to burst. Additionally, wash step can be
208
performed in high-salt condition (100-400mM) in order to remove transcription
factors.
4. Always include no-methyltransferase control to help determine the level of
endogenous DNA methylation in ambiguous regions such as the GCG
trinucleotide.
5. A highly concentrated version of M.CviPI, 50U/µl, can also be specially ordered
from New England Biolabs. If using this version of the enzyme, use the following
reaction mixture:
1M Sucrose 45 µl
10x GpC Buffer 5 µl
Nuclei (250,000) 94.5 µl
32 mM SAM 1.5 µl
50 U/µl M.CviPI 4 µl
----------------------
Total 150 µl

6. If using 50U/µl of enzyme, boost with the following amount:
32 mM SAM 1.5 µl
50 U/µl M.CviPI 2 µl
----------------------
Total 3.5 µl

7. If using 50U/µl of enzyme as described in Note 5, add 153.5 µl of stop solution.
8. The standard method described in section 3.1 and 3.2 may also be used for
treating nuclei freshly released from single cell suspension of primary tissue
which is prepared by enzymatic or mechanical disaggregation. However, the
nuclei isolation procedure often varies depending on the tissue type and
composition and is more laborious compared to isolating nuclei from cell-lines. In
order to minimize the technical variability between tissues, we isolated chromatin
from cross-linked tissue and treated the chromatin with M.CviPI for a longer
incubation time.
209
9. Use different clean scalpels for each sample when processing multiple tissues at
the same time to avoid contamination. If using frozen tissue (-80°C storage),
avoid freeze-thawing the tissue and minimize the time when dissecting.
10. Over-treatment is not a concern when using fixed chromatin, however, it is
critical to incubate the reaction for a minimum of 1 hour and replenish the SAM
periodically when performing longer incubation in order to ensure that all
accessible GpC sites are methylated.
11. The primers should not include CpG and GpC sites in the sequence which will
result in a biased PCR amplification. The size of amplicons should be around 350-
700bp so as to provide enough resolution for the region being analyzed. Longer
amplicons, on the other hand, are difficult to amplify and sequence due to the
fragmentation that occurs during bisulfite conversion.
12. In order to avoid PCR bias, we recommend performing the PCR and cloning in
duplicates while combining the sequencing results. Alternatively, spike-ins of
known ratios of methylated and unmethylated amplicons may be included in the
PCR reaction.
13. The manufacturer’s instruction for the TOPO kit recommends a total reaction
volume of 6 µl (4 µl PCR product, 1 µl of TA vector and 1 µl of 150mM NaCl),
however, we routinely halve the reaction without adverse affect. We subsequently
transform all 3 µl of reaction in 25 µl of competent cells.
14. The number of clones required for clear footprinting pattern will depend on the
regions and condition being assessed. We recommend sequencing a minimum of
10 clones to start which generally suffices for most regions being interrogated.
15. Preparation of WGBS libraries outlined in this section is based on previously
described methods (Berman et al., 2012; Hawkins et al., 2010; Kelly et al., 2012).
Alternatively, Truseq DNA Methyl Prep kit can also be used to generate libraries.
We also recommend validating a few known control regions at loci-specific level
before proceeding to this step in order to confirm that the enzyme treatment is
successful.
16. We regularly use the Covaris S2 system to fragment DNA for WGBS in order to
generate tightly distributed DNA fragments. DNA is diluted to a total volume of
210
130 µl and transferred into 6x16mm microtubes, taking care to avoid air bubbles.
Sonication is performed at 10% duty cycle, 5 intensity and 200 cycles per burst
for 6 minutes.
17. The methylated adaptor listed here is the same as the Truseq adaptor used by
Illumina. If barcoding libraries, indexed adaptors from Illumina or Bioo Scientific
can be used.
18. If performing gel extraction, we recommend using 2% low melting gel or a 50:50
mixture of low and regular melting gel. Excise tightly in order to limit excess gel
that may interfere with subsequent purification step and use clean scalpel each
time to avoid cross-contamination between samples. In either case, gel should be
melted slowly at room temperature as to not bias library complexity based on CG
density.
19. At this stage, DNA can be safely stored long term to -20°C if not proceeding to
the next step.
20. It is critical to minimize the PCR cycles to be less than 10 cycles so as to avoid
excessive PCR duplicates in the libraries which will bias subsequent analysis.
21. During gel extraction, PCR products may be difficult to see when present in low
amount. DNA ladder should be used as a guide when excising size-selected
regions.
22. Take care to avoid contamination from self-ligated adapter dimers which usually
appear at ~150bp.
23. For genome-wide analysis, we generally aim to have a minimum of 300-400
million uniquely mapped sequencing reads. Due to the reduced complexity of
bisulfite converted genomes, 50-bp SE reads are likely not sufficient for obtaining
a large number of uniquely mappable reads.

211

Abstract (if available)

Abstract In the mammalian genome, the biochemical addition of a methyl group on the cytosine of CpG dinucleotides, or DNA methylation, is a component of the epigenetic landscape that is most frequently associated with gene silencing. The pattern of DNA methylation is established during development, is unique for each cell type and can be faithfully copied or inherited during somatic cell divisions. Significant disruption of the normal methylome has emerged as a key signature of human malignancies and thus, an attractive target for therapy. Nevertheless, the factors driving DNA methylation changes and the molecular implication of these changes are still poorly understood. This dissertation aims to address the contribution of environmental cues on the alteration of DNA methylation patterns and delineate the effects of DNA methylation changes on the structure of the cancer epigenome. ❧ Using an in vivo model, I hereby demonstrate that specific signals from the local tissue environment may be required for the maintenance of a cell-type specific methylome and that changes in the tissue environment drastically alter the epigenetic landscape of human intestinal epithelial cells. Comprehensive investigation of the loss of DNA methylation, using a global hypomethylation colorectal cancer cell line model and techniques such as NOMe-seq, ChIP-seq and RNA-seq, reveal that the effects of DNA methylation are highly variable depending on genomic context. Specifically, I show that DNA methylation controls CpG islands (CGI), but not non-CGI, promoters at the chromatin level such that the loss of methylation in cancer may result in the reestablishment of a permissive and/or poised chromatin landscape, characterized by the presence of certain histone modifications and well-phased nucleosomes. By treating cancer cells with the FDA-approved demethylating agent, 5-Aza-CdR, I also illustrate that DNA methylation contributes to the silencing of distal regulatory regions, which abrogates the necessary crosstalks between promoter and enhancers for gene activation. Furthermore, I expand my work to generate the first integrated map of DNA methylation and nucleosome positioning of colon adenocarcinomas, which is used to examine the relationship between the epigenome and the underlying disease. Taken together, this dissertation presents a detailed view of the human epigenome, normal and cancer, and illustrates an often-overlooked influence of DNA methylation in shaping the chromatin landscape.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding DNA methylation and nucleosome organization in cancer cells using single molecule sequencing

PDF

Identification of novel epigenetic biomarkers and microRNAs for cancer therapeutics

PDF

Effects of chromatin regulators during carcinogenesis

PDF

The relationship between DNA methylation and transcription factor binding in colon cancer cells

PDF

Functional role of chromatin remodeler proteins in cancer biology

PDF

Comparative analysis of DNA methylation in mammals

PDF

DNA methylation changes in the development of lung adenocarcinoma

PDF

CpG poor promoter SULT1C2 regulated by DNA methylation and is induced by cigarette smoke condensate in lung cell lines

PDF

Identification of DNA methylation markers in diffuse large B-cell lymphoma

PDF

DNA methylation inhibitors and epigenetic regulation of microRNA expression

PDF

Using epigenetic toggle switches to repress tumor-promoting gene expression

PDF

Role of DNA methyltransferases 3A and 3B in inheritance of DNA methylation patterns

PDF

DNA hypermethylation: its role in colorectal tumorigenesis and potential clinical applications

PDF

Identification and characterization of cancer-associated enhancers

PDF

Integrated genomic & epigenomic analyses of glioblastoma multiforme: Methods development and application

PDF

Breast epithelial cell type specific enhancers and functional annotation of breast cancer risk loci

PDF

Preprocessing and analysis of DNA methylation microarrays

PDF

DNA methylation as a biomarker in human reproductive health and disease

PDF

DNA methylation and gene expression profiles in Vidaza treated cultured cancer cells

PDF

An analysis of conservation of methylation

Asset Metadata

Creator Lay, Fides D. (author)

Core Title Functional DNA methylation changes in normal and cancer cells

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Genetic, Molecular and Cellular Biology

Publication Date 10/15/2016

Defense Date 10/15/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cancer,DNA methylation,environment,epigenetic therapy,epigenetics,histone modification,nucleosome positioning,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Jones, Peter Anthony (committee chair), Coetzee, Gerhard (Gerry) A. (committee member), Farnham, Peggy J. (committee member)

Creator Email flay@usc.edu,lay.fides@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-507662

Unique identifier UC11298225

Identifier etd-LayFidesD-3016.pdf (filename),usctheses-c3-507662 (legacy record id)

Legacy Identifier etd-LayFidesD-3016.pdf

Dmrecord 507662

Document Type Dissertation

Format application/pdf (imt)

Rights Lay, Fides D.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA