<%BANNER%>

WIF1 is a Frequent Target for Aberrant Hypermethyation in Cervical Squamous Cell Carcinoma

Permanent Link: http://ufdc.ufl.edu/UFE0042181/00001

Material Information

Title: WIF1 is a Frequent Target for Aberrant Hypermethyation in Cervical Squamous Cell Carcinoma
Physical Description: 1 online resource (57 p.)
Language: english
Creator: Delmas, Amber
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: cervical, dna, epigenetics, mapit, wif1, wnt
Medicine -- Dissertations, Academic -- UF
Genre: Medical Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN CERVICAL SQUAMOUS CELL CARCINOMA By Amber Delmas August 2010 Aberrant activation of the Wnt/?-catenin signaling pathway is a prominent oncogenic mechanism in numerous cancer types, and evidence indicates this pathway is of importance in human cervical cancer. Wnt Inhibitory Factor 1 (WIF1) is one of several secreted antagonists that bind to Wnt. Although it has been described as a tumor suppressor in various types of cancer, regulation of the WIF1 gene has not been examined in human cervical cancer. Here we show that WIF1 was unmethylated and expressed in normal cervical epithelium and aberrantly methylated in high-grade, squamous cell carcinomas. Three of four cervical tumor cell lines showed low or no WIF1 expression. In all three, expression increased when cells were cultured with a DNA demethylating drug, indicating epigenetic silencing of WIF1. Interestingly, differences were seen in chromatin structure between expressing and non-expressing cell lines as shown by the chromatin footprinting technique, MAPit. The WIF1 promoter was aberrantly methylated in ten of seventeen high-grade squamous cell cervical tumors, compared to paired normal tissue. Immunohistochemistry of normal stratified squamous cervical epithelium revealed that WIF1 expression was high in peribasal cells, but gradually diminished toward the superficial epithelial layer. WIF1 protein was not detectable in tumors with high WIF1 methylation and was absent in two of the seven unmethylated tumors. Thus, although WIF1 was hypermethylated in most cervical cancer samples, other mechanisms may also contribute to its repression. Our findings establish the WIF1 gene as commonly hypermethylated and a possible target for epigenetic silencing in squamous cell carcinoma.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amber Delmas.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Kladde, Michael P.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042181:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042181/00001

Material Information

Title: WIF1 is a Frequent Target for Aberrant Hypermethyation in Cervical Squamous Cell Carcinoma
Physical Description: 1 online resource (57 p.)
Language: english
Creator: Delmas, Amber
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: cervical, dna, epigenetics, mapit, wif1, wnt
Medicine -- Dissertations, Academic -- UF
Genre: Medical Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN CERVICAL SQUAMOUS CELL CARCINOMA By Amber Delmas August 2010 Aberrant activation of the Wnt/?-catenin signaling pathway is a prominent oncogenic mechanism in numerous cancer types, and evidence indicates this pathway is of importance in human cervical cancer. Wnt Inhibitory Factor 1 (WIF1) is one of several secreted antagonists that bind to Wnt. Although it has been described as a tumor suppressor in various types of cancer, regulation of the WIF1 gene has not been examined in human cervical cancer. Here we show that WIF1 was unmethylated and expressed in normal cervical epithelium and aberrantly methylated in high-grade, squamous cell carcinomas. Three of four cervical tumor cell lines showed low or no WIF1 expression. In all three, expression increased when cells were cultured with a DNA demethylating drug, indicating epigenetic silencing of WIF1. Interestingly, differences were seen in chromatin structure between expressing and non-expressing cell lines as shown by the chromatin footprinting technique, MAPit. The WIF1 promoter was aberrantly methylated in ten of seventeen high-grade squamous cell cervical tumors, compared to paired normal tissue. Immunohistochemistry of normal stratified squamous cervical epithelium revealed that WIF1 expression was high in peribasal cells, but gradually diminished toward the superficial epithelial layer. WIF1 protein was not detectable in tumors with high WIF1 methylation and was absent in two of the seven unmethylated tumors. Thus, although WIF1 was hypermethylated in most cervical cancer samples, other mechanisms may also contribute to its repression. Our findings establish the WIF1 gene as commonly hypermethylated and a possible target for epigenetic silencing in squamous cell carcinoma.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amber Delmas.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Kladde, Michael P.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042181:00001


This item has the following downloads:


Full Text





WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN
CERVICAL SQUAMOUS CELL CARCINOMA





















By

Amber Delmas


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2010



































2010 Amber Delmas









ACKNOWLEDGMENTS

I am most thankful to both of my mentors, Mike and Kevin, for believing in me and constantly

stimulating my brain. Mike is an extremely intelligent individual who has continuously inspired

me to think outside of the box, while Kevin is one of the most dedicated and hardworking

individuals I know. Together you make the perfect boss.



To my committee members: Jim Resnick and Paul Gulig, thank you for your guidance and

support.



My deepest appreciation to my lab family: Carolina Pardo, Russell Darst, Nancy Nabilsi, Santhi

Pondugula, and Chris Hoffman. Thank you for putting up with my endless questions and

constant interruptions. I also wish to thank my extended lab family: Lisa Dyer for the mass of

cell culture work and Ryan Skehan for his wealth of computer knowledge.



I want to acknowledge the Moffitt Cancer Center and Erin Siegel in Tampa, Florida and wish to

thank Bridget Riggs for her hard work and dedication.



I am thankful to Joyce Conners and all of her hard work in assisting every Master's student along

the road from acceptance to graduation.



I thank my Dad for trying to understand what exactly I'm researching and to my Mom for not

trying to have me explain what I do. I wish to express thanks to John Barli for allowing me the

opportunity to support myself through school, and to everyone else who has come in and out of

my life to get me where I am today.









TABLE OF CONTENTS page


A CK N O W LED G M EN T S ...................................................... ........................................ 3

LIST OF FIGURES ................................... .. .... .... ................. .6

L IST O F A B B R E V IA TIO N S ....................................................................... ... ....................... 7

A B S T R A C T ............ ................... ............................................................ 10

CHAPTER

1 BACKGROUND ...................................... ....................... 12

C erv ical C an cer ................................................................12
H um an Papillom virus .................. ....................................... .......... .... 13
H um an C ell C ulture........... ...................................................................... ....... .. 15
E pig en etics..... ...................................................16
DN A M ethylation ..................................................... ............... ...... ... ... 16
H stone M modifications ..................................... ........ ............ .. .......... .. 17
N uclesom e Positioning .................. ........................................ .. ...... .... 18
Epigenetics in Cancer ..................................... .................. ............ .. ............ 20
A berrant M ethylation ............................................ .. .. .... ........ ......... 20
C hrom atin C changes in C ancer.............................................................. .....................2 1
W n t P ath w ay ................... .. .............. ..... ........ ...................................... .. 2 3
W nt A gonists and A ntagonists .............................................. .............................. 23
W n t R ecep to rs ......................................................................... 2 4
C anonical Pathw ay ........................ .. ........................ .. ............. ........ 24
N oncanonical P athw ay .......................................................................... ....................26

2 M A TER IA L S A N D M ETH O D S ........................................ .............................................28

C ervical Specim ens .......... ...... .................................................................. .... ...... .... 28
Cell Culture and 5-aza-2'-deoxycytidine (5-azadC) Treatment............................................28
R N A and D N A Isolation ............................................................................. .....................29
Sodium B isulfite M odification ...................................................................... ...................30
Polymerase Chain Reaction and Pyrosequencing............................................ .............31
R reverse Transcriptase PCR ............................. ............................................. ............... 31
Methyltransferase Accessibility Protocol for Individual Templates (MAPit) ........................32
Im munohistochem istry ..................................... .. ........... .. ............33

3 R E SU L T S .......................................................................................... 3 5

WIF1 is Epigenetically Silenced in Cervical Cancer Cell Lines ................. .................35
Aberrant Chromatin Footprint in CaSki and SiHa Cell Lines..............................................39
WIF1 Promoter is Aberrantly Methylated in Squamous Cell Cervical Tumors.....................41
WIF1 Methylation is Not Correlated with WIF 1 Protein Levels ............... ............ .....43









4 DISCU SSION ........... .... ..... ... .... ............................ ............... 46

L IST O F R E F E R E N C E S ......... ...... ........... ................. .................... ..................... ...............5 1

B IO G R A PH IC A L SK E T C H ............................................................................... .....................57









LIST OF FIGURES


Figure page

1-1 Normal Cervical Epithelium, CIN1 and CIN2-3 .................................... ............... 13

1-2 D iagram of O pen and Closed Chrom atin............................... ............... .....................19

1-3 Diagram of Canonical W nt Pathw ay ........................................... .......................... 25

3-1 RT-PCR of Cervical Cell Lines Before and After 5-azadC Treatment .............................36

3-2 WIF1 Promoter and Pyrograms of Cervical Cell Lines................. ............... ..............37

3-3 Average Methylation of the WIF1 Promoter Before and After 5-azadC Treatment .........38

3-4 Chromatin Probing of CaSki and SiHa Cell Line..................................40

3-5 Average CpG methylation of Cervical SSC and Adjacent Normal Tissue at WIF1 .........42

3-6 Pyrograms of Pap Smear Samples at WIF1 ............................................ ............... 42

3-7 WIF1 Immunohistochemistry of Normal Cervical Tissue and SSC...............................44









LIST OF ABBREVIATIONS

APC adenomatous polyposis coli

ATP adenosine-5'-triphosphate

CCD charged couple device

CK1 casein kinase 1

CLB cell lysis buffer

CIN cervical intraepithelial neoplasia

CIS carcinoma in situ

CpG cytosine guanine dinucleotide

CRB cell resuspension buffer

DAB diaminobenzidine

dg.dH20 degassed distilled water

dH20 distilled water

DKK Dickkopf

DNA deoxyribonucleic acid

DNMT deoxyribonucleic acid methyltransferase

dNTP deoxyribonucleotide triphosphate

Dvl Dishevelled

EGF epidermal growth factor

FBS fetal bovine serum

FFPE formalin-fixed paraffin-embedded

Fz or Fzd Frizzled receptor

GSK3p glycogen synthase kinase 30

GTPase guanine triphosphotase

HAT histone acetyltransferase









HDAC histone deacetylase

HDM histone demethylase

HMGA2 high mobility group AT-hook 2

HMT histone methyltransferase

HPV human papillioma virus

hr hours)

IHC immunohistochemistry

JNK c-jun N-terminal kinases

LEF lymphoid enhancer factor

LRP lipoprotein receptor-related protein

MB methylation buffer

MAPit methyltransferase accessibility protocol for individual templates

MAZ MYC-associated zinc finger

min minutes)

NF-kB nuclear factor-kappa B

NFR nucleosome-free region

NuRD nucleosome remodeling and deacetylase

Pap Papanicolau

PCR polymerase chain reaction

PPi pyrophosphate

Rb retinoblastoma protein

RNA ribonucleic acid

RT-PCR reverse transcriptase polymerase chain reaction

SCC squamous cell carcinoma

sec seconds)









sFRP secreted Frizzled receptor protein

SWI/SNF switch/sucrose non-fermentable

TCF T cell factor

TSG tumor suppressor gene

TSS transcription start site

UICC International Union Against Cancer

WD WIF1 domain

WIF 1 Wnt inhibitor factor 1

WHO World Health Organization









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN
CERVICAL SQUAMOUS CELL CARCINOMA

By

Amber Delmas

August 2010

Chair: Michael Kladde
Major: Medical Sciences

Aberrant activation of the Wnt/p-catenin signaling pathway is a prominent oncogenic

mechanism in numerous cancer types, and evidence indicates this pathway is of importance in

human cervical cancer. Wnt Inhibitory Factor 1 (WIF ) is one of several secreted antagonists

that bind to Wnt. Although it has been described as a tumor suppressor in various types of

cancer, regulation of the WIF 1 gene has not been examined in human cervical cancer. Here we

show that WIF1 was unmethylated and expressed in normal cervical epithelium and aberrantly

methylated in high-grade, squamous cell carcinomas. Three of four cervical tumor cell lines

showed low or no WIF1 expression. In all three, expression increased when cells were cultured

with a DNA demethylating drug, indicating epigenetic silencing of WIF1. Interestingly,

differences were seen in chromatin structure between expressing and non-expressing cell lines as

shown by the chromatin footprinting technique, MAPit. The WIF1 promoter was aberrantly

methylated in ten of seventeen high-grade squamous cell cervical tumors, compared to paired

normal tissue. Immunohistochemistry of normal stratified squamous cervical epithelium revealed

that WIF1 expression was high in peribasal cells, but gradually diminished toward the superficial

epithelial layer. WIF 1 protein was not detectable in tumors with high WIF1 methylation and was

absent in two of the seven unmethylated tumors. Thus, although WIF1 was hypermethylated in









most cervical cancer samples, other mechanisms may also contribute to its repression. Our

findings establish the WIF1 gene as commonly hypermethylated and a possible target for

epigenetic silencing in squamous cell carcinoma.









CHAPTER 1
BACKGROUND

Cervical Cancer

In 2009, the American Cancer Society estimated within a year approximately 4,070

women would die from cervical cancer in the United States. About 11,270 new cases of invasive

cervical cancer would be diagnosed, and four times as many women would have a non-invasive

cervical cancer [1]. Although cervical cancers can develop in the mucus-producing gland cells of

the endocervix, most form in the lining of the ectocervix, which is composed of stratified

squamous epithelium. The area defining the border between the endocervix and ectocervix is

most susceptible to dysplasia, and is termed the transformation zone (Figure 1-1A). Dysplasia is

characterized by four major pathological microscopic changes: unequally sized cells,

abnormally-shaped cells, an unusual number of cells that are currently dividing and

hyperchromatism (degeneration of cell nuclei or increased staining capacity) [2]. If molecular

changes occur in the basal cells where these cells originate from, they can promote the normal

tissue through increasing grades of dysplasia and if not caught, cervical carcinoma. The naming

and classification of cervical carcinoma precursor lesions has changed several times. Originally,

the World Health Organization (WHO) classified lesions as mild, moderate, and severe dysplasia

or carcinoma in situ (CIS). These precursor lesions were graded depending on the degree of

disruption of epithelial differentiation. These classifications were later standardized as cervical

intraepithelial neoplasia (CIN), with mild dysplasia designated CIN1 and severe as CIN3. The

most recent classification is the Bethesda System, which divides all cervical epithelial precursor

lesions into two groups [3]: low-grade squamous intraepithelial lesions that correspond to CIN1

(Figure 1-1B) and high-grade squamous intraepithelial lesions that include CIN2 and 3 (Figure

1-1C). These cellular changes can be detected by the Papanicolau test (Pap test) and treated to










prevent the development of cancer. Treatment consists of surgery in early stages or

chemotherapy and radiotherapy in advanced stages of the disease.



A Transformalion zone
Normal squemoIs epithelium
Norrnal glalncular
opilhalium







B .. C







Y. I t Q, P.
-AW
I : -t a


CINI with HPV Hig-grade CIN (C I N2-3)




Figure 1-1. Normal Cervical Epithelium, CIN1 and CIN2-31 (A) Staining of the transformation
zone showing both squamous and glandular epithelium on the surface and the
underlying parenchyma. Nuclei are stained dark purple. (B) Squamous cells that have
progressed to CIN1: moderate nuclear enlargement and rounded cells. (C) CIN 2-3
cells have large nuclei occupying the majority of the cell and increased staining.

Human Papillomavirus

Infection with oncogenic human papillomavirus (HPV) has been shown to be the primary

etiological factor for cervical cancer and its precursor lesions [5]. Papillomavirus is a small

deoxyribonucleic acid (DNA) virus that induces a variety of proliferative lesions. HPV infections

I adapted with permission from Ref. 4. Lie, A.K. and G. Kristensen, Human papillomavirus E6/E7 mRNA;,. ;,'i,; as
a predictive marker for cervical carcinoma. Expert Rev Mol Diagn, 2008. 8(4): p. 405-415.









are often transient, but persistent infections, especially with oncogenic HPV or high-risk types,

can further increase the likelihood of cervical dysplasia [6]. HPV-induced oncogenesis in

cervical carcinoma derives from the properties of the expressed viral gene products. Integration

of the viral circular DNA into the patient's genome often disrupts the viral E2 gene, which leads

to deregulation and expression of the E6 and E7 viral proteins [7].

The E6 gene product is able to interact with many cellular proteins, but it primarily binds

with the human tumor suppressor p53 protein. E6 forms a complex with the transcription factor

AP1, binds to p53, and induces its degradation through the ubiquitin protein ligase pathway. The

p53 protein has many names, including "the guardian of the genome", due to its role in

maintenance of genome stability by preventing the transmission of DNA mutations that arising

due to cellular stress. Cells with dysfunctional p53 are unable to respond normally to DNA

damage and thus allow accumulation of genomic mutations [8, 9].

Viral E7 disrupts normal cell cycle regulation through its interaction with the tumor

suppressor retinoblastoma (Rb) protein. Under normal cell cycle regulation, Rb forms complexes

with the E2F-1 family of transcription factors thereby inhibiting their function. Viral E7 binds to

Rb thus disrupting the inhibitory complex resulting in the release of active E2F-1. Subsequently

E2F-1 binds and initiates transcription of target genes that facilitate the G1/S transition and begin

S phase. The E7 proteins of the high risk HPVs, such as HPV-16 and HPV-18, bind Rb with

10-fold higher affinity than do the E7 proteins of the low risk HPV types. The difference in

binding affinity correlates with the transforming potential of the different E7 proteins. HPV

infection alone is not sufficient to induce malignant transformation. Other significant cofactors

contribute to the multi-step process of tumor formation, such as individual genetic variations as

well as environmental factors; however, such cofactors are not important in the absence of HPV









[10]. The combined dysregulation of p53 and Rb allows cells with DNA damage to proceed into

S phase facilitating the accumulation of genetic mutations, which is the basis for carcinogenesis.

Human Cell Culture

Much of what is known about the correlation between cervical cancer and HPV infection

comes from cell culture studies [11]. The practice of tissue culture is commonly used to study all

types of cancer, including those of the cervix. Cell culture is a useful tool for the study of in vitro

animal cell biology because it allows access to a highly-controlled, easily-manipulated

environment. There are a number of cultured human cervical cell lines that are commonly used

to study cervical cancer. The cell lines used in this study are C-33A, CaSki, HeLa 229 and SiHa.

The C-33A cell line is one of a series of lines derived by N. Auersperg from cervical cancer

biopsies [12]. The line, when originally cultured, exhibited an epithelial morphology, and

karyological instability that was observed with continued passage. The line is pseudodiploid with

a modal chromosome number of forty-six, occurring in 70% of cells examined. The cells are

negative for human papillomavirus DNA and RNA. The CaSki line was established with cells

from a metastasis in the small bowel mesentery of a woman with epidermoid carcinoma of the

cervix. These cells are reported to contain integrated HPV-16 as well as sequences related to

HPV-18. HeLa 229 cells were created from an epithelial cervical adenocarcinoma that contains

HPV-18. They have a modal chromosome number of 82. HeLa 229 cells are essentially the same

as the original HeLa cell line; however, they are relatively resistant to infection by polioviruses.

The SiHa cell line was established from fragments of a primary tissue sample obtained after

surgery from a Japanese patient with grade-II squamous cell carcinoma that contains one to two

copies of HPV-16 per cell [12]. The HPV-positive cervical cancer cell lines express normal Rb

and low levels of wild-type p53 proteins, which are presumed to be altered in function as a

consequence of association with HPV E7 and E6 oncoproteins, respectively. In C-33A, the Rb









protein is present but abnormal in size. These cells also overexpress a p53 DNA binding mutant

with an arginine to cysteine substitution in the DNA binding domain [13]. Characterization of

these cell lines further supports that inactivation of the normal functions of the tumor suppressor

proteins Rb and p53 are important steps in human cervical carcinogenesis, either by genetic

mutation or through inactivation by the HPV E6 and E7 oncoproteins.

Epigenetics

Both genetic and epigenetic changes are components of tumorigenesis [14]. Epigenetics is

the study of the heritable changes in gene expression that occur independent of the DNA coding

sequence. Epigenetic mechanisms that modify chromatin structure can be divided into four main

categories: DNA methylation, covalent histone modifications, non-covalent mechanisms such as

nucleosome remodeling, and small, non-coding ribonucleic acids (RNAs) [15]. Collectively,

these modifications regulate chromatin accessibility and its compactness. Their interplay controls

how the genome manifests itself in different cell types and developmental stages. Inappropriate

alterations of components of these mechanisms are commonly found in diverse disease states,

including cancer.

DNA Methylation

In mammals, DNA methylation primarily occurs at the five position of the cytosine ring

when directly followed by guanine (CpG). The majority of the genome is CpG poor; however,

there are variable stretches of DNA that contain a high frequency of CpG dinucleotides within

the genome and are thus designated CpG islands [16]. CpG dinucleotides that are not located in

CpG islands are sparsely distributed, highly methylated, and generally found in regions of large

repetitive sequences such as retrotransposon elements and centromeric repeats [17]. CpG islands

are more commonly found in the 5' end of the gene proximal to the transcription start site (TSS)

and occupy approximately sixty percent of all human gene promoters [18]. The majority of CpG









islands are generally unmethylated and found in differentiated tissues. Promoters containing CpG

islands that become hypermethylated during development result in long-term transcriptional

silencing. Classic examples of such naturally occurring epigenetic silencing mechanisms are X-

chromosome inactivation and gene imprinting [19, 20].

In mammals, DNA methylation is catalyzed by two families of enzymes designated DNA

methyltransferases (DNMTs): DNMT1 and DNMT3. The threeknown membersof the DNMT3

family are DNMT3A, 3B and 3L. DNMT3A and DNMT3B can modify unmethylated and

hemimethylated CpG base pairs at the same rate, and are thus referred to as de novo

methyltransferases. Little is known about the regulation of de novo DNA methylation or which

proteins are involved. It is known that DNMT3L, although catalytically inactive, interacts with

DNMT3a and DNMT3b and the complexes co-localize in the nucleus [21]. DNMT3L may also

participate in transcriptional repression [22]. Maintenance methyltransferases maintain the

methylation patterns established by the de novo methyltransferase by methylating the nascent or

hemimethylated DNA strand following replication. DNMT1 is known as the maintenance

methyltranferase because it localizes to the replication foci and interacts with the proliferating

cell nuclear antigen to methylate the newly-synthesized DNA strand [23]. DNMT1 and DNMT3s

work together in a complex relationship to maintain existing and establish new cytosine

methylation in mammals.

Histone Modifications

The majority of the mammalian genome is normally packaged in a transcriptionally-

repressive chromatin state. This type of chromatin is heavily methylated and forms repeating

units of nucleosomes. Nucleosomes are made up of an octamer of four core histone proteins with

147 base pairs of DNA wrapped around the octamer. H3, H4, H2A and H2B are the core histone

proteins within the nucleosome [24]; however, several histone variants also exist. For instance, it









has been seen at active or poised genes that H2A and H3 are replaced by H2A.Z and H3.3

histone variants, respectively [25]. The histone amino-terminal tails are sites of frequent post-

translational covalent modifications including acetylation, methylation, ubiquitylation,

sumoylation and phosphorylation at specific residues [26]. These modifications regulate key

cellular processes such as transcription, replication and repair.

Nucleosome Positioning

Histone modifications work by either changing the accessibility of chromatin or by

recruiting and/or inhibiting binding by non-histone effector proteins. Families of enzymes

coordinate covalent histone modifications to precise residues. Histone acetyltransferases (HATs)

and histone methyltransferases (HMTs) add acetyl and methyl groups, respectively, while

histone deacetylases (HDACs) and histone demethylases (HDMs) remove such marks from the

histone tails. Interactions between the DNA methylation machinery and histone-modifying

enzymes occur at multiple levels to affect gene expression. DNA methylation itself can affect

histone methylation and acetylation through proteins such as methyl CpG binding protein 2

(MeCP2) recruiting HMTs and HDACs [27, 28]. Conversely, HMTs can direct DNA

methylation to specific genomic targets by directly recruiting DNMTs to stably silence genes

[29, 30], and by regulating the stability of DNMT-DNA interaction [31]. DNMTs can also

recruit HDACs and methyl-binding proteins to achieve gene silencing and chromatin

condensation [32, 33]. The histone modification mechanism of inheritance is still not fully

understood. Whether histone marks survive replication has yet to be answered, but it is thought

that chromatin states are metastable because they self perpetuate due to the network of DNMTs,

HMTs, and HDACs that are recruited by the heritable DNA methylation. Regardless, it is widely

accepted that the deacetylated and methylated state of the histone tails maintain nucleosomes in a

tightly-compacted and transcriptionally-silent state, though exceptions have been found [34].















..






Figure 1-2. Diagram of Open and Closed Chromatin2 Chromatin is shown in a closed state that is
representative of the proposed 30 nm chromatin fiber (far right). This tight packing of
nucleosomes is refractory to transcription because it impairs DNA binding by
transcription factors. The cell is able to change the state of chromatin to allow access
to DNA by a combination of nucleosome-modifying enzymes, DNA binding factors
and nucleosome-remodeling activities. Open chromatin is represented as a beads-on-
a-string configuration with a close up of two nucleosomes (far left). In open
chromatin, variable lengths of linker DNA are observed between nucleosomes.

Only a small fraction of the genome, known as euchromatin, is transcriptionally active.

Both covalent and non-covalent mechanisms, such as nucleosome remodeling, regulate gene

expression by altering the accessibility of chromatin to transcription factors. Nucleosomes

around the promoter are more widely spaced than in heterochromatin and generally contain

heavily acetylated histones. It is generally thought that nucleosomal DNA resists transcriptional

activation, which requires the assembly of bulky transcription machinery onto the TSS. It has

long been hypothesized that nucleosomes need to be removed for active transcription [35], and

there are current models depicting this in yeast [36, 37] and mammalian systems [38, 39]. In the

human genome, it has been shown that nucleosomes immediately upstream of TSSs are depleted

upon activation of inducible genes. The adenosine-5'-triphosphate (ATP)-dependent chromatin


2
adapted from 44. Gilbert, N. and B. Ramsahoye, The relationship between chromatin structure and transcriptional

activity in mammalian genomes. Brief Funct Genomic Proteomic, 2005. 4(2): p. 129-42.









remodeling complex switch/sucrose non-fermentable (SWI/SNF) is thought to regulate promoter

nucleosome eviction by sliding and/or removing nucleosomes [40, 41]. Indeed, nucleosome-free

regions (NFRs) are found at the 5' and 3' ends of genes [42, 43]. Packaging of DNA into

chromatin by the formation of nucleosomes is vital to gene regulation and the organization of

such is only beginning to be understood.

Epigenetics in Cancer

Epigenetic alterations involve genome-wide losses and regional gains of DNA methylation

as well as altered patterns of histone modifications and the enzymes that regulate them; however,

the heritability of histone modifications is inconclusive since their mechanism of inheritance is

currently unknown [45]. These aberrant epigenetic events have been observed in cancer, but each

is currently being debated as to whether it is a foundation for or rather the consequence of

cancer.

Aberrant Methylation

The least understood of these epigenetic alterations is genome-wide loss of methylated

cytosine residues. Global DNA hypomethylation is most often observed within repetitive

sequences dispersed throughout the genome. Recently, hypomethylation has also been observed

within the body of genes [46]. Aberrant gene silencing is the more commonly studied cancer-

associated epigenetic event. Aberrant gene silencing is the abnormal transcriptional silencing of

genes by either direct methylation of the gene promoter or modifications to the chromatin

structure. The study of hypermethylation is generally focused on tumor suppressor genes (TSG).

TSGs are genes that normally inhibit cell growth. TSG proteins are found in many different

cellular pathways involving cell cycle checkpoint responses, detection and repair of DNA

damage, mitogenic signaling, differentiation and migration, and tumor angiogenesis [47]. In

most mammalian genes, the CpG islands are normally free of DNA methylation. In cancer cells,









however, various TSG promoters have densely methylated CpG islands resulting in a loss of

gene transcription [48]. This loss of transcription can act as one or possibly as both hits needed in

Knudson's 'two hit hypothesis' [49]. Knudson's hypothesis that two hits are required for the full

inactivation of a tumor-suppressor gene has been shown to be fundamentally correct in almost all

cases of human cancer, but it is now known that the hit is not always a genetic mutation. The

mechanism for aberrant methylation is largely unknown. It might be caused by dysregulation of

the DNMTs or other chromatin binding proteins. Despite its unknown cause, aberrant DNA

methylation is an important mechanism that is exploited by cancer.

Chromatin Changes in Cancer

Aberrant DNA methylation is one of a series of complex epigenetic silencing mechanism

that lead to alterations in gene expression through changes in chromatin structure. Experiments

showing that epigenetic gene silencing occurs in the absence of detectable CpG methylation

indicated that mechanisms other than DNA methylation modulate gene silencing [50, 51]. It is

now hypothesized that nucleosome remodeling works in concert with DNA methylation and

histone modifications to drive gene silencing. There are many possible mechanisms by which

this could take place in cancer. For example, the nucleosome remodeling and deacetylase

(NuRD) complex facilitates recruitment of the polycomb repressive complex 2 and DNMT3A to

target gene promoters leading to their permanent silencing by establishing a repressive chromatin

state [52]. Components of the NuRD complex have been linked directly to oncogenesis [53]. In

addition, it has been shown that alterations in the SWI/SNF complex are associated with cancer

development [54]. Interactions between the DNA methylation machinery and histone-modifying

enzymes further enhance the complexity of epigenetic regulation of gene expression. More is

known about the maintenance of tumor suppressor gene (TSG) methylation, but we are slowly

starting to understand how aberrant DNA methylation may be initiated.









Epigenetic alterations are important for the development and progression of malignant

diseases. In cervical cancer, alterations in DNA methylation are thought to be associated with

integration of viral DNA. The integrated viral DNA of Hepatitis B/C virus infections of the liver

and Epstein-Barr virus infections in stomach cancer are known to alter the DNA methylation

status of the host genome [55, 56]. HPV E7 protein has been reported to directly interact with

DNMT1 and stimulate its enzymatic activity [57]. This may be the cause of hypermethylation of

TSGs that has been seen in cervical carcinoma [58]. Reports that DNMT1 protein expression is

increased even in low-grade cervical intra-epithelial neoplasias compared with normal squamous

epithelium, and further increased in higher-grade cervical intra-epithelial neoplasias and

squamous cell carcinomas of the uterine cervix, indicate that this may act as an initiating event in

cervical carcinogenesis [59]. The regulation of DNA methyltransferases may be altered by viral

oncogenes through interactions with Rb-associated proteins [60]. Furthermore, E6, E7, and their

interactions with host proteins drive the cell cycle to ignore normal DNA regulation checkpoints,

which leads to genomic instability [61]. These virally-induced alterations in host gene structure

contribute to the molecular pathogenesis of HPV-associated cancers.

As the role of epigenetics in cancer becomes clearer and the interrelationships between

DNA methylation and chromatin components are increasingly understood, the possibility of

epigenetic approaches to cancer prevention, detection and therapy arise. A number of epigenetic

alterations occur during all stages of cervical carcinogenesis. From a diagnostic point of view,

because epigenetic abnormalities occur very early in the carcinogenic process they can

potentially be exploited as molecular markers for early detection. Assessment of

hypermethylated genes in primary tumors could potentially serve as a prognostic factor or as a

means of predicting response to radiation, chemotherapy or transcriptional agents. The focus of









this project is to determine if Wnt inhibitory factor, an antagonist of the Wnt pathway, exhibits

epigenetic aberrations in cervical cancer that may be exploited for future cancer therapies.

Wnt Pathway

The Wnt extracellular signaling pathway, wingless in Drosophila, is a highly-conserved

signal transduction pathway that mediates a wide range of biological processes. From hydra to

humans, Wnt signals control multiple aspects of development, including the proliferation, fate

specification, polarity and migration of cells [62, 63]. There are at least two distinct pathways of

Wnt signaling. The first is the Wnt/p-catenin pathway named for the molecule that activates it,

Wnt. The p-catenin pathway of Wnt signaling is also commonly referred to as the 'canonical'

pathway. The second is the non-canonical or p-catenin-independent pathway. Two decades of

research have given us a firm understanding of the Wnt pathway. In addition, activating

mutations in components of the Wnt signaling pathway have been found to play a role in

oncogenesis [64, 65].

Wnt Agonists and Antagonists

There are various agonists and antagonists that act on the Wnt signaling pathway. To

date, nineteen Wnt ligands have been found in mammals. They are cysteine-rich proteins of

approximately 350-400 amino acids and contain an N-terminal signal peptide for secretion. Wnts

are glycosylated and lipid modified in the endoplasmic reticulum, and subsequently escorted by

Wntless from the Golgi to the plasma membrane for secretion. After secretion, mature Wnts

either bind to heparin sulfate proteoglycans and lipoprotein particles, or form multimers with

each other. Wnt antagonists can be divided into two functional classes [66]: the secreted

Frizzled-related protein (sFRP) class and the Dickkopf class. Members of the sFRP class include

the sFRP family, Wnt inhibitory protein (WIF 1) and Cerberus all of which bind directly to Wnts,

thereby altering Wnt's ability to bind to one of its receptor complexes. Only sFRPs are able to









bind both Wnts and the Wnt receptor Frizzled. The Dickkopf (DKK) class of inhibitors binds

directly to the LRP5/6 co-receptor of the Wnt receptor complex (Figure 1-3B).

Wnt Receptors

The Wnt receptor complex that activates the canonical pathway contains two

components: the family of seven-pass transmembrane Frizzled (Fz or Fzd) receptors and a co-

receptor from the low-density lipoprotein receptor-related protein 5/6 family (LRP5 or LRP6).

There are ten Fz genes in mammals which likely have functional redundancy and have varying

capabilities to activate Wnt signaling when co-overexpressed with Wnt and LRP5/6 [67]. Either

one of two single-span transmembrane proteins, LRP5 or 6, are also needed. In vitro studies

support the model that Wnt induces the formation of Fz-LRP5/6 complex [68]; however, this has

yet to be clearly shown in vivo. A particular Wnt may activate canonical and/or noncanonical

pathways depending on the receptor complement [69], though it remains unclear whether the

noncanonical Wnt pathway also requires LRP5/LRP6.

Canonical Pathway

The Wnt/p-catenin pathway, also referred to as the 'canonical' pathway, is activated when

a Wnt ligand binds a member of the Fz receptor and the co-receptor LRP5 or 6 (Figure 1-3A).

Formation of the Wnt-Fz-LRP complex recruits the scaffolding protein Dishevelled (Dvl),

resulting in LRP phosphorylation and activation which recruits Axin. This leads to inhibition of

Axin-mediated P-catenin degradation. Without constant degradation, 0-catenin accumulates in

the cytoplasm and translocates to the nucleus. In the nucleus, P-catenin complexes with the

family of T cell factor/lymphoid enhancer factors (TCF/LEF). P-catenin binding to TCF/LEF

proteins provides a transcription activation domain. As a result of this complex, Wnt target gene

expression is activated [65].










A ON B B p OFF
sFRP'
LRLRP )j LRP -
Frized Frizzed





ltindil a Int le Ii

c om l form-aioin In te ab rlen c o n p -ctnn
multiprotei ,- om-p- ex an. i pe b CK ( e an s e t....


T/ TCF -o \ / [Td
I wmr9 PrnrhlOlV ga ii Wnl raipogivn getiu I


Figure 1-3. Diagram of the Canonical Wnt Pathway 3 (A) Wnts are the primary agonists and
form a complex with LRP5/6 and Fz to activate signaling. In the presence of Wnt
ligand, a receptor complex forms between Fz and LRP5/6. Dvl recruitment by Fz
leads to LRP5/6 phosphorylation and Axin recruitment. This disrupts Axin-mediated
phosphorylation/degradation of p-catenin, allowing 3-catenin to accumulate in the
nucleus where it serves as a coactivator for TCF to activate Wnt-responsive genes.
(B) WIF 1 and sFRP antagonists bind directly to secreted Wnts and/or Fz binds
directly to secreted sFRP. DKK antagonist proteins bind LRP5/6 to prevent Fz-LRP6
complex formation. In the absence of Wnt, cytoplasmic P-catenin forms a
multiprotein complex and is phosphorylated by CK1 (blue) and subsequently by
GSK3 (yellow). Phosphorylated P-catenin is recognized by the ligase P-Trcp, which
targets P-catenin for proteasomal degradation. Wnt target genes are repressed by
TCF-TLE/Groucho and HDAC.

In the absence of Wnt (Figure 1-3B), cytoplasmic P-catenin interacts with the adenomatous

polyposis coli gene product (APC) as well as Axin scaffold proteins and is a substrate for casein

kinase 1 (CK1) and glycogen synthase kinase 33 (GSK33). CK1 and GSK33 sequentially

phosphorylate the amino terminal region of 3-catenin, resulting in 3-catenin ubiquitination by

P-Trcp and proteasomal degradation. Constant elimination of P-catenin prevents it from reaching



3 Modified from Ref. 65. MacDonald, B.T., K. Tamai, and X. He, Wnt/beta-catenin signaling: components,
mechanisms, and diseases. Dev Cell, 2009. 17(1): p. 9-26.









the nucleus; thereby, allowing TCF/LEF to stay bound to and repress Wnt target genes.

TCF/LEF factors, when bound to DNA at Wnt-responsive genes, are able to interact with other

factors, such as Groucho and histone deacetylase, to repress transcription [70].

Noncanonical Pathway

Noncanonical Wnt signaling pathways are less understood, but appear to function in a

P-catenin-independent manner to regulate processes such as convergent extension during

vertebrate gastrulation. In vertebrates, the noncanonical pathways have been termed the

Wnt/Calcium and Wnt/c-Jun N-terminal kinases (JNK) pathways. In brief, activation of the

Wnt/Calcium pathway involves Wnt binding to a Frizzled receptor, leading to the release of

intracellular calcium and the activation of enzymes such as Ca2+/Calmodulin-Dependent Protein

Kinase II and Protein Kinase C. The Wnt/JNK pathways appear to similarly use Frizzled

receptors, Dishevelled, JNK and Rho family guanosine triphosphatases (GTPases). Recent data

have also implicated components of noncanonical Wnt pathway in diverse forms of human

cancer [71].

There are numerous components of the Wnt network, from the plasma membrane to the

cell nucleus. While an understanding of the overall organization of the network is beginning to

emerge, much is still unknown. Wnt signals ultimately promote cell proliferation and tissue

expansion. Numerous Tcf target genes have been identified in diverse biological systems and

include genes involved in cancer, such as cMyc and cyclin Dl.

The Wnt signaling pathway is frequently altered in cancer. While no direct mutations of

Wnt have been found in cancer, several of the downstream molecules in the Wnt signaling

pathway (i.e., p-catenin, APC, Axin) are dysregulated in a variety of human tumors. Probably the

most characterized of them are inactivating, germline-mutations in APC found in colorectal

cancer. Activation of Wnt signaling is believed to play a role in the progression and pathogenesis









of cervical cancer. It has been shown that p-catenin is increased within the cytoplasm and

nucleus in invasive cervical carcinomas. In addition, in vivo activation of the Wnt pathway is

sufficient to induce transformation of HPV-immortalized human keratinocytes [72].

We focused specifically on Wnt inhibitory protein (WIF 1), an antagonist of the Wnt

pathway. WIF 1 has a unique N-terminal signal sequence, WIF domain (WD), and five epidermal

growth factor (EGF)-like repeats which mediate Wnt binding to WIF1 [73]. WIF1 mainly

functions as a secreted inhibitor of Wnt signaling by binding to Wnt proteins and thus competing

with the binding of Wnt proteins to the frizzled receptor. WIF1 expression is downregulated in

prostate, breast, lung and bladder tumors [74], suggesting that WIF1 is a potential TSG. Loss of

WIF1 expression in tumors may lead to unrestricted binding of Wnt ligands to the frizzled

receptor, followed by enhanced transcription of target genes in the Wnt pathway. Several groups

have correlated decreased WIF1 expression with aberrant promoter hypermethylation. While the

WIF1 gene is a target for epigenetic silencing in some tumor types and its expression is

downregulated in several cancers, it is currently unknown if this gene is subject to epigenetic

silencing in cervical cancer. To determine if WIF1 is epigenetically silenced in cervical cancer,

we have analyzed the expression and DNA methylation status of the WIF1 promoter in a panel of

cultured human cervical tumor cell lines and surgically-obtained human cervical tumors. Our

results demonstrate that WIF1 hypermethylation is a frequent event in cervical cancer and lend

strength to the idea that dysregulation of Wnt signaling is an important contributor of human

cervical tumorigenesis.









CHAPTER 2
MATERIALS AND METHODS

Cervical Specimen

All human tissue samples were obtained from patients under protocols approved by

Institutional Review Boards at the Moffitt Cancer Center (Tampa, FL). Tumor tissue (n = 22)

and adjacent matched normal stratified epithelium (n = 22) were obtained from archived

formalin-fixed paraffin-embedded (FFPE) blocks from the tumor repository. The archived tissue

came from women diagnosed with squamous cell cervical cancer and having a surgical

procedure between 1993 and 1999 at the Moffitt Cancer Center. Cases were restricted to those

that did not receive radiation treatment before surgery. Histological diagnosis, tumor stage and

grade were reviewed by a pathologist at Moffitt Cancer Center according to the WHO and the

international union against cancer (UICC) classification of tumors.

The Pap samples (n = 8) are scrapings of the cervix obtained from patients from four sites:

the local Health Department, Tampa General Hospital, Lifetime Cancer Screening and the

Moffitt Cancer Center. Routine gynecological examinations of healthy, disease-free women were

conducted. Exfoliated cells were placed in 20 mL of PreserveCyt solution, mixed and separated

into 4 ml aliquots and stored at -800C. One milliliter of the solution was transferred to a new

tube with a pipette and pelleted. To resuspend the pellet, 200 [il of phosphate-buffered saline

(PBS) was added, and DNA was extracted using the QIAamp DNA extraction kit (Qiagen,

Valencia, CA). Sample #3 is the only one that is HPV negative. The other seven samples were

found to have one or more HPV types: 1, 16, 51, 58, 59 and 62.

Cell Culture and 5-azadC Treatment

The cervical tumor cell lines C-33A, CaSki, HeLa 229 and SiHa were purchased from the

American Type Culture Collection and cultured according to their specifications. In short, HeLa,









SiHa and C-33 lines were grown in Eagle's Minimal Essential medium (Mediatech, Manassas,

VA) with 10% fetal bovine serum (FBS) (Invitrogen Life Technologies, Grand Island, NY) and

penicillin-streptomycin (Mediatech). The CaSki line was grown in Roswell Park Memorial

Institute 1640 medium (Meditech) with 10% FBS. All cell lines were cultured and maintained in

monolayers at 370C in a humidified incubator under 95% 02 and 5% CO2. For the drug

treatment, cells were seeded, allowed to recover overnight, then 5-azadC (Sigma-Aldrich, St.

Louis, MO) was added to complete growth media at a final concentration of 5 itM. Fresh drug

was added every 24 hr for 5 d. After treatment, the cells were washed with PBS and allowed to

recover in their respective medium for 2 d prior to collection.

RNA and DNA Isolation

One 10 cm plate of confluent cells cultured with and without 5-azadC was harvested for

RNA and DNA extraction with 1 mL of TRizol (Invitrogen) reagent for 5 min at room

temperature. After phenol/chloroform extraction and phase separation, RNA was digested

according to the manufacturer's instructions for on-column DNase I digestion with RNase-free

DNase I (Qiagen) and RNAeasy Kit (Qiagen). The DNA was isolated according to

manufacturer's instructions from the organic phase of the RNA isolation. The 260/280 nm

absorbance ratio and sample concentrations of the RNA and DNA was determined using the

NanoDrop ND-8000 (Thermo Scientific, Wilmington, DE).

All genomic DNA isolated from patient samples was conducted at the Moffitt Cancer

Center by either centrifugation of 1-4 ml of Thin Prep solutions into a pellet that was

resuspended in PBS (Pap samples) or by sectioning tissues from paraffin blocks into small pieces

and deparaffinized by xylene extraction. The genomic DNA was then extracted using QIAmp

mini DNA Kit (Qiagen) according to manufacturer's protocol. The concentration of the extracted

DNA was then determined by the NanoDrop 1000 Spectrophotometer (Thermo Scientific).









Sodium Bisulfite Modification

At the Moffitt Cancer Center, sodium bisulfite conversion of 1 tg of patient tissue-

derived DNA was carried out with the EZ DNA Methylation-Direct kit (Zymo Research

Corporation, Orange, CA), following the manufacturer's protocol. Bisulfite-modified DNA was

resuspended in 30 tl of water and stored at -800C in six 5 tl aliquots to avoid several

freeze/thaws. Cell line DNA was bisulfite treated using a specific in-laboratory protocol that

consistently yields a 99.8% efficiency of cytosine conversion [75]. The protocol is: the day

before bisulfite treatment, distilled water was degassed (dg.dH20) by boiling -200 mL of

distilled H20 (dH20) for 20 min in a glass beaker, and carefully poured into a 125 mL bottle

until it was completely full (above the lip). The cap was screwed on the bottle tightly and the

dg.dH20 was cooled overnight on the bench top. Immediately prior to bisulfite treatment, 3 N

NaOH and 100 mM hydroquinone were prepared by dissolving each chemical in dg.dH20. Total

genomic DNA (1-2 Gg) in 20 pl 0. 1X TE was pipetted into a 0.65 ml micro-centrifuge tube (that

fits in a thermocycler) containing 10 pl of denaturation buffer (6.5 ptL dg.dH20, 3.0 ptL 3N

NaOH, and 0.5 ptL 0.5 M EDTA, pH 8.0). Saturated sodium metabisulfite solution was made as

follows: to a 20 ml-capacity glass scintillation vial containing a small stir bar, 0.1 ml of 100 mM

hydroquinone, 5 g sodium metabisulfite (Sigma, previously aliquoted into air-tight vials), and

7 ml dg.dH20 were added, and the solution was immediately stirred. Then, 1.0 ml of 3 N NaOH

was quickly added. The solution was adjusted to pH 5.0 at room temperature with additional 3 N

NaOH and preheated to 500C in a water bath. A saturated solution was employed to promote

sulfonation of cytosine, the first step of bisulfite deamination. Samples of genomic DNA in

denaturation solution were heated at 980C for 5 min with the tube caps open in a thermocycler.

With one minute remaining in the denaturation step, the preheated sodium metabisulfite solution









was transferred to a stir-plate and stirred gently. After 5 min denaturation, 0.2 ml of sodium

metabisulfite was rapidly added to each sample while in the thermocycler. Samples were capped,

vortexed and incubated for 6 hr in a covered water bath at 500C. Following deamination, the

DNA was desulfonated and cleaned using the EZ Bisulfite DNA Clean-up kit (Zymo Research

Corporation) according to manufacturer protocol and eluted in 10 pl of M-Elution Buffer.

Polymerase Chain Reaction (PCR) and Pyrosequencing

Following bisulfite modification, a 20 tl PCR with a final concentration of Ix Coral

Buffer, 250 nM deoxyribonucleotide triphosphates (dNTPs), 250 nM forward and reverse

primer, 1 U of HotStarTaq Plus (Qiagen) and 2 il of bisulfite-treated DNA was incubated in the

thermocycler as follows: 95C for 5 min, once; 940C for 30 sec, 59C for 30 sec, 720C for

30 sec, repeated 45 cycles. The specific primers used were (Forward, F) 5'-GTA GGT TTT TTG

GTA TTT AGG T -3' and (Reverse, R) 5'-CAT ACT ACT CAA AAC CTC CT-3'. A second

nested PCR was preformed using 2 il of the original PCR product in the same reaction mixture.

Amplification was checked by agarose gel electrophoresis. A series of controls were used to

validate the assay: no template, no sequencing primer, sequencing primer only, sequencing

primer with reverse primer only, beads only and a blank well. Also, in vitro methylated bisulfite-

treated DNA (Zymo Research Corporation) was used as a methylation positive control for all

pyrosequencing runs. All samples and controls for each reaction were then analyzed using the

PyroMark MD system (Biotage, Charlotte, NC). Methylation density was quantified using the

PyroMark Pyro Q-CpG 1.0.9 software.

Reverse Transcriptase PCR (RT-PCR)

Total RNA was used in first-strand complementary DNA (cDNA) synthesis reactions

using GoScript Reverse Transcriptase System and random hexamer primers (Promega, Madison,

WI). WIF1 transcript levels were subsequently analyzed by PCR. The cDNA WIF 1-specific









primers used were (F) 5'-CCG AAA TGG AGG CTT TTG TA-3' and (R) 5'-TGG TTG AGC

AGT TTG CTT TG-3'. They were designed to span a gap between exons to ensure amplification

of cDNA and not genomic DNA. p-actin served as a control for RNA integrity and was amplified

using primers (F) 5'-CCC TGG CAC CCA GCA C-3' and (R) 5'-GCC GAT CCA CAC GGA

GTA C -3'. PCR thermocycler conditions for WIF1 were 95C 5 min, once; 940C for 30 sec,

59C for 30 sec, 720C for 30 sec, repeated 45 cycles. PCR conditions for p-actin mRNA were the

same as for WIF 1 transcript except that an annealing temperature of 600C was used and PCR was

conducted for 20 cycles. APCR reaction substituting dH20 for cDNA was conducted as a

negative control.

Methyltransferase Accessibility Protocol for Individual Templates (MAPit)

Nuclei isolation and DNMT probing of nuclei from CaSki, HeLa and SiHa cells were

isolated from -4-7 x 106 cells at 40C under buffer conditions which preserve the integrity of

nuclei and chromatin structure. After harvesting, cells were washed twice with PBS and

resuspended in cell resuspension buffer (CRB: 20 mM HEPES, pH 7.5, 70 mM NaC1, 0.25 mM

EDTA, 0.5 mM EGTA, 0.5% glycerol (v/v), 10 mM DTT, 0.25 mM phenylmethylsulfonyl

fluoride). After pelleting by centrifugation at 1000 x g, cells were resuspended in cell lysis buffer

(CLB: Ix CRB plus 0.19% (v/v) NP-40) for ten minutes on ice. Nuclei were then washed twice

with CRB, and a 1-21L aliquot was stained with 4% (w/v) trypan blue and visualized by light

microscopy to confirm their integrity. Nuclei were finally resuspended in methylation buffer

(MB; Ix CRB plus 160 [M S-adenosylmethionine). After pre-warming nuclei to 370C for 5 min,

60 U and 100 U of M.CviPI were added for 15 min at 370C. Methylation reactions were stopped

by adding two volumes of methylation stop buffer (MSB; 100 mM NaC1, 10 mM EDTA, pH 8.0,

1% SDS (w/v)), and then incubated overnight with 100 [g/ml proteinase K at 500C. DNA was









isolated by extraction with phenol:chloroform:isoamyl alcohol (25:24:1) and concentrated by

ethanol precipitation.

After isolation, DNA was treated with sodium bisulfite as described above. PCR primers

designed to amplify a 712 basepair segment of the promoter were designed. Primer sequences

were (F) 5'- ATT ATT ATT ATT ATT ATT AGY ATT TAG TT-3' and (R) 5'- CAA RCA CAA

AAA AAT RCT CCA AA -3'. Three 20 pl mixtures with a final concentration of Ix Coral

Buffer, 250 nM dNTPs, 250 nM forward and reverse primers, 275 nM MgC12, 1 U of HotStarTaq

Plus (Qiagen) and 4 .il of bisulfite-treated DNA were incubated in the thermocycler as follows:

5 min at 950C, once; 95C for 30 sec, 530C for 30 sec, 720C for 2 min, repeated for 40 cycles.

Following PCR, products were digested with restriction enzymes R.HindIII and R.XhoI. After

digestion the products were pooled and resolved by an agarose gel electrophoresis. Subsequently,

the bands were excised and gel purified using the QiaexII Gel Extraction kit (Qiagen). The

digested, purified PCR products were then ligated at 160C overnight into a R.HindIII- and

R.XhoI-digested BlueScript SK minus plasmid (Stratagene, Santa Clara, CA) with T4 ligase

(New England Biolabs, Ipswich, MA). Recombinants were identified by colony PCR with Apex

Taq (Genesee Scientific, San Diego, CA) in the same reaction mixture as above. The inserts were

sequenced at the University of Florida Interdisciplinary Center for Biotechnology Research DNA

sequencing core laboratory using a vector-encoded primer (M13-Rev).

Immunohistochemistry

Two tissue microarrays were assembled at Moffitt with 4 .im sections of FFPE tissue

sectioned from each cervical tumor with matching normal adjacent tissue. After sections were

deparaffinized by xylene and rehydrated through graded alcohols, the samples were blocked

using the Avidin/Biotin Blocking Kit (Vector Laboratories, Burlingame, CA) following the









manufacturer's recommended protocol. Antigen retrieval by steam heat for 20 min with 10 mM

sodium citrate, pH 6.0 was conducted prior to overnight incubation with the rabbit anti-C-

terminus primary antibody diluted 1:50 (Abcam, Cambridge, MA). Samples were then washed

and incubated with biotinylated-goat anti-rabbit antibody from the Rabbit IgG ABC kit (Vector

Laboratories). Detection of the antigen was carried out by avidin-biotin complex following

standard procedures. The tissue microarray was developed with peroxidase and

3,3'-diaminobenzidine tetrahydrochloride (reddish-brown reaction product). Sections were then

counterstained with haematoxylin to stain the nuclei purple and mounted in an aqueous mounting

medium (Fisher). A control section was performed where the primary antibodies was replaced by

irrelevant rabbit monoclonal antibody of the same isotype and at the same concentration as the

specific primary antibody. Analysis was performed under 100x magnification on a Leica

DM6000B microscope (Leica Microsystems, Bannockburn, IL), and images were taken with a

Retiga SRV Fast 1394 digital camera (QImaging, Surrey, British Columbia, Canada).









CHAPTER 3
RESULTS

WIF1 is Epigenetically Silenced in Cervical Cancer Cell Lines

To determine if the WIF1 gene is targeted for epigenetic silencing in cervical cancer, we

cultured a panel of four cervical cancer lines (C-33A, CaSki, HeLa, and SiHa) in the presence or

absence of the DNA demethylating drug 5-azadC. Using RT-PCR we observed in the untreated

cells that C-33A, CaSki, and HeLa cell lines expressed varying amounts of WIF1 transcript

while SiHa did not. However, when cultured in the presence of 5 mM 5-azadC, the CaSki, HeLa,

and SiHa lines displayed elevated levels of WIF1 transcript (Figure 3-1). These data suggest that

WIF1 expression may be epigenetically repressed by DNA methylation in CaSki, HeLa and SiHa

cells.

The varied reactions of the cell lines to 5-azadC were of note. The SiHa cell line continued

to divide during 5-azadC treatment, while the HeLa cell remained quiescent and quickly divided

during the two-day recovery period in the absence of drug. Similarly, CaSki cell numbers

remained constant during 5-azadC treatment, but slowly divided afterward. In contrast some

C-33A cells died during 5-azadC treatment, and there was little cell division during the two-day

recovery period. The difference in response to the drug treatment appeared to correlate with the

presence of the HPV genome and inversely correlate with the methylation status at this locus,

which may or may not be of significance.

To determine if the difference in WIF1 expression levels was correlated with promoter

DNA hypermethylation, we analyzed the methylation status of WIF1 by pyrosequencing.

Pyrosequencing is essentially 'sequencing by synthesis' to quantitatively analyze cytosine









C-33A CaSki HeLa SiHa H20
5-azadC tx: + + + +


WIF1




B-actin




Figure 3-1. RT-PCR of Cervical Cell Lines Before and After 5-azadC Treatment. Cultures of the
cervical tumor cell lines C-33A, CaSki, HeLa and SiHa were untreated (-) or treated
with 5 pM 5-azadC (+) for 5 d (fresh medium and drug were added every 24 hr).
Following this treatment, total RNA and genomic DNA were prepared from the cells.
RT-PCR analysis of WIF1 expression (top panel) and the housekeeping gene P-actin
as a control for RNA integrity (bottom panel).

methylation at individual CpG dinucleotides after PCR amplification of a bisulfite-converted

sequence. Briefly, bisulfite conversion of unmethylated cytosines in CpG dinucleotides results in

conversion of C to U while methylated cytosines remain as C. Pyrosequencing requires PCR of a

segment of the gene of interest from bisulfite-converted genomic DNA using primers that

amplify in a methylation-independent manner by avoiding areas with CpG present. In addition,

one primer is designed with a 5' biotin tag. PCR amplicons are subsequently purified using

streptavidin-coated beads, and then denatured to single-stranded DNA. A sequencing primer is

annealed to the captured strand and added to a 96 well reaction plate containing DNA

polymerase and dNTPs. As nucleotides are incorporated onto the 3' end of the oligonucleotide

primer, pyrophosphate (PPi) is given off as a byproduct. The reaction mix also contains ATP

sulfurylase which quantitatively converts PPi to ATP in the presence of adenosine 5'

phosphosulfate. This ATP drives a luciferase-mediated conversion of luciferin to oxyluciferin

that generates visible light in amounts that are proportional to the ATP concentration. The light










produced in the luciferase-catalyzed reaction is detected by a charge coupled device (CCD)

camera and is recorded by the pyrosequencer. Each light signal is proportional to the number of

nucleotides incorporated, and thus DNA methylation is quantified at each CpG dinucleotide.


A
TATA AT
.i3Iih bp n
CpG MMit f- n--:-1111 i I i ll II+b l iii1 EI
Pyro: ur
MAPit: r RBW
Fof RZ


B Avg.
C-33A .
S3anrisPealrywd; YGGGAGOYGAYVGTTTTAGTYGTTTAAAVGOOAAT
%CpG
twemhylUlon 4% 4% &% 5% 4% 4%4






dNTP I
5 ii s 1" ,' C 15


CaSki


Avg = .= -


3a~ % A s% IW a% x%






A-g


SiHa = 145a


t 1J If KE 5 au P V '; v J W 9 Prj Oa
Figure 3-2. WIF1 Promoter and Pyrograms of Cervical Cell Lines. (A) Schematic representation
of the WIF1 promoter. Each vertical line represents a CpG dinucleotide in the
sequence. Positioned below the diagram are the primers used for pyrosequencing and
MAPit, respectively. (B) Data output from the pyrosequencer, termed a program.
Shown in the programs are the proportions of nucleotides incorporated along with
the sequence of nucleotides dispensed shown on the x-axis. The percent methylation
at each CpG dinucleotide (gray rectangles) is shown above the CpG site. Average
CpG methylation (Avg. Meth.) within the segment of the gene analyzed is displayed
in the box at the top of the program. There are several quality controls where a
nucleotide that should not be incorporated is dispensed (site 6,14, 28 and 29). A
bisulfite conversion control dispenses a C directly after a converted T to check the
bisulfite conversion efficiency (tan rectangle).









We designed pyrosequencing primers to amplify a 164 basepair region of the WIF1

promoter. This assay quantitatively measures the first seven CpGs following the transcriptional

start site using the forward PCR primer as the sequencing primer (Figure 3-2A). When

pyrosequencing was conducted on bisulfite-modified genomic DNA harvested from the four

cervical cancer cell lines, we observed that the cell lines that express WIF] in the absence of

5-azadC (C-33A and CaSki), exhibited very low levels of DNA methylation (<3%). Conversely,

cell lines which expressed little to no WIF1 in the absence of 5-azadC, HeLa and SiHa cells,

exhibited high and moderate levels of methylation, respectively (21% and 45%) (Figure 3-2B).

Genomic DNA harvested from HeLa and SiHa cells cultured with 5-azadC displayed a 10-15%

decrease in average methylation, while C-33A and CaSki showed little change (Figure 3-3).

These data demonstrate that the expression of WIF1 is correlated to the methylation status of its

promoter in cervical cell lines.


C:



LII -5-azadC







C-33A CaSki HeLa SiHa
Cell Lines


Figure 3-3. Average Methylation of the WIFI Promoter Before and After 5-azadC Treatment.
Cultures of the cervical tumor cell lines C-33A, CaSki, HeLa and SiHa were
untreated (white bars) or treated with 5 [iM 5-azadC (black bars). Shown are the
averages and standard deviation for the six CpGs assayed by pyrosequencing as
outlined in Materials and Methods.









Aberrant Chromatin Footprint in CaSki and SiHa Cell Lines

To determine if promoter methylation might be repressing WIF1 expression by altering

chromatin accessibility, we preformed a methyltransferase accessibility protocol for individual

templates (MAPit) on the CaSki and SiHa cell lines. This technique probes chromatin structure

with minimal disruption to nuclei integrity to capture an accurate footprint of chromatin structure

and nucleosome dynamics in vivo. Nuclei were extracted and probed with M.CviPI, a DNA

methyltransferase (DNMT) that only methylates GC dinucleotides. Accessibility of DNMTs to

GC sites is impaired based on the strength and span of the protein-DNA interaction to be probed.

Methylated cytosines that resist chemical deamination are detected positively by the PCR-based

technique of bisulfite genomic sequencing. This unique capability to score methylation at

multiple sites in single molecules permits detection of inherent structural variability in chromatin

structure and non-histone protein-DNA interactions. MAPit analysis was conducted on the active

and repressed WIF1 promoter of CaSki and SiHa lines, respectively (Figure 3-4).

In the CaSki cell line, which expressed WIF1, two populations of cells were evident: one

population with two distinct areas of methylation and one population with nearly no methylation

(Figure 3-4A). After probing with M.CviPI, there were three distinct regions of open chromatin

(Figure 3-4B). However, there was no difference in accessibility between the methylated and

unmethylated populations. It is of note that sequence inspection revealed two consensus Spl

binding sites (5'-GGCGGG-3') within the two regions devoid of endogenous methylation and

inaccessible to the methyltransferase. This is consistent with other reports showing that Sp

binding to DNA may be able to protect sequences from DNA methylation [76]. Near the

upstream SP1 site, it has also been shown that NF-kP and MYC-associated zinc finger (MAZ)

transcription factors bind [77]. In the SiHa cell line in which WIF1 was highly methylated and

not expressed, high levels of endogenous methylation were observed throughout the body of the











promoter and accessibility to the methyltransferase seemed to vary more, thus suggesting


dynamic chromatin reorganization (Figure 3-4C).


NF-Kp MAZ
1. hC
= -== SP1


il gl 4 '"*
,H *
i F.


ds4&1 7

FU! !ai.
7 ', S'

t,.





: U '.:.i'.

6' *we

W d a:& V
'. iA09 %V
Se jamw e
*j &1.4:


oCG VGC


'* i


m .a .1
*~t *w* w
* &I& 'q'r!1 I
* .&':i::.. W

i.: i
; .,


t
* hLt

* itet
I %flti
RAM


d Ii. a H ,:M:l


I, ,,a, i i 0*. f4

W. m' *a n5

*m TCG v Gm5c


. rt


ATG

SE,
S00%



II.

I 100%
1 3%
'00


"wy..PW v :i::WM


* 4ffYr,'r .
* W OT V


*'-. ,,,, '-.. \. .s L
t e .-. Vt a. t Bt
I .f.' W *1 .'(, g 4. 4-



<. H:::Jr k


*=VT 'AISm


Figure 3-4. Chromatin Probing of CaSki and SiHa Cell Line. Nuclei from both cells lines were
isolated and chromatin structure was probed as stated in the Materials and Methods.
Endogenously methylated CpG (CG) sites are depicted with black circles and
unmethylated CpG with white circles. All GC sites are denoted by triangles, and
those that were methylated by M.CviPI probe are red. Regions that were not protected
against methylation are indicated by rectangles spanning at least two contiguous GC
sites (i.e., one methylated site does constitute accessibility). Methylated GCG
(Gm5CG) sites are indicated in gray and cannot be discerned from endogenous
methylation or M.CviPI modified sites. The bisulfite conversion efficiency (C.E.) for
each molecule is shown at the right. Transcription factor binding sites for Spl (blue)
and both NF-KB and MYC-associated zinc finger (MAZ) (tan) are indicated.


A


B


7. i.
I.



I t.


61

U
3
+ 7-:
9 :"


C


*i **
% .tt



)" '" '**
q g

9 t
1 '
'. I+









WIF1 Promoter is Aberrantly Methylated in Squamous Cell Cervical Tumors

To determine if promoter methylation of the WIF1 gene occurs in primary cervical tumors,

we used pyrosequencing to analyze the methylation of WIF1 in patient-derived cervical

squamous cell carcinoma samples. Matched adjacent normal cervical stratified epithelium and

Pap smear samples from different patients were also analyzed. Cervical tumors and normal

tissues were obtained from the Moffitt Cancer Center Tissue Bank. DNA was extracted, treated

with sodium bisulfite and PCR amplified with the WIF1 pyrosequencing primers. PCR

amplicons where then analyzed with the PyroMark MD and Pyro Q-CpG software. All cases

samples were re-analyzed with repeated PCR and pyrosequencing reactions to validate the initial

pyrosequencing findings. Subsets of bisulfite-treated samples were also analyzed multiple times

by bisulfite genomic sequencing, with analogous results (data not shown).

To ascertain if aberrant methylation of the WIF1 promoter occurs in cervical cancer, we

pyrosequenced 22 high-grade squamous cell carcinomas. Of these 22 tumors, only 17 produced

PCR amplicons of sufficient quality for pyrosequencing. We grouped these specimens into three

categories based on their average methylation of six CpG sites as followed: little or no

methylation (0-9%), moderate methylation (10-19%) and heavily methylated (>20%). Of the 17

tumor samples, 6 were not methylated, 3 had moderate methylation, and 8 were heavily

methylated (Figure 3-5).

We also analyzed a set of 8 Pap smear samples that had been previously genotyped and 7

of which were found to be HPV positive. We found no significant methylation, <5%, in all 8 of

the Pap smear samples (Figure 3-6). In addition we analyzed the paired adjacent

normal tissue from each tumor sample. Unfortunately, several of theses samples failed to














c 50-
0

S40-

30.


I0.


Tumr:11 12 12 13 14 1i 17 1 20 21 22 23 2 2 2 27 31 32
Adj. Normal : 37 43 45 47 49 51 53 55 WS
Sample ID
Figure 3-5. Average CpG methylation of Cervical SSC and Adjacent Normal Tissue at WIF1.
The tumors and adjacent matching normal tissues were given arbitrary numbers. Both
the PCR and the pyrosequencing were run in duplicate, and the average methylation
status of the WIF1 promoter is shown along with the standard deviation for each
sample. Black bars represent the methylation of the tumor samples and the white bars
show the methylation of the paired normal stratified epithelium.



Pap #1 ^g.1 I Pap #7 ^.. =

mMyIt ,1: t1% 1% 1% 1% 2% 1% 1% (3% 2 .% 2% 2%

i i .. .................. .. ... ... .... .. ..... ....... ......
dI *p. ] rM. .... ... ., T......
40 ) .. .



d*llpl,,d; E- ,:TT jT*A. ^+JA Tit i i L^T ,IES*T ,' TOlTf':-.X "T','rGT';Y "T.T'T'13GAT; A TE
Figure 3-6. Programs of Pap Smear Samples at WIF1. The percent methylation at each CpG
dinucleotide (gray rectangles) is shown above the CpG site. Average CpG
methylation (Avg. Meth.) within the segment of the gene analyzed is displayed in the
box at the top of the programs. Seven additional pap smear samples were analyzed
with similar results (data not shown).

yield usable amounts of DNA, and we were not able to assess WIF1 promoter methylation.

However, the 9 matched normal samples that were analyzed all had little or no methylation

(<5%) (Figure 3-5). These data show that this region of the WIF1 promoter is essentially









methylation-free in normal cervical tissue and supports the notion that the DNA methylation that

we observed in the tumor samples resulted from the disease and was not part of the normal

biology of cervical tissue.

WIF1 Methylation is not Correlated with WIF1 Protein Levels

To assess if the methylation status of the WIF1 promoter correlated with the expression

of WIF1 protein, we conducted immunohistochemical staining on a tissue microarray including

22 tumors and their adjacent normal tissue. Immunohistochemistry (IHC) is a method for

demonstrating the presence and location of proteins in the context of intact tissue. IHC staining is

accomplished with antibodies that recognize the target protein. The antibody-antigen interaction

is visualized using chromogenic detection, in which an enzyme conjugated to the antibody

cleaves a substrate to produce a colored precipitate, in this case brownish-red, at the location of

the protein.

Immunostaining of normal tissue with anti-WIF 1 antibody indicated that the protein was

expressed in the stratified squamous epithelium of the ectocervix. No staining of the underlying

cervical parenchyma (Figure 3-7 Al) was observed, except within the endometrial cell layer of

blood vessels, which stained intensely with anti-WIF1 antibody. Close examination of the

staining indicated a gradient of WIF1 expression within epithelial tissue (Figure 3-7A).

Specifically, nearly no staining was observed in the basal epithelium or underlying basement

membrane (Figure 3-7A, zones 1 and 2). However, strong staining was evident within the

peribasal epithelial layer (Figure 3-7A, zone 3). We observed notable extracellular staining

within the layers of differentiated non-keratinized epithelium with diminished staining in cells at

the apical surface of the ectocervix (Figure 3-7A, zone 5). A negative control for background

staining was incubated with only rabbit IgG and demonstrated that there was no background

staining of the WIF1 antibody within the cervical epithelium (Figure 3-7 B).




















2 -. -' -
i,,,Z,. ~-- .. ,' "7' ,.' ,,' ... .



Figure 3-7. WIF1 Immunohistochemistry of Normal Cervical Tissue and SSC. Normal cervical
epithelium incubated with (A) anti-WIF 1 rabbit antibody or (B) rabbit IgG. No
staining from rabbit IgG and the location of WIF 1 staining in the cervical epithelium
was observed. (C) Tumor section with intense staining and low DNA methylation.
(D) Tumor section with mild staining and low methylation. (E) Tumor section with
light staining and high methylation. (F) Heterogeneous tumor with a mixed
population of stained and unstained cells. Bracketed at the left are the basal
epithelium (or underlying basement membrane, zones 1 and 2), peribasal epithelial
layer (zone 3), epithelial layer (zone 4) and apical surface of the ectocervix (zone 5).

When IHC was conducted on samples of invasive squamous cell carcinoma (SCC) with an

anti-WIF rabbit antibody, we observed intense staining (Figure 3-7C) in tumor samples that had

little methylation, such as sample 20 for which pyrosequencing showed an average CpG

methylation of 2%. However, there was a notable lack of staining within two samples, and this

was independent of the methylation status of the WIF1 gene promoter. For example, little

staining was seen in tumor sample 13 (Figure 3-7D) that possessed a mean methylation value of

2%. Moreover, the majority of methylated tumors did not express or had mixed expression of

WIF1. For example, tumor sample 32, which possessed a high level of WIF1 methylation (41%),

also showed no clear indication of expression of WIF1 protein (Figure 3-7E). We also observed a

number of tumors with heterogeneous population of cells that possessed mixed staining for

WIF1 within the same lesion (Figure 3-7F), which is a common feature of solid tumors. Of note,









we observed more intense WIF1 staining in less de-differentiated SCC lesions. These studies

show that low or absent WIF1 expression is commonly observed in invasive SCC of the cervix,

but that this tumor feature may occur independent of WIF1 promoter methylation status. These

findings also suggest that mechanisms) other then promoter hypermethylation may serve to

inhibit WIF1 expression. We conclude that hypermethylation of the WIF1 gene is a common

event in cervical squamous cell carcinoma but that promoter methylation is not the only

necessary cause of its silencing.









CHAPTER 4
DISCUSSION

The rationale for examining DNA methylation at the WIF1 promoter in cervical cancer

was based on three primary factors. First, exploitation of the Wnt pathway is a common

mechanism of tumorigenesis. Uncontrolled p-catenin signaling, either from disruption of

function of its regulatory proteins (i.e., Axin, APC, WIF1) or genetic mutations that prevent P-

catenin degradation, leads to excessive proliferation that predisposes cells to tumorigenesis [65].

Second, WIF1 expression has been shown to be downregulated in a variety of cancer types, such

as bladder, breast, lung and prostate [74]. WIF1 downregulation is an epigenetic event correlated

to promoter hypermethylation in many of these cancers. Third, while some studies have

examined the epigenetic regulation of WIF1 and other genes encoding molecular components of

the Wnt signaling pathway, no research has examined WIF1 silencing in cervical carcinomas.

Using both cervical cancer cell lines and primary formalin-fixed/paraffin-embedded

squamous cell carcinomas samples, we established that the WIF1 gene is commonly subjected to

DNA hypermethylation in cervical tumors. In vitro studies in cervical cell lines indicated that

WIF1 expression correlated with promoter methylation. Furthermore, we showed that the DNA

demethylating agent 5-azadC increased WIF1 expression. In an effort to show that WIF1

expression is epigenetically silenced in primary tissue, we conducted pyrosequencing and

immunostaining on cervical tumor samples. We found that the WIF1 promoter is unmethylated

and WIF 1 protein is expressed in the peribasal and the less differentiated epithelial layers of

normal cervical epithelium. In contrast, within high-grade squamous cell carcinomas, WIF1

protein expression was often low to nonexistent. This loss of WIF1 protein expression correlated

with promoter hypermethylation in several of the tumors; however, loss of WIF 1 protein

expression in two of the tumors could not be attributed to WIF1 promoter hypermethylation. This









suggests that WIF1 is epigenetically silenced in cervical cancer, but other mechanisms in

addition to promoter hypermethylation may be involved.

Lack of complete concordance between WIF1 promoter hypermethylation and lack of

protein expression could be due to several factors. First, pyrosequencing analyses revealed the

six CpG sites in the WIF1 promoter in CaSki cells were unmethylated; however, transcript levels

increased after treatment with 5-azaC as assayed by RT-PCR. These findings suggest that the

WIF1 gene is epigenetically silenced in these cells, but by methylation of CpG sites other then

the six downstream of the TSS we assayed by pyrosequencing. This is supported by MAPit

methylation footprinting experiments, which showed two areas of endogenous methylation

further upstream of the TSS in CaSki cells. PCR using the MAPit primers was attempted on

several of the tumor samples without success. This failure is likely because FFPE- and bisulfite-

treated DNA is very fragmented and thus unable to yield large PCR products. One method to

examine if the tumor samples had similar patches of methylation to the CaSki cell line further

upstream of the TSS would be to create primers specific for the areas of interest that yield

smaller PCR products.

Pyrosequencing of small bisulfite-converted amplicons at the two upstream methylated

regions is needed to test the extent to which hypermethylation correlates with WIF1

downregulation in the primary tumor samples. Methylation of either of these upstream regions

correlating with the loss of WIF1 expression as determined by IHC would lend more support to

the hypothesis that the WIF1 gene is directly epigenetically silenced in cervical cancer.

HPV may also play a role in the epigenetic silencing of the WIF1 promoter in cervical

cancer. This is based on the observation that the three cell lines (CaSki, SiHa and HeLa) that

displayed varying levels of WIF1 epigenetic silencing were all derived from HPV-infected









specimens. By contrast, the cell line derived from a non-HPV-infected specimen did not appear

to be epigenetically silenced. HPV viral proteins can directly effect DNA methylation by binding

to and recruiting DNMTs [57]. Knockdown of expression of the viral E6 and E7 genes may help

to understand if the virus plays a role in WIF1 hypermethylation. Furthermore, HPV genotyping

of the patient cervical tumors should be done to look for a correlation between WIF1 expression

and the presence of HPV.

MAPit footprinting of the CaSki and SiHa cell lines may yield insight into how WIF1

silencing might occur regardless of promoter methylation. In the WIF]-expressing CaSki cells,

we observed a region of protection against methylation by M.CviPI just upstream of the

transcriptional start site. We hypothesized that this area would be nucleosome free and hence

accessible to the DNMT, as has been observed at other transcriptionally-active promoters. The

size of the inaccessible region is consistent with a nucleosome (-147 base pairs) that occupies a

variety of positions. The variable location of the protected region on each molecule seems

inconsistent with protection of the region against methylation by a site-specific transcription

factor bound to DNA, such as SP1. It is possible that SP1 or another transcription factor is

forming a ternary complex with a nucleosome that occupies several positions.

Interestingly, two different populations of molecules with distinct endogenous DNA

methylation patterns were observed in the CaSki cell line. In one group, nearly no endogenous

methylation was observed, whereas the other group had two distinct areas of endogenous

methylation straddling the upstream SP1 site. Despite these two patterns of methylation by

endogenous enzymes, there appears to be little difference in accessibility to M.CviPI between

them. This suggests that whatever is bound to the DNMT accessible area binds regardless of the

methylation status of the gene. It should be noted, however, that we have only examined a









limited number of molecules from one experiment and there exists the possibility the molecules

shown were selectively sequenced due to PCR or cloning bias. Similarly, the heavily methylated

WIF1 promoter in SiHa cells showed clear regions of M.CviPI accessibility but with greater

variability (Figure 7C). Often, areas of heavily methylated DNA are inaccessible to the

methyltransferase and unable to provide clear footprints. This suggests that, whereas the WIF1

promoter is heavily methylated in the SiHa cell line, it is not in a fully compacted

heterochromatin state.

Further experiments are needed to clarify what is binding to the area upstream of the WIF1

TSS. Knockdown experiments of the various transcription factors, particularly SP1, in the cell

line followed by MAPit might help to clarify what is binding to the DNA. It would be of interest

to see if knockdown of SP1 alters WIF1 endogenous methylation or chromatin accessibility,

particularly in the highly methylated SiHa cell line. However, preferential association of NF-KB

or MYC-associated with the WIF1 promoter in Caski, but not in SiHa, cells may explain the

difference in expression between the cell lines. Chromatin immunoprecipitation experiments

probing for histones might help to create a clearer picture of regulation of the WIF1 promoter

and establish if there is a bias in the MAPit footprint. Better understanding of the WIF1 promoter

region and its associated proteins could potentially lead to use as a cervical cancer drug therapy.

In addition to technical limitations that may have precluded detection of DNA methylation

in cervical tumors lacking WIF 1 protein expression, it is also possible that non-epigenetic factors

are also responsible for loss of WIF1 expression in these tumors. For example, chromosome

rearrangements involving 12ql3-15, where WIF1 is located, are frequent occurrences among

tumors. While no DNA mutations of WIF1 have been reported, one group demonstrated salivary

gland pleomorphic adenoma expressing an high mobility group AT-hook 2 (HMGA2)/WIF1









fusion transcript, which resulted in re-expression of wild-type HMGA2 transcripts and very low

levels of WIF1 expression [78]. An alternative explanation for the lack of correlation between

WIF1 promoter methylation and protein expression is that WIF1 transcription may require an

additional activator which could be inhibited or after chromosome rearrangement the gene may

be excised. These are a few alternative mechanisms aside from epigenetic silencing that could

result in downregulation of WIF 1 protein in cervical cancer.

A common aim of TSG research is to evaluate the possibility of a gene being used as either

a diagnostic or prognostic tool. We showed that the WIF1 gene is aberrantly methylated in high-

grade squamous cell carcinomas of the cervix, which makes it a potential prognostic marker.

However, we failed to obtain any low-grade or mildly dysplastic cervical squamous cells for

analysis within the confines of the study. Examining the status of WIF1 expression at pre-

cancerous stages or within low-grade carcinomas may help determine when during tumor

progression aberrant methylation of the WIF1 promoter occurs, and if hypermethylation is an

indicator of high-grade squamous cell carcinoma. There exists a possibility that a correlation

between expression of WIF1 and the pathological state or grade of the tumor could be made to

provide insight into WIF1 being a possible biomarker for diagnosis.

It is clear that cancer cells undergo global changes in DNA methylation that lead to

chromatin alterations involving the whole epigenome and that entire pathways relevant to cell

division are subject to epigenetic dysregulation. As the role of epigenetics in cancer becomes

clearer new approaches to cancer prevention, detection, and therapy will begin to emerge.









LIST OF REFERENCES


1. Detailed Guide: Cervical Cancer. What are the key statistics about cervical cancer?
Available from: http://www.cancer.org/docroot/CRI/CRI_2_3x.asp?rnav=cridg&dt=8.

2. Schoell, W.M.J., M.F. Janicek, and R. Mirhashemi, Epidemiology and biology of
cervical cancer. Seminars in Surgical Oncology, 1999. 16(3): p. 203-11.

3. Kurman, R.J., et al., From Papanicolaou to Bethesda: the rationale for a new cervical
cytologic classification. Obstet Gynecol, 1991. 77(5): p. 779-82.

4. Lie, A.K. and G. Kristensen, Human papillomavirus E6/E7 mRNA testing as a predictive
marker for cervical carcinoma. Expert Rev Mol Diagn, 2008. 8(4): p. 405-15.

5. Walboomers, J.M.M., et al., Human papillomavirus is a necessary cause of invasive
cervical cancer worldwide. J Pathol, 1999. 189(1): p. 12-9.

6. Ho, G.Y.F., et al., Persistent Genital Human Papillomavirus Infection as a Risk Factor for
Persistent Cervical Dysplasia. J Natl Cancer Institute, 1995. 87(18): p. 1365-71.

7. Choo, K.B., C.C. Pan, and S.H. Han, Integration of human papillomavirus type-16 into
cellular DNA of cervical carcinoma Preferential Deletion of the E2 gene and invariable
retention of the long control region and the E6/E7 Open Reading Frames. Virology, 1987.
161(1): p. 259-61.

8. Di Leonardo, A., et al., DNA damage triggers a prolonged p53-dependent G1 arrest and
long-term induction of Cipl in normal human fibroblasts. Genes Dev, 1994. 8(21): p.
2540-51.

9. Smith, M.L., et al., Involvement of the p53 tumor suppressor in repair of U.V.-type DNA
damage. Oncogene, 1995. 10(6): p. 1053-9.

10. Subramanya, D. and P.D. Grivas, HPV and cervical cancer: updates on an established
relationship. Postgrad Med, 2008. 120(4): p. 7-13.

11. Gissmann, L., Human papillomaviruses and genital cancer. Semin Cancer Biol, 1992.
3(5): p. 253-61.

12. American Type Culture Collection., Catalogue of strains II, American Type Culture
Collection.: Rockville, Md.,. p. v.

13. Scheffner, M., et al., The state of the p53 and retinoblastoma genes in human cervical
carcinoma cell lines. Proc Natl Acad Sci USA, 1991. 88(13): p. 5523-7.

14. Jones, P.A. and S.B. Baylin, The epigenomics of cancer. Cell, 2007. 128(4): p. 683-92.










15. Jones, P.A., Overview of cancer epigenetics. Semin Hematol, 2005. 42(3 Suppl 2): p. S3-
8.

16. Cooper, D.N. and M. Krawczak, Cytosine methylation and the fate of CpG dinucleotides
in vertebrate genomes. Hum Genet, 1989. 83(2): p. 181-8.

17. Bird, A., DNA methylation patterns and epigenetic memory. Genes Dev, 2002. 16(1): p.
6-21.

18. Larsen, F., et al., CpG islands as gene markers in the human genome. Genomics, 1992.
13(4): p. 1095-107.

19. Mohandas, T., R.S. Sparkes, and L.J. Shapiro, Reactivation of an inactive human X
chromosome: evidence for X inactivation by DNA methylation. Science, 1981.
211(4480): p. 393-6.

20. Li, E., C. Beard, and R. Jaenisch, Role for DNA methylation in genomic imprinting.
Nature, 1993. 366(6453): p. 362-5.

21. Suetake, I., et al., DNMT3L stimulates the DNA methylation activity of Dnmt3a and
Dnmt3b through a direct interaction. J Biol Chem, 2004. 279(26): p. 27816-23.

22. Otani, J., et al., Structural basis for recognition of H3K4 methylation status by the DNA
methyltransferase 3A ATRX-DNMT3-DNMT3L domain. EMBO Rep, 2009. 10(11): p.
1235-41.

23. Hermann, A., H. Gowher, and A. Jeltsch, Biochemistry and biology of mammalian DNA
methyltransferases. Cell Mol Life Sci, 2004. 61(19-20): p. 2571-87.

24. Luger, K., et al., Crystal structure of the nucleosome core particle at 2.8 A resolution.
Nature, 1997. 389(6648): p. 251-60.

25. Sarma, K. and D. Reinberg, Histone variants meet their match. Nat Rev Mol Cell Biol,
2005. 6(2): p. 139-49.

26. Kouzarides, T., Chromatin modifications and their function. Cell, 2007. 128(4): p. 693-
705.

27. Fuks, F., et al., The methyl-CpG-binding protein MeCP2 links DNA methylation to
histone methylation. J Biol Chem, 2003. 278(6): p. 4035-40.

28. Nan, X., et al., Transcriptional repression by the methyl-CpG-binding protein MeCP2
involves a histone deacetylase complex. Nature, 1998. 393(6683): p. 386-9.









29. Tachibana, M., et al., G9a/GLP complexes independently mediate H3K9 and DNA
methylation to silence transcription. EMBO J, 2008. 27(20): p. 2681-90.

30. Zhao, Q., et al., PRMT5-mediated methylation of histone H4R3 recruits DNMT3A,
coupling histone and DNA methylation in gene silencing. Nat Struct Mol Biol, 2009.
16(3): p. 304-11.

31. Esteve, P.O., et al., Regulation ofDNMT1 stability through SET7-mediated lysine
methylation in mammalian cells. Proc Natl Acad Sci USA, 2009. 106(13): p. 5076-81.

32. Jones, P.L., et al., Methylated DNA and MeCP2 recruit histone deacetylase to repress
transcription. Nat Genet, 1998. 19(2): p. 187-91.

33. Robertson, K.D., et al., DNMT1 forms a complex with Rb, E2F1 and HDAC1 and
represses transcription from E2F-responsive promoters. Nat Genet, 2000. 25(3): p. 338-
42.

34. Wang, Z., et al., Genome-wide mapping of HATs and HDACs reveals distinct functions
in active and inactive genes. Cell, 2009. 138(5): p. 1019-31.

35. Lorch, Y., J.W. LaPointe, and R.D. Kornberg, Nucleosomes inhibit the initiation of
transcription but allow chain elongation with the displacement of histones. Cell, 1987.
49(2): p. 203-10.

36. Bernstein, B.E., et al., Global nucleosome occupancy in yeast. Genome Biol, 2004. 5(9):
p. R62.

37. Jiang, C. and B.F. Pugh, Nucleosome positioning and gene regulation: advances through
genomics. Nat Rev Genet, 2009. 10(3): p. 161-72.

38. Ozsolak, F., et al., High-throughput mapping of the chromatin structure of human
promoters. Nat Biotechnol, 2007. 25(2): p. 244-8.

39. Heintzman, N.D., et al., Distinct and predictive chromatin signatures of transcriptional
promoters and enhancers in the human genome. Nat Genet, 2007. 39(3): p. 311-8.

40. Boeger, H., J. Griesenbeck, and R.D. Kornberg, Nucleosome retention and the stochastic
nature of promoter chromatin remodeling for transcription. Cell, 2008. 133(4): p. 716-26.

41. Dechassa, M.L., et al., SWI/SNF has intrinsic nucleosome disassembly activity that is
dependent on adjacent nucleosomes. Mol Cell, 2010. 38(4): p. 590-602.

42. Shivaswamy, S., et al., Dynamic remodeling of individual nucleosomes across a
eukaryotic genome in response to transcriptional perturbation. PLoS Biol, 2008. 6(3): p.
e65.









43. Schones, D.E., et al., Dynamic regulation of nucleosome positioning in the human
genome. Cell, 2008. 132(5): p. 887-98.

44. Gilbert, N. and B. Ramsahoye, The relationship between chromatin structure and
transcriptional activity in mammalian genomes. Brief Funct Genomic Proteomic, 2005.
4(2): p. 129-42.

45. Ptashne, M., On the use of the word 'epigenetic'. Curr Biol, 2007. 17(7): p. R233-6.

46. Zemach, A., et al., Genome-wide evolutionary analysis of eukaryotic DNA methylation.
Science, 2010. 328(5980): p. 916-9.

47. Sherr, C.J., Principles of tumor suppression. Cell, 2004. 116(2): p. 235-46.

48. Jones, P.A. and S.B. Baylin, The fundamental role of epigenetic events in cancer. Nat
Rev Genet, 2002. 3(6): p. 415-428.

49. Garinis, G.A., et al., DNA hypermethylation: when tumour suppressor genes go silent.
Hum Genet, 2002. 111(2): p. 115-27.

50. Boyes, J. and A. Bird, Repression of genes by DNA methylation depends on CpG density
and promoter strength: evidence for involvement of a methyl-CpG binding protein.
EMBO J, 1992. 11(1): p. 327-33.

51. Wolffe, A.P., Chromatin remodeling: why it is important in cancer. Oncogene, 2001.
20(24): p. 2988-90.

52. Morey, L., et al., MBD3, a component of the NuRD complex, facilitates chromatin
alteration and deposition of epigenetic marks. Mol Cell Biol, 2008. 28(19): p. 5912-23.

53. Ramirez, J. and J. Hagman, The Mi-2/NuRD complex: a critical epigenetic regulator of
hematopoietic development, differentiation and cancer. Epigenetics, 2009. 4(8): p. 532-6.

54. Reisman, D., S. Glaros, and E.A. Thompson, The SWI/SNF complex and cancer.
Oncogene, 2009. 28(14): p. 1653-68.

55. Etoh, T., et al., Increased DNA methyltransferase 1 (DNMT1) protein expression
correlates significantly with poorer tumor differentiation and frequent DNA
hypermethylation of multiple CpG islands in gastric cancers. Am J Pathol, 2004. 164(2):
p. 689-99.

56. Muller, K., H. Heller, and W. Doerfler, Foreign DNA integration. Genome-wide
perturbations of methylation and transcription in the recipient genomes. J Biol Chem,
2001. 276(17): p. 14271-8.









57. Burgers, W.A., et al., Viral oncoproteins target the DNA methyltransferases. Oncogene,
2007. 26(11): p. 1650-5.

58. Dong, S.M., et al., Promoter hypermethylation of multiple genes in carcinoma of the
uterine cervix. Clin Cancer Res, 2001. 7(7): p. 1982-6.

59. Sawada, M., et al., Increased expression of DNA methyltransferase 1 (DNMT1) protein
in uterine cervix squamous cell carcinoma and its precursor lesion. Cancer Lett, 2007.
251(2): p. 211-9.

60. Bigey, P., et al., Transcriptional regulation of the human DNA Methyltransferase
(DNMT1) gene. Gene, 2000. 242(1-2): p. 407-18.

61. Ghittoni, R., et al., The biological properties of E6 and E7 oncoproteins from human
papillomaviruses. Virus Genes, 2010. 40(1): p. 1-13.

62. Cadigan, K.M. and R. Nusse, Wnt signaling: a common theme in animal development.
Genes Dev, 1997. 11(24): p. 3286-305.

63. Hobmayer, B., et al., WNT signalling molecules act in axis formation in the diploblastic
metazoan Hydra. Nature, 2000. 407(6801): p. 186-9.

64. Reya, T. and H. Clevers, Wnt signalling in stem cells and cancer. Nature, 2005.
434(7035): p. 843-50.

65. MacDonald, B.T., K. Tamai, and X. He, Wnt/beta-catenin signaling: components,
mechanisms, and diseases. Dev Cell, 2009. 17(1): p. 9-26.

66. Kawano, Y. and R. Kypta, Secreted antagonists of the Wnt signalling pathway. J Cell Sci,
2003. 116(Pt 13): p. 2627-34.

67. Cong, F., L. Schweizer, and H. Varmus, Wnt signals across the plasma membrane to
activate the P-catenin pathway by forming oligomers containing its receptors, Frizzled
and LRP. Development, 2004. 131(20): p. 5103-15.

68. He, X., et al., LDL receptor-related proteins 5 and 6 in Wnt/p-catenin signaling: arrows
point the way. Development, 2004. 131(8): p. 1663-77.

69. van Amerongen, R., A. Mikels, and R. Nusse, Alternative Wnt signaling is initiated by
distinct receptors. Sci Signal, 2008. 1(35): p. re9.

70. Arce, L., K.T. Pate, and M.L. Waterman, Groucho binds two conserved regions of LEF-1
for HDAC-dependent repression. BMC Cancer, 2009. 9: p. 159.

71. Jessen, J.R., Noncanonical Wnt signaling in tumor progression and metastasis. Zebrafish,
2009. 6(1): p. 21-8.










72. Uren, A., et al., Activation of the canonical Wnt pathway during genital keratinocyte
transformation: a model for cervical cancer progression. Cancer Res, 2005. 65(14): p.
6199-206.

73. Hsieh, J.C., et al., A new secreted protein that binds to Wnt proteins and inhibits their
activities. Nature, 1999. 398(6726): p. 431-6.

74. Wissmann, C., et al., WIF a component of the Wnt pathway, is down-regulated in
prostate, breast, lung, and bladder cancer. J Pathol, 2003. 201(2): p. 204-12.

75. Warnecke, P.M., et al., Identification and resolution of artifacts in bisulfite sequencing.
Methods, 2002. 27(2): p. 101-7.

76. Macleod, D., et al., Spl sites in the mouse aprt gene promoter are required to prevent
methylation of the CpG island. Genes Dev, 1994. 8(19): p. 2282-92.

77. Reguart, N., et al., Cloning and characterization of the promoter of human Wnt inhibitory
factor-1. Biochem Biophys Res Commun, 2004. 323(1): p. 229-34.

78. Persson, F., et al., High-resolution genomic profiling of adenomas and carcinomas of the
salivary glands reveals amplification, rearrangement, and fusion of HMGA2. Genes
Chromosomes Cancer, 2009. 48(1): p. 69-82.









BIOGRAPHICAL SKETCH

Amber Lyn Delmas was born in Stuart, Florida. Through dual enrollment during high

school, she attended Indian River Community College. She received an Associates of Arts

degree in science the summer after graduating from high school in 2003. She then attended the

University of Florida and earned a Bachelor of Science in microbiology and cell science in 2006.

After graduation, she worked as a laboratory technician in the Department of Biochemistry and

Molecular Biology at the University of Florida for two years. In August 2010, she received a

Master of Science in medical sciences from the University of Florida.





PAGE 1

WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN CERVICAL SQUAMOUS CELL CARCINOMA By Amber Delmas A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2010 1

PAGE 2

2010 Amber Delmas 2

PAGE 3

ACKNOWLEDGMENTS I am most thankful to both of my mentors, Mike and Kevin, for believing in me and constantly stimulating my brain. Mike is an extremely inte lligent individual who ha s continuously inspired me to think outside of the box, while Kevin is one of the most dedicated and hardworking individuals I know. Together you make the perfect boss. To my committee members: Jim Resnick and Paul Gulig, thank you for your guidance and support. My deepest appreciation to my lab family: Caro lina Pardo, Russell Darst, Nancy Nabilsi, Santhi Pondugula, and Chris Hoffman. Thank you for pu tting up with my endless questions and constant interruptions. I also wish to thank my extended lab family: Lisa Dyer for the mass of cell culture work and Ryan Skehan fo r his wealth of computer knowledge. I want to acknowledge the Moffitt Cancer Center a nd Erin Siegel in Tampa, Florida and wish to thank Bridget Riggs for her hard work and dedication. I am thankful to Joyce Conners and all of her ha rd work in assisting every Masters student along the road from acceptance to graduation. I thank my Dad for trying to understand what ex actly Im researching and to my Mom for not trying to have me explain what I do. I wish to express thanks to John Ba rli for allowing me the opportunity to support myself thr ough school, and to everyone else who has come in and out of my life to get me where I am today. 3

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3 LIST OF FIGURES.........................................................................................................................6 LIST OF ABBREVIATIONS.......................................................................................................... 7 ABSTRACT...................................................................................................................................10 CHAPTER 1 BACKGROUND................................................................................................................... .12 Cervical Cancer......................................................................................................................12 Human Papillomavirus....................................................................................................13 Human Cell Culture.........................................................................................................15 Epigenetics..............................................................................................................................16 DNA Methylation............................................................................................................16 Histone Modifications.....................................................................................................17 Nuclesome Positioning....................................................................................................18 Epigenetics in Cancer.......................................................................................................... ...20 Aberrant Methylation......................................................................................................20 Chromatin Changes in Cancer.........................................................................................21 Wnt Pathway...........................................................................................................................23 Wnt Agonists and Antagonists........................................................................................23 Wnt Receptors.................................................................................................................2 4 Canonical Pathway..........................................................................................................24 Noncanonical Pathway....................................................................................................26 2 MATERIALS AND METHODS...........................................................................................28 Cervical Specimens............................................................................................................. ...28 Cell Culture and 5-aza-2'-deoxycytidine (5-azadC) Treatment..............................................28 RNA and DNA Isolation........................................................................................................29 Sodium Bisulfite Modification...............................................................................................30 Polymerase Chain Reaction and Pyrosequencing...................................................................31 Reverse Transcriptase PCR....................................................................................................31 Methyltransferase Accessibility Protoc ol for Individual Templates (MAPit)........................32 Immunohistochemistry........................................................................................................... 33 3 RESULTS...................................................................................................................... .........35 WIF1 is Epigenetically Silenced in Cervical Cancer Cell Lines............................................35 Aberrant Chromatin Footprint in CaSki and SiHa Cell Lines................................................39 WIF1 Promoter is Aberrantly Methylated in Squamous Cell Cervical Tumors.....................41 WIF1 Methylation is Not Correlated with WIF1 Protein Levels...........................................43 4

PAGE 5

4 DISCUSSION................................................................................................................... ......46 LIST OF REFERENCES...............................................................................................................51 BIOGRAPHICAL SKETCH.........................................................................................................57 5

PAGE 6

LIST OF FIGURES Figure page 1-1 Normal Cervical Epithelium, CIN1 and CIN2-3...............................................................13 1-2 Diagram of Open and Closed Chromatin...........................................................................19 1-3 Diagram of Canonical Wnt Pathway.................................................................................25 3-1 RT-PCR of Cervical Cell Lines Before and After 5-azadC Treatment.............................36 3-2 WIF1 Promoter and Pyrograms of Cervical Cell Lines.....................................................37 3-3 Average Methylation of the WIF1 Promoter Before and After 5-azadC Treatment.........38 3-4 Chromatin Probing of CaSki and SiHa Cell Line..............................................................40 3-5 Average CpG methylation of Cervi cal SSC and Adjacent Normal Tissue at WIF1 .........42 3-6 Pyrograms of Pap Smear Samples at WIF1 .......................................................................42 3-7 WIF1 Immunohistochemistry of Normal Cervical Tissue and SSC..................................44 6

PAGE 7

LIST OF ABBREVIATIONS APC adenomatous polyposis coli ATP adenosine-5'-triphosphate CCD charged couple device CK1 casein kinase 1 CLB cell lysis buffer CIN cervical intraepithelial neoplasia CIS carcinoma in situ CpG cytosine guanine dinucleotide CRB cell resuspension buffer DAB diaminobenzidine dg.dH20 degassed distilled water dH20 distilled water DKK Dickkopf DNA deoxyribonucleic acid DNMT deoxyribonucleic acid methyltransferase dNTP deoxyribonucleotide triphosphate Dvl Dishevelled EGF epidermal growth factor FBS fetal bovine serum FFPE formalin-fixed paraffin-embedded Fz or Fzd Frizzled receptor GSK3 glycogen synthase kinase 3 GTPase guanine triphosphotase HAT histone acetyltransferase 7

PAGE 8

HDAC histone deacetylase HDM histone demethylase HMGA2 high mobility group AT-hook 2 HMT histone methyltransferase HPV human papillioma virus hr hour(s) IHC immunohistochemistry JNK c-jun N-terminal kinases LEF lymphoid enhancer factor LRP lipoprotein receptor-related protein MB methylation buffer MAPit methyltransferase accessibility protocol for individual templates MAZ MYC-associated zinc finger min minute(s) NF-kB nuclear factor-kappa B NFR nucleosome-free region NuRD nucleosome remodeling and deacetylase Pap Papanicolau PCR polymerase chain reaction PPi pyrophosphate Rb retinoblastoma protein RNA ribonucleic acid RT-PCR reverse transcriptase polymerase chain reaction SCC squamous cell carcinoma sec second(s) 8

PAGE 9

sFRP secreted Frizzled receptor protein SWI/SNF switch/sucrose non-fermentable TCF T cell factor TSG tumor suppressor gene TSS transcription start site UICC International Union Against Cancer WD WIF1 domain WIF1 Wnt inhibitor factor 1 WHO World Health Organization 9

PAGE 10

Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science WIF1 IS A FREQUENT TARGET FOR ABERRANT HYPERMETHYLATION IN CERVICAL SQUAMOUS CELL CARCINOMA By Amber Delmas August 2010 Chair: Michael Kladde Major: Medical Sciences Aberrant activation of the Wnt/ -catenin signaling pathway is a prominent oncogenic mechanism in numerous cancer types, and eviden ce indicates this pathway is of importance in human cervical cancer. Wnt Inhibito ry Factor 1 (WIF1) is one of several secreted antagonists that bind to Wnt. Although it has been describe d as a tumor suppressor in various types of cancer, regulation of the WIF1 gene has not been examined in human cervical cancer. Here we show that WIF1 was unmethylated and expressed in normal cervical epithelium and aberrantly methylated in high-grade, squamous cell carcino mas. Three of four cervical tumor cell lines showed low or no WIF1 expression. In all three, expression increased when cells were cultured with a DNA demethylating drug, indi cating epigenetic silencing of WIF1 Interestingly, differences were seen in chromatin structure between expressing and non -expressing cell lines as shown by the chromatin footprin ting technique, MAPit. The WIF1 promoter was aberrantly methylated in ten of seventeen high-grade squa mous cell cervical tumors, compared to paired normal tissue. Immunohistochemistry of normal stratified squamous cervical epithelium revealed that WIF1 expression was high in peribasal cells, but gr adually diminished toward the superficial epithelial layer. WIF1 protein was not detectable in tumors with high WIF1 methylation and was absent in two of the seven unme thylated tumors. Thus, although WIF1 was hypermethylated in 10

PAGE 11

most cervical cancer samples, ot her mechanisms may also contri bute to its repression. Our findings establish the WIF1 gene as commonly hypermethylat ed and a possible target for epigenetic silencing in squamous cell carcinoma. 11

PAGE 12

CHAPTER 1 BACKGROUND Cervical Cancer In 2009, the American Cancer Society estim ated within a year approximately 4,070 women would die from cervical cancer in the United States. About 11,270 new cases of invasive cervical cancer would be diagnosed, and four ti mes as many women would have a non-invasive cervical cancer [1]. Although cervical cancers can develop in the mucus-producing gland cells of the endocervix, most form in the lining of the ectocervix, which is composed of stratified squamous epithelium. The area defining the bord er between the endocer vix and ectocervix is most susceptible to dysplasia, and is termed the transformation zone (Figure 1-1A). Dysplasia is characterized by four major pa thological microscopic chan ges: unequally sized cells, abnormally-shaped cells, an unusual number of cells that are currently dividing and hyperchromatism (degeneration of cell nuclei or in creased staining capacity) [2]. If molecular changes occur in the basal cells where these cel ls originate from, they can promote the normal tissue through increasing grades of dysplasia and if not caught, cervical carcinoma. The naming and classification of cervical carcinoma precursor lesions has changed several times. Originally, the World Health Organization (WHO) classified le sions as mild, moderate and severe dysplasia or carcinoma in situ (CIS). These precursor lesions were graded depending on the degree of disruption of epithelial differentiation. These classi fications were later st andardized as cervical intraepithelial neoplasia (CIN), with mild dyspl asia designated CIN1 and severe as CIN3. The most recent classification is the Bethesda System which divides all cervical epithelial precursor lesions into two groups [3]: lowgrade squamous intraepithelial le sions that correspond to CIN1 (Figure 1-1B) and high-grade squamous intraepithe lial lesions that includ e CIN2 and 3 (Figure 1-1C). These cellular changes can be detected by the Papanicolau te st (Pap test) and treated to 12

PAGE 13

prevent the development of cancer. Treatment consists of surgery in early stages or chemotherapy and radiotherapy in adva nced stages of the disease. Figure 1-1. Normal Cervical Epithelium, CIN1 and CIN2-31 (A) Staining of the transformation zone showing both squamous and gla ndular epithelium on the surface and the underlying parenchyma Nuclei are stained dark purple. (B) Squamous cells that have progressed to CIN1: moderate nuclear enlargement and rounded cells. (C) CIN 2-3 cells have large nuclei occupying the majo rity of the cell and increased staining. Human Papillomavirus Infection with oncogenic human papillomavirus (HPV) has been shown to be the primary etiological factor for cervical cancer and its pr ecursor lesions [5]. Papillomavirus is a small deoxyribonucleic acid (DNA) virus that induces a variety of proliferative lesions HPV infections 1 adapted with permission from Ref. 4. Lie, A.K. and G. Kristensen, Human papillomavirus E6/E7 mRNA testing as a predictive marker for cervical carcinoma. Expert Rev Mol Diagn, 2008. 8 (4): p. 405-415. 13

PAGE 14

are often transient, but persistent infections, es pecially with oncogenic HPV or high-risk types, can further increase the likelihood of cervical dysplasia [6]. HPV-induced oncogenesis in cervical carcinoma derives from the properties of the expressed viral gene products. Integration of the viral circular DNA into the patients ge nome often disrupts the viral E2 gene, which leads to deregulation and expression of the E6 and E7 viral proteins [7]. The E6 gene product is able to interact with many cellular pr oteins, but it primarily binds with the human tumor suppressor p53 protein. E6 forms a complex with the transcription factor AP1, binds to p53, and induces its degradation through the ubiquitin protein ligase pathway. The p53 protein has many names, including "the guard ian of the genome", due to its role in maintenance of genome stability by preventing the transmission of DNA mutations that arising due to cellular stress. Cells with dysfuncti onal p53 are unable to respond normally to DNA damage and thus allow accumulati on of genomic mutations [8, 9]. Viral E7 disrupts normal cell cycle regulati on through its interaction with the tumor suppressor retinoblastoma (Rb) protein. Under nor mal cell cycle regulation, Rb forms complexes with the E2F-1 family of transcription factors th ereby inhibiting their function. Viral E7 binds to Rb thus disrupting the inhi bitory complex resulting in the release of active E2F-1. S ubsequently E2F-1 binds and initiates transcrip tion of target genes that facilitate the G1/S transition and begin S phase. The E7 proteins of the high risk HPVs, such as HPV-16 and HPV-18, bind Rb with 10-fold higher affinity than do the E7 proteins of the low risk HPV types. The difference in binding affinity correlates with the transforming potential of the different E7 proteins. HPV infection alone is not sufficient to induce malignant transformation. Ot her significant cofactors contribute to the multi-step process of tumor form ation, such as individual genetic variations as well as environmental factors; however, such cofactors are not important in the absence of HPV 14

PAGE 15

[10]. The combined dysregulation of p53 and Rb a llows cells with DNA damage to proceed into S phase facilitating the a ccumulation of genetic mutations, whic h is the basis for carcinogenesis. Human Cell Culture Much of what is known about the correlation between cervical cancer and HPV infection comes from cell culture studies [ 11]. The practice of tissue cultur e is commonly used to study all types of cancer, including those of the cervix. Cell culture is a useful tool for the study of in vitro animal cell biology because it allows access to a highly-controlled, easily-manipulated environment. There are a number of cultured hum an cervical cell lines that are commonly used to study cervical cancer. The cell lines used in this study are C-33A, CaSki, HeLa 229 and SiHa. The C-33A cell line is one of a series of lin es derived by N. Auersp erg from cervical cancer biopsies [12]. The line, when originally cu ltured, exhibited an epithelial morphology, and karyological instability that was observed with continued passage. The line is pseudodiploid with a modal chromosome number of forty-six, occurr ing in 70% of cells examined. The cells are negative for human papillomavirus DNA and RNA. The CaSki line was es tablished with cells from a metastasis in the small bowel mesentery of a woman with epidermoid carcinoma of the cervix. These cells are reported to contain integr ated HPV-16 as well as sequences related to HPV-18. HeLa 229 cells were created from an epith elial cervical adenocar cinoma that contains HPV-18. They have a modal chromosome number of 82. HeLa 229 cells are essentially the same as the original HeLa cell line; however, they are relatively resistant to infection by polioviruses. The SiHa cell line was established from fragments of a primary tissue sample obtained after surgery from a Japanese patient with grade-II squamous cell carcinom a that contains one to two copies of HPV-16 per cell [12]. The HPV-positive cervical cancer cell lines express normal Rb and low levels of wild-type p53 proteins, which are presumed to be altered in function as a consequence of association with HPV E7 and E6 oncoproteins, respectively. In C-33A, the Rb 15

PAGE 16

protein is present but abnormal in size. These cells also overexpress a p53 DNA binding mutant with an arginine to cysteine substitution in the DNA binding domain [13]. Characterization of these cell lines further supports th at inactivation of the normal f unctions of the tumor suppressor proteins Rb and p53 are important steps in hum an cervical carcinogenesis, either by genetic mutation or through inactivation by the HPV E6 and E7 oncoproteins. Epigenetics Both genetic and epigenetic changes are com ponents of tumorigenesis [14]. Epigenetics is the study of the heritable changes in gene expr ession that occur independent of the DNA coding sequence. Epigenetic mechanisms that modify chro matin structure can be divided into four main categories: DNA methylati on, covalent histone modi fications, non-covalent mechanisms such as nucleosome remodeling, and small, non-coding ribonucleic acids (RNAs ) [15]. Collectively, these modifications regulate chromatin accessibility and its compactness. Their interplay controls how the genome manifests itself in different cell types and deve lopmental stages. Inappropriate alterations of components of these mechanisms are commonly found in diverse disease states, including cancer. DNA Methylation In mammals, DNA methylation pr imarily occurs at the five position of the cytosine ring when directly followed by guanine (CpG). The majority of the genome is CpG poor; however, there are variable stretches of DNA that contain a high frequenc y of CpG dinucleotides within the genome and are thus designate d CpG islands [16]. CpG dinucleo tides that are not located in CpG islands are sparsely distri buted, highly methylated, and gene rally found in regions of large repetitive sequences such as retrotransposon elem ents and centromeric repeats [17]. CpG islands are more commonly found in the 5' end of the gene proximal to the transcription start site (TSS) and occupy approximately sixty percent of all hu man gene promoters [18]. The majority of CpG 16

PAGE 17

islands are generally unmethylated and found in differentiated tissues. Promoters containing CpG islands that become hypermethyl ated during development result in long-term transcriptional silencing. Classic examples of such naturally occurring epigenetic silencing mechanisms are Xchromosome inactivation a nd gene imprinting [19, 20]. In mammals, DNA methylation is catalyzed by two families of enzymes designated DNA methyltransferases (DNMTs): DNMT1 and DNMT3. The threeknown membersof the DNMT3 family are DNMT3A, 3B and 3L. DNMT3A and DNMT3B can modify unmethylated and hemimethylated CpG base pairs at the sa me rate, and are thus referred to as de novo methyltransferases. Little is known about the regulation of de novo DNA methylation or which proteins are involved. It is know n that DNMT3L, although catalytical ly inactive, interacts with DNMT3a and DNMT3b and the complexes co-localize in the nucleus [21]. DNMT3L may also participate in transcriptional repression [22]. Maintenance methyltransferases maintain the methylation patterns established by the de novo methyltransferase by methylating the nascent or hemimethylated DNA strand following replication. DNMT1 is known as the maintenance methyltranferase because it localizes to the repli cation foci and interacts with the proliferating cell nuclear antigen to methyl ate the newly-synthesized DNA strand [23]. DNMT1 and DNMT3s work together in a complex relationship to maintain existing and establish new cytosine methylation in mammals. Histone Modifications The majority of the mammalian genome is normally packaged in a transcriptionallyrepressive chromatin state. This type of chro matin is heavily methylated and forms repeating units of nucleosomes. Nucleosomes are made up of an octamer of four core histone proteins with 147 base pairs of DNA wrapped around the octamer. H3, H4, H2A and H2B are the core histone proteins within the nucleosome [24]; however, several histone variants also exist. For instance, it 17

PAGE 18

has been seen at active or poised genes that H2A and H3 are replaced by H2A.Z and H3.3 histone variants, respectively [25]. The histone amino-terminal tails are sites of frequent posttranslational covalent modifi cations including acetylation, methylation, ubiquitylation, sumoylation and phosphorylation at specific residues [26]. These modifications regulate key cellular processes such as transc ription, replication and repair. Nucleosome Positioning Histone modifications work by either changing the accessibility of chromatin or by recruiting and/or inhibiting binding by non-histone effector proteins. Families of enzymes coordinate covalent histone modi fications to precise residues. Hi stone acetyltransferases (HATs) and histone methyltransferases (HMTs) add acetyl and methyl groups, respectively, while histone deacetylases (HDACs) and histone demet hylases (HDMs) remove such marks from the histone tails. Interactions between the DNA methylation ma chinery and histone-modifying enzymes occur at multiple levels to affect gene expression. DNA methyla tion itself can affect histone methylation and acetylation through prot eins such as methyl CpG binding protein 2 (MeCP2) recruiting HMTs and HDACs [ 27, 28]. Conversely, HMTs can direct DNA methylation to specific genomic targets by dire ctly recruiting DNMTs to stably silence genes [29, 30], and by regulating the stability of DNMT-DNA interaction [31]. DNMTs can also recruit HDACs and methyl-binding proteins to achieve gene silencing and chromatin condensation [32, 33]. The histon e modification mechanism of i nheritance is still not fully understood. Whether histone marks survive replica tion has yet to be answ ered, but it is thought that chromatin states are metastable because they self perpetuate due to the network of DNMTs, HMTs, and HDACs that are recruited by the herita ble DNA methylation. Regardless, it is widely accepted that the deacetylated and methylated state of the histone tails maintain nucleosomes in a tightly-compacted and transcriptionally-silent st ate, though exceptions have been found [34]. 18

PAGE 19

Figure 1-2. Diagram of Open and Closed Chromatin2 Chromatin is shown in a closed state that is representative of the proposed 30 nm chromatin fiber (far ri ght). This tight packing of nucleosomes is refractory to transc ription because it impairs DNA binding by transcription factors. The cell is able to ch ange the state of chromatin to allow access to DNA by a combination of nucleosomemodifying enzymes, DNA binding factors and nucleosome-remodeling activ ities. Open chromatin is represented as a beads-ona-string configuration with a close up of two nucleosomes (far left). In open chromatin, variable lengths of linke r DNA are observed between nucleosomes. Only a small fraction of the genome, known as euchromatin, is transcriptionally active. Both covalent and non-covalent mechanisms, such as nucleosome remodeling, regulate gene expression by altering the accessibi lity of chromatin to transcription factors. Nucleosomes around the promoter are more widely spaced than in heterochromatin and generally contain heavily acetylated histones. It is generally thoug ht that nucleosomal DNA resists transcriptional activation, which requires the asse mbly of bulky transc ription machinery onto the TSS. It has long been hypothesized that nucle osomes need to be removed fo r active transcription [35], and there are current models depicti ng this in yeast [36, 37] and ma mmalian systems [38, 39]. In the human genome, it has been shown that nucleosomes immediately upstream of TSSs are depleted upon activation of inducible genes. The adenosine-5'-triphosphate (ATP)-dependent chromatin 2 adapted from 44. Gilbert, N. and B. Ramsahoye, The relationship between chromatin structure and transcriptional activity in mammalian genomes. Brief Funct Genomic Proteomic, 2005. 4 (2): p. 129-42. 19

PAGE 20

remodeling complex switch/sucrose non-fermentable (SWI/SNF) is thought to regulate promoter nucleosome eviction by sliding and/or removing nucleosomes [40, 41]. Indeed, nucleosome-free regions (NFRs) are found at the 5' and 3' e nds of genes [42, 43]. Packaging of DNA into chromatin by the formation of nucleosomes is v ital to gene regulation and the organization of such is only beginning to be understood. Epigenetics in Cancer Epigenetic alterations involve genome-wide losses and regiona l gains of DNA methylation as well as altered patterns of histone modifications and the enzy mes that regulate them; however, the heritability of histone modifi cations is inconclusive since their mechanism of inheritance is currently unknown [45]. These aberra nt epigenetic events have been observed in cancer, but each is currently being debated as to whether it is a foundation for or rath er the consequence of cancer. Aberrant Methylation The least understood of these ep igenetic alterations is genome-wide loss of methylated cytosine residues. Global DNA hypomethylation is most often observed within repetitive sequences dispersed throughout the genome. Rece ntly, hypomethylation has also been observed within the body of genes [46]. Aberrant gene silencing is th e more commonly studied cancerassociated epigenetic event. Aberrant gene sile ncing is the abnormal tran scriptional silencing of genes by either direct methylation of the gene promoter or modificat ions to the chromatin structure. The study of hypermethylation is gene rally focused on tumor suppressor genes (TSG). TSGs are genes that normally inhibit cell grow th. TSG proteins are found in many different cellular pathways involving cell cycle checkpoi nt responses, detection and repair of DNA damage, mitogenic signaling, differentiation and migration, and tumor angiogenesis [47]. In most mammalian genes, the CpG islands are normally free of DNA methylati on. In cancer cells, 20

PAGE 21

however, various TSG promoters have densely me thylated CpG islands resulting in a loss of gene transcription [48]. This loss of transcription can act as one or possibly as both hits needed in Knudsons two hit hypothesis [49]. Knudsons hypothesis that two hits ar e required for the full inactivation of a tumor-suppressor ge ne has been shown to be fundame ntally correct in almost all cases of human cancer, but it is now known that the hit is not always a genetic mutation. The mechanism for aberrant methylation is largely unknown. It might be caused by dysregulation of the DNMTs or other chromatin binding proteins. Despite its unknown cause, aberrant DNA methylation is an important mechanism that is exploited by cancer. Chromatin Changes in Cancer Aberrant DNA methylation is one of a series of complex epigenetic silencing mechanism that lead to alterations in gene expression through changes in chromatin structure. Experiments showing that epigenetic gene silencing occurs in the absence of detectable CpG methylation indicated that mechanisms other than DNA me thylation modulate gene silencing [50, 51]. It is now hypothesized that nucleosome remodeling works in concert with DNA methylation and histone modifications to drive gene silencing. There are many possible mechanisms by which this could take place in cancer. For example, the nucleosome remodeling and deacetylase (NuRD) complex facilitates recruitment of th e polycomb repressive complex 2 and DNMT3A to target gene promoters leading to their permanen t silencing by establishing a repressive chromatin state [52]. C omponents of the NuRD complex have been linked directly to oncogenesis [53]. In addition, it has been shown that alterations in the SWI/SNF comp lex are associated with cancer development [54]. Interactions between the DNA methylation machinery and histone-modifying enzymes further enhance the complexity of epig enetic regulation of ge ne expression. More is known about the maintenance of tumor suppressor gene (TSG) me thylation, but we are slowly starting to understand how aberrant DNA methylation may be initiated. 21

PAGE 22

Epigenetic alterations are important for the development and progression of malignant diseases. In cervical cancer, a lterations in DNA methylation are thought to be associated with integration of viral DNA. The integrated viral DNA of Hepatitis B/C virus infections of the liver and Epstein-Barr virus infecti ons in stomach cancer are known to alter the DNA methylation status of the host genome [55, 56]. HPV E7 protein has been reporte d to directly interact with DNMT1 and stimulate its enzymatic activity [57]. This may be the cause of hypermethylation of TSGs that has been seen in cervical carcinoma [58]. Reports that DNMT1 protein expression is increased even in low-grade cervical intra-epithelial neoplasias compared with normal squamous epithelium, and further increas ed in higher-grade cervical in tra-epithelial neoplasias and squamous cell carcinomas of the uterine cervix, indi cate that this may act as an initiating event in cervical carcinogenesis [59]. The regulation of DNA methyltransf erases may be altered by viral oncogenes through interactions with Rb-associated proteins [60]. Furthermore, E6, E7, and their interactions with host proteins drive the cell cycle to ignore no rmal DNA regulation checkpoints, which leads to genomic instability [61]. These virally-induced alterati ons in host gene structure contribute to the molecular pathog enesis of HPV-associated cancers. As the role of epigenetics in cancer become s clearer and the interrelationships between DNA methylation and chromatin components are increasingly understood, the possibility of epigenetic approaches to cancer prevention, detection and therapy ar ise. A number of epigenetic alterations occur during all stages of cervical carcinogenesis. Fr om a diagnostic point of view, because epigenetic abnormalities occur very early in the carcinogenic process they can potentially be exploited as molecular ma rkers for early detection. Assessment of hypermethylated genes in primary tumors could poten tially serve as a prognostic factor or as a means of predicting response to radiation, chemot herapy or transcriptional agents. The focus of 22

PAGE 23

this project is to determine if Wnt inhibitory f actor, an antagonist of the Wnt pathway, exhibits epigenetic aberrations in cervical cancer that may be exploited for future cancer therapies. Wnt Pathway The Wnt extracellular signa ling pathway, wingless in Drosophila is a highly-conserved signal transduction pa thway that mediates a wide range of biological processes. From hydra to humans, Wnt signals control multiple aspects of development, including the proliferation, fate specification, polarity and migrati on of cells [62, 63]. There are at least two distinct pathways of Wnt signaling. The first is the Wnt/ -catenin pathway named for the molecule that activates it, Wnt. The -catenin pathway of Wnt signaling is also commonly referred to as the canonical pathway. The second is the non-canonical or -catenin-independent pathway. Two decades of research have given us a firm understanding of the Wnt pathway. In addition, activating mutations in components of the Wnt signaling pa thway have been found to play a role in oncogenesis [64, 65]. Wnt Agonists and Antagonists There are various agonists and antagonists that act on the Wnt signaling pathway. To date, nineteen Wnt ligands have been found in mammals. They are cysteine-rich proteins of approximately 350-400 amino acids and contain an N-terminal signal peptide for secretion. Wnts are glycosylated and lipid modified in the e ndoplasmic reticulum, and subsequently escorted by Wntless from the Golgi to the plasma membrane for secretion. After s ecretion, mature Wnts either bind to heparin sulfate proteoglycans a nd lipoprotein particles, or form multimers with each other. Wnt antagonists can be divided in to two functional classes [66]: the secreted Frizzled-related protein (sFRP) class and the Dickkopf class. Members of the sF RP class include the sFRP family, Wnt inhibitory protein (WIF1) a nd Cerberus all of which bind directly to Wnts, thereby altering Wnts ability to bind to one of its receptor complexes. Only sFRPs are able to 23

PAGE 24

bind both Wnts and the Wnt receptor Frizzled. Th e Dickkopf (DKK) class of inhibitors binds directly to the LRP5/6 co-receptor of the Wnt receptor complex (Figure 1-3B). Wnt Receptors The Wnt receptor complex that activat es the canonical pathway contains two components: the family of seven-pass transmembrane Frizzled (Fz or Fzd) receptors and a coreceptor from the low-density lipoprotein receptor-related protein 5/6 family (LRP5 or LRP6). There are ten Fz genes in mammals which likel y have functional redundancy and have varying capabilities to activate Wnt signa ling when co-overexpressed with Wnt and LRP5/6 [67]. Either one of two single-span transmembrane proteins, LRP5 or 6, are also needed. In vitro studies support the model that Wnt induces the formation of Fz-LRP5/6 co mplex [68]; however, this has yet to be clearly shown in vivo. A particular Wnt may activ ate canonical and/or noncanonical pathways depending on the receptor complement [69], though it remains unclear whether the noncanonical Wnt pathway also requires LRP5/LRP6. Canonical Pathway The Wnt/ -catenin pathway, also referred to as th e canonical pathway, is activated when a Wnt ligand binds a member of the Fz receptor and the co-receptor LRP5 or 6 (Figure 1-3A). Formation of the Wnt-Fz-LRP complex recruits the scaffolding protein Dishevelled (Dvl), resulting in LRP phosphorylation an d activation which recruits Axi n. This leads to inhibition of Axin-mediated -catenin degradation. Wit hout constant degradation, -catenin accumulates in the cytoplasm and translocates to the nucleus. In the nucleus, -catenin complexes with the family of T cell factor/lymphoid enhancer factors (TCF/LEF). -catenin binding to TCF/LEF proteins provides a transcripti on activation domain. As a result of this complex, Wnt target gene expression is activated [65]. 24

PAGE 25

Figure 1-3. Diagram of the Canonical Wnt Pathway 3 (A) Wnts are the primary agonists and form a complex with LRP5/6 and Fz to activate signaling. In the presence of Wnt ligand, a receptor complex forms between Fz and LRP5/6. Dvl recruitment by Fz leads to LRP5/6 phosphorylation and Axin r ecruitment. This disrupts Axin-mediated phosphorylation/degradation of -catenin, allowing -catenin to accumulate in the nucleus where it serves as a coactivator for TCF to activate Wnt-responsive genes. (B) WIF1 and sFRP antagonists bind directly to secreted Wnts and/or Fz binds directly to secreted sFRP. DKK antagonist proteins bind LRP5/6 to prevent Fz-LRP6 complex formation. In the absence of Wnt, cytoplasmic -catenin forms a multiprotein complex and is phosphorylated by CK1 (blue) and subsequently by GSK3 (yellow). Phosphorylated -catenin is recognized by the ligase -Trcp, which targets -catenin for proteasomal degradati on. Wnt target genes are repressed by TCF-TLE/Groucho and HDAC. In the absence of Wnt (F igure 1-3B), cytoplasmic -catenin interacts with the adenomatous polyposis coli gene product (APC) as well as Axin scaffold proteins and is a substrate for casein kinase 1 (CK1) and glycogen synthase kinase 3 (GSK3 ). CK1 and GSK3 sequentially phosphorylate the amino terminal region of -catenin, resulting in -catenin ubiquitination by -Trcp and proteasomal degradatio n. Constant elimination of -catenin prevents it from reaching 3 Modified from Ref. 65. MacDonald, B.T., K. Tamai, and X. He, Wnt/beta-catenin signaling: components, mechanisms, and diseases. Dev Cell, 2009. 17 (1): p. 9-26. 25

PAGE 26

the nucleus; thereby, allowing TCF/LEF to stay bound to and repress Wnt target genes. TCF/LEF factors, when bound to DNA at Wnt-responsiv e genes, are able to interact with other factors, such as Groucho and histone deacet ylase, to repress transcription [70]. Noncanonical Pathway Noncanonical Wnt signaling pathways are less understood, but appear to function in a -catenin-independent manner to regulate proc esses such as converg ent extension during vertebrate gastrulation. In vertebrates, the noncanonical pathways have been termed the Wnt/Calcium and Wnt/c-Jun N-terminal kinases (J NK) pathways. In brief, activation of the Wnt/Calcium pathway involves Wnt binding to a Frizzled receptor leading to th e release of intracellular calcium and the activ ation of enzymes such as Ca2+/Calmodulin-Dependent Protein Kinase II and Protein Kinase C. The Wnt/JNK pa thways appear to similarly use Frizzled receptors, Dishevelled, JNK and Rho family guanos ine triphosphatases (GTPases). Recent data have also implicated components of noncanoni cal Wnt pathway in diverse forms of human cancer [71]. There are numerous components of the Wnt ne twork, from the plasma membrane to the cell nucleus. While an understa nding of the overall organization of the network is beginning to emerge, much is still unknown. Wnt signals ultimately promote cell proliferation and tissue expansion. Numerous Tcf target genes have been identified in diverse biological systems and include genes involved in cancer, su ch as cMyc and cyclin D1. The Wnt signaling pathway is frequently altered in cancer. While no direct mutations of Wnt have been found in cancer, several of the downstream molecules in the Wnt signaling pathway (i.e., -catenin, APC, Axin) are dysregulated in a variety of human tumors. Probably the most characterized of them are inactivating, germline-mutations in APC found in colorectal cancer. Activation of Wnt signaling is believed to play a role in the progression and pathogenesis 26

PAGE 27

of cervical cancer. It has been shown that -catenin is increased within the cytoplasm and nucleus in invasive cervical carcinomas. In addition, in vivo activation of the Wnt pathway is sufficient to induce transformation of HPV-immortalized human keratinocytes [72]. We focused specifically on Wnt inhibitory protein (WIF1), an antagonist of the Wnt pathway. WIF1 has a unique N-terminal signal sequen ce, WIF domain (WD), and five epidermal growth factor (EGF)-like repeats whic h mediate Wnt binding to WIF1 [73]. WIF1 mainly functions as a secreted inhibito r of Wnt signaling by binding to Wn t proteins and thus competing with the binding of Wnt protei ns to the frizzled receptor. WIF1 expression is downregulated in prostate, breast, lung and bladde r tumors [74], suggesting that WIF1 is a potential TSG. Loss of WIF1 expression in tumors may lead to unrestr icted binding of Wnt ligands to the frizzled receptor, followed by enhanced transcription of target genes in the Wnt pathway. Several groups have correlated decreased WIF1 expression with aberrant prom oter hypermethylation. While the WIF1 gene is a target for epigenetic silenci ng in some tumor types and its expression is downregulated in several cancers, it is currently unknown if this gene is su bject to epigenetic silencing in cervical cancer. To determine if WIF1 is epigenetically silenc ed in cervical cancer, we have analyzed the expression and DNA methylation status of the WIF1 promoter in a panel of cultured human cervical tumor cell lines and surgically-obtained human cervical tumors. Our results demonstrate that WIF1 hypermethylation is a frequent event in cervical cancer and lend strength to the idea that dysre gulation of Wnt signaling is an important contributor of human cervical tumorigenesis. 27

PAGE 28

CHAPTER 2 MATERIALS AND METHODS Cervical Specimen All human tissue samples were obtained from patients under protocols approved by Institutional Review Boards at the Moffitt Cancer Center (Tampa, FL). Tumor tissue (n = 22) and adjacent matched normal stratified epith elium (n = 22) were obtained from archived formalin-fixed paraffin-embedded (FFPE) blocks from the tumor repository. The archived tissue came from women diagnosed with squamous cell cervical cancer and having a surgical procedure between 1993 and 1999 at the Moffitt Can cer Center. Cases were restricted to those that did not receive radiation treatment before surgery. Histol ogical diagnosis, tumor stage and grade were reviewed by a pathologist at Moff itt Cancer Center according to the WHO and the international union against cancer (U ICC) classification of tumors. The Pap samples (n = 8) are scrapings of the cervix obtained from patie nts from four sites: the local Health Department, Tampa General Ho spital, Lifetime Cancer Screening and the Moffitt Cancer Center. Routine gynecological ex aminations of healthy, disease-free women were conducted. Exfoliated cells were placed in 20 mL of PreserveCyt solution, mixed and separated into 4 ml aliquots and stored at 80oC. One milliliter of the solution was transferred to a new tube with a pipette and pelleted. To resuspend the pellet, 200 l of phosphate-buffered saline (PBS) was added, and DNA was extracted usi ng the QIAamp DNA extraction kit (Qiagen, Valencia, CA). Sample #3 is the only one that is HPV negative. The other seven samples were found to have one or more HPV types: 1, 16, 51, 58, 59 and 62. Cell Culture and 5-azadC Treatment The cervical tumor cell lines C-33A, CaSki, HeLa 229 and SiHa were purchased from the American Type Culture Collection and cu ltured according to their specifications. In short, HeLa, 28

PAGE 29

SiHa and C-33 lines were grown in Eagle' s Minimal Essential medium (Mediatech, Manassas, VA) with 10% fetal bovine serum (FBS) ( Invitrogen Life Technologies, Grand Island, NY ) and penicillin-streptomycin (Mediatech). The CaSki line was grown in Roswell Park Memorial Institute 1640 medium (Meditech) with 10% FBS. All cel l lines were cultured and maintained in monolayers at 37C in a humidi fied incubator under 95% O2 and 5% CO2. For the drug treatment, cells were seeded, allowed to recover overnight, then 5-azadC (Sigma-Aldrich, St. Louis, MO) was added to complete growth media at a final concentration of 5 M. Fresh drug was added every 24 hr for 5 d. After treatment, the cells were washed with PBS and allowed to recover in their respective medi um for 2 d prio r to collection. RNA and DNA Isolation One 10 cm plate of confluent cells cultured with and without 5-azad C was harvested for RNA and DNA extraction with 1 mL of TRizol (Invitrogen) reagent for 5 min at room temperature. After phenol/chloroform extrac tion and phase separati on, RNA was digested according to the manufacturers instructions fo r on-column DNase I digestion with RNase-free DNase I (Qiagen) and RNAeasy Kit (Qia gen). The DNA was isolated according to manufacturers instructions from the organic phase of the RNA isolation. The 260/280 nm absorbance ratio and sample concentrations of the RNA and DNA was determined using the NanoDrop ND-8000 (Thermo Scientific, Wilmington, DE). All genomic DNA isolated from patient sa mples was conducted at the Moffitt Cancer Center by either centrifugation of 1-4 ml of Thin Prep solutions into a pellet that was resuspended in PBS (Pap samples) or by sectioning tissues from paraffin blocks into small pieces and deparaffinized by xylene extraction. The genomic DNA was then extracted using QIAmp mini DNA Kit (Qiagen) according to manufacturers protocol. The concentr ation of the extracted DNA was then determined by the NanoDrop1000 Sp ectrophotometer (Thermo Scientific). 29

PAGE 30

Sodium Bisulfite Modification At the Moffitt Cancer Center, sodium bisulfite conversion of 1 g of patient tissuederived DNA was carried out with the EZ DNA Methylation-Direct kit (Zymo Research Corporation, Orange, CA), following the manufacturer's protocol. Bisulfite-modified DNA was resuspended in 30 l of water and stored at 80C in six 5 l aliquo ts to avoid several freeze/thaws. Cell line DNA was bisulfite treated using a specific in-laboratory protocol that consistently yields a 99.8% efficiency of cytosine conversion [75]. The protocol is: the day before bisulfite treatment, distilled water was degassed (dg.dH2O) by boiling ~200 mL of distilled H2O (dH2O) for 20 min in a glass beaker, and carefully poured in to a 125 mL bottle until it was completely full (above the lip). The cap was screwed on the bottle tightly and the dg.dH2O was cooled overnight on the bench top. I mmediately prior to bisulfite treatment, 3 N NaOH and 100 mM hydroquinone were prepared by dissolving each chemical in dg.dH2O. Total genomic DNA (1-2 g) in 20 l 0.1X TE was pipette d into a 0.65 ml micro-centrifuge tube (that fits in a thermocycler) containi ng 10 l of denaturation buffer ( 6.5 L dg.dH2O, 3.0 L 3N NaOH, and 0.5 L 0.5 M EDTA, pH 8.0) Saturated sodium metabisu lfite solution was made as follows: to a 20 ml-capacity glass scintillation vial containing a small stir bar, 0.1 ml of 100 mM hydroquinone, 5 g sodium metabisulfite (Sigma, previously aliquoted into air-tight vials), and 7 ml dg.dH2O were added, and the solution was immedi ately stirred. Then, 1.0 ml of 3 N NaOH was quickly added. The solution was adjusted to pH 5.0 at room temperature with additional 3 N NaOH and preheated to 50C in a water bath. A saturated solution was employed to promote sulfonation of cytosine, the firs t step of bisulfite deaminati on. Samples of genomic DNA in denaturation solution were heated at 98C for 5 mi n with the tube caps open in a thermocycler. With one minute remaining in the denaturation st ep, the preheated sodium metabisulfite solution 30

PAGE 31

was transferred to a stir-plate and stirred gen tly. After 5 min denaturation, 0.2 ml of sodium metabisulfite was rapidly added to each sample while in the thermocycler. Samples were capped, vortexed and incubated for 6 hr in a covered water bath at 50C. Following deamination, the DNA was desulfonated and cleaned using the EZ Bisulfite DNA Clean-up kit ( Zymo Research Corporation ) according to manufacturer protocol and eluted in 10 l of M-Elution Buffer. Polymerase Chain Reaction (PCR) and Pyrosequencing Following bisulfite modification, a 20 l PCR with a final concentration of 1 Coral Buffer, 250 nM deoxyribonucleot ide triphosphates (dNTPs), 250 nM forward and reverse primer, 1 U of HotStarTaq Plus (Qiagen) and 2 l of bisulfite-treated DNA was incubated in the thermocycler as follows: 95C for 5 min, once; 94C for 30 sec, 59C for 30 sec, 72C for 30 sec, repeated 45 cycles. The specif ic primers used were (Forward, F) 5'-GTA GGT TTT TTG GTA TTT AGG T -3' and (Rever se, R) 5'-CAT ACT ACT CAA AAC CTC CT-3'. A second nested PCR was preformed using 2 l of the orig inal PCR product in the same reaction mixture. Amplification was checked by agar ose gel electrophoresis. A series of controls were used to validate the assay: no template, no sequenci ng primer, sequencing primer only, sequencing primer with reverse primer only, beads only and a blank well. Also, in vitro methylated bisulfitetreated DNA (Zymo Research Corporation ) was used as a methylation positive control for all pyrosequencing runs. All samples and controls fo r each reaction were then analyzed using the PyroMark MD system (Biotage, Charlotte, NC). Methylation density wa s quantified using the PyroMark Pyro Q-CpG 1.0.9 software. Reverse Transcriptase PCR (RT-PCR) Total RNA was used in first-strand comp lementary DNA (cDNA) synthesis reactions using GoScript Reverse Tr anscriptase System and random hexamer primers (Promega, Madison, WI). WIF1 transcript levels were subsequently analyzed by PCR. The cDNA WIF1-specific 31

PAGE 32

primers used were (F) 5'-CCG AAA TGG AGG CTT TTG TA -3' and (R) 5'-TGG TTG AGC AGT TTG CTT TG-3'. They were designed to span a gap between exons to ensure amplification of cDNA and not genomic DNA. -actin served as a control for RNA integrity and was amplified using primers (F) 5'-CCC TGG CAC CCA GCA C-3' and (R) 5'-GCC GAT CCA CAC GGA GTA C -3'. PCR thermocycler conditions for WIF1 were 95C 5 min, once; 94C for 30 sec, 59C for 30 sec, 72C for 30 sec, repeated 45 cycles. PCR conditions for -actin mRNA were the same as for WIF1 transcript excep t that an annealing temperature of 60C was used and PCR was conducted for 20 cycles. A PCR reaction substituting dH2O for cDNA was conducted as a negative control. Methyltransferase Accessibility Protocol for Individual Templates (MAPit) Nuclei isolation and DNMT probing of nuclei from CaSki, HeLa and SiHa cells were isolated from ~4-7 x 106 cells at 4C under buffer conditions which preserve the integrity of nuclei and chromatin structure. After harvesting, cells were washed twice with PBS and resuspended in cell resuspension buffer (CRB: 20 mM HEPES, pH 7.5, 70 mM NaCl, 0.25 mM EDTA, 0.5 mM EGTA, 0.5% glyc erol (v/v), 10 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride). After pelleting by centrifugation at 1000 g, cells were resuspende d in cell lysis buffer (CLB: 1 CRB plus 0.19% (v/v) NP-40) for ten minutes on ice. Nuclei were then washed twice with CRB, and a 1-2L aliquot was stained with 4% (w/v) trypan blue and visualized by light microscopy to confirm their integrity. Nuclei we re finally resuspended in methylation buffer (MB; 1 CRB plus 160 M S-adenosylmethionine). After pre-wa rming nuclei to 37C for 5 min, 60 U and 100 U of M.CviPI were added for 15 min at 37C. Methylation reactions were stopped by adding two volumes of methylation stop buf fer (MSB; 100 mM NaCl, 10 mM EDTA, pH 8.0, 1% SDS (w/v)), and then incubated overnight with 100 g/ml proteina se K at 50C. DNA was 32

PAGE 33

isolated by extraction with phenol:chloroform:is oamyl alcohol (25:24:1) and concentrated by ethanol precipitation. After isolation, DNA was treated with sodium bisulfite as described above. PCR primers designed to amplify a 712 basepair segmen t of the promoter were designed. Primer sequences were (F) 5'ATT ATT ATT ATT ATT ATT AGY ATT TAG TT-3' and (R) 5'CAA RCA CAA AAA AAT RCT CCA AA -3'. Thr ee 20 l mixtures with a final concentration of 1 Coral Buffer, 250 nM dNTPs, 250 nM forward and reverse primers, 275 nM MgCl2, 1 U of HotStarTaq Plus (Qiagen) and 4 l of bisu lfite-treated DNA were incubated in the thermocycler as follows: 5 min at 95C, once; 95C for 30 sec, 53C for 30 sec, 72C for 2 min, repeated for 40 cycles. Following PCR, products were digested with restri ction enzymes R.HindIII and R.XhoI. After digestion the products were pooled and resolved by an agarose gel electro phoresis. Subsequently, the bands were excised and gel purified usi ng the QiaexII Gel Extraction kit (Qiagen). The digested, purified PCR products were then li gated at 16C overnight into a R.HindIIIand R.XhoI-digested BlueScript SK minus plasmid (S tratagene, Santa Clara, CA) with T4 ligase (New England Biolabs, Ipswich, MA). Recombin ants were identified by colony PCR with Apex Taq (Genesee Scientific, San Diego, CA) in the sa me reaction mixture as above. The inserts were sequenced at the University of Florida Interd isciplinary Center for Biotechnology Research DNA sequencing core laboratory using a vector-encoded primer (M13-Rev). Immunohistochemistry Two tissue microarrays were assembled at Moffitt with 4 m sections of FFPE tissue sectioned from each cervical tumor with matching norma l adjacent tissue. After sections were deparaffinized by xylene and rehydrated through graded alcohols, the samples were blocked using the Avidin/Biotin Blocking Kit (Vector Laboratories, Burlingame, CA) following the 33

PAGE 34

manufacturers recommended protocol. Antigen retrieval by steam heat for 20 min with 10 mM sodium citrate, pH 6.0 was conducted prior to overnight incubation wi th the rabbit anti-Cterminus primary antibody diluted 1:50 (Abcam, Cambridge, MA ). Samples were then washed and incubated with biotinylat ed-goat anti-rabbit antibody from the Rabbit IgG ABC kit (Vector Laboratories). Detection of the antigen was carried out by avidin-biotin complex following standard procedures. The tissue microarray was developed with peroxidase and 3,3'-diaminobenzidine tetrahydrochloride (reddish-b rown reaction product). Sections were then counterstained with haematoxylin to stain the nucl ei purple and mounted in an aqueous mounting medium (Fisher). A control section was performe d where the primary antibodies was replaced by irrelevant rabbit monoclonal antibody of the same isotype and at the same concentration as the specific primary antibody. Analysis was performed under 100 magnification on a Leica DM6000B microscope (Leica Microsystems, Bannockbur n, IL), and images were taken with a Retiga SRV Fast 1394 digital camera (QImaging, Surrey, British Columbia, Canada). 34

PAGE 35

CHAPTER 3 RESULTS WIF1 is Epigenetically Silenced in Cervical Cancer Cell Lines To determine if the WIF1 gene is targeted for epigenetic silencing in cervical cancer, we cultured a panel of four cervical cancer lines (C-33A, CaSki, HeLa and SiHa) in the presence or absence of the DNA demethylating drug 5-azadC. Using RT-PCR we observed in the untreated cells that C-33A, CaSki, and HeLa cell lines expressed varying amounts of WIF1 transcript while SiHa did not. However, when cultured in the presence of 5 mM 5-azadC, the CaSki, HeLa, and SiHa lines displayed elevated levels of WIF1 transcript (Figur e 3-1). These data suggest that WIF1 expression may be epigenetic ally repressed by DNA methylati on in CaSki, HeLa and SiHa cells. The varied reactions of the cell lines to 5-azadC were of note. The SiHa cell line continued to divide during 5-azadC treatment, while the HeLa cell remained quiescent and quickly divided during the two-day recovery period in the ab sence of drug. Similarly, CaSki cell numbers remained constant during 5-azadC treatment, but slowly divided afterward. In contrast some C-33A cells died during 5-azadC treatment, and there was little cell division during the two-day recovery period. The difference in response to the drug treatment a ppeared to correlate with the presence of the HPV genome and inversely correlat e with the methylation status at this locus, which may or may not be of significance. To determine if the difference in WIF1 expression levels was correlated with promoter DNA hypermethylation, we analyzed the methylation status of WIF1 by pyrosequencing. Pyrosequencing is essentially sequencing by synt hesis to quantitatively analyze cytosine 35

PAGE 36

Figure 3-1. RT-PCR of Cervical Cell Lines Before and After 5-azadC Treatment. Cultures of the cervical tumor cell lines C-33A, CaSki, HeLa and SiHa were untreated () or treated with 5 M 5-azadC (+) for 5 d (fresh me dium and drug were added every 24 hr). Following this treatment, total RNA and genomic DNA were prepared from the cells. RT-PCR analysis of WIF1 expression (top panel) a nd the housekeeping gene -actin as a control for RNA integrity (bottom panel). methylation at individual CpG di nucleotides after PCR amplification of a bisulfite-converted sequence. Briefly, bisulfite conversion of unmethyl ated cytosines in CpG di nucleotides results in conversion of C to U while methylated cytosines remain as C. Pyrosequencing requires PCR of a segment of the gene of interest from bisulf ite-converted genomic DNA using primers that amplify in a methylation-independent manner by avoiding areas with CpG present. In addition, one primer is designed with a 5' biotin tag. PCR amplicons are subsequently purified using streptavidin-coated beads, and then denatured to single-stranded DNA. A sequencing primer is annealed to the captured strand and added to a 96 well reactio n plate containing DNA polymerase and dNTPs. As nucleotides are incorp orated onto the 3' e nd of the oligonucleotide primer, pyrophosphate (PPi) is given off as a byp roduct. The reaction mix also contains ATP sulfurylase which quantitatively converts PPi to ATP in the presence of adenosine 5 phosphosulfate. This ATP drives a luciferase-med iated conversion of luciferin to oxyluciferin that generates visible light in amounts that are proportional to the ATP concentration. The light 36

PAGE 37

produced in the luciferase-catalyzed reaction is detected by a charge coupled device (CCD) camera and is recorded by the pyrosequencer. Each light signal is proporti onal to the number of nucleotides incorporated, and t hus DNA methylation is quantified at each CpG dinucleotide. Figure 3-2. WIF1 Promoter and Pyrograms of Cervical Ce ll Lines. (A) Schematic representation of the WIF1 promoter. Each vertical line repr esents a CpG dinucleotide in the sequence. Positioned below the diagram are the primers used for pyrosequencing and MAPit, respectively. (B) Da ta output from the pyrosequencer, termed a pyrogram. Shown in the pyrograms are the proportions of nucleotides incorporated along with the sequence of nucleotides dispensed show n on the x-axis. The percent methylation at each CpG dinucleotide (gray rectangles ) is shown above the CpG site. Average CpG methylation (Avg. Meth.) within the segm ent of the gene analyzed is displayed in the box at the top of the pyrogram. Ther e are several quality controls where a nucleotide that should not be incorporat ed is dispensed (site 6,14, 28 and 29). A bisulfite conversion control dispenses a C directly after a converted T to check the bisulfite conversion efficiency (tan rectangle). 37

PAGE 38

We designed pyrosequencing primers to amplify a 164 basepair region of the WIF1 promoter. This assay quantitatively measures th e first seven CpGs following the transcriptional start site using the forward PCR primer as the sequencing primer (Figure 3-2A). When pyrosequencing was conducted on bisulfite-modified genomic DNA harvested from the four cervical cancer cell lines, we observe d that the cell lines that express WIF1 in the absence of 5-azadC (C-33A and CaSki), exhibited very low levels of DNA methylation (<3%). Conversely, cell lines which expressed little to no WIF1 in the absence of 5-azadC, HeLa and SiHa cells, exhibited high and moderate levels of methylat ion, respectively (21% an d 45%) (Figure 3-2B). Genomic DNA harvested from HeLa and SiHa ce lls cultured with 5-az adC displayed a 10-15% decrease in average methylation, while C-33A and CaSki showed little change (Figure 3-3). These data demonstrate that the expression of WIF1 is correlated to the methylation status of its promoter in cervical cell lines. Figure 3-3. Average Methylation of the WIF1 Promoter Before and After 5-azadC Treatment. Cultures of the cervical tumor cell lines C-33A, CaSki, HeLa and SiHa were untreated (white bars) or treated with 5 M 5-azadC (black bars). Shown are the averages and standard deviation for the six CpGs assayed by pyrosequencing as outlined in Materials and Methods. 38

PAGE 39

Aberrant Chromatin Footprint in CaSki and SiHa Cell Lines To determine if promoter me thylation might be repressing WIF1 expression by altering chromatin accessibility, we preformed a methyltran sferase accessibility protocol for individual templates (MAPit) on the CaSki and SiHa cell line s. This technique probes chromatin structure with minimal disruption to nuclei in tegrity to capture an accurate footprint of chromatin structure and nucleosome dynamics in vivo Nuclei were extracted and probed with M.CviPI, a DNA methyltransferase (DNMT) that only methylates GC dinucleotides. Accessibility of DNMTs to GC sites is impaired based on the strength and sp an of the protein-DNA in teraction to be probed. Methylated cytosines that resist chemical deam ination are detected pos itively by the PCR-based technique of bisulfite genomi c sequencing. This unique capabili ty to score methylation at multiple sites in single molecules permits detection of inherent structural variability in chromatin structure and non-histone proteinDNA interactions. MAPit analysis was conducted on the active and repressed WIF1 promoter of CaSki and SiHa lin es, respectively (Figure 3-4). In the CaSki cell line, which expressed WIF1 two populations of cells were evident: one population with two distinct area s of methylation and one populati on with nearly no methylation (Figure 3-4A). After probing with M.CviPI, there were three dis tinct regions of open chromatin (Figure 3-4B). However, there was no difference in accessibility between the methylated and unmethylated populations. It is of note that sequence inspection revealed two consensus Sp1 binding sites (5'-GGCGGG-3') within the two regions devoid of endogenous methylation and inaccessible to the methyltransfer ase. This is consistent with other reports showing that Sp1 binding to DNA may be able to protect sequences from DNA methylation [76]. Near the upstream SP1 site, it has also been shown that NF-k and MYC-associated zinc finger (MAZ) transcription factors bind [77]. In the SiHa cell line in which WIF1 was highly methylated and not expressed, high levels of endogenous methyla tion were observed thro ughout the body of the 39

PAGE 40

promoter and accessibility to the methyltransf erase seemed to vary more, thus suggesting dynamic chromatin reorganization (Figure 3-4C). Figure 3-4. Chromatin Probing of Ca Ski and SiHa Cell Line. Nuclei from both cells lines were isolated and chromatin structure was probed as stated in the Materials and Methods. Endogenously methylated CpG (CG) sites are depicted with black circles and unmethylated CpG with white circles. A ll GC sites are denoted by triangles, and those that were methylated by M.CviPI probe are red. Regions that were not protected against methylation are indi cated by rectangles spanning at least two contiguous GC sites (i.e., one methylated site does c onstitute accessibility). Methylated GCG (Gm5CG) sites are indicated in gray a nd cannot be discerned from endogenous methylation or M.CviPI modified sites. The bisulfite conversion efficiency (C.E.) for each molecule is shown at the right. Transc ription factor binding sites for Sp1 (blue) and both NFB and MYC-associated zinc fi nger (MAZ) (tan) are indicated. 40

PAGE 41

WIF1 Promoter is Aberrantly Methylated in Squamous Cell Cervical Tumors To determine if promoter methylation of the WIF1 gene occurs in primary cervical tumors, we used pyrosequencing to analyze the methylation of WIF1 in patient-derived cervical squamous cell carcinoma samples. Matched adj acent normal cervical stratified epithelium and Pap smear samples from different patients were also analyzed. Cervical tumors and normal tissues were obtained from the Moffitt Cancer Center Tissue Bank. DNA was extracted, treated with sodium bisulfite an d PCR amplified with the WIF1 pyrosequencing primers. PCR amplicons where then analyzed with the PyroMark MD and Pyro Q-CpG software. All cases samples were re-analyzed with repeated PCR and pyrosequencing reactions to validate the initial pyrosequencing findings. Subsets of bisulfite-treated samples were also analyzed multiple times by bisulfite genomic sequencing, with analogous results (data not shown). To ascertain if aberra nt methylation of the WIF1 promoter occurs in cervical cancer, we pyrosequenced 22 high-grade squamous cell carci nomas. Of these 22 tumors, only 17 produced PCR amplicons of sufficient quality for pyroseq uencing. We grouped these specimens into three categories based on their average methylation of six CpG sites as followed: little or no methylation (0-9%), moderate methylation (1019%) and heavily methylat ed (>20%). Of the 17 tumor samples, 6 were not methylated, 3 ha d moderate methylation, and 8 were heavily methylated (Figure 3-5). We also analyzed a set of 8 Pap smear samp les that had been previously genotyped and 7 of which were found to be HPV positive. We fou nd no significant methylation, <5%, in all 8 of the Pap smear samples (Figure 3-6). In a ddition we analyzed the paired adjacent normal tissue from each tumor sample. Unfortuna tely, several of theses samples failed to 41

PAGE 42

Figure 3-5. Average CpG methylation of Cerv ical SSC and Adjacent Normal Tissue at WIF1. The tumors and adjacent matching normal ti ssues were given arbitrary numbers. Both the PCR and the pyrosequencing were run in duplicate, and the average methylation status of the WIF1 promoter is shown along with the standard deviation for each sample. Black bars represent the methylati on of the tumor samples and the white bars show the methylation of the paired normal stratified epithelium. Figure 3-6. Pyrograms of Pap Smear Samples at WIF1. The percent methylation at each CpG dinucleotide (gray rectan gles) is shown above the CpG site. Average CpG methylation (Avg. Meth.) within the segment of the gene analyzed is displayed in the box at the top of the pyrograms. Seven additional pap smear samples were analyzed with similar results (data not shown). yield usable amounts of DNA, and we were not able to assess WIF1 promoter methylation. However, the 9 matched normal samples that were anal yzed all had little or no methylation (<5%) (Figure 3-5). These data show that this region of the WIF1 promoter is essentially 42

PAGE 43

methylation-free in normal cervical tissue and su pports the notion that the DNA methylation that we observed in the tumor samples resulted from the disease and was not part of the normal biology of cervical tissue. WIF1 Methylation is not Correlated with WIF1 Protein Levels To assess if the methylation status of the WIF1 promoter correlated with the expression of WIF1 protein, we conducte d immunohistochemical staining on a tissue microarray including 22 tumors and their adjacent normal tissue. Immunohistochemistry (IHC) is a method for demonstrating the presence and location of proteins in the context of intact tissue. IHC staining is accomplished with antibodies that recognize the ta rget protein. The antibody-antigen interaction is visualized using chromogenic detection, in which an en zyme conjugated to the antibody cleaves a substrate to produce a colored precipitat e, in this case brownish-red, at the location of the protein. Immunostaining of normal tissue with anti-WI F1 antibody indicated that the protein was expressed in the stratified squamous epithelium of the ectocervix. No staining of the underlying cervical parenchyma (Figure 3-7 A1) was observed, except within the endometrial cell layer of blood vessels, which stained intensely with anti-WIF1 antibody. Close examination of the staining indicated a gradient of WIF1 expres sion within epithelial tissue (Figure 3-7A). Specifically, nearly no staining was observed in the basal epithelium or underlying basement membrane (Figure 3-7A, zones 1 and 2). However, strong staining was evident within the peribasal epithelial layer (Fi gure 3-7A, zone 3). We observed notable extracellular staining within the layers of differentiate d non-keratinized epithelium with di minished staining in cells at the apical surface of the ectoc ervix (Figure 3-7A, zone 5). A negative control for background staining was incubated with only rabbit IgG a nd demonstrated that there was no background staining of the WIF1 antibody within the cervical epithelium (Figure 3-7 B). 43

PAGE 44

Figure 3-7. WIF1 Immunohistochemi stry of Normal Cervical Ti ssue and SSC. Normal cervical epithelium incubated with (A) anti-WIF 1 rabbit antibody or (B) rabbit IgG. No staining from rabbit IgG and the location of WI F1 staining in the cervical epithelium was observed. (C) Tumor section with in tense staining and lo w DNA methylation. (D) Tumor section with mild staining a nd low methylation. (E) Tumor section with light staining and high me thylation. (F) Heterogeneous tumor with a mixed population of stained and unstained cells. Bracketed at the le ft are the basal epithelium (or underlying basement membra ne, zones 1 and 2), peribasal epithelial layer (zone 3), epithelial layer (zone 4) a nd apical surface of the ectocervix (zone 5). When IHC was conducted on samples of invasive squamous cell carcinoma (SCC) with an anti-WIF1 rabbit antibody, we obser ved intense staining (Figure 3-7C ) in tumor samples that had little methylation, such as sample 20 for wh ich pyrosequencing showed an average CpG methylation of 2%. However, ther e was a notable lack of staining within two samples, and this was independent of the me thylation status of the WIF1 gene promoter. For example, little staining was seen in tumor sample 13 (Figure 37D) that possessed a mean methylation value of 2%. Moreover, the majority of methylated tumors did not express or had mixed expression of WIF1 For example, tumor sample 32, which possessed a high level of WIF1 methylation (41%), also showed no clear indication of expression of WIF1 protein (Figure 3-7E). We also observed a number of tumors with heterogeneous populatio n of cells that possessed mixed staining for WIF1 within the same lesion (Figure 3-7F), which is a common feature of solid tumors. Of note, 44

PAGE 45

we observed more intense WIF1 staining in le ss de-differentiated SCC lesions. These studies show that low or absent WIF1 expression is commonly observed in invasive SCC of the cervix, but that this tumor feature may occur independent of WIF1 promoter methylation status. These findings also suggest that mechanism(s) othe r then promoter hypermethylation may serve to inhibit WIF1 expression. We conclude th at hypermethylation of the WIF1 gene is a common event in cervical squamous cel l carcinoma but that promoter methylation is not the only necessary cause of its silencing. 45

PAGE 46

CHAPTER 4 DISCUSSION The rationale for examining DNA methylation at the WIF1 promoter in cervical cancer was based on three primary factors. First, exploitation of the Wnt pathway is a common mechanism of tumorigenesis. Uncontrolled -catenin signaling, either from disruption of function of its regulatory proteins (i.e., Axin, APC, WIF1) or genetic mutations that prevent catenin degradation, leads to excessive proliferation that pred isposes cells to tumorigenesis [65]. Second, WIF1 expression has been shown to be downregulated in a variety of cancer types, such as bladder, breast, l ung and prostate [74]. WIF1 downregulation is an ep igenetic event correlated to promoter hypermethylation in many of thes e cancers. Third, while some studies have examined the epigenetic regulation of WIF1 and other genes encoding molecular components of the Wnt signaling pathway, no research has examined WIF1 silencing in cerv ical carcinomas. Using both cervical cancer cell lines and primary formalin-fixed/paraffin-embedded squamous cell carcinomas samples, we established that the WIF1 gene is commonly subjected to DNA hypermethylation in cervical tumors. In vitro studies in cervical ce ll lines indicated that WIF1 expression correlated with promoter methyl ation. Furthermore, we showed that the DNA demethylating agent 5-azadC increased WIF1 expression. In an effort to show that WIF1 expression is epigenetically silenced in primary tissue, we conducted pyrosequencing and immunostaining on cervical tumor samples. We found that the WIF1 promoter is unmethylated and WIF1 protein is expressed in the peribasal and the less differentiated epithelial layers of normal cervical epithelium. In contrast, within high-grade sq uamous cell carcinomas, WIF1 protein expression was often low to nonexistent. This loss of WIF1 protein expression correlated with promoter hypermethylation in several of the tumors; however, loss of WIF1 protein expression in two of the tumors could not be attributed to WIF1 promoter hypermethylation This 46

PAGE 47

suggests that WIF1 is epigenetically silenced in cerv ical cancer, but other mechanisms in addition to promoter hyperm ethylation may be involved. Lack of complete concordance between WIF1 promoter hypermethylation and lack of protein expression could be due to several factors. First, pyrose quencing analyses revealed the six CpG sites in the WIF1 promoter in CaSki cells were unmet hylated; however, transcript levels increased after treatment with 5-azaC as assa yed by RT-PCR. These findings suggest that the WIF1 gene is epigenetically silenced in these ce lls, but by methylation of CpG sites other then the six downstream of the TSS we assayed by pyrosequencing. This is supported by MAPit methylation footprinting experiments, which showed two areas of endogenous methylation further upstream of the TSS in CaSki cells. PCR using the MAPit primers was attempted on several of the tumor samples wit hout success. This failure is li kely because FFPEand bisulfitetreated DNA is very fragmented and thus unabl e to yield large PCR products. One method to examine if the tumor samples had similar patches of methylation to the CaSki cell line further upstream of the TSS would be to create primers specific for the areas of interest that yield smaller PCR products. Pyrosequencing of small bisulfite-converted amplicons at the two upstream methylated regions is needed to test the extent to which hypermethylation correlates with WIF1 downregulation in the primary tumor samples. Me thylation of either of these upstream regions correlating with the loss of WIF1 expression as determined by IHC would lend more support to the hypothesis that the WIF1 gene is directly ep igenetically silenced in cervical cancer. HPV may also play a role in th e epigenetic silencing of the WIF1 promoter in cervical cancer. This is based on the observation that the three cell lines (CaSki, SiHa and HeLa) that displayed varying levels of WIF1 epigenetic silencing were al l derived from HPV-infected 47

PAGE 48

specimens. By contrast, the cell line derived from a non-HPV-infected specimen did not appear to be epigenetically si lenced. HPV viral protei ns can directly effect DNA methylation by binding to and recruiting DNMTs [57]. Knockdown of expression of the viral E6 and E7 genes may help to understand if the vi rus plays a role in WIF1 hypermethylation. Furthermore, HPV genotyping of the patient cervical tumors should be done to look for a correlation between WIF1 expression and the presence of HPV. MAPit footprinting of the CaSki and SiHa cell lines may yield insight into how WIF1 silencing might occur regardless of promoter methylation. In the WIF1 -expressing CaSki cells, we observed a region of protec tion against methylation by M.CviPI just upstream of the transcriptional start site. We hypothesized that this area woul d be nucleosome free and hence accessible to the DNMT, as has been observed at other transcriptionally-active promoters. The size of the inaccessible region is consistent with a nucleosome (~147 base pairs) that occupies a variety of positions. The variable location of the protected region on each molecule seems inconsistent with protection of the region agai nst methylation by a site -specific transcription factor bound to DNA, such as SP1. It is possible that SP1 or another tr anscription factor is forming a ternary complex with a nucleo some that occupies several positions. Interestingly, two different populations of molecules with distinct endogenous DNA methylation patterns were observed in the CaSk i cell line. In one gr oup, nearly no endogenous methylation was observed, whereas the other group had two distinct areas of endogenous methylation straddling the upstream SP1 site. Despite these two patterns of methylation by endogenous enzymes, there appears to be little di fference in accessibility to M.CviPI between them. This suggests that whatever is bound to the DNMT accessible area binds regardless of the methylation status of the gene. It should be noted, however, that we have only examined a 48

PAGE 49

limited number of molecules from one experiment and there exists the possibility the molecules shown were selectively sequenced due to PCR or cloning bias. Similarly, the heavily methylated WIF1 promoter in SiHa cells showed clear regi ons of M.CviPI accessibility but with greater variability (Figure 7C). Often, areas of heavily methylated DNA are inaccessible to the methyltransferase and unable to provide clea r footprints. This sugge sts that, whereas the WIF1 promoter is heavily methylated in the SiHa cell line, it is not in a fully compacted heterochromatin state. Further experiments are needed to clarify wh at is binding to the area upstream of the WIF1 TSS. Knockdown experiments of th e various transcription factors, particularly SP1, in the cell line followed by MAPit might help to clarify what is binding to the DNA. It would be of interest to see if knockdown of SP1 alters WIF1 endogenous methylation or chromatin accessibility, particularly in the highly methyl ated SiHa cell line. However, pr eferential association of NFB or MYC-associated with the WIF1 promoter in Caski, but not in SiHa, cells may explain the difference in expression between the cell lin es. Chromatin immunoprecipitation experiments probing for histones might help to create a clearer picture of regulation of the WIF1 promoter and establish if there is a bias in the MA Pit footprint. Better understanding of the WIF1 promoter region and its associated proteins could potentially lead to use as a cervical cancer drug therapy. In addition to technical limitations that ma y have precluded detec tion of DNA methylation in cervical tumors lacking WIF1 protein expression, it is also possi ble that non-epigenetic factors are also responsible for loss of WIF1 expression in these tumors. For example, chromosome rearrangements involving 12q13-15, where WIF1 is located, are frequent occurrences among tumors. While no DNA mutations of WIF1 have been reported, one gr oup demonstrated salivary gland pleomorphic adenoma expressing an high mobility group AT-hook 2 (HMGA2)/WIF1 49

PAGE 50

fusion transcript, which resulted in re-expression of wild-type HMGA2 transcripts and very low levels of WIF1 expression [78]. An altern ative explanation for the l ack of correlation between WIF1 promoter methylation and protein expression is that WIF1 transcription may require an additional activator which could be inhibited or after chromoso me rearrangement the gene may be excised. These are a few alternative mechanisms aside from epigenetic silencing that could result in downregulation of WIF1 protein in cervical cancer. A common aim of TSG resear ch is to evaluate the possibility of a gene being used as either a diagnostic or prognostic tool. We showed that the WIF1 gene is aberrantly methylated in highgrade squamous cell carcinomas of the cervix, which makes it a potential prognostic marker. However, we failed to obtain any low-grade or mildly dysplastic cervical squamous cells for analysis within the confines of the study. Examining the status of WIF1 expression at precancerous stages or within low-grade carcinomas may help determine when during tumor progression aberrant methylation of the WIF1 promoter occurs, and if hypermethylation is an indicator of high-grade squamous cell carcinoma. There exists a possibility that a correlation between expression of WIF1 and the pathological state or grad e of the tumor could be made to provide insight into WIF1 being a possible biomarker for diagnosis. It is clear that cancer cells undergo global changes in DNA methyla tion that lead to chromatin alterations involving the whole epigenom e and that entire pathways relevant to cell division are subject to epigenet ic dysregulation. As the role of epigenetics in cancer becomes clearer new approaches to cancer prevention, de tection, and therapy wi ll begin to emerge. 50

PAGE 51

LIST OF REFERENCES 1. Detailed Guide: Cervical Cancer. What ar e the key statistics about cervical cancer? Available from: http://www.cancer.org/docroot/CRI/CRI_2_3x.asp?rnav=cridg&dt=8 2. Schoell, W.M.J., M.F. Janicek, and R. Mirhashemi, Epidemiology and biology of cervical cancer. Seminars in Surgical Oncology, 1999. 16(3): p. 203-11. 3. Kurman, R.J., et al., From Papanicolaou to Bethesda: the rationale for a new cervical cytologic classification. Obstet Gynecol, 1991. 77(5): p. 779-82. 4. Lie, A.K. and G. Kristensen, Human papill omavirus E6/E7 mRNA testing as a predictive marker for cervical carcinoma. Expe rt Rev Mol Diagn, 2008. 8(4): p. 405-15. 5. Walboomers, J.M.M., et al., Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol, 1999. 189(1): p. 12-9. 6. Ho, G.Y.F., et al., Persistent Genital Human Papillomavirus Infection as a Risk Factor for Persistent Cervical Dysplasia. J Natl Cancer Institute, 1995. 87(18): p. 1365-71. 7. Choo, K.B., C.C. Pan, and S.H. Han, Integrat ion of human papillomavirus type-16 into cellular DNA of cervical carcinoma Preferentia l Deletion of the E2 gene and invariable retention of the long contro l region and the E6/E7 Open Reading Frames. Virology, 1987. 161(1): p. 259-61. 8. Di Leonardo, A., et al., DNA damage trigge rs a prolonged p53-depe ndent G1 arrest and long-term induction of Cip1 in normal hum an fibroblasts. Genes Dev, 1994. 8(21): p. 2540-51. 9. Smith, M.L., et al., Involvement of the p53 tumor suppressor in repair of U.V.-type DNA damage. Oncogene, 1995. 10(6): p. 1053-9. 10. Subramanya, D. and P.D. Grivas, HPV and cervical cancer: updates on an established relationship. Postgrad Med, 2008. 120(4): p. 7-13. 11. Gissmann, L., Human papillomaviruses and genital cancer. Semin Cancer Biol, 1992. 3(5): p. 253-61. 12. American Type Culture Collection., Catalogue of strains II, American Type Culture Collection.: Rockville, Md.,. p. v. 13. Scheffner, M., et al., The state of the p53 and retinoblastoma genes in human cervical carcinoma cell lines. Proc Natl A cad Sci USA, 1991. 88(13): p. 5523-7. 14. Jones, P.A. and S.B. Baylin, The epigenomics of cancer. Cell, 2007. 128(4): p. 683-92. 51

PAGE 52

15. Jones, P.A., Overview of cancer epigenet ics. Semin Hematol, 2005. 42(3 Suppl 2): p. S38. 16. Cooper, D.N. and M. Krawczak, Cytosine met hylation and the fate of CpG dinucleotides in vertebrate genomes. Hum Genet, 1989. 83(2): p. 181-8. 17. Bird, A., DNA methylation patterns and ep igenetic memory. Genes Dev, 2002. 16(1): p. 6-21. 18. Larsen, F., et al., CpG islands as gene markers in the human genome. Genomics, 1992. 13(4): p. 1095-107. 19. Mohandas, T., R.S. Sparkes, and L.J. Sh apiro, Reactivation of an inactive human X chromosome: evidence for X inactiva tion by DNA methylation. Science, 1981. 211(4480): p. 393-6. 20. Li, E., C. Beard, and R. Jaenisch, Role for DNA methylation in genomic imprinting. Nature, 1993. 366(6453): p. 362-5. 21. Suetake, I., et al., DNMT3L stimulates th e DNA methylation activity of Dnmt3a and Dnmt3b through a direct interaction. J Biol Chem, 2004. 279(26): p. 27816-23. 22. Otani, J., et al., Structural basis for recognition of H3K4 methylation status by the DNA methyltransferase 3A ATRX-DNMT3-DNMT3L domain. EMBO Rep, 2009. 10(11): p. 1235-41. 23. Hermann, A., H. Gowher, and A. Jeltsc h, Biochemistry and biology of mammalian DNA methyltransferases. Cell Mol Life Sci, 2004. 61(19-20): p. 2571-87. 24. Luger, K., et al., Crystal structure of th e nucleosome core particle at 2.8 resolution. Nature, 1997. 389(6648): p. 251-60. 25. Sarma, K. and D. Reinberg, Histone varian ts meet their match. Nat Rev Mol Cell Biol, 2005. 6(2): p. 139-49. 26. Kouzarides, T., Chromatin modifications and their function. Ce ll, 2007. 128(4): p. 693705. 27. Fuks, F., et al., The methyl-CpG-bindi ng protein MeCP2 links DNA methylation to histone methylation. J Biol Chem, 2003. 278(6): p. 4035-40. 28. Nan, X., et al., Transcriptional repressi on by the methyl-CpG-b inding protein MeCP2 involves a histone deacetylase co mplex. Nature, 1998. 393(6683): p. 386-9. 52

PAGE 53

29. Tachibana, M., et al., G9a/GLP complexes independently mediate H3K9 and DNA methylation to silence transcri ption. EMBO J, 2008. 27(20): p. 2681-90. 30. Zhao, Q., et al., PRMT5-mediated methyl ation of histone H4R3 recruits DNMT3A, coupling histone and DNA methylation in gene silenci ng. Nat Struct Mol Biol, 2009. 16(3): p. 304-11. 31. Esteve, P.O., et al., Regulation of DNMT 1 stability through SET7-mediated lysine methylation in mammalian cells. Proc Natl Acad Sci USA, 2009. 106(13): p. 5076-81. 32. Jones, P.L., et al., Methylated DNA and Me CP2 recruit histone deacetylase to repress transcription. Nat Genet, 1998. 19(2): p. 187-91. 33. Robertson, K.D., et al., DNMT1 forms a complex with Rb, E2F1 and HDAC1 and represses transcription from E2F-responsiv e promoters. Nat Genet, 2000. 25(3): p. 33842. 34. Wang, Z., et al., Genome-wide mapping of HA Ts and HDACs reveals distinct functions in active and inactive genes. Cell, 2009. 138(5): p. 1019-31. 35. Lorch, Y., J.W. LaPointe, and R.D. Kornberg, Nucleosomes inhib it the initiation of transcription but allow chain elongation with the displacement of histones. Cell, 1987. 49(2): p. 203-10. 36. Bernstein, B.E., et al., Global nucleosome occupancy in yeast. Genome Biol, 2004. 5(9): p. R62. 37. Jiang, C. and B.F. Pugh, Nucleosome positi oning and gene regulation: advances through genomics. Nat Rev Genet, 2009. 10(3): p. 161-72. 38. Ozsolak, F., et al., High-throughput mapping of the chromatin structure of human promoters. Nat Biotechnol, 2007. 25(2): p. 244-8. 39. Heintzman, N.D., et al., Distinct and predic tive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nat Genet, 2007. 39(3): p. 311-8. 40. Boeger, H., J. Griesenbeck, and R.D. Kornbe rg, Nucleosome retention and the stochastic nature of promoter chromatin remodeling for transcription. Cell, 2008. 133(4): p. 716-26. 41. Dechassa, M.L., et al., SWI/SNF has intrinsi c nucleosome disassembly activity that is dependent on adjacent nucleosomes. Mol Cell, 2010. 38(4): p. 590-602. 42. Shivaswamy, S., et al., Dynamic remodeling of individual nucleosomes across a eukaryotic genome in response to transcrip tional perturbation. PL oS Biol, 2008. 6(3): p. e65. 53

PAGE 54

43. Schones, D.E., et al., Dynamic regulati on of nucleosome positioning in the human genome. Cell, 2008. 132(5): p. 887-98. 44. Gilbert, N. and B. Ramsahoye, The rela tionship between chromatin structure and transcriptional activit y in mammalian genomes. Brief Funct Genomic Proteomic, 2005. 4(2): p. 129-42. 45. Ptashne, M., On the use of the word 'e pigenetic'. Curr Biol, 2007. 17(7): p. R233-6. 46. Zemach, A., et al., Genome-wide evolutiona ry analysis of eukaryotic DNA methylation. Science, 2010. 328(5980): p. 916-9. 47. Sherr, C.J., Principles of tumo r suppression. Cell, 2004. 116(2): p. 235-46. 48. Jones, P.A. and S.B. Baylin, The fundamental role of epigenetic events in cancer. Nat Rev Genet, 2002. 3(6): p. 415-428. 49. Garinis, G.A., et al., DNA hypermethylation: when tumour suppressor genes go silent. Hum Genet, 2002. 111(2): p. 115-27. 50. Boyes, J. and A. Bird, Repression of ge nes by DNA methylation depends on CpG density and promoter strength: evidence for invol vement of a methyl -CpG binding protein. EMBO J, 1992. 11(1): p. 327-33. 51. Wolffe, A.P., Chromatin re modeling: why it is important in cancer. Oncogene, 2001. 20(24): p. 2988-90. 52. Morey, L., et al., MBD3, a component of the NuRD complex, facilitates chromatin alteration and deposition of epigenetic marks. Mol Cell Biol, 2008. 28(19): p. 5912-23. 53. Ramirez, J. and J. Hagman, The Mi-2/NuRD co mplex: a critical epig enetic regulator of hematopoietic development, differentiation a nd cancer. Epigenetics, 2009. 4(8): p. 532-6. 54. Reisman, D., S. Glaros, and E.A. T hompson, The SWI/SNF complex and cancer. Oncogene, 2009. 28(14): p. 1653-68. 55. Etoh, T., et al., Increased DNA methyltr ansferase 1 (DNMT1) protein expression correlates significantly with poorer tu mor differentiation and frequent DNA hypermethylation of multiple CpG islands in gastric cancers. Am J Pathol, 2004. 164(2): p. 689-99. 56. Muller, K., H. Heller, and W. Doerfler, Foreign DNA integration. Genome-wide perturbations of methylation and transcription in the reci pient genomes. J Biol Chem, 2001. 276(17): p. 14271-8. 54

PAGE 55

57. Burgers, W.A., et al., Viral oncoproteins target the DNA methyltransferases. Oncogene, 2007. 26(11): p. 1650-5. 58. Dong, S.M., et al., Promoter hypermethylati on of multiple genes in carcinoma of the uterine cervix. Clin Cancer Res, 2001. 7(7): p. 1982-6. 59. Sawada, M., et al., Increased expression of DNA methyltransferase 1 (DNMT1) protein in uterine cervix squamous cell carcinoma and its precursor lesion. Cancer Lett, 2007. 251(2): p. 211-9. 60. Bigey, P., et al., Transcriptional regula tion of the human DN A Methyltransferase ( DNMT1 ) gene. Gene, 2000. 242(1-2): p. 407-18. 61. Ghittoni, R., et al., The biological properti es of E6 and E7 oncoproteins from human papillomaviruses. Virus Genes, 2010. 40(1): p. 1-13. 62. Cadigan, K.M. and R. Nusse, Wnt signaling: a common theme in animal development. Genes Dev, 1997. 11(24): p. 3286-305. 63. Hobmayer, B., et al., WNT signalling molecule s act in axis formation in the diploblastic metazoan Hydra. Nature, 2000. 407(6801): p. 186-9. 64. Reya, T. and H. Clevers, Wnt signalling in stem cells and cancer. Nature, 2005. 434(7035): p. 843-50. 65. MacDonald, B.T., K. Tamai, and X. He, Wnt/beta-catenin signaling: components, mechanisms, and diseases. Dev Cell, 2009. 17(1): p. 9-26. 66. Kawano, Y. and R. Kypta, Secreted antagonist s of the Wnt signalling pathway. J Cell Sci, 2003. 116(Pt 13): p. 2627-34. 67. Cong, F., L. Schweizer, and H. Varmus, Wn t signals across the plasma membrane to activate the -catenin pathway by forming oligomer s containing its receptors, Frizzled and LRP. Development, 2004. 131(20): p. 5103-15. 68. He, X., et al., LDL receptor-related proteins 5 and 6 in Wnt/ -catenin signaling: arrows point the way. Development, 2004. 131(8): p. 1663-77. 69. van Amerongen, R., A. Mikels, and R. Nusse, Alternative Wnt signaling is initiated by distinct receptors. Sci Signal, 2008. 1(35): p. re9. 70. Arce, L., K.T. Pate, and M.L. Waterman, Groucho binds two conserved regions of LEF-1 for HDAC-dependent repressi on. BMC Cancer, 2009. 9: p. 159. 71. Jessen, J.R., Noncanonical Wnt signaling in tu mor progression and metastasis. Zebrafish, 2009. 6(1): p. 21-8. 55

PAGE 56

72. Uren, A., et al., Activation of the canoni cal Wnt pathway during genital keratinocyte transformation: a model for cervical cancer progression. Cancer Res, 2005. 65(14): p. 6199-206. 73. Hsieh, J.C., et al., A new secreted protein th at binds to Wnt protei ns and inhibits their activities. Nature, 1999. 398(6726): p. 431-6. 74. Wissmann, C., et al., WIF1, a component of the Wnt pathway, is down-regulated in prostate, breast, lung, and bladder cancer. J Pathol, 2003. 201(2): p. 204-12. 75. Warnecke, P.M., et al., Identification and re solution of artifacts in bisulfite sequencing. Methods, 2002. 27(2): p. 101-7. 76. Macleod, D., et al., Sp1 sites in the mouse aprt gene promoter are required to prevent methylation of the CpG island. Genes Dev, 1994. 8(19): p. 2282-92. 77. Reguart, N., et al., Cloning and characterizati on of the promoter of human Wnt inhibitory factor-1. Biochem Biophys Res Commun, 2004. 323(1): p. 229-34. 78. Persson, F., et al., High-resol ution genomic profiling of ade nomas and carcinomas of the salivary glands reveals am plification, rearrangement, and fusion of HMGA2. Genes Chromosomes Cancer 2009. 48(1): p. 69-82. 56

PAGE 57

BIOGRAPHICAL SKETCH Amber Lyn Delmas was born in Stuart, Fl orida. Through dual enrollment during high school, she attended Indian River Community College. She received an Associates of Arts degree in science the summer after graduating fr om high school in 2003. She then attended the University of Florida and earned a Bachelor of Science in microbiology and cell science in 2006. After graduation, she worked as a laboratory technician in the De partment of Biochemistry and Molecular Biology at the University of Florida for two years. In August 2010, she received a Master of Science in medical sciences from the University of Florida. 57