m."
’
. nu. HIE...“

: 5
km... rs...w.m. e 2%;
.KII.’IIIA! 'u.lcn‘3\.-—Q}
m..3....5........t 51...}...5...‘
$1; '17.. 31,15)?

éJMMm-mlgzr.g

. b.1131!
... . 45:15”..th
£643».th

 

- w-v-Ovt’I

 

_

..‘ialtimufletil I'M-.6 ..I.
. 1...;Hxltl."

Jan-n.6,. ... .
1.11.53. .. .J ... Did-g...

 

 

.2}:
H.122:
. ..t...“...

5.. v...

. ..14T£_.:,a.s
11...-..
a 5....

 

“In...”..w.“uqoaoou-upumnumuv ..- nun-unpu- ~-

 

.uo..un-0u..nu.

,‘Il‘¢«u.~a

 

....11

., . . . . V V V . “ V............¢u.x ..

 

2 LIBRA liY
10°?” Michigan state ‘
Unive r“: ! ty

This is to certify that the
dissertation entitled

BMl-1 COLLABORATES WITH H-RAS TO PROMOTE MAMMARY
EPITHELIAL CELL TRANSFORMATION, TUMORIGENESIS, AND
METASTASIS

presented by

Mark James Hoenerhoff

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Pathobiology and Diagnostic
Investigation

 

 

/fl£,qflm3 [L/ltbgfa4mfl

 

Major Professor’s Sigature

ficfgf/Tg m 3"

Date

 

MSU is an affinnative-action, equal-opportunity employer

a--..-a—.-.-.-.-.-l-.—.---

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K‘IProlecc8-PreleIRC/Dateoue lndd

BMI-l COLLABORATES WITH H-RAS TO PROMOTE MAMMARY EPITHELIAL
CELL TRANSFORMATION, TUMORIGENESIS, AND METASTASIS

By

Mark James Hoenerhoff

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Pathobiology and'Diagnostic Investigation

2008

ABSTRACT

BMI-l COLLABORATES WITH H-RAS TO PROMOTE MAMMARY EPITHELIAL
CELL TRANSFORMATION, TUMORIGENESIS, AND METASTASIS

By
Mark James Hoenerhoff

The polycomb group protein Bmi-l is a transcription repressor reported to
regulate self-renewal of normal and cancer stem cells, and prevent cellular senescence
through inhibition of cyclin-dependent kinase inhibitors pl6Ink4a and pl9Arf. Originally
discovered as a cooperating oncogene with c—Myc in a murine model of leukemia,
overexpression of Bmi-l has since been reported in several forms of cancer, including
breast cancer. A direct or collaborative role of Bmi-l in the pathogenesis of breast cancer
would be critical in the development of novel biomarkers for diagnosis and treatment of
this disease. We demonstrate that overexpression of Bmi-l alone and in combination with
H—Ras in normal mammary epithelial cells (MCFIOA) markedly changes cellular
morphology, increases proliferative indices, and decreases . the apoptotic response to
apoptosis-inducing agents in vitro. Our data also suggests that overexpression of Bmi-l
confers stem cell-like properties, including the formation of spheroid structures
resembling mammospheres in culture and overexpression of stem cell markers. Xenografi
experiments in immunodeficient (SCID) mice demonstrate that 1) Bmi-l overexpression
alone in MCFlOA cells cannot induce tumors; 2) H-ras overexpression induces
development of tumors with vascular, smooth muscle, and mast cell components;
whereas 3) the combination of Bmi-l and I-I-Ras induces development of poorly-,
differentiated aggressive mammary neoplasms with greatly increased propensity for

spontaneous distant metastasis to liver, spleen, and brain. Moreover, by using tail vein

injection experiments in SCID mice as a model of direct hematogenous spread to the
lungs, combination of Bmi-l and H-ras overexpression in mammary epithelial cells
results in fulminant pulmonary metastasis leading to clinical signs of weight loss and
dyspnea, whereas injection of cells overexpressing H-ras or Bmi-l alone does not result
in this aggressive metastatic phenotype. Additionally, our data suggests that Bmi-l
overexpression allows cells with H-ras overexpression to bypass senescence, avoid
anoikis, and essentially overcome dormancy in order to form pulmonary metastases.
These findings suggest augmented stem cell properties through overexpression of Bmi-l
in conjunction with H—ras overexpression that results in greater metastatic potential.
Therefore, Bmi-l may be an important marker for diagnosis and a target for the treatment

of aggressive breast cancer.

ACKNOWLEDGEMENTS

I would like to thank everyone who had a part in the development of this
dissertation research, including my Principal Investigator, Jeffery E. Green from the
Transgenic Oncogenesis Group at the National Cancer Institute, my collaborators Dr.
Goberdhan Dimri and Dr. Sonal Datta at the Feinberg School of Medicine and Robert H.
Lurie Comprehensive Cancer Center at Northwestern University, my committee members
Mark R. Simpson from the Comparative Molecular Pathology Unit at the National
Cancer Institute, Thomas P. Mullaney, Kurt J. Williams, Michael Scott, Vilma
Yuzbasiyan-Gurkan, and RS. MohanKumar from the Department of Pathobiology and
Diagnostic Investigation at Michigan State University. A special thanks to my fellow
laboratory members at the National Cancer Institute who were instrumental in my
introduction to benchwork and my education in molecular biology. I would like to thank
my family and friends for their support, and most importantly I would like to thank my
wife for her support of my scholarly endeavors, her patience during my research trials
and tribulations, and her confidence in my success. I could not have achieved this goal

without her.

iv

TABLE OF CONTENTS

LIST OF FIGURES .......................................................................................................... vii
I. EPIGENETIC REGULATION OF GENE EXPRESSION ......... , ........................ 1
II. BMI-l IN STEM CELL REGULATION AND CARCWOGENESIS .............. 3
III. COLLABORATING ONCOGENES IN CARCINOGENESIS ........................ 5
IV. EXPERIMENTAL DESIGN TO STUDY THE COLLABORATION OF
H-RAS AND BMI-l .................................................................................... 9
CHAPTER 1:

THE TRANSCRIPTION REPRESSOR POLYCOMB BMI-l: DIFFERENTIAL
EXPRESSION HISTOLOGIC SUBTYPES OF HUMAN BREAST CANCER,
WILDTYPE MOUSE TISSUE, AND GENETICALLY ENGINEERED MOUSE
MODELS OF BREAST CANCER ....................................................................... 12
APPENDIX A/FIGURES .......................................................................... 28

CHAPTER 2:
BMI—l COOPERATES WITH H-RAS TO TRANSFORM HUMAN
MAMMARY EPITHELIAL CELLS VIA DYSREGULATION OF MULTIPLE
GROWTH REGULATORY PATHWAYS ........................................................... 38
APPENDIX B/FIGURES .......................................................................... 62

CHAPTER 3:
BMI-l AND H-RAS COLLABORATION IN MAMMARY EPITHELIAL
CELLS PROMOTES AN AGGRESSIVE AND METASTATIC PHENOTYPE,
AND BMI-l KNOCKDOWN IN XENOGRAFT TUMORS RESULTS IN
DELAYED TUMOR ONSET ............................................................................... 75
APPENDIX C/FIGURES .......................................................................... 97

CHAPTER 4:

BMI-l CONTRIBUTES TO THE OUTGROWTH OF EMBOLIC BREAST
CANCER CELLS IN THE LUNGS THROUGH INCREASED

PROLIFERATION AND PREVENTION OF APOPTOSIS .............................. 108
APPENDIX D/FIGURES ........................................................................ 120
CHAPTER 5:
SIGNIFICANCE AND FUTURE DIRECTIONS ............................................... 126
1. CONCLUSIONS AND SIGNIFICANCE ............................................ 126
II. MICROARRAY ANALYSIS OF MCF 10A-DERIV ED
CELL LINES ............................................................................... 131

III. DEVELOPMENT OF A TISSUE-SPECIFIC CONDITIONAL
TRANSGENIC MOUSE MODEL OF BMI-l OVER-
EXPRESSION IN THE MAMMARY GLAND ......................... I36

IV. FUTURE DIRECTIONS ................................................................. 146
APPENDIX E/FIGURES ........................................................................ 150

REFERENCES ................................................................................................................ l 57

vi

LIST OF FIGURES
Figure l: Endogenous expression of Bmi-l in wildtype mouse (FVB) tissue ................. 28

Figure 2: Endogenous Bmi—l expression in wildtype mouse tissues and GEM tumors by

QRT-PCR normalized to brain ............................................................ ' .............................. 29

Figure 3: Bmi-l expression during phases of the estrus Cycle, pregnancy, and

lactation .............................................................................................................................. 30

Figure 4: Bmi-l expression in genetically engineered mouse models of human breast

cancer ................................................................................................................................. 34
Figure 5: Bmi-l expression in human breast cancer samples ........................................... 36

Figure 6: Bar graph illustrating the relative levels of Bmi-l expression in histologic

subtypes of human breast cancers by immunohistochemistry ........................................... 37

Figure 7: Bmi-l and H-Ras co-expression transforms human mammary epithelial cells

(HMECs) ............................................................................................................................ 62

Figure 8: Various growth regulatory pathways are dysregulated in cellsco—expressing

Bmi-l and H-Ras ............................................................................................................... 63

vii

Figure 9: Late-passage H-Ras—expressing MCF 1 0A cells exhibit transformed

features ............................................................................................................................... 64

Figure 10: Expression level of H-Ras determines proliferation in MCFlOA cells

overexpressing H-Ras ........................................................................................................ 67

Figure 11: Gross morphology, histopathology, and immunohistochemistry of tumors

originating from xenografis ............................................................................................... 69

Figure 12: Analysis of p53 pathway in control MCF 10A and MCFlOA-derived

late-passage cells ................................................................................................................ 72
Figure 13: Morphology of MCF lOA-derived cell lines in two dimensional culture ........ 97

Figure 14: Morphology of MCFlOA-derived cell lines in three-dimensional Matrigel

culture ................................................................................................................................ 98

Figure 15: Proliferative indices, apoptotic response to DNA damage, and invasive

properties of MCF lOA-derived cell lines in vitro.......- ....................................................... 99

Figure 16: MTS-formazan proliferation assay in siRNA-Bmi-li induced established

breast cancer cell lines ..................................................................................................... 101

viii

Figure 17: Anti-histone cell death detection ELISA and Matrigel Invasion Assay in

siRNA-Bmi-li induced established breast cancer cell lines ............................................ 102

Figure 18: Spontaneous metastasis from intra-mammary fat pad MCF 10A+H-Ras and

MCF 10A+H-Ras+Bmi-1 xenografts ............................................................................... 103

Figure 19: MCF 10A+H-Ras and MCF 10A+H-Ras+Bmi-l tail vein xenograft

model ................................................................................................................................ 1.04

Figure 20: Mammary fat pad and tail vein model using MCFlOA+H-Ras+Bmi-

1+pLL3.7 and MCFl 0A+H-Ras+Bmi- 1+Bmi- 1 i cells ................................................... 105

Figure 21: Mammary fat pad and tail vein model using MDA231+pLL3.7 and

MDA23 1+Bmi- 1i cells .................................................................................................... 106

Figure 22: Western blot illustrating continued siRNA knock down of Bmi-l expression

in MCF 10A+H-Ras+Bmi-1 and MDA231 xenografts .................................................... 107

Figure 23: MCFlOA+H-Ras and MCFlOA+H-Ras+Bmi-l GFP expressing cells in the

lungs of mice at 2 hours, 2 days, one week, and four weeks after injection .................... 120

ix

Figure 24: Ki67 immunohistochemistry in MCFlOA+H-Ras and MCF lOA+H-Ras+Bmi-

1 mammary fat pad (SQ) and intravenous xenograft models at end-stage disease ......... 122

Figure 25: TUNEL staining in MCFlOA+H-Ras and MCFlOA+H-Ras+Bmi-l mammary

fat pad and intravenous xenograft models at end-stage disease ...................................... 123

Figure 26: Ki67 immunolabeling and TUNEL staining in MCFlOA+H-Ras and

MCF 10A+H-Ras+Bmi-1 intravenous xenografts during early progression of disease... 124

Figure 27: Bmi-l prevents cellular senescence, cell cycle arrest and apoptosis by

inhibiting the Ink4a-Arf tumor suppressor locus ............................................................. 125
Figure 28: All enriched BioCarta pathways shared by Bmi-l overexpressing cells ...... 150

Figure 29: Genes up- and down-regulated in the ATM and G2/M cell cycle checkpoint

pathways .......................................................................................................................... 151
Figure 30: Gene-Ontology Biological Processes associated with Bmi-l signature ....... 152

Figure 31: Vectors and expression plasmids used in development of inducible

mammary-specific transgenic mouse model of Bmi-l overexpression ........................... 152

Figure 32: Possible orientation of integrations of the transgene and their respective

fragment sizes following restriction enzyme digestion ................................................... 154

Figure 33: Southern blot analysis of lOug of genomic DNA from pTRE-Bmi-l

transgenic founders .......................................................................................................... 155
Figure 34: Tetracycline induced mammary-specific mRN A expression of Bmi-l ........ 156

Figures in this dissertation are presented in color.

xi

INTRODUCTION

1. Epigenetic Regulation of Gene Expression

According to the American Cancer Society, breast cancer is the most common
cancer in women in the United States, and is the second leading cause of cancer
mortality. Traditionally breast cancer, and cancer in general, has been viewed as a genetic
disease, in which the accumulation of genetic mutations over time results in the activation
of oncogenes, or the inactivation of tumor suppressor genes, culminating in cellular
transformation and a cancer phenotype [1]. However, it has become evident that non-
genetic, or epigenetic, factors play a role in tumorigenesis. Epigenetic factors influence
gene expression in a heritable way without altering the DNA sequence [1-3]. Examples
include chromatin modification [4-7], epithelial-stromal interactions [8, 9], and RNA
interference [10, 11], to name a few.

Regarding chromatin modification, it is well known that changes in the structure
of chromatin can greatly influence gene expression. For example, when chromatin is
associated with histones and condensed, the DNA is not accessible to transcription
factors, and transcription does not occur, effectively “turning off” genes. When chromatin
is in an open conformation on the other hand, promoters are accessible to transcription
factors, and genes may be “turned on” as required [12, 13].

There are two major mechanisms of chromatin modification as a means of
epigenetic regulation of gene expression; DNA methylation and modification of histones
[3, 13, 14]. Methylation is the addition of methyl groups to areas of high cytosine and

guanine concentration, or CpG islands, which prevents transcription binding to

promoters, resulting in gene silencing [15]. Histones are the basic protein units that
associate with DNA to form chromatin. Four pairs of histones associate with 146 base
pairs of DNA to form nucleosomes, the fundamental repeating units of eukaryotic
chromatin [16]. When DNA is associated with histones, access to promoters by
transcription factors is prevented. Through modification of histones, gene expression is
suppressed by preventing disassociation of the DNA from the histone unit, preventing
transcription factor binding, and effectively silencing gene expression [17].

One of the primary mediators of suppression of gene expression through histone
modification is the polycomb proteins. These proteins bind to and modify chromatin to
suppress transcription of Homeotic genes that are essential for basic body plan
establishment and morphogenesis during development [18], and have been shown to be
involved in epigenetic gene regulation, acting as heritable gene repressors to achieve a
“memory” of cell identity by preserving transcriptional patterns in the face of dividing
cells [19—21]. The polycomb proteins exist in two complexes, 'Polycomb-Repressive
Complex 1 (PRCl), containing Bmi-l, BVX, PHCl-3, RNF 1-2, and SCMLl-Z, [22, 23]
and a second complex, Polycomb-Repressive Complex 2 (PRC2), containing the
polycomb proteins ENX/EZHZ, EED, and SUZlZ [20, 22, 24, 25]. The PRC2 is
responsible for the initiation of transcriptional repression, and recruits the PRCl to the
site of gene suppression. The PRCl is necessary for sustaining this repression by binding

to chromatin and antagonizing the remodeling of nucleosomes [19, 20, 24, 26].

II. Bmi-l in Stem Cell Regulation and Carcinogenesis

B lymphoma Moloney murine leukemia virus insertion region-l (Bmi-l) is a
member of the polycomb group of transcription repressors, first identified as a c-myc-
cooperating oncogene in murine B and T-cell lymphoma [18, 27-30]. The bmi-l gene has
conserved homology across several species, and the gene structure in the human is very
similar to that of the mouse; at the nucleotide level, it is 92.4% homologous with mouse
bmi-l in the open reading frame (ORF) [28]. Additionally, the ORF codes for a nuclear
protein of 326 amino acids that shares 98% homology with the mouse bmi-l protein at
the amino acid level [28]. Bmi-l has been shown to be.a key regulator of the self renewal
of both normal and malignant hematopoietic and neuronal stem cells [31, 32], including
I-type neuroblastoma cells, thought to be derived fiom malignant neural crest stem cells
[33, 34]. Importantly, bmi-l has also been shown to regulate self-renewal of normal and
malignant human mammary stem cells with the hedgehog signaling pathway [35]. In
addition, several recent studies have suggested a link between Bmi-l and mammary
carcinogenesis [29, 31]. It has been shown that Bmi-l is over-expressed in several human
breast cancer cell lines, and that Bmi-l regulates telomerase and can immortalize human
mammary epithelial cells and fibroblasts [18, 29, 31, 36]. Additionally, overexpression
of Bmi-l has been reported in human cancers, including nasopharyngeal carcinoma [37],
oral squamous cell carcinoma [38], neuroblastoma [39], medulloblastoma, [40], non-
small cell lung cancer, [41], basal cell carcinoma [42], liver carcinoma [43], colorectal
carcinoma [44] and other gastrointestinal tumors [45] and precancerous lesions of the
gastrointestinal tract [46], in addition to human breast cancer [29], where it has been

correlated with axillary lymph node metastasis in invasive ductal breast carcinoma [47].

In multiple types of human cancer, including prostate, breast, and lung cancer,
overexpression of Bmi-l is associated with a stem cell-like 11 gene expression
microarray signature that has been determined to be a powerful predictor of short interval
to distant metastasis in patients diagnosed with early-stage disease [48]. Bmi-l and EZH2
have been shown to be tumor-associated antigens [49], and together their coexpression is
associated with the degree of malignancy in B-cell non-Hodgkin lymphoma [25].

Bmi-l has been shown to be critical in the maintenance of stem cell populations
through a process known as self-renewal, as well as through prevention of senescence
through suppression of the Inka-ARF tumor suppressor locus [18, 27, 32], which encodes
the cyclin-dependent kinase (CDK) inhibitors p16 and 14 (p19 in mice). The Ink4a locus
encodes p16, a CDK that prevents the association of cyclin D with its cyclin dependent
kinases, CDK4 and 6, thus preventing phosphorylation of retinoblastoma protein (pr),
ultimately leading to cell cycle arrest, apoptosis, or senescence. The Arf tumor suppressor
locus encodes 1514, which promotes p53 activation by inactivating Mdm2, likewise
leading to cell cycle arrest, apoptosis, or senescence [18, 27, 30, 32, 50]. In this way,
Bmi-l can act as an oncogene through binding of chromatin, resulting in inactivation of
this tumor suppressor gene locus [51-53], and when overexpressed, can bypass
senescence and extend the replicative lifespan of primary cells [18, 32]. Inactivation of
these senescence pathways contributes significantly to carcinogenesis. Inactivation of
Ink4a leads to several forms of human cancer including lymphoma, pulmonary adenoma,
brain tumors, and melanoma [54-56], and deletion of the Arf locus is associated with
increased incidence of sarcomas, lymphomas, and gliomas, as well as other types of

cancer [57]. Given the important function in regulation of the stem cell pool in multiple

tissues as well as the implications in tumorigenesis, the role of Bmi-l in regulation of the
mammary epithelial stem cell compartment and further elucidation of the role this gene
plays in mammary carcinogenesis of the mouse will be paramount in shedding light on
the comparative pathology of the complex disease of human breast cancer, as well as

provide powerful mouse models by which to further study the disease.

111. Collaborating Oncogenes in Carcinogenesis

Cooperation between oncogenes to produce neoplasia is a well-documented
occurrence in many types of human cancers. Whether resulting from a combination of
proliferative or of anti-apoptotic effects from overexpression of multiple oncogenes, or
the abrogation of tumor suppressor genes, the process of oncogenesis commonly involves
mutations of multiple genetic loci. In the multi-step process of carcinogenesis, it is
generally thought that malignant phenotypes arise via a recurrent mechanism of clonal
expansion triggered by epigenetic and genetic lesions that may frequently require
multiple alterations affecting several levels of growth control [58-63]. Whether this
includes a mutation inducing loss of function of one or more tumor suppressor genes that
inhibit proliferation or detect and/or repair DNA damage, or activation of a
protooncogene resulting from a number of possible mutagenic events depends on the cell
type and inducing mutation. Collaborations between oncogenes indicate that regulatory
mechanisms that preclude transformation by one oncogene can be circumvented by a
second oncogene [62] Several reports have demonstrated this concept of collaborating
oncogenes both in vitro in primary cell lines and through the use of gene transfer

experiments and in vivo using transgenic mouse models [58, 60, 64-69].

In vitro, it has been shown that overexpression of a single oncogene is often
insufficient to transform cells to a malignant state, and that this transformation often
requires two or more oncogenes acting together [70-72]. Several cellular and viral
oncogenes require the presence of additional oncogenes or loss of tumor suppressor genes
to transform cells, although expression of these oncogenes alone may immortalize cells.
For example, while polyomavirus large T-antigen is sufficient to immortalize primary
cells, it is unable to cause transformation [71, 73]. Additionally, whereas transformation
of established fibroblasts by polyomavirus only requires the middle T-antigen,
transformation of non-established fibroblasts requires the small, middle, and large T-
antigens collectively [71, 74]. Furthermore, transformation of rat embryonic fibroblasts
by the ras oncogene requires concurrent expression of adenovirus-S or SV40 large T-
antigen [62][58]. Other viral and cellular oncogenes known to collaborate with ras to
transform primary cells include PyMT, HPV16 E7, Adenovirus ElA, p53, c-fos, Notch,
and myc [62, 71, 75] In fact, unless coexpressed with such collaborating oncogenes,
nonestablished or early passage rodent cells are transformed by ras at very low
frequencies [62, 71, 76, 77]. Additionally, transformation of primary Schwann cells by
ras is a synergistic process that is dependent on coexpression of the simian virus 40 large
T antigen [60, 78].

Constitutively active mutant forms of Notch appear to be involved in human T-
cell leukemia and mammary carcinomas in mice, and transformation of these cells
requires signals from the ERK-MAPK pathway downstream from Ras [75]. In addition,
cell lines obtained from tumors from Notch4 transgenic mice require ras signaling for

anchorage independent growth [75]. These data show that the effects of other oncogenes

on the transforming activity of ras are complementary, and that in combination,
collaborating oncogenes allow release of cells from controls that prevent transformation
by ras alone [71].

Overexpression of oncogenic Ras or c-myc in the mammary gland induces
mammary tumors in rodent models, and when expressed together, the rate of tumor onset
is significantly increased [79]. Additionally, by utilizing retroviral insertional
mutagenesis, identification of oncogenes that collaborate with the transgene of interest in
carcinogenesis has been achieved, and remains a powerful tool in the identification of
such collaboration in the process of carcinogenesis [59].

Through the use of transgenic mouse models, much insight has been gained
regarding the function of Bmi-l and its role in collaboration with other oncogenes in
oncogenesis. As a result of retroviral insertional mutagenesis using the Moloney Murine
Leukemia Virus (MoMLV), Bmi-l was shown to collaborate with c-myc in a Err-myc
transgenic mouse model of B-cell lymphomagenesis [30, 52, 53, 80, 81]. As a result of
Bmi-l overexpression in this model, the incidence of B-cell lymphoma increased by 15%
[28, 52, 82, 83]. Furthermore, breeding transgenic lines expressing Bmi-l under the Eu
promoter (to target expression to the lymphoid compartment) with Err-myc transgenic
mice caused a marked increase in early onset lymphoma compared with Eu-myc or Eu-
Bmi-l transgenic mice alone. Over 75% of double transgenic mice developed B-cell
lymphoma by ten weeks of age, compared to zero in the single transgenic E u-myc or Eu-
Bmi-l groups. Alternatively, heterozygosity for Bmi-l results in inhibition of lymphoma

development in Eu-myc as a result of enhanced c-myc-induced apoptosis [30]. These data

Show that Bmi-l collaborates with c-myc in lymphomagenesis, and that elevated
expression of Bmi-l is required to achieve the full oncogenic capability of c-myc.

Guney et al. (2006) found that there is a canonical c-myc binding site within the
promoter of Bmi-l and that direct binding of the c-myc protein to this site occurs;
therefore, the bmi-l gene is a direct transcriptional target of c-myc. Furthermore, direct
downregulation of c-myc results in downregulation of Bmi-l, and inversely, upregulation
of c-myc results in upregulation of Bmi-l, with resulting up- and downregulation of p16,
respectively [19]. This provides firrther strong evidence that there is a direct relationship
and synergistic collaborative effect between these two genes. In addition, through its
effects on the Ink4a-Arf tumor suppressor locus, as a result of proviral insertion, Bmi-l
causes repression of p19 (p14) and p16, antagonizing the growth-inhibitory and
proapoptotic effects of c-myc overexpression [19, 30]. In this way, Bmi—l acts in concert
and as a result of c-myc overexpression to contribute to dysregulation of cell cycle
dysregulation and tumorigenesis.

Dysregulation of c-myc expression has been observed in several different human
cancers [30, 84] and has been shown to induce tumorigenesis in transgenic mouse models
[30, 85-87]. Given its collaborative role with c-myc in the induction of
lymphomagenesis, we were interested in the potential collaborative role Bmi-l may play
with oncogenic pathways commonly dysregulated in human breast cancer, such as occurs
through H-ras activation. While the incidence of ras mutation in breast cancer is less than
2% [88-90], overexpression of Ras protein and the downstream effectors is common due
to activation of growth pathways or aberrant estrogen dependent effects [91—93]

Overexpression of this oncogene has been shown to be present in up to 20% of human

breast cancer patients, and has been associated with concurrent overexpression of c-myc
and her2neu, often arising late in the course of disease [94]. Additionally, Ras
overexpression has been associated with a poor prognosis, [89, 94, 95], and
overexpression of downstream mediators has been correlated with regional lymph node
metastasis and more aggressive forms of breast cancer [89, 96, 97]. Interestingly, Bmi-l
is phosphorylated by 3pK, a MAPKAP kinase of the ERK and p38 pathways downstream
of Ras that regulates association of Bmi-l with chromatin [98], providing further

evidence of an interaction between these two oncogenes.

IV. Experimental Design to Study the Collaboration of H-Ras and Bmi-l

In order to evaluate the effects of Bmi-l overexpression alone and in combination
with H-ras overexpression, we developed an overarching hypothesis stating that
overexpression of Bmi-l in the mammary epithelium will promote proliferation and
invasion in vitro and primary and metastatic tumor formation in vivo through cooperation
with other cellular oncogenes, in this case H-Ras. To answer this question, several
specific aims were developed; specific aim 1 was to determine the effect of Bmi-l
overexpression alone and in combination with H-Ras overexpression on human
mammary epithelial cells in vitro. Specific aim 2 was developed to investigate this effect
in vivo, and specific aim 3 was developed to determine the effect of depression of Bmi-l
expression in established breast cancer cell lines that overexpress Bmi-l. To address
specific aims 1 and 2, an overexpression model was developed using a retroviral
expression vector to overexpress Bmi-l or H-Ras alone and in combination in MCFlOA

cells. These cells are normal human mammary epithelial cells that have undergone

spontaneous immortalization in culture through loss of the Ink4a-ARF tumor suppressor
locus by homologous recombination [99]. In this way, our model enables us to
investigate Ink4a-ARF independent mechanisms of Bmi-l function on oncogenesis. In
vitro, this cell line forms plates of polygonal cells with well differentiated cell-cell
junctions and has a relatively low rate of proliferation. In vivo, injection of these cells
into the mammary fat pad results in a small residual population of well differentiated
mammary ductal structures and epithelium, but does not result in tumorigenesis. These
cells are used to represent the behavior of normal mammary epithelial cells, and are
commonly used in the study of mammary gland biology [99]. Changes in cellular
morphology, proliferation, apoptosis in response to DNA damage, and invasion were
evaluated in vitro, and a severe combined immunodeficient (SCID) mouse xenograft
model was used to evaluate the cells for oncogenic and metastatic properties.

Specific aim 3 was addressed by depressing, or “knocking down” expression of
Bmi-l in established human breast cancer cell lines that overexpress the oncogene. We
hypothesized that depression of Bmi-l expression in these cells would lead to decreased
proliferation and invasion, and increased apoptosis in response to DNA damage in vitro,
and decreased tumorigenesis and metastasis in vivo. Suppression of Bmi-l expression
was established by using shRNA technology, and knock down was confirmed by western
blot. shRNA is commonly used as a means to determine functional effects of suppression
of genes of interest by effectively targeting and destroying endogenous mRNA. In this
way, expression of a gene of interest is suppressed by ablating or reducing its mRNA
signal. The following data show that Bmi-l collaborates with H-ras in determining an

aggressive cell phenotype in vitro and in vivo, with an increased metastatic propensity to

10

distant sites. These data have relevance because they suggest that in breast cancers with '
H-Ras overexpression, Bmi-l may play a pro-tumorigenic role, and breast cancers that
overexpress both oncogenes may be more aggressive and metastatic. Furthermore,

overexpression of Bmi-l may be associated with more aggressive forms of breast cancer.

11

CHAPTER ONE

The transcription repressor polycomb Bmi-l: Differential expression in histologic
subtypes of human breast cancer, wildtype mouse tissue, and genetically engineered

mouse models of breast cancer

Abstract: Bmi-l is a member of the polycomb group (PcG) of transcription repressors,
which have been demonstrated to regulate stem cell self-renewal and prevention of
senescence through inhibition of the cyclin-dependent kinase inhibitors p16Ink4a and
p19”. Stem cells are a critical subpopulation of cells in tissues that regulate the genesis
of progenitors and mature resident cells through proliferation. Overexpression of Bmi-l
has been reported in several human cancers, and has been found to induce telomerase
activity and immortalize human mammary epithelial cells. However, little is known about
the endogenous expression of Bmi-l in murine and human tissues, spontaneous and
induced tumors, genetically engineered mouse (GEM) models of human breast cancer,
and in histologic subtypes of breast cancer. Using human and mouse tissue microarrays,
we determined the expression of Bmi-l by immunohistochemistry in endogenous mouse
tissues, tumors generated from GEM models, and histologic subtypes of breast cancer. In
these studies, we show that there is differential expression of Bmi-l that correlates with
histologic phenotype in certain human breast cancers, which suggests that Bmi-l may be
involved in the pathogenesis of these tumors. Furthermore, we show that some aggressive
genetically engineered mouse models of human breast cancer have upregulation of Bmi-

1, suggesting that Bmi-l may be involved in more aggressive and metastatic forms of

12

breast cancer. Lastly, we show that there is differential expression of endogenous Bmi-l

in mouse tissues, as well as among the different stages of estrus, pregnancy, lactation, and

involution that implies a role of Bmi-l in stem cell activation, proliferation, and

inhibition of apoptosis at different stages of development of the mammary gland.
Introduction

Several lines of evidence exist that a variety of cancers including breast cancer
may result from transformation of both stem and progenitor cells during carcinogenesis,
and that cancer may be a stem cell disorder [27, 31, 100-107]. Transformed stem cells, or
‘cancer stem cells’ are defined as cancer cells that have the ability to self-renew to give
rise to a malignant stem cell as well as undergo differentiation to give rise to
phenotypically diverse cells within a tumor [27, 108]. The cancer stem cell hypothesis
suggests that a small population of cells within a tumor have stem cell properties that
maintain malignant populations [109]. In addition, tumors that are clonally derived yet
display markers for multiple cellular lineages provide evidence for a multipotent cancer
stem/progenitor cell of origin [103, 110]. Maintenance of normal stem cell and cancer
stem cell populations has been shown to be regulated by the transcription repressor, B-
cell Moloney murine leukemia virus insertion region-1 (Bmi-l) [31, 40, 111, 112].

Bmi-l belongs to the polycomb group of transcription factors, critical for the
regulation of gene expression during development [18]. By binding to cis-acting silencer
regions called polycomb response elements (PRE), Bmi-l acts to modify chromatin
structure and regulate transcription of Homeobox (Hox) genes, which are necessary for
the establishment of proper body morphogenesis [28, 113]. The Bmi-l gene has

conserved homology across several species, and the gene structure in the human is very

l3

similar to that of the mouse: 92.4% homologous at the nucleotide level in the open
reading frame (ORF) [28]. Additionally, the ORF codes for a nuclear protein of 326
amino acids that shares 98% homology with the mouse Bmi-l protein at the amino acid
level [28].

Bmi-l has been shown to be a key regulator of the self renewal of hematopoietic
stem cells [114, 115], central and peripheral nervous system neuronal stem cells, [31,
116], mesenchyrnal stem cells [117], and mammary stem cells [35, 103]. Several recent
studies have suggested a link between Bmi-l and mammary carcinogenesis [29, 118].
Given the important firnction of Bmi-l in regulating‘the stem cell pool in multiple tissues
as well its associations in tumorigenesis, defining the role of Bmi-l in regulating the
mammary epithelial stem cell compartment and understanding of the role this gene plays
in mammary carcinogenesis will be paramount in shedding light on the pathophysiology
of the complex disease of human breast cancer. Here we report the differential
endogenous expression of Bmi-l in various human breast cancer histologic subtypes,
wildtype mouse tissues, and in genetically engineered mouse (GEM) models of

mammary cancer that recapitulate genetic mutations found in the human disease.
Materials and Methods

Tissue microarrays. To characterize the immunohistochemical labeling of Bmi-l in

human breast cancer tissues, tissue microarray slides were obtained from two different

commercial sources (US Biomax, Inc., Rockville MD and Super Bio Chips, San Diego,

CA). Prognostic marker data including estrogen and progesterone receptor status and p53

14

status were provided for one of these (US Biomax, Inc., Rockville MD), as was survival
information. To characterize Bmi-l immunoreactivity in GEM model mammary tumors,
two mouse-based tissue microarrays developed by the NCI Mouse Models of Mammary
Cancer Collective were used that include tissue sections 80f mammary gland development,
GEM model mammary tumors, and normal control tissues.

Human arrays: Routine pathological methods were employed to examine histopathology
of carcinoma simplex (n=35), infiltrative ductal carcinoma (n=33), scirrhous mammary
carcinoma (n=9), atypical medullary carcinoma (n=10), infiltrating lobular carcinoma
(n=3), Signet ring carcinoma ,(n=1), and solid papillary carcinoma (n=l).

Mouse arrays: Tissue sections including various stages of mammary gland development
(proestrus [n=10], estrus [n=10)], diestrus [n=20], metestrus [n=10], pregnancy [n=15],
lactation [n=15], involution [n=40]), spontaneous mammary tumors (n=10), GEM model
mammary tumors, (MMTV—neu [n=24], MMTV-wnt [n=23], MMTV—myc [n=12], PyMT
[n=38], C(1)3-SV40-Tag [n=8], BRCAl [n=15], Notch 4 [n=9]) and normal mouse tissue
(liver [n=2], brain [n=3], spleen [n=2], thymus [n=2], salivary gland [n=8], lung [n=2],
heart [n=2], .skeletal muscle [n=2], small intestine [n=2], and kidney [n=2]) were

examined.

Histology and immunohistochemistry. All sections were examined using routine light
microscopy, and upon confirmation of the diagnosis, samples were prepared for
immunohistochemistry. Paraffin-embedded sections were routinely deparaffinized and
rehydrated, followed by steaming in sodium citrate buffer pH 6.0 for antigen retrieval.

Endogenous peroxidase activity was blocked with 3% hydrogen peroxidase in methanol,

15

and following 5% bovine serum albumin (BSA) block, a polyclonal rabbit anti-Bmi-l
antibody (Abgent, San Diego CA) at a dilution of 1:300 diluted in PBS containing 3%
BSA was applied and tissues were incubated at 4 °C overnight. Following avidin—biotin
blocking, sections were incubated in goat anti-rabbit secondary antibody (Vectashield,
Burlingame CA) at a dilution of 1:10,000, and amplification of primary antibody binding
was achieved using the labelled avidin-biotinylated enzyme complex (ABC) technique.
Following washing, tissue sections were visualized using diaminobenzidine (DAB)
chromogen substrate, and counterstained with Mayer’s hematoxylin, washed in double
distilled water, followed by immersion in bluing solution (Scott’s tapwater) for 2 minutes
followed by distilled water wash. The sections were then dehydrated through graded
alcohols (70%, 95%, 100%) and mounted with coverslips following xylene immersion.
Negative controls were obtained by substitution of the primary antibody with buffer
solution. Selected areas were photographed using an Olympus DP70 (Olympus) digital
camera, and labeling intensity measured as tinctorial change in the Cytoplasm and nucleus
was entered into a MicrosoftTM Excel Spreadsheet. Labeling intensity was graded based
on a 0-3 scale (0 = no labeling, l = mild labeling, 2 = moderate labeling, 3 = marked

labeling). One way ANOVA was performed on all samples.

Quantitative real-time polymerase chain reaction (QRT-PCR).
RT-PCR was performed on brain, thymus, spleen, C(1)3-SV40-Tag tumor, PyMT tumor,
MMTV-neu and MMTV-myc tumors. Brain, thymus, and spleen were chosen because

they have previously been shown to express varying levels of Bmi-l [30, 32]. Tissues

l6

were suspended in liquid nitrogen and pulverized with a mortar and pestel, followed by
mechanical grinding using a tissue macerator. For RNA samples for RT-PCR, 1 mL of
Trizol reagent (Invitrogen, Carlsbad CA) was added to each sample, and RNA was

extracted following the manufacturer’s protocol.

Results
Endogenous Bmi-l protein is differentially expressed in mouse tissues. Normal
mouse tissues examined histologically and immunohistochemically were brain, liver,
spleen, thymus, salivary gland, lung, heart, skeletal muscle, small intestine, kidney, testis,
ovary, uterus, and mammary gland (Figure l). Bmi-l was expressed in splenic
lymphocytes in the marginal zone of the lymphoid follicles and in megakaryocytes. Cells
that did not stain in the spleen include endothelial and vascular smooth muscle, smooth
mucle in the splenic cords, and lymphocytes around the periarteriolar lymphocyte sheath
(PALS). In the thymus, Bmi-l was expressed in lymphocytes within the medulla and
outer cortex, but was absent in lymphocytes immediately surrounding cortical blood
vessels. Similarly, by RT-PCR, Bmi-l was expressed in the spleen and thymus (Figure
2). Immunohistochemically, Bmi-l expression was present in the granule cell layer and
Purkinje cell layer of the cerebellum and in occasional neurons of the cerebrum, but was
absent from the white matter. Bmi-l was expressed diffusely within hepatocytes, and in
sections of kidney, Bmi-l was expressed diffusely within renal tubular epithelium,
whereas occasional glomerular mesangial cells had weak positive immunoreactivity.
Sections of medulla were not included in the tissue microarray and were thus not

examined. In sections of lung, there was scant Bmi-l expression, limited to the

17

bronchiolar epithelium and occasional alveolar epithelial cells, but most alveolar
epithelial cells did not have immunoreactivity to Bmi-l. Small intestinal sections showed
Bmi-l expression within the intestinal epithelium, with less immunoreactivity within the
underlying lamina propria. Bmi-l was expressed within. ovarian follicular epithelium and
corpora lutea, but was absent to scant within the surrounding theca cells and vascular
endothelium. Within the testis, positive immunoreactivity was present within the
cytoplasm and nucleus of the germ cell layers and Sertoli cells, but was weak or absent
from interstitial Leydig cells. Expression of Bmi-l in sections of uterus was limited to the
endometrial epithelium, and was absent in the surrounding smooth muscle layers. In
salivary gland, Bmi-l was expressed diffusely within salivary epithelial cells but was

absent from vascular endothelial cells.

Bmi-l is differentially expressed within the mouse mammary gland depending on
the stage of estrus, gestation, lactation, and involution. Within normal, non-pregnant,
non-lactating mammary gland, expression was limited to the ductular epithelium.
Expression was increased during proestrus, then decreased during estrus, and increased
again reaching maximum expression during metestrus, and then decreased again during
diestrus. There was a statistically significant difference in Bmi-l expression between
stages of proestrus, metestrus, and diestrus (*p < 0.05) (Figure 3A). During early
pregnancy, Bmi-l was expressed at moderate to high levels through day 10, and then
waned by day 18-21. During lactation, Bmi-l expression was moderate early (day l) and
remained steady through day 5 to day 10 (Figure 3B). At the time of weaning (involution

day 0), Bmi-l expression remained high in the mammary epithelium, but by day l of

18

involution, Bmi-l levels dropped dramatically to almost undetectable, and remained very
low until about one week after the onset of involution. By two weeks involution, Bmi-l
levels were increasing back to proestrus levels. There was a statistically significant
elevation in expression (*p < 0.01) during stages of lactation compared to pregnancy and

involution (Figure 3C).

Bmi-l overexpression is associated with aggressive GEM models of human breast
cancer. In GEM mouse models of human cancer, Bmi-l overexpression by real time
quantitative polymerase chain reaction (QRT-PCR) was associated with the C(1)3Tag
and polyoma middle T antigen models, two particularly aggressive GEM models that
have a rapidly progressive and metastatic disease course. However, the c-myc and
her2neu (cErbB2) models, two more slowly progressive GEM models, showed lower
levels of Bmi-l expression (Figure 2). Immunohistochemically, of the GEM mouse
models examined (C3(l)SV40 T/t antigen, PyMT, c-myc, her2neu, BRCAl knockout
(KO), Notch, Wnt-l and MMTV-induced), there was increased immunolabeling in each
model examined, with little variation in the staining intensity (Figure 4). Of the GEM
models examined, the BRCA1 KO model expressed the highest levels of Bmi-l by

immunohistochemistry (p < 0.01) using the grading scheme for labeling intensity.

Bmi-l is differentially expressed in histologic subtypes of human breast cancer and
high expression correlates with medullary breast carcinoma. In human breast tumor
tissue, there was differential Bmi-l expression (Figure 5), which was significantly

correlated with histologic tumor subtype by immunohistochemistry based upon a grading

l9

scale of 0-3 (0 = no labeling, 1 = mild labeling, 2 = moderate labeling, 3 = marked
labeling). Of the histologic subtypes examined, atypical medullary carcinoma expressed
the highest level of Bmi-l, followed by carcinoma simplex, infiltrating ductal carcinoma,
and scirrhous carcinoma in descending order (Figure 6) (*p < 0.05, "p < 0.01). There

was no evidence of a significant correlation between luminal and basal subtypes.

Discussion

This report demonstrates the level of endogenous Bmi-l expression in wild type
mouse tissues, expression in several genetically engineered mouse models of breast
cancer, and expression in human breast cancer samples. From these data, we have shown
that expression of Bmi-l is associated with different stages of the estrus cycle,
pregnancy, lactation, and involution in the mouse, aggressive subtypes of GEM models,
and certain histologic subtypes in spontaneous human breast cancer based upon tumor
morphology.

The differential endogenous expression of Bmi-l in wild type mouse tissues is
consistent with its role in the maintenance of stem cell populations through self-renewal
[105]. Self-renewal is a process by which a stem cell pool maintains its numbers through
symmetric and asymmetric division. In symmetric cell division, at least one of the
progeny is identical to the initial stem cell; in asymmetric stem cell self-renewal, one of
the two progeny is identical to the initial stem cell, whereas the other cell is a committed
progenitor cell, which undergoes cellular differentiation [103, 105]. Through the tightly
regulated process of self-renewal, stem cells are able to function over the lifespan of the

host [27, 114].

20

We have shown that in many tissues, Bmi-l expression is correlated with cell
populations with more capacity for regeneration or proliferation, such as intestinal villous
epithelium, hepatocytes, and renal tubular epithelium, whereas in tissues with lower
regenerative capacity, such as skeletal muscle and brain, either few cells express this
protein, or only those cells that function as a stable pool for adult tissue stem cell
progenitors express it, as has been reported in the subventricular zone (SVZ) of the brain
[28]. It has been shown that Bmi-l‘l' have a reduced number of neurons in the external
granular cell layer of the cerebellum [40]. In the normal developing post-natal mouse,
signals from the Purkinje cell layer stimulate a proliferation of cerebellar granule neuron
progenitors in this layer as a result of Sonic Hedgehog (Shh) signaling [119, 120]. As a
result, granule neuron progenitors differentiate and migrate through the molecular layer
into the internal granular layer where they remain for the life of the host [121]. Thus, the
lack of Bmi-l in the Purkinje cell layer leads to a loss of signaling to the cerebellar
granular cell neurons, ultimately resulting in cerebellar aplasia.

We found that the spleen and thymus expressed high levels of Bmi-l. These
findings are expected, as these are sites of centers of B and T lymphocyte genesis and
propagation, and Bmi-l has been. reported to be highly expressed in these tissues and in
certain types of leukemia in humans [31, 112, 122]. The germ cell layers of the
seminiferous tubules and layers of the ovarian follicles have particularly prominent Bmi-
1 expression, consistent with the germ/stem cell enrichment of these organs. Similarly, in
the cycling, pregnant, lactating, and involuting mammary gland there is differential Bmi-
1 expression supporting its firnction in the maintenance of stem cell pools and its

proliferative and anti-apoptotic effects in development.

21

Within normal, non-pregnant, non-lactating mammary gland, expression is
limited to the ductular epithelium. During the estrus stage, expression is minimal within
the mammary ductular epithelium, and increases during proestrus, reaching maximum
expression during metestrus, and then decreases during diestrus (Figure 3A).
Interestingly, it has been reported that peak BrDU labeling, indicating the highest degree
of proliferation, occurs during metestrus and early diestrus [123, 124]. Increased Bmi-l
expression may be important during these stages of the estrus cycle when proliferation of
the ductular epithelium and lobuloalveolar formation is critical.

During early pregnancy, Bmi-l is expressed at moderate to high levels through
day 10, and then wanes by days 18-21. Increased expression during early to mid
pregnancy supports the role of Bmi-l in the mobilization of stem cells and progenitors to
proliferate during states of physiologic hyperplasia. Bmi-l is expressed at moderate to
high levels throughout lactation, again supporting a role in proliferation and mobilization
of stem cell pools in the mammary gland (Figure 4B).

Bmi-l is highly expressed at the time of weaning, and drops dramatically during
the first day of involution. Expression remains minimal through day 5 of involution, but
eventually increases again to proestrus levels by two weeks following the start of
involution (Figure 4C). Consistent with its role in preventing cell cycle arrest and
apoptosis, these data suggest that during periods of time when the mammary gland is
proliferating or mobilizing stem and progenitor cells to differentiate into luminal milk-
producing cells, Bmi-l is upregulated. Then, when the gland needs to involute following
weaning, the massive apoptosis that characterizes this stage correlates with a drop in

Bmi-l expression to minimal levels. Taken together, these data show that Bmi-l

22

expression correlates with cell populations with a greater potential for regeneration and
proliferation suggesting a function in mobilization of stem cell pools. Additionally, the
loss of Bmi-l expression during involution is consistent with the role of Bmi-l as an anti-
apoptotic mediator. These data are consistent with the function of Bmi-l in mobilizing
stem cell populations during growth and differentiation, and .in preventing cell cycle
arrest and apoptosis. Further correlation of Bmi-l expression with markers of cellular
proliferation (Ki67), apoptosis (Bcl-2, Bad), and mammary gland differentiation
(estrogen receptor [ER], whey acidic protein [WAP], and beta-casein) will be important
in characterizing the role of Bmi-l in mammary gland proliferation, apoptosis, and
differentiation.

By examining Bmi-l expression by quantitative RT-PCR in GEM models of
human breast cancer, we have determined that high levels of expression are correlated
with two particularly aggressive models, the C3(l)SV40 T/tag model and the PyMT
model. Each of these models is extremely useful in the study of the human disease
because each possess molecular events that culminate in the transformation of mammary
epithelial cells and subsequent metastatic disease in the host. The C3(1)SV40 T/t antigen
model is characterized by progressive development of mammary lesions leading to
invasive carcinoma formation through inactivation of the tumor suppressors p53 and Rb
[125-127]. These mice develop lesions of atypical ductal hyperplasia by 8 weeks of age,
progressing to mammary intraepithelial neoplasia (MIN) by 12 weeks, which is
histologically similar to human ductal carcinoma in situ (DCIS). After 16 weeks of age,
these animals develop invasive carcinomas, and approximately 15% of these animals

develop metastases to the lungs in the FVB/N background, and 50% in the FVB/N X

23

Sv129 hybrid background [127-129]. This model is very useful‘in the study of the human
disease because it progresses through multiple stages from precancerous to invasive
carcinoma with pulmonary metastasis in a predictable manner, and represents a basal
subtype of human breast cancer. The basal subtype of breast cancer is frequently very
poorly differentiated, estrogen receptor (ER) and progesterone receptor (PR) negative,
and associated with a poor prognosis [130, 131]. Interestingly, in our
immunohistochemical evaluation of the various GEM models of mammary cancer, one
such basal model, the BRCA1 KO model, consistently expressed high levels of Bmi-l.
The BRCA-l mutation is a well characterized mutation in humans that is responsible for
25-30% of familial cases of breast cancer [132]. Women harboring this mutation have a
56-84% estimated cumulative lifetime risk of developing breast cancer [132-134].
Histologically, BRCA-l associated human breast carcinomas are poorly differentiated
tumors that often have morphologic features of medullary or atypical medullary
carcinoma [132, 134-136]. It has been proposed by F oulkes (2004) that BRCAl functions
as a mammary epithelial stem cell regulatory gene in the transition from the primitive
basal to the glandular epithelial phenotype. This basal phenotype is also consistent with
the genotype of medullary carcinoma by molecular profiling [137]. The correlation of
Bmi-I with these Mo basal type GEM models (C(1)3-Tag and BRCA1 KO) suggest that
expression of Bmi-l may promote a more aggressive and less differentiated phenotype,
possibly related to a transformed stem cell phenotype. Correlation of basal markers such
as cytokeratin 5/6 and 14, smooth muscle actin (SMA), and negative ER, PR, and
Her2neu status in these GEM models will be important in further characterizing the role

of Bmi-l with the pathophysiology of the basal phenotype.

24

We also found elevated levels of Bmi-l mRNA by real time QRT-PCR in
MMTV-PyMT GEM model tumors. MMTV-PyMT transgenic mice express the mouse
polyomavirus middle—T antigen (PyMT) under the control of the mouse mammary tumor
virus (MMTV) promoter [138]. These animals develop multifocal mammary tumors that
have a high incidence of metastasis to the lungs [139]. Similar to the C3(1)SV40 T/t-
antigen model, these animals develop mammary lesions that develop progressively
through distinct stages from in-situ proliferative precancerous lesions to invasive
carcinoma with pulmonary metastasis. More importantly, this model shares morphologic
similarities with the human disease, as well as several biomarkers that are indicative of a
poor prognosis in humans such as the loss of estrogen and progesterone receptors,
integrin-Bl expression, and expression of ErbBZ/Neu and cyclinDl [140]. These
characteristics make this model particularly useful when studying the biological
progression of breast cancer in humans. The association of a high level of Bmi-l
expression with these particularly aggressive transgenic models of human breast cancer
suggest that overexpression of Bmi-l may play a role in the acquisition of an aggressive
breast cancer phenotype. Study of the molecular mechanisms that occur between the
stages of precancerous to invasive metastatic carcinoma may bear important light on the
mechanism of Bmi-l in the pathogenesis of these models as they relate to the human

disease.

In general, human breast cancer can be divided into familial and sporadic causes
[141]. Within sporadic and familial breast cancers, characterization is based on several
factors, including histologic subtype. For invasive breast cancer, infiltrating ductal

carcinoma is the most common histologic type of breast cancer, accounting for

25

approximately 70% of cases [132, 141]. Scirrhous carcinoma is a variant of infiltrative
ductal carcinoma that is characterized by cords and packets of tumor cells embedded in
an abundant desmoplastic matrix [142]. Carcinoma simplex is a subtype of infiltrative .
ductal carcinoma that is characterized by solid sheets 'of poorly differentiated epithelial
cells with large vesicular nuclei and prominent nucleoli, ofien forming clusters with an
indistinct cell border (syncitial growth pattern), and a broad pushing margin with a
prominent lymphocytic infiltrate [132, 137, 143, 144]. The diagnosis of atypical
medullary carcinoma is made when there is less lymphocytic infiltrate, or there is an
invasive margin in part of the tumor [136]. Medullary or atypical medullary carcinomas
are the predominant histologic types seen in patients with BRCAl mutation, and they are
typically higher grade tumors when associated with the BRCA1 mutation [134, 136]. Our
immunohistochemical examination of Bmi-l in this series of breast tumor specimens
showed that Bmi-l expression was statistically correlated with the histologic subtype of
atypical medullary carcinoma (Figure 6). This supports our previous findings of increased
Bmi-l in the BRCA1 KO mouse model, and suggests that Bmi-l overexpression is
associated with a more aggressive basal tumor phenotype possibly related to
transformation of a primitive precursor, such as a stem cell.

In summary, we showed that Bmi-l expression is differentially expressed in wild
type moues tissues, is associated with more aggressive GEM models of human breast
cancer (PyMT, C3(1)SV40 T/t-antigen, and BRCA1 K0), and is associated with the
BRCA1 tumor phenotype, medullary/atypical medullary carcinoma. These results suggest
that 1) Bmi-l functions in normal development to promote proliferation and prevent

apoptosis in the mammary gland and 2) Bmi—l plays a role in the promotion of more

26

aggressive types of breast cancer, possibly through the promotion of stem or progenitor
cell pathways. It has recently been shown that BRCAl is involved in human mammary
stem/progenitor cell fate, and that loss of BRCA1 may result in the accumulation of
genetically unstable breast stem cells that aretargets‘ for further transformation [145].
Furthermore, BRCAl tumors contain mammary epithelial cells with a stem cell
immunophenotype (CD44: CD24', CD133I) with cancer stem cell characteristics [146].
Further evaluation of Bmi-l with the basal subtype of breast cancer and in tumors
associated with stem cell pathways associated with the basal phenotype (Wnt, Notch,
BRCAl KO), would be important in determining the role of Bmi-l in the pathogenesis of

mutations leading to transformation of mammary epithelial cells.

27

APPENDIX A

Figure l: Endogenous expression of Bmi-l in wildtype mouse (FVB) tissue. Anti-Bmi-l
polyclonal antibody, hematoxylin counterstain (Bar = 100 pm).

 

28

Figure 2: Endogenous Bmi-l expression in wildtype mouse tissues and GEM tumors by
QRT-PCR normalized to brain. There is a five-fold and an eight-fold increase in Bmi-l
mRNA expression in the thymus and spleen, respectively, compared to the brain. The

‘ C(1)3-SV40-T/t-antigen and MMTV-PyMT GEM models have 15 and four-fold
increases in Bmi-l mRNA expression, respectively, normalized to expression levels in
brain.

16

14.

—l
'9

Fold Change
co 3

 

Brain Thymus Spleen her2neu PyMT c-myc C3Tag

29

Figure 3: Bmi-l expression during phases of the estrus cycle, pregnancy, and lactation.
3A: Bmi-l expression during phases of the estrus cycle, immunohistochemical
expression (above) and graphical representation (below) indicating a significant
difference in Bmi-l expression between stages of proestrus, metestrus, and diestrus (a vs.
c and b vs. c, p < 0.05). 3B-C: Bmi-l expression during stages of pregnancy, lactation,
and involution. Above: immunohistochemical expression of Bmi-l (Bar=100 um).
Below: graphic representation illustrating statistically significant elevation in expression
(0 vs. b and b vs. c, p < 0.01) during stages of involution (0 = no labeling; l = mild
labeling; 2 = moderate labeling; 3 = marked labeling).

 
  

’iI‘AIFII I ‘ ."' {‘v' '
. A ,> .. a» ., -.
(«”ch ’ "‘ I J:- '3‘,” .4 3’

v ‘0 ‘r

O

m proestrus
m estrus
E metestrus
[HID diestrus

Average staining
intensity

proestrus estrus metestrus diestrus

30

Figure 3 (continued)

B. Pregnancy Lactation

         

 

31

Figure 3 (continued)

 

32

Figure 3 (continued)

 

 

2 1
36:35
95022 ommho><

«>2. S. 5320?:
get N. 53202:
gnu m co=:_o>_.__
gee m :o=:_o>:_
>3 F co=:_o>:_
goo c 53202:
m>mo 2. cozfiom.
goo m .5358.
get To cozflofi
gap 5.2. moi
gap Saw moi

33

Figure 4: Bmi-l expression in genetically engineered mouse models of human breast
cancer. Top: anti-Bmi-l polyclonal antibody, hematoxylin counterstain (Bar = 100 pm).
Each GEM model expresses Bmi-l diffusely within the nucleus and cytoplasm, with
some minor variation in expression levels between tumors. Bottom: graphic
representation of Bmi-l expression in GEM tumors, indicating statistically significant
difference in Bmi-l expression in BRCAl tumors compared to her2neu, p53 KO, Wnt,
and Notch models, and between PyMT and her2neu models (a vs. b, p < 0.05; d vs. b, p <
0.001; d vs. c and e, p < 0.01, d vs]: p < 0.05) (0 = no labeling; 1 = mild labeling; 2 =
moderate labeling; 3 = marked labeling).

:9 " .
. ..wl'r': ,. ‘95.

c
.0
-...

-‘
..

\
.’“-n

. I . ‘; 1 1;. I
. '-~.I‘ '
c 1&Qigdfqlb. .

, ‘ :7.

I. , r p.‘.
. as»);

 

Figure 4 (continued)

2.75-
3. 2.50-
2 2.25-
2.00‘
1.75-
1.50-
1.25-
; 1.00-
0.75-
0.50-
0.25-
0.00-

te

mg m

abel

Averag

 

 

PyMT C3Tag Neu Myc Notch BRCA Wnt p53

35

Figure 5: Bmi-l expression in human breast cancer samples. A; Carcinoma simplex, B;
atypical medullary carcinoma, C; infiltrative ductal carcinoma, D; scirrhous carcinoma.
Anti-Bmi-l polyclonal antibody, hematoxylin counterstain (Bar = 100 um). Each tumor
expresses Bmi-l within the nucleus and cytoplasm, with some variation in staining
intensity between tumors.

  

Histopathol y (

nfitf”

  

   

     
    

 

.- é:- $.75? 2.};
" '. r: .- ‘3‘
$1.3" ~_ ”<7?" ." -f‘\
. w <‘x-A’g‘w:
‘3", .’ -'\ '7 ’5 v

36

Figure 6: Bar graph illustrating the relative levels of Bmi-l expression in histologic
subtypes of human breast cancers by immunohistochemistry. There is a significant
increase in Bmi-l protein expression associated with the atypical medullary carcinoma
phenotype (0 = no labeling; 1 = mild labeling; 2 = moderate labeling; 3 = marked
labeling) (0 vs. c or d, p < 0.01; b vs d, p < 0.05).

2.0-

_.L —L
m 'o:
l 1

Average labeling intensity
.9
CD

0.4 ‘

 

 

0..
g gm 2 mg
=(u X 3
35 as 09 go
$0 038 .Eo- t.E
EE .2'6 9g 59
39 *6; (DU) (08
.98 £8 0
o. t:
3‘ E
<

37

CHAPTER TWO

Hoenerhoff, M.J.*, Datta, S.*, Bommi, P., Sainger, R., Guo, W., Dimri, M., Band, H.,
Band, V., Green, J.E., Dimri, GP. (2007) Bmi-l cooperates with H-Ras to transform
human mammary epithelial cells via dysregulation of multiple growth regulatory

pathways. Cancer Research, 67: 10286-10295.

*These authors contributed equally to this work.

38

CHAPTER TWO

Bmi-l cooperates with H-Ras to transform human mammary epithelial cells via

dysregulation of multiple growth regulatory pathways

Abstract: Elevated expression of Bmi-l is associated with the malignant progression of
many cancers including breast cancer. Here we examined the oncogenic potential of Bmi-
l oncoprotein in MCF 10A, a normal but spontaneously immortalized strain of human
mammary epithelial cells (HMECs). Bmi-l overexpression alone in MCF 10A cells did
not result in oncogenic transformation. However, Bmi-l co-overexpression with activated
H-Ras (RasGlZV) resulted in efficient transformatibn of MCFIOA cells in vitro. When
injected into nude mice, Bmi-l and H-Ras overexpressingMCFIOA cells formed poorly
differentiated carcinomas with spindle cell features. Although, early passage H-Ras alone
expressing MCF 10A cells were not transformed, late passage 'H-Ras expressing cells
exhibited features of transformation in vitro. Injection of these late passage H-Ras
expressing MCFIOA cells formed carcinomas with leiomatous, hemangiomatous, and
mast cell components, which were quite distinct from Bmi-l and H-Ras induced tumors.
Bmi-l and Ras co-overexpression in MCFlOA cells also induced features of Epithelial-
Mesenchymal Transition (EMT). Examination of various growth regulatory pathways
suggested that Bmi-l and H-Ras co-expression leads to activation of AKT, upregulation
of CDK4 and cyclin D1, and downregulation of p53 expression in MCF 10A cells.
Furthermore, compared to control cells, late passage H-Ras expressing, and Bmi-l and

H-Ras co—expressing MCF 10A cells showed an attenuated DNA damage response due to

'39

substantially less p53 phosphorylation at ser-15 and ser-37 residues. Thus, our studies
suggest that Bmi—l overexpression together with H-Ras transforms HMECs, and

promotes breast oncogenesis by deregulating multiple growth regulatory pathways.

Introduction

Polycomb group (PcG) proteins play an important role as epigenetic gene silencers
during development [147]. In addition to their role in development, these proteins were
recently reported to be overexpressed in various human cancers such as malignant
lymphomas and various other solid tumors [148]. In particular Bmi-l oncogene is
overexpressed in a number of malignancies such as mantle cell lymphoma [149], B-cell
non-Hodgkin lymphoma [25], myeloid leukemia [122], non-small cell lung cancer [41],
colorectal cancer [44], breast and prostate cancers [44, 48]. In addition, we have recently
reported that Bmi-l is also overexpressed in nasopharyngeal carcinoma and oral cancer
[37, 38]. Apart, from its role in oncogenesis, Bmi-l has been shown to be required for
self-renewal of hematopoietic stem cells and neural stem cells [31, 32, 115, 116]. In
addition, recently it was shown that Bmi-l regulates self-renewal of normal and cancer
stem cells in breast, and that modulation of Bmi-l expression in mammosphere-initiating
cells alters mammary development in a humanized nonobese diabetic-severe combined
immunodeficient mouse model [35, 103].

The polycomb proteins which are overexpressed in tumors, such as Bmi-l and
EZH2, appear to deregulate cell cycle progression [148]. Recent studies using in vivo

mouse and in vitro cell culture models have shown that Bmi-l regulates the expression of

4A A F
6INK 9 R

INK4A/ARF locus, which encode two important tumor suppressors p1 and pl

40

(p14ARF in human) [18, 150]. By downregulating p16lNK4A and pl9ARF, Bmi-l can
potentially regulate pl6-pr and p53-p21 pathways of senescence [151]. Indeed,
overexpression of Bmi-l bypasses senescence in human and rodent fibroblasts, human
mammary epithelial cells (HMEC), nasopharyngeal. epithelial cells and normal oral
keratinocytes [18, 29, 37, 150, 152]. Along these lines, we have recently reported that
.Bmi-l downregulation by another PcG protein Mel-18, and Bmi-l knockdown using an
RNA interference (RNAi) approach, induce premature senescence via upregulation of
p16INK4A [153]. Interestingly, in normal human oral keratinocytes (NHOK), the
mechanism of extension of replicative life span by Bmi-l does not appear to be via

6‘NK4A suggesting the possible role of other unidentified targets of

downregulation of p1
Bmi-l that are involved in cellular proliferation [152].

Our recent data suggest that independent of its effect on p16Ink4a, Bmi-l also
regulates AKT activity [153], which may play a role in cell proliferation and survival of
breast cancer cells. It is thought that the precursor cells for breast cancer are p16INK4A
negative due to promoter methylation and silencing [154], suggesting that overexpression
of Bmi-l in p16INK4A negative tumors may contribute to oncogenesis via p16INK4A-
independent growth regulatory pathways. Here, we examined the oncogenic potential of
Bmi-l in an immortal but untransformed HMEC line MCF 10A, which does not express
p16INK4A due to deletion of the INK4A/ARF locus [155]. In addition to p161NMA and
p14ARF deletion, this cell line also does not express plSlNMB [155], which is another

CDK4 inhibitor and a potential tumor suppressor. Using in vitro cell culture and in vivo

mouse models, we show that overexpression of Bmi-l alone is not sufficient for

K4A INK4B
6'N 5

oncogenic transformation of immortal HMECs, which lack pl and

.p1

41

pl4ARF. However, the combined overexpression of the G12V mutant of H-Ras and Bmi-l
transformed HMECs in culture as determined by transformation assays. Furthermore,
orthotopic injection of cells co-overexpressing Bmi-l and activated H-Ras resulted in the
formation of poorly differentiated and invasive tumors in severe combined

immunodeficient (SCID) mice.

Materials and Methods
Cells, cell culture, expression vectors, retrovirus production and infection of
HMECs. MCFlOA and MCF lOA-derived cell lines were cultured as described [29]. A
retroviral vector overexpressing Bmi-l has been described earlier [29, 153]. A retroviral
vector pMSCV—Ras expressing H-Ras G12V mutant was constructed by subcloning
cDNA of H-Ras from pcDNA3.1 obtained from UMR cDNA Resource Center
(University of Missouri, MO). Stable cell lines expressing gene(s) of interest were
generated by infection of the retroviral vector(s) expressing a particular gene and
selecting cells in either puromycin, G418 or hygromycin as described [29, 153]. The
retroviruses were produced by transient transfection of the retroviral vector together with

pIK packaging plasmid into tsa 54 packaging cell line as described [29, 153].

Antibodies, western blot analysis, immunolabeling Matrigel, and soft-agar and
wound healing assays. Bmi-l was detected using either F6 mouse monoclonal antibody
(mAb) from Upstate Cell Signaling Solutions (Charlottesville, Virginia), or lH6BlOG7
mAb from Zymed (S. San Francisco, CA). For the analysis of AKT pathway, antibodies

to phospho-AKT 1/2/3 (Ser-473) (sc-7985-R), AKT-l (B-l, sc-5298) and AKT-2 (F-7,

42

sc-5270) were obtained from Santa Cruz Biotech (Santa Cruz, CA). Among other
antibodies, CDK4 (C-22, sc260), Cyclin D1 (A-12, sc-8396), H-Ras (F-235, sc-29), p21
(F-5, sc-6246), p53 (DO-l, sc-126), p53-Ser15 (sc-11764-R), PUMA (FL-193, sc-
28226), Bax (6A7, sc-23959), ERK (C-16, sc-93), phospho-ERK (E-4, sc-7383), and
PRAK (A-7, sc-46667) antibodies were obtained from Santa Cruz Biotech, Santa Cruz,
CA. p53-Ser-37 mAb was obtained from Cell Signaling Technology (Beverly, MA).
Vimentin, F ibronectin and E-Cadherin mAbs were obtained from BD Transduction -
Laboratories (Franklin Lake, NJ). B-Actin and B-Tubulin mAbs were obtained from
Sigma-Aldrich (St. Louis, MO), and DSHB (University of Iowa, IA) respectively. To
determine the AKT and ERK activity in synchronized cells, MCF 10A cells were growth
factor deprived using D3 medium [156] for 48 hrs and stimulated for 40 min by addition
of D medium, which contains 12.5 ng/ml epidermal growth factor (EGF) [156]. Western
blot analyses of total cell extracts using antibodies that detect various proteins were
performed as described [29, 153]. Irnmunostaining for EMT markers such as E-Cadherin,
fibronectin and Vimentin, and soft-agar, matrigel and wound healing assays were

performed as described [29, 153].

Mice and in vivo experiments. For mammary fat pad injection experiments, four cohorts
of ten, six-week-old, female SCID mice each were used. Each cohort was injected in the
right axillary mammary fat pad with 1><106 cells from each cell line. Tumor growth was
measured weekly by caliper, and mice were euthanized by C02 asphyxiation once tumors
reached 2 cm in diameter, or until mice became clinically ill. All animal work was

performed following NIH guidelines under an approved animal protocol.

43

Necropsy, histopathology, histochemistry and immunohistochemistry. At necropsy
examination, tumor tissue, brain, lung, heart, liver, spleen, and kidney were collected and
fixed in 4% paraforrnaldehyde and routinely processed into paraffin blocks from which 4
pm sections were cut and stained with hematoxylin and eosin (H&E). Selected sections
were stained using Masson’s Trichrome and Giemsa histochemical methods. For
immunohistochemical analysis, following deparaffinization, rehydration, antigen
retrieval, and quenching of endogenous peroxidase activity, polyclonal and monoclonal
primary antibodies were applied. Negative controls were obtained by substitution of the
primary antibody with buffer solution. Sections for vimentin were pretreated by
microwaving in sodium citrate buffer, sections for CD31 were pretreated by microwaving
in EDTA, and sections for alpha smooth muscle actin (aSMA) were not subjected to
antigen retrieval. Following washing, the labeled avidin-biotinylated enzyme complex
technique was employed for amplification of primary antibody binding. For visualization,
3, 3-diaminobenzidine (DAB) was applied and counterstained with Mayer’s hematoxylin.
The sections were dehydrated through graded alcohols, immersed in xylene, and mounted
with coverslips. Selected areas were photographed using an Olympus DP70 (Olympus)

digital camera.

Results
Bmi-l overexpression does not lead to transformation of HMECs. Although Bmi-l is
known to be overexpressed in many cancers including breast cancer, its exact role in

oncogenesis, in particular breast cancer is unclear. To define the role of Bmi-l in breast

44

cancer progression, we overexpressed Bmi-l in MCF10A, a non-tumorigenic but
immortal HMEC cell line (Figure 7A). Next, we examined the oncogenic potential of
,MCF 10A cells overexpressing Bmi-l. Consistent with recent observation that 4 or more
oncogenic events are required for the in vitro transtrmation of HMECs [157], Bmi-l
overexpressing MCF10A cells did not form colonies in soft-agar. This indicates that
Bmi-l is insufficient to cause transformation of immortal HMECs which also do not
express p16INK4A and p14ARF tumor suppressors. Similar results were obtained using Bmi-

INK4A
6

l-overexpressing 76N HMECs, which do not express p1 tumor suppressor and are

immortal [29].

Overexpression of H-Ras together with Bmi-l transforms MCF10A cells via
deregulation of multiple growth regulatory pathways. Next, ‘we overexpressed a
constitutively active mutant GlZV of H-Ras [158], in control MCF10A and Bmi-l
overexpressing MCF10A cells (Figure 7B). The pool populations of cells expressing H-
Ras (MCF 10A+H-Ras), Bmi-l (MCF 10A+Bmi-l) or both Bmi-l and H-Ras
(MCFlOA+H-Ras+Bmi-l) were studied for transformed phenotype using soft agar and
Matrigel assays (Figure 7C and 7D). The soft-agar assay indicated that cells expressing
either Bmi-l or H-Ras alone did (not exhibit anchorage-independent growth, however,
cells expressing both Bmi-l and H-Ras readily formed colonies in soft-agar (Figure 7C),
indicating that Bmi-l and H-Ras co-overexpression leads to transformation of HMECs.
To confirm, in vitro transformation potential of Bmi-l and H-Ras expressing cells, Bmi-
1, H-Ras and Bmi-1+H-Ras expressing cells were seeded in matrigel. After one week,

cells were assessed for growth in matrigel. The results indicated that control MCF10A,

45

MCF10A+Bmi-l and MCFlOA+H-Ras cells formed normal spherical acini, while
MCFlOA- Bmi-l+H-Ras cells formed large irregular branched structures indicative of
transformed phenotype of seeded cells (Figure 7D). According to the normal process of
differentiation of MCF10A cells in Matrigel described by Grubbe et al [159], MCF10A
cells grow in spherical colonies that ultimately undergo central apoptosis with lumen
formation and polarization of epithelial cells around this central lumen. The formation of
irregular branched structures indicates a block in this differentiation and a highly
proliferative and transformed phenotype.

We have recently shown that Bmi-l overexpression leads to activation of
AKT/PKB [153], which is also a downstream target of H-Ras. The other known
downstream effector targets of Ras include extracellular signal-regulated kinases ERK1
and ERK2. To assess the mechanism of Bmi-l and H-Ras induced transformation of
HMECS, we analyzed MCF10A and MCF10A derived cells for the expression of AKT
and ERK kinases. The cells were starved for 48 hrs and then EGF containing D-medium
was added for 40 min. The activities of ERK and AKT kinases were determined by
westem—blot analysis of phosphorylated and unphosphorylated forms.

The results indicate that control MCF10A and MCF10A-R38 cells have very little
or no basal phospho-AKT (pAKT) expression, while MCF10A+Bmi-l and MCF 10A+H-
Ras+Bmi-l cells contained higher levels of activated AKT (pAKT)‘even under EGF
starved conditions (Figure 8A). AKT activity was induced in all cells after EGF addition,
and the induction of AKT activity was most noticeable in Bmi-1+H-Ras expressing cells.
On the other hand, ERK activity was constitutively high in H-Ras and Bmi-l+H-Ras

expressing cells regardless of EGF (Figure 8A). It was induced after addition of EGF in

46

MCF10A and MCF10A+Bmi-l cells (Figure 8A). These results suggested that Bmi-l
and H-Ras could transform HMECs by activating AKT and ERK kinases.

Next, we determined the expression of cyclin D1 and CDK4, as the
overexpression of these two cell cycle regulatory proteins has been linked to breast
cancer progression [160, 161]. Our results indicate that, compared to control cells, Bmi-l
or H-Ras overexpression upregulated Cyclin D1, while Bmi-l and H-Ras co-
overexpression upregulated CDK4 as well as Cyclin D1 expression in MCF10A cells
(Figure 8B). We also noticed a higher molecular weight form of CDK4 in Bmi-l and H-
Ras co-expressing cells, which may represent increased phosphorylation of CDK4 in
these cells. We also determined the expression of pr and p53 tumor suppressor in
control and MCF10A-derived cells. Although, MCF10A cells are already pl6mK"A
negative and contain high amounts of hyper-phosphorylated pr, MCF lOA+Bmi-l cells
contained even higher amount of hyper-phosphorylated pr (Figure 8B). On examining
p53 expression, we found that MCF 10A+H-Ras cells contained slightly higher p53
protein levels, while MCF10A+Bmi-l and MCF10A+H-Ras+Bmi-l cells showed
downregulation of p53 (Figure SB). Collectively, our data indicate that Bmi-l together
with H-Ras overexpression leads to activation of ERK and AKT, and upregulation of

Cyclin D1 and CDK4 expression and downregulation of p53 protein.

H-Ras expressing late passage (LP) HMECs cells exhibit a transformed phenotype.
It has been reported in the literature that in some instances, H-Ras overexpression alone
can lead to transformation of MCF10A cells, while other reports suggest that

overexpression of H-Ras alone cannot transform MCF10A cells [162-166]. In our case,

47

the early passage H-Ras expressing cells were not transformed. These early passage
cultures of cells were also heterogeneous and exhibited mixed morphologies with some
enlarged senescent cells and some small normal proliferating cells. The late passage (>5
passages) (LP) culture of H-Ras expressing cells, On the other hand, exhibited more
uniform morphology with most cells proliferating. We considered whether these LP cells
have undergone selection for rapidly proliferating cells and that during this selection may
have acquired transformed properties.

To probe this hypothesis, late passage H-Ras and Bmi-l+H-Ras expressing
MCF10A cells were plated on soft-agar and allowed to form colonies for 10-14 days. The
results indicated that, similar to Bmi-l and H-Ras expressing cells, MCF 10A- H-Ras
(LP) cells formed colonies in soft-agar indicating that H-Ras (LP) cells have also
undergone transformation (Figure 9A). The late passage control and Bmi-l
overexpressing MCF10A cells still did not make any colonies in soft-agar, indicating that
Bmi-l expression alone is not sufficient to cause transformation of HMECs. Transformed
phenotype of LP H-Ras cells was also confirmed by matrigel assays. These assays
indicated that LP H-Ras and H-Ras+Bmi-1 MCF10A cells form very disorganized,
highly branched structures, which ultimately connect through branches and appear like a

big sieved structure (Figure 9B).

Bmi-l expression together with H-Ras induces EMT phenotype in HMECs. When
examining the morphology of MCF10A-derived cells, we noticed that cells expressing
both H-Ras and Bmi-l exhibited fibroblastic morphology, while control and Bmi-l or H-

Ras expressing cells exhibited more epithelial morphology. The fibroblastic morphology

48

of MCF10A+H-Ras+Bmi-l cells suggested the induction of an EMT phenotype in these
cells. To confirm the induction of an EMT phenotype, we examined control MCF10A
and MCF10A derived cells for the presence of EMT markers by irnmunostaining and
western-blot analysis (Figure 9D). The results indicated that control MCF10A and
MCF 10A+Bmi-1(LP) cells expressed E-Cadherin, a cell junction protein characteristic of
epithelial cells, while MCFlOA+H-Ras (LP) and MCF10A+H-Ras+Bmi-l (LP) lost the
expression of E-Cadherin expression. On the other hand, MCF 10A+H-Ras (LP) and
MCFlOA+H-Ras+Bmi-l (LP) cells expressed fibroblastic markers such as vimentin and
fibronectin (Figure 9C). Similar results were obtained using western blot analysis using
antibodies specific for E-Cadherin, fibronectin and vimentin (Figure 9D). These data
indicate that Bmi-l and H-Ras co-expression induces a strong EMT phenotype.
Although, the early passage H-Ras cells did not show an EMT phenotype, late passage H-
Ras cells also exhibit the EMT phenotype, although the late passage H-Ras cells still
appear less fibroblastic compared to Bmi-1+H-Ras cells.

As H-Ras and Bmi-1+H-Ras expressing cells exhibited the EMT phenotype,
which is closely linked to migration and invasion, we performed a wound healing assay
(scratch test), which was repeated three times. The control MCF10A, and Bmi-l, H-Ras
and Bmi-1+H-Ras overexpressing MCF10A cells were grown to 80% confluence, starved
in D3 medium for 48 hrs. A wound was made in the middle of culture dish containing
near-confluent cells and the cells were stimulated with EGF containing D medium for 15
hrs. Cells were photographed at 0 hr, before adding D medium and 15 hr after stimulating
with. D medium. The results indicated that Bmi-l+H-Ras cells have the highest migration

potential and that these cells filled the wound quickly compared to other cells (Figure

49

9E). H-Ras expressing cells also showed migration, although much less compared to
Bmi-1+H-Ras cells (Figure 9E). These cells tend to undergo cell death during migration.
Control cells showed no migration, while Bmi-l expressing cells exhibited minimal
migration (Figure 9E). Thus, these data suggest that Bmi-l and H-Ras co-overexpressing
cells have acquired migration and invasion potential typical of highly transformed

HMECs.

Expression level of H-Ras determines proliferation in H-Ras—expressing MCF10A
cells. The differential ability of MCF10A—H-Ras (EP) and MCF10A—H-Ras (LP) cells to
undergo transformation could be related to the different levels of Ras, which in turn may
determine the proliferation in these cells. To examine this possibility, we determined
expression of H-Ras by Western blot analysis in control MCF10A, and MCF10A-derived
early- and late-passage cells, and. did Ras and Ki-67 co-immunostaining in these cells
(Figure 10). The Western blot analysis of control, early-, and late-passage cells indicated
that H-Ras (EP) cells expressed a high level of Ras, whereas H-Ras (LP) cells expressed a
low level of Ras (Figure 10A). On the other hand, Bmi-1+H-Ras (LP) cells expressed a
high level of Ras (Figure 10A and B). Bmi-l+H-Ras (EP) cells and H-Ras (EP) cells
expressed similar levels of Ras (Figure 10A and B). Because early-passage cultures are
heterogeneous with cells expressing variable levels of Ras, it is possible that cells
expressing Ras above a certain threshold level are not proliferating. At increasing number
of population doublings, there may be selection for cells expressing a lower level of Ras,
which permits continued proliferation. Accordingly, H-Ras (LP) cells will have low
expression of Ras. Consistent with this hypothesis, on a single-cell basis, we observed

that in H-Ras (EP) cultures, most cells with high Ras stained negative for Ki-67, a

50

proliferation marker, whereas cells with low Ras stained positive for Ki-67 (Figure 10C
and D). On the other hand, H-Ras (LP) culture mostly contained cells with low Ras,
which stained positive with Ki-67 (Figure 10C and D). The percentage of low Ras—
expressing cells, which were Ki-67 positive, was also high in MCF10A—Bmi-1+H-Ras
(EP) culture, although some cells in this culture also expressed high Ras, which were
positive for Ki-67 (Figure 10C and D). Importantly, most Bmi-1+H-Ras (LP) cells
expressed high Ras and stained positive for Ki-67, indicating that Bmi-l permits
proliferation of these cells despite high Ras (Figure 10C and D). In all cultures, variable
percentages of low Ras—expressing cells were Ki-67 negative. Because of growth

asynchrony in culture, such cells may not be proliferating at the time of staining.

MCF10A cells expressing H-Ras and MCF10A cells expressing Bmi-l + H-Ras form
histologically distinct tumors in vivo. As late passage MCF 10A+H-Ras cells also
exhibited features of transformation, we hypothesized that these cells will also give rise
to tumors in vivo, as has been reported in some instances. However, such tumors may be
different than the tumors that arise after injecting MCFlOA+H-Ras+Bmi-1 cells. To
examine the hypothesis that both late passage MCFlOA+H-Ras and MCF 10A+H-
Ras+Bmi-1 cells will give‘ rise to tumors in vivo, and that these tumors may differ in
behavior due to the overexpression of Bmi-l, we injected MCF10A, MCF10A+Bmi-1
(LP), MCFlOA+H-Ras (LP), MCF10A+H—Ras+Bmi-l (LP) cells into the mammary fat
pad. As expected, MCF10A control cells did not produce tumors in vivo. Injection of
MCF10A+Bmi-l cells also did not result in tumor formation even after 60 days,
indicating that overexpression of Bmi-l alone is not sufficient for neoplastic

transformation of mammary epithelial cells in vivo.

51

In contrast, MCF10A+H-Ras and MCF10A+H-Ras+Bmi-1 cells produced
progressively enlarging tumors in the mammary fat pad upon injection. Grossly, these
tumors were strikingly different (Figure 11A); MCF10A+H-Ras+Bmi-l cells formed
tumors which were solid, firm, and irregular, and were whitish tan on. the cut surface. In
contrast, tumors formed by MCF10A+H-Ras cells were variably hemorrhagic and often
cystic, composed predominantly of large thin cysts filled with clotted and/or unclotted
blood (Figure 11A). Tumor growth rates when measured by tumor volume were not
significantly different between these two groups; however, this was due to the general
volume of tumors being similar, rather than the actual tumor mass.

Histologically, MCF10A+H-Ras and MCF10A+H-Ras+Bmi-1 tumors differed
strikingly. MCF10A+H-Ras tumors consisted of variable populations of poorly to fairly
well-differentiated smooth muscle, variably cystic irregular vascular spaces lined by
poorly- to fairly well-differentiated endothelial cells, and multifocal clusters and nests of
poorly- to well-differentiated mast cells (Figure llB-C). In contrast, MCF10A+H-
Ras+Bmi-l tumors were composed of streams and bundles of poorly-differentiated
spindle-shaped cells with scant, faintly eosinophilic fibrillar cytoplasm embedded in
scant eosinophilic stroma, large round to oval hyperchromatic nuclei with multiple
prominent nucleoli, and numerous mitotic figures (approximately 2-3/hpf) (Figure 11B-
C). These cells often infiltrated into the surrounding fat pad, effacing normal ducts and
adipose tissue, and, in one case, infiltrating and destroying the cortical bone of a
subjacent rib and invading and effacing the bone marrow (Figure 11B).

Both tumor types were immunoreactive to antibodies against vimentin, whereas

the MCF10A+H-Ras tumors were also immunoreactive to aSMA and CD31 (PECAM),

52

illustrating the smooth muscle and hemangiomatous components of these tumors (Figure
11C). Geimsa staining confirmed the multifocal mast cell clusters of varying
differentiation in the MCF10A+H-Ras tumors, and Masson’s Trichrome histochemical
staining confirmed that the MCF10A+H-Ras+Bmi-l tumors were composed of spindle
cells with scant collagen production, more suggestive of a myogenic phenotype than a
fibrosarcomatous one (Figure 11C). Animals with tumors formed by MCF10A+H-Ras
cells were often very hemorrhagic, resulting in early morbidity possibly due to anemia as
a result of bleeding from the tumors, rather than tumor burden in contrast to mice bearing
tumors formed by MCF10A+H-Ras+Bmi-l cells, which as a group lived longer with

tumors than MCF10A+H-Ras tumor-bearing mice (Figure 11D).

MCF10A+H-Ras (LP) and MCF10A+H-Ras+Bmi-l (LP) cells display defective p53
phosphorylation and attenuated DNA damage response. H-Ras is known to cause
premature senescence or oncogene induced senescence (018) in primary cells, which is
mediated by p16INK4A and p53-p21 pathways of senescence [167-169]. Using senescence-
associated B-galactosidase marker, we noticed senescence induction in a large proportion
(40—50%) of MCF10A cells by H-Ras overexpression at early passages. Because
MCF10A cells are p16INK4A negative, the partial 018 in these cells may depend on p53
and its target genes. Consistent with partial OIS, early-passage MCF10A-Ras cells also
showed slower growth compared with vector control MCF10A and MCF10A cells co-
overexpressing H-Ras and Bmi-l.

The senescent cells in MCF10A—H-Ras and MCF10A—Bmi-1+H-Ras cells were

progressively lost, and rapidly proliferating cells were selected in later passages. We

53

hypothesized that the selection of rapidly proliferating cells in late-passage cultures of
MCF10A—H-Ras and MCF10A—Bmi-l+H-Ras cells may depend on a defect in p53
pathway in these cells. To examine this hypothesis, we determined p53 expression in
control MCF10A, MCF10A—Bmi-l (LP), MCF10A—H-Ras (LP), and MCF10A—Bmi-
l+H-Ras (LP) cells. The results indicated that unlike in MCF10A—H-Ras (EP) cells
(Figure 8B), p53 was down-regulated in MCF10A—H-Ras (LP) cells (Figure 12A). To
determine the mechanism of p53 down-regulation and its possible significance with
respect to transformed phenotype of MCF10A—H-Ras (LP) and MCF10A—Bmi-1+H-Ras
(LP) cells, we further studied p53 pathway in these cells.

MCF10A control and MCF10A-derived late-passage cells were treated with the
DNA-damaging agent camptothecin (500 nmol/L) for the indicated amount of time, and
expression of p53, phosphorylated p53, and p53 target genes was studied by Western blot
analysis (Figure 128). The results indicated that although MCF10A—H-Ras (LP) and
MCF10A-Bmi-l+H-Ras (LP) cells had overall low p53 compared with control MCF10A
and MCF10A—Bmi—1 cells, p53 remained inducible by camptothecin in all four set of
cells, although the induced levels of p53 was still low in MCF10A—H-Ras (LP) and
MCF 10A—Bmi-1+H-Ras (LP) cells (Figure 12B). Further analysis of phosphorylated p53
indicated that MCF10A—H-Ras (LP) and MCF10A—Bmi-l+H-Ras (LP) cells were
partially defective in phosphorylation of p53 at Ser-15 and Ser-37 residues (Figure 128).
Quantification of Western blot data showed reduced phosphorylation of p53 at Ser-15 in
both MCF10A—H-Ras (LP) and MCF10A—Bmi-l+H-Ras (LP) cells at 4 and 8 h time
points, whereas the basal levels of p53 Ser-lS were similar in all MCF10A-derived cells.

Ser—37 phosphorylation was also compromised in MCF10A-H-Ras (LP) and MCF10A-

54

Bmi-l+H-Ras (LP) cells. Neither of these cell lines showed any induction of Ser-37
phosphorylation of p53 by camptothecin treatment.

Because it has been reported that PRAK mediates Ser—3-7 phosphorylation of p53
induced by H-Ras and that PRAK mediates Ras-induc'ed 018 [168], we hypothesized that
PRAK may be lost during selection of rapidly proliferating cells in H-Ras (LP) cells in
culture. To examine this possibility, we determined PRAK expression in these cells by
Western blot analysis. The results indicated that regardless of DNA damage, PRAK
expression is not lost in control or H-Ras (LP) cells (Figure 12B). Interestingly, PRAK
expression was up-regulated in H-Ras (LP) cells (Figure 123). The up-regulation of
PRAK is consistent with the notion that PRAK is an H-Ras target, which acts negatively
to suppress H-Ras—induced proliferation [170]. Nonetheless, it seems that it is possible
that this PRAK-mediated negative feedback regulation of H-Ras—mediated proliferation
may be deficient in MCF10A—H-Ras (LP) cells, which may have allowed these cells to
undergo transformation in culture.

Next, we studied the induction of p21 and PUMA (p53 up-regulated modulator of
apoptosis), two well-known transcriptional targets of p53 [171, 172]. Our results
indicated that both p21 and PUMA induction by camptothecin is partially compromised
in MCF10A—H-Ras (LP) cells (Figure 12B), and p21 induction was more compromised in
MCF10A—Bmi-l+H-Ras (LP) cells. Attenuated response of these targets of p53 is
consistent with defective phosphorylation at Ser-15 and Ser-37 residues. We also
examined expression of Bax and P163 (p53-inducible gene 3), two other known targets of
p53 [171]. Analysis of these two genes indicated that Bax is expressed at very low levels

and is inducible in control MCF10A cells. However, MCF10A—H-Ras (LP) cells had

55

higher levels of Bax, which were not inducible by DNA damage (Figure 12B).
Interestingly, among all four cell types, MCF10A—H-Ras (LP) cells expressed high
BCL2, which may be related to transformed properties of these cells. PIG3, which
usually has a delayed kinetics of induction by p53 [173], was not inducible in any of the
cell types within the time frame used in our experiment (Figure 128). Interestingly,
compared with control MCF10A cells, MCF10A-derived (LP) cells showed significant

down-regulation of P103 (Figure 123).

Discussion

Several recent studies have suggested that PcG proteins, in particular EZH2 and
Bmi-l are overexpressed in human cancers. Recent elegant studies have clearly shown
that oncogenic transformation of human cells is a multi-step process [174]. It is very
likely that overexpression of EZH2 or Bmi-l alone is not sufficient to cause
transformation of human cells. To gain an insight into breast cancer progression, we
examined the transformation potential of Bmi-l oncoprotein in immortalized HMECs.
Although immortalized HMECs that we studied lack pl6mK4A and pl4ARF, Bmi-l
expression still provides an oncogenic signal in these cells by the activation of PI3K-
AKT pathway [153]. However, the oncogenic signal provided by Bmi-l alone does not

appear to be sufficient to cause transformation of HMECS, despite these cells being

INK4B INK4A ARF
5 6 4

immortal, lacking three important tumor suppressors- p1 , pl and pl
[155], and overexpressing the Bmi-l oncogene. This observation underscores the

stringency of transformation in HMECS. Nonetheless, Bmi-l overexpression is frequently

56

observed in invasive breast tumors [47, 48, 170], suggesting the involvement of
additional oncogenic events during breast cancer progression in such tumors.

To understand the genetic basis of these presumptive additional oncogenic events,
we overexpressed a constitutively active mutant, G12V of H-Ras [158], in Bmi-l
overexpressing MCF10A cells. G12V mutant of H-Ras promotes proliferation and
oncogenesis via activation of mitogen-activated protein kinase (MAPK) kinase
(MEK)/MAPK and the phosphoinositide 3-kinase (PI-3K)/AKT pathways. However, the
activation of these pathways and their outcome is cell-type specific. For example, in
primary cells, activation of these pathways led to induction of 018, while in immortalized
cells with compromised p53-p21 and/or p16INK4A pathways, H-Ras G12V may promote
proliferation. Our reasoning behind using H-Ras Gl2V in these assays was based on its
relevance to breast cancer, and its reported use in oncogenic assays involving HMECs
[157]. Although, the direct mutational activation of H-Ras is rare in breast cancer, its
hyperactivation by persistent growth factor signaling caused by EGFR and HER2/neu
overexpression occurs in a large proportion (20-40%) of breast cancers [17 5, 176].

018 caused by 612V mutant of H-Ras may require both functional pl6mmA and
p53. In MCF10A cells, which have functional p53, we initially noticed the appearance of
a heterogeneous culture with approximately 40-50% cells exhibiting senescent
morphology upon H-Ras overexpression. Consistent with partial senescence induction,
our western blot data also indicated upregulation of p53 protein in these cells. Senescence
acts as a strong barrier to oncogenesis [151], hence the initial induction of senescence in a
high proportion of MCF10A cells by H-Ras indicate an anti-oncogenic response. As

expected, these early passage cells were not transformed by soft-agar and Matrigel

57

assays. However, late passage cultures, which were much more homogeneous and did not
contain cells with senescent morphology, displayed transformed phenotype in Matrigel
and soft-agar assays. Ras and Ki-67 costaining data also suggest that early passage
culture of MCF10A-H-Ras are more heterogeneous in terms of Ras expression, whereas
the late-passage culture of these cells are homogeneous in terms of Ras expression.
Importantly, only low Ras—expressing cells tend to be Ki-67 positive, suggesting that low
Ras permits proliferation, whereas high Ras blocks proliferation, possibly via OIS. This
differential effect of Ras on proliferation explains the emergence of low Ras—expressing
culture at late passages.

The H-Ras overexpression in Bmi-l overexpressing MCF10A cells, on the other
hand caused senescence in minority of cells when compared to H-Ras alone cells, and
homogeneous culture with proliferating cells appeared much more rapidly from these
cells.,These data indicate that to some extent, Bmi-l can overcome H-Ras induced OIS,
even in p16'NK4A negative cells, presumably via pl6mK4A- and p14ARF-independent targets
of Bmi-l. The homogeneous culture that rapidly emerged from Bmi-1+H-Ras—expressing
cells continued to express high Ras. Most cells in this culture were Ki-67 positive despite
expressing high Ras, suggesting that Bmi-l permits proliferation of cells despite high
Ras, and thus there is no selection for cells expressing low Ras. The biochemical basis for
proliferation of MCF10A—Bmi-l+H-Ras (LP) cells despite high Ras remains to be
elucidated.

On examination of Ser-37 and Ser-15 phosphorylation of p53 in response to DNA
damage, we found that Ser—37 phosphorylation of p53 is significantly low and not

inducible in both late-passage H-Ras and Bmi-1+H-Ras—expressing cells. In addition,

58

these cells also had much lower induction of Ser-15 phosphoryated p53, suggesting a
possible defect in other p53-activating kinases such as ATM. A detailed analysis of
various p53 phosphorylating kinases in late-passage MCF10A—H-Ras and MCF10A—
Bmi-1+H-Ras remains to be done. Nevertheless, our data clearly indicate that these late-
passage H-Ras— and Bmi-l+H—Ras—expressing cells have defects in p53 phosphorylating
pathways, which results in attenuation of induction of p53 targets such as p21 and
PUMA. This compromised induction of p53 targets may contribute to a transformed
phenotype of MCF 10A cells expressing Bmi-l and H-Ras.

The differential behavior of early- and late-passage H-Ras—overexpressing
MCF10A cells with respect to the transformed phenotype explains the different results
that are reported in the literature[l62-l66]. Our data suggest that in cases where H-Ras—
expressing MCF10A cells showed a transformed phenotype and gave rise to tumors in
nude mice assays, late-passage H-Ras—expressing cells with defective p53 regulation may
have been used. In other studies, where transformation of H-Ras—expressing MCF10A
cells was not reported, early-passage H-Ras—expressing MCF10A cells may have been
used. Alternatively, the transforming potential of H-Ras cells could be correlated with the
level of expression of H-Ras. In studies where H-Ras alone was reported to be
transforming, the expression of H-Ras may be low, which permits proliferation. On the
other hand, in cases where Ras was reported to be insufficient for transformation, the
expression of Ras may be very high, which causes proliferation arrest and 018. Neither of
these possibilities is mutually exclusive and both possibilities are likely to contribute to
transformation of HMECs by H-Ras. Recently, it was shown that low levels of K-Ras

induce proliferation and mammary epithelial cell hyperplasias, whereas high expression

59

of K-Ras induces proliferation arrest and 018 in doxycycline-inducible K-Ras transgenic
mice [177]. In this report, it was also shown that inactivation of p53 permits
transformation of mammary epithelial cells and tumor formation by high expression of
Ras [177]. Our in vitro data are consistent with this report.

The results of histopathology, including special stains and immunohistochemistry,
confirm that the MCF10A+H-Ras tumors are composed of multiple different populations
of varying phenotypes (smooth muscle, hemangiomatous, and mast cells). The multiple
populations present were differentiated along multiple mesenchymal lines, suggesting at
least in part that portions of these tumors may be in part an in vivo response to the
xenografted tumor population rather than original components of the neoplastic
population that have undergone dedifferentiation and redifferentiation along multiple
lines. For example, the large clusters well- to poorly-differentiated mast cells may have
been recruited to the site of tumor formation by cytokines or other grth factors known
to recruit mast cells, such as GM-CSF or IL-3 [178], produced by the xenografted tumor
cells, rather than being of xenograft origin. This would explain the unusual phenotype of
multiple differentiated populations. The MCF10A-Bmi-1+H-Ras tumors, on the other
hand, represent a pure population of highly atypical, poorly differentiated, and infiltrative
spindle cells. Although a-SMA immunohistochemistry was negative in these tumors,
Masson's trichrome stain along with positive immunohistochemistry for vimentin would
suggest that these cells may represent a myoepithelial phenotype consistent with EMT.

Although MCF10A—Bmi-1+H-Ras (LP) and MCF10A—H-Ras (LP) cells give rise
to histologiCally distinct type of tumors, biochemically these cells Show only minor

differences in regulation of growth-regulatory pathways. The only significant difference

60

between these two cell lines is that H-Ras (LP) cells expressed higher levels of BCL2,
which may contribute to the oncogenicity of these cells. In any case, we did not observe
tumor formation by MCF10A—Bmi-l cells, suggesting the involvement of additional
oncogenic events such as down-regulation of p53, oVerexpression of CDK4 and cyclin
D1, and up-regulation of AKT and ERK activities in the transformation of HMECs and
breast cancer progression. Our data also indicate that Bmi-l may cooperate with Ras in
transformation by simply allowing high Ras—expressing cells to proliferate. The
additional oncogenic events then may be largely contributed by H-Ras in the experiments
described here. It remains to be determined which of these oncogenic lesions, together

with Bmi-l, are sufficient to transform HMECs and form tumors in vivo.

61

APPENDIX B

Figure 7: Bmi-l and H-Ras co-expression transforms HMECs. (A) Bmi-l
overexpressing MCF10A cells were generated by stable overexpression of Bmi-l, and
cells (as indicated) were analyzed for Bmi-l overexpression by western blot analysis. (B)
H-Ras was introduced in control MCF10A and MCF10A+Bmi-1 cells, and cells were
analyzed for H-Ras expression by western blot analysis. (C) MC F 10A and MCF10A cells
expressing H-Ras alone, Bmi-l alone or Bmi-l together with H-Ras (as indicated) were
analyzed for anchorage-independent growth using soft-agar assays. MCF10A+Bmi-1+H-
Ras cells form colonies in soft agar, indicating anchorage independent growth. (D)
MCF10A and MCF10A-derived cells (as indicated) were analyzed for acini formation
using Matrigel assays. While all other MCF10A-derived cell lines developspherical
colonies consistent with differentiated cells, MCF10A+Bmi-l+H-Ras cells develop
highly branched proliferative structures indicative of a poorly differentiated population.

Q99
Y. B NOV. x’<\

Q
Q" - I‘ Q It? - 3‘

C. MCF10A Bmi-1

A.

 

62

Figure 7 (continued)

D. MCF10A Bmi-1

‘f§.'--I
3"

 

H-Ras Bmi-1+H-Ras

Figure 8: Various growth regulatory pathways are dysregulated in cells co-expressing
Bmi-l and H-Ras. (A) Western blot analysis of phospho-AKT, total AKT (AKTI and
AKT2), phospho-ERK and total ERK in control MCF10A and MCF10A-derived cells (as
indicated) with and without EGF stimulation. Western blot analysis using a-actin served
as a loading control. (B) Western blot analysis of CDK4, H-Ras, phosphorylated pr (P-
pRB), pRB, Cyclin D1, and p53 in asynchronously growing MCF10A and MCF10A-
derived cells (as indicated). B-actin and QM are loading controls.

at
A. «gt» ,3? B.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.'\ .N Y‘ 9’
ifl‘ii a“ a»: .. "
0 4o 0 4o 0 40 0 4o +EGF(min) ‘f f ’2’ 9:“
[ - -~-'--'l-pAKT 88...;100'“
----——-- AKT1 '-
----?:-- AKTZ - - ”4;“
P R8
-‘3'-’.' pERK --!-i dig
"" “‘- nyolinD1
-----oool'lfiwn - - - -“W

 

 

 

 

 

63

Figure 9: Late-passage H-Ras—expressing MCF10A cells exhibit transformed features.
All MCF10A-derived cells were analyzed at passage 8. (A) Control MCF10A and
MCF10A-derived late-passage cells (as indicated) were grown in soft agar to determine
anchorage-independent growth potential of these cells. Cells were photographed (x4,
final magnification) at day 14. (B) Three-dimensional growth of MCF10A—H-Ras (LP)
and MCF10A—Bmi-l+H-Ras (LP) was analyzed using Matrigel assays as described in
Materials and Methods. Cells in Matrigel were photographed (x10) at day 7. (C) EMT
phenotype of MCF10A and MCF10A-derived late-passage cells was analyzed by
immunostaining using antibodies specific for E-cadherin, vimentin, and fibronectin (as
indicated). To visualize nuclei, cells were stained with 4',6-diamidino-2-phenylindole,
and immunostained cells were visualized and photographed using Zeiss LSM510 UV
META confocal microscope (x60). (D) EMT markers expressing in MCF10A and
MCF10A-derived LP cells were analyzed by western blot analysis using antibodies
specific for E-Cadherin, vimentin (lower bands), and fibronectin as described in materials
and methods. a-tubulin is a loading control. (E) The migration potential of MCF10A and
MCF10A-derived cells was determined by wound-healing assay. The control MCF10A,
and Bmi-1—, H-Ras—, and Bmi-l+H-Ras—overexpressing MCF10A cells were grown to
80% confluence, starved in D3 medium for 48hr. A wound was made in the middle of
culture dish containing near-confluent cells and the cells were stimulated with EGF-
containing D medium for 15 h. Cells were photographed at 0 h, before adding D medium
and at 15 h, after stimulating with D medium. Cells were photographed using a light
microscope (x4).

 

MCF10A Bmi-1 (LP)

    

4X

 

 

 

 

   

H-Ras+Bmi-1 (LP)

64

Figure 9 (continued)

MCF10A

H-Ras (LP)

Bmi-1 (LP)

H-Ras+
Bmi-1 (LP)

 

 

, N
\’
D gr 325°
6" eff \' e”?
s ‘2‘ a“ e'
"""' J—E-cad herin

 

 

.
! '

m j-Fibronectin

was JVImentin
. .1,
J— or-Tubulin

65

 

 

 

___‘fi

 

 

_—

Figure 9 (continued)

E- 0hr

 
 
  
 
 

    

MCF10A

H-Ras+
Bmi-1 (LP)

66

Figure 10: Expression level of H-Ras determines proliferation in MCF10A cells
overexpressing H-Ras. (A) H-Ras expression in MCF10A control and MCF10A-derived
early-passage (passage 2 after Ras selection) and late-passage cells (passage 8) was
determined by Western blot and repeated four times. (B) To determine the relative
expression of H-Ras in MCF10A and MCF10A-derived cells, its signal in each lane was
quantified by densitometric analysis using ImageJ 1.3 software (NIH) and normalized to
a-tubulin signal. (C) H-Ras and Ki-67 co-immunostaining was done to determine
proliferation in MCF10A-derived early-passage (passage 2) and late-passage (passage 8)
cells. MCF10A cells were used as control because they do not express detectable Ras but
are Ki-67 positive under our experimental conditions. Representative photos (60X
magnification) of costaining in each cell line (as indicated). (D) Quantification of Ras-
and Ki-67—expressing cells in MCF10A-derived early-passage (passage 2) and late-
passage (passage 8) culture of H-Ras and Bmi-l+H-Ras cells. Co—staining was done in
triplicates and a total of 100 to 200 stained cells were counted in multiple fields.

\ \
A. (c3 as?
\ \ '\' \ '\’
\QY. «3 83 36‘ QY‘ WQ §\ x9®
(5‘ e"? «“5 <21”9 (55‘ 89 'x’" Q99
.3: Q) Q" 4‘ ‘2‘ 9‘0 Q“

 

ma ... .2.” a W JBml-‘l
[ ‘ ...... ... .....- J lH-Ras

 

 

 

 

 

 

 

 

A
O

(A)
O
I
I
-1

m

. (D

-L
O
. t t

 

Relative Expression
N
O

O

 

67

FCigure 10 (continued)

H- Ras
(EP)
MCF10A - I I
H- Ras
(LP)
H-Ras+
Bmi-1 (LP)

H-Ras+
Bmi-1 (EP)

 

 

 

Ki67 H- Ras Merge
D. 1001 -High Ras, Ki67-ve
90-] -High Ras. K167+ve
80: -Low Ras, Ki67-ve
70 ‘ -Low Ras, Ki67+ve
2 .
8
'O
o
E
S
a)
o\°

 

 

H-Ras H-Ras+Bmi-1 H-Ras H-Ras+Bmi-1
(EP) (EP) (LP) (LP)

68

Figure 11: Gross morphology, histopathology, and immunohistochemistry of tumors
originating from xenografts. (A) Gross morphology of tumors resulting fi'om injection of
MCF10A—H-Ras (LP) and MCF10A—Bmi-l+H-Ras cells (LP; as indicated). Left, tumors
induced by MCF10A—H-Ras._ Right, tumors induced by MCF10A—Bmi-1+H-ras cells.
(B) Histopathology of tumors resulting from injection of MCF10A-H-Ras (left) and
MCF10A—Bmi-l+H-Ras (right) cells. Left, tumors induced by MCF10A—H-Ras cells
were composed of variable populations of poorly differentiated to well-differentiated
endothelial cells forming haphazard vascular channels (top), spindle-shaped cells
resembling smooth muscle (middle), and multiple variably-sized clusters of poorly-
differentiated to well—differentiated mast cells (bottom) (Bar=100 um). Right, tumors
induced by MCF10A—Bmi-l+H-Ras cells were composed of a homogeneous population
of sheets and intersecting bundles of poorly differentiated spindle cells (top) that
infiltrated adjacent adipose tissue and bone (middle) (Bar=100 pm). Cells were poorly
differentiated with large pleomorphic nuclei and frequent mitoses (bottom) (Bar=50 pm).
(C) Histochemical and immunohistochemical staining of tumors induced by MCF10A—
H-Ras (left) and MCF10A-Bmi-1+H-Ras (right) cells. Left, tumors induced by
MCF10A—H-Ras were multifocally immunoreactive for antibodies against or-SMA and
CD31; mast cell clusters were diffusely positive with Giemsa staining for mast cell
granules, and tumors were diffusely negative for collagen by Masson's trichrome staining
(Bar=100 pm). Right, tumors induced by MCF10A—H-Ras+Bmi-1 cells were diffusely
negative for a-SMA and CD31 except for the presence of intratumoral capillaries
(arrowheads), diffusely negative with Giemsa staining except for occasional resident
mast cells (arrowhead), and showed very little collagen production with Masson's
Trichrome (MT) stain (Bar=100 um). (D) Kaplan-Meier survival curve. Whereas
MCF10A and MCF10A—Bmi-l xenografted mice did not develop tumors and survived
throughout the course of the study, mice xenografted with MCF10A—H-Ras and
MCF10A—Bmi-l+H-Ras had decreased survival after the development of palpable
tumors. MCF10A—H-Ras xenografted mice had significantly decreased survival
compared with MCF10A—Bmi-l+H-Ras mice (P < 0.002) (n=10 mice per group).

A. MCF10A+H-Ras MCF10A+H-Ras+Bmi-1

         

   

69

Figure 11 (continued)
B. MCF10A+H-Ras MCF10A+H-Ras+Bmi-1

 

70

Figure 11 (continued)

C- MCF10A+H-Ras MCF10A+H-Ras+Bmi-1

,9 ... . 4'3“ - - .41“
. 3 fi“: .
to :7. D‘ f‘;‘%.* ’03 ’4

 
 
   
  
   

  

a. 3

-..£. ' v A
"‘ '.f‘t‘i“‘.’,<*.’ ... “I“?
2". ‘ $191; 9",‘1 M?
I;‘ I‘k'é:0*.“iv a MS\ a
N . . a . ‘... i? . .~ ,
99° térvl'w" (“a ‘-‘ 1“
212' ~11: first. - ...

   
 

9 Giemsa

71

Figure 11 (continued)

-- MCF10A
Bmi-1
-- H-Ras

-‘- H-Ras+Bmi-1

 

 

 

0 1'0 2'0 3'0 40 5'0 6'0 7'0 60
Trme (days post-injection)

Figure 12: Analysis of p53 pathway in control MCF10A and MCF10A-derived late-
passage cells. (A) Top, western blot analysis of Bmi-l, H-Ras, and p53 in control
MCF10A and MCF10A-derived (LP) cells (as indicated) was done as described in Figure
2. Bottom, densitometric analysis of signals (of p53 and H-Ras) present in each lane was
done, normalized to corresponding a-tubulin signal, and plotted to determine the
expression levels of p53 and H-Ras as indicated. (B) Analysis of DNA damage response
in MCF10A and MCF10A-derived late-passage cells. The cells were treated with
camptothecin (CPT) for indicated amount of time, harvested, and analyzed by Western
blot analysis for total p53, phosphorylated p53 (Ser-15 and Ser-37), p53 target genes
(p21, PUMA, Bax, and PIG3), PRAK, and BCL2. B-Actin was used as a loading control.

(0"
,8)
A ' «\Q?‘ QQQ _\"\ Q50?

s" ~2~’ <22“ 8’

 

textural r-jr ' ] Bmi'1

l

l w l H-Ras
[ ---... lp53
L _

 

 

 

 

 

 

l a-Tubulin

72

Figure 12 (continued)

Relative expression

Relative expression

 

1.2 -

O—L
000

0.6
0.4"
0.2

40-
35.
30-
25'
20-
15'
10'

5-

 

 

0-

 

MCF10A H-Ras Bmi-1 Bmi-1 + H—

Cell line Ras

MCF10A H-Ras Bmi-1 Bmi-1+H-
Cellline R35

73

Figure 12 (continued)

B. MCF10A H-Ras (LP) Bmi-1 (LP) Bmi-1+H-Ras (LP)
_ —

04s

 

.3

 

048 04 04BCPT(hrs)
F.’""." g . ~, .aT 41s:-

   
 

      
 
 
  

 

p53 ear-37

p53 ser-15

CHAPTER THREE

Bmi-l and H-ras collaboration in mammary epithelial cells promotes an aggressive
and metastatic phenotype, and Bmi-l knockdoWn in xenograft tumors results in

delayed tumor onset

Abstract: The polycomb group protein Bmi-l is a transcription repressor reported to
regulate self-renewal of normal and cancer stem cells, and prevent cellular senescence
through inhibition of cyclin-dependent kinase inhibitors p161nk4a and p19ARF. Originally
discovered as a cooperating oncogene with c-szc in a murine model of leukemia,
overexpression of Bmi—I has since been reported in several forms of cancer, including
breast cancer, where it has been associated with regional lymph node metastasis, short
interval to relapse, and poor prognosis. A direct or collaborative role of Bmi-l in the
pathogenesis of metastatic breast cancer would be critical in the development of novel
biomarkers for diagnosis and treatment of this aggressive disease. We have previously
shown a collaborative role of Bmi-l with H-Ras exists in the induction of proliferation,
transformation and invasion of mammary epithelial cells (MEC) in vitro and in vivo.
These current data show that through the use of xenograft and tail vein experiments,
simultaneous overexpression of Bmi-l and H-Ras in MCF10A mammary epithelial cells
promotes proliferation and invasion, and inhibits apoptosis due to detachment from the
extracellular matrix, greatly potentiating a metastatic phenotype to the liver, spleen, and
brain in vivo. These findings suggest that overexpression of Bmi-l in conjunction with

H-Ras overexpression results in greater metastatic potential. Additionally, we show that

75

knockdown of Bmi-l in several breast cancer cell lines leads to decreased proliferation
and invasion, and increased apoptosis in response to DNA damage induced by etoposide.
Furthermore, knockdown of Bmi-l expression delays tumor onset in two mammary fat
pad xenograft models. These results suggest that Bmi-l plays an important role in the
behavior of several breast cancer cells lines, and may be an important marker for

diagnosis and a target for the treatment of aggressive breast cancer.

Introduction

Cooperation between oncogenes to produce neoplasia is a well-documented
occurrence in many types of human cancers [70-72]. Whether resulting from a
combination of proliferative or of anti-apoptotic effects from overexpression of multiple
oncogenes, or the abrogation of tumor suppressor genes, the process of oncogenesis
commonly involves mutations of multiple genetic loci. Bmi-l is a member of a group of
transcription repressors called polycombs that inhibit transcription of various homeobox
genes that are critical for body plan establishment and morphogenesis [20, 28, 113].
Through epigenetic mechanisms, it is possible that Bmi-l may interact with various
genetic loci and exert a suppressive effect on gene transcription throughout the genome.
It has been shown that Bmi-l is critical in the maintenance of stem cell populations by
preventing senescence through suppression of the Ink4a-Arf tumor suppressor locus [18,
27, 150]. By suppressing this locus, the effects of the cyclin-dependent kinase inhibitors
p16 and p19 are abrogated, resulting in prevention of p53 degradation and
phosphorylation of retinoblastoma (Rb), respectively [18, 27, 150]. This results in

prevention of cell cycle arrest and the resultant cellular senescence and apoptosis.

76

Therefore, if dysregulated or overexpressed, Bmi-l may act as an oncogene through the
suppression of gene transcription of tumor suppressor genes. In fact, Haupt et a1. (1991)
initially described Bmi-l as a cooperating oncogene in a Eu-myc murine model of B-cell
lymphomagenesis, as a result of retroviral induced mutagenesis by the Moloney murine
leukemia virus. It was found that Bmi-l was a frequent target of retroviral insertional
mutagenesis, and its resultant overexpression in this model increased lymphomagenesis
by 75% [52, 83].

Since its discovery, Bmi-l has also been implicated in the development of many
different types of human cancers, including nasopharyngeal [37], liver [43], colorectal
[44], brain [40], lung [41], and breast cancer [47]. One study has shown that Bmi-l
overexpression correlates with prognosis and regional lymph node metastasis in patients
with infiltrative ductal carcinoma [47]. Given the collaborative role Bmi-l plays with c-
myc in lymphomagenesis, the following studies were undertaken to determine the
relationship of Bmi-l with protooncogenes commonly overexpressed in breast cancer (H-
Ras) and its relationship in the pathogenesis of breast cancer metastasis.

The MCF10A (Ml) cell line was derived as a non-transformed breast epithelial
cell line derived from the spontaneous immortalization of normal breast epithelial cells
obtained from a patient with fibrocystic disease [99]. In vitro, this cell line forms plates
of polygonal cells with well differentiated cell-to-cell junctions and has a relatively low
rate of proliferation. In vivo, injection of these cells into the mammary fat pad results in a
small residual population of well differentiated mammary ductal structures and

epithelium, but does not result in tumorigenesis [99, 179]. These cells are commonly used

77

to represent the behavior of normal mammary epithelial cells, and are commonly used in
the study of mammary gland biology.

In this study, we used MCF10A cell lines engineered to overexpress Bmi-l
(MCF10A+Bmi-l), H-Ras (MCF10A+H-Ras), and'Bmi-l in combination with H-Ras
(MCF lOA+H-Ras+Bmi-1), to characterize the effects of overexpression of these
oncogenes alone and in combination on the behavior of mammary epithelial cells in vitro
and in vivo, including the ability to survive and establish metastasis in distant sites, either
spontaneously or by introduction via tail vein injection. Our results demonstrate that
Bmi-l collaborates with H-Ras in the induction of an aggressive cell phenotype in vitro
and in vivo, with a greatly increased metastatic propensity to distant sites. These data
suggest that Bmi-l may be an important target for diagnosis and treatment of aggressive
breast cancer.

Materials and Methods
Cell lines, vectors, and plasmids. MCF10A+H-Ras and MCF10A+H-Ras+Bmi-l cells
were cultured in Dulbecco’s modified eagle medium supplemented with 5% horse serum
(Gibco, Carlsbad CA), penicillin/streptomycin (Gibco), human recombinant insulin (10
pg/ml, Gibco), epidermal growth factor (EGF) (20 ng/ml, Upstate), choleratoxin (100
ng/ml, Sigma), and hydrocortisone (500 ng/ml, Sigma, St. Louis MO). MDA-MB-231
and MCF7 cells were grown in DMEM high glucose 1X (Gibco) supplemented with 10%
fetal bovine serum (FBS) (Gibco), and penicillin/streptomycin (Gibco). ZR75-l cells
were grown in RPMI 1640 with 2 mM L-glutamine (Gibco) supplemented with 10% F BS
(Gibco), penicillin/streptomycin (Gibco), sodium bicarbonate (1.5 g/L, Gibco), glucose

(4.5 g/L, Sigma), HEPES (10 mM, Gibco), and sodium pyruvate (1 mM, Gibco). Bmi-l

78

shRNA lentiviral vector (LV-pLL3.7+Bmi-li) and empty shRNA lentiviral vector (LV-
pLL3.7) were kindly provided by Max Wicha and have been described previously [35].
The shRNA lentiviral vectors LV-pLL3.7+Bmi-li and LV-pLL3.7 were used to transfect
293T packaging cells, and viral supernatant was used to transduce each of four cell lines
(MDA-MB-23l, ZR75, MCF 7 and MCF10A+H-Ras+Bmi-l). Significant knockdown of

Bmi-l was confirmed by western blot.

Western blot analysis. Bmi-l was detected using F6 mouse monoclonal antibody (mAb)
from Upstate Cell Signaling Solutions (Charlottesville, Virginia), and H-Ras was
detected using the F-235 mouse monoclonal antibody from Santa Cruz Biotech, Santa

Cruz, CA, as previously described [118].

Cell proliferation, apoptosis, and invasion assays. Cell proliferation assays were
performed using the Cell Titer 96G Aqueous MTS-formazan proliferation assay
(Promega, Madison WI). Cells were cultured in a 96-well plate at 1X103 cells per well
(with eight repeats), and growth curves were established by measuring the absorbance at
490 nm at 24 hr time intervals following the addition of 20ul of MTS reagent. The MTS
reagent is converted by proliferating cells into a forrnazan dye that is read
colorimetrically on a spectrophotometer at 490nm. The absorbance reading corresponds
with the number of viable proliferating cells in the assay.

Apoptosis assays were performed using the Cell Death Detection ELISAPLUS

sandwich ELISA (Roche, Indianapolis IN). Cells were cultured in a 96 well plate at

1X103 cells per well (with eight repeats) for 24 hours, then etoposide was added.

79

Apoptosis was measured following treatment of each cell line with 100 mM etoposide at
12 and 24 hr for MCF10A-derived cell lines and 48 hr for siRNA-induced cell lines.
Cells were lysed with 200ul of lysis buffer, and following centrifugation, 20ul of
supernatant was added to a strepavidin-coated 96-well plate containing 80ul
immunoreagent (anti-histone biotin and anti-DNA-horseradish peroxidase) and incubated
for two hours. During incubation, the anti-histone antibody binds to the histones
contained in nucleosomes in each sample, and also binds the entire complex to the
streptavidin-coated plate by biotinylation. The anti-histone antibody reacts with histones
H1, H2A, H28, H3, and H4 from several different species (man, mouse, rat, hamster,
cow), and the anti-DNA-horseradish peroxidase binds to single- and double-stranded
DNA. ABTS substrate is added to visualize the anti-DNA-horseradish peroxidase, and
the amount of nucleosomes remaining in the sample (apoptosis) correlates with the
absorbance measurement at 490nm of the horseradish peroxidase retained in the complex.

Invasion assays were performed using the BD Matrigel invasion chamber assay
(BD Bioscences, Bedford MA) for the overexpression experiments, or the CytoSelectTM
24-well Cell Invasion Assay (Cell Biolabs Inc, San Diego CA) for the shRNA
knockdown experiments. Cells were cultured in a 24-well Boyden chamber format.
Boyden chambers are composed of an insert well containing a membrane with 4pm
pores, overlain with Matrigel basement membrane extract. Cells were cultured at 1X103
cells per chamber in serum-free media, and inserts were placed in a 24 well plate
containing 10% fetal bovine serum to establish a chemoattractant serum gradient.
Invasion was measured either manually and expressed as a percentage of invading cells

(overexpression experiments) or measured colorimetrically using a spectrophotometer

80

(shRNA knockdown experiments). For manual measurement of invasion, cells were
stained with 400ul of crystal violet and membranes were excised from the Boyden
chambers and mounted on microscope slides. Invasion was measured as the number of
cells present on the underside of the membrane (able to migrate through the Matrigel
matrix and 4pm pores) of the Matrigel inserts compared to the number of cells present on
. the underside of control inserts (migrating through 4pm pores alone). For colorimetric
measurements, cells were stained with crystal violent and then lysed from the membrane
with 200ul lysis buffer. Then lOOul of lysate per sample was measured colorimentrically
as absorbance on a spectrophotometer at 490nm. The absorbance measurement
corresponds with the number of cells that migrated through the Matrigel and/or control
inserts. Overexpression experiments were measured manually and shRNA knockdown
experiments were measured colorimetrically. For statistical analysis of proliferation,
apoptosis, and invasion assays, one-way ANOVA wasperformed on MCF10A derived
cells (comparing four cell lines), and T-tests were performed on shRNA-induced cell

lines (knockdown compared to empty vector control).

Animals, necropsy, and histopathology. We established six cohorts composed of 10
six-week-old female SCID mice (NCI Frederick, MD). Mice from each cohort were
injected in the right axillary mammary fat pad with 1><106 cells from one of the
previously described cell lines (MCF10A+H-Ras, MCF10A+H-Ras+Bmi-l,
MCF10A+H-Ras+Bmi-l+Bmi-li, MCF10A+H-Ras+Bmi-1+pLL3.7, MDA-MB-
231+Bmi-li, MDA-MB-23l+pLL3.7). Tumor growth was measured weekly by caliper.

Additionally, six cohorts of 10 six-week-old female mice each were injected with 1><106

81

cells via the tail vein. Mice were followed for six weeks for signs of clinical illness
(weight loss, decreased body weight, roughened haircoat, difficulty breathing). Mice
injected in the mammary fat pad were humanely euthanized by CO; asphyxiation once
tumors reached 2 cm in diameter, or when mice beCame clinically ill. Tail vein injected
mice were euthanized by C02 asphyxiation once clinical symptoms of weight loss,
roughened hair coat, rapid respiration, or hunched posture were noted. All animal work
was performed following NIH guidelines under an approved animal protocol. Following
euthanasia, tumor (intramammary fat pad injected mice), brain, lung, heart, liver, spleen,
and kidney from intramammary fat pad injected mice were collected, either frozen in
OCT freezing medium or fixed in 4% paraformaldehyde (PF A), processed into paraffin
. blocks, sectioned at 4 pm, and stained with hematoxylin and eosin (H&E). Sections of
tumors were snap frozen in liquid nitrogen for further protein and RNA analysis. Selected
areas were photographed using an Olympus DP70 (Olympus) digital camera. Statistical
analysis of tumor grth was analysed using a Two-Way ANOVA with a Bonferroni test
for multiple comparisons.
Results

Bmi-l and H-Ras collaborate to transform MCF10A human mammary epithelial
cell phenotype in vitro. Overexpression of Bmi-l or H-Ras alone and in combination
markedly altered MCF10A morphology (Figure l3A-E). MCF10A cells were polygonal
to polyhedral, with abundant cytoplasm and large round centralized nuclei, and grew in
plates with clear cell-to-cell junctions (A). Overexpression of Bmi-l resulted in a spindle-
shaped morphology, without cell-to-cell contact (B). H-Ras overexpression resulted in

rounded, triangular, and tear-drop shaped cells (C), whereas overexpression of H-Ras and

82

Bmi-l together resulted in a more pronounced spindle-shaped phenotype, with thin cell
processes often radiating from central clusters (D). After several passages in culture,
MCF10A+H-Ras+Bmi-l cells began to form spherical structures that detach and float
freely in culture (E), similar to mammospheres that .occur in suspension culture and have
been shown to be enriched for stem cells. These structures remained viable in non-
adherent culture, and when disassociated and re-cultured on adherent substrata, returned
to an adherent phenotype; however, additional spheroid structures formed more readily
on subsequent passages (Data not shown).

In three dimensional Matrigel culture, MCF10A cells formed round clusters with
central cells undergoing apoptosis to form a central lumen, and well-polarized cells
around this central lumen (Figure 14A). This phenotype is used to represent normal
differentiation of human mammary epithelial cells into a luminal phenotype [159]. When
Bmi-l or H-Ras were overexpressed alone, clusters remained disorganized, without
central lumen formation or polarization of cells (Figure l4B-C), indicating a block in the
normal MCF10A differentiation. With Bmi-l and H-Ras co-overexpression, cells grew in
haphazard, highly disorganized and proliferative masses with a somewhat spindle-shaped
morphology and frequent mitoses (D), indicating loss of differentiation as well as a
highly proliferative phenotype. Phalloidin staining for f-actin with a DAPI overlay better

illustrates the structure of these clusters by confocal microscopy in Figure 14A.
Bmi-l cooperates with H-Ras to significantly increase cell proliferation and

invasion, and inhibit apoptosis in MCF10A cells. Using the Cell Titer 96 Aqueous

MTS-formazan proliferation assay, overexpression of H-Ras alone in MCF10A cells

83

resulted in a statistically significant increase (p < 0.01) in proliferation in comparison to
parental MCF10A cells at l and 3 days (Figure 15A). At two days, overexpression of
Bmi-l or H-Ras alone led to a statisitically significant increase in proliferation over
MCF10A alone (p < 0.01), and concurrent overeXpression of both Bmi-l and H-Ras
resulted in statistically significant increase in proliferation over all cell lines at all time
points (p < 0.01). Conversely, overexpression of Bmi-l or H-Ras in MCF10A cells led to
a statistically significant decrease in apoptosis in response to DNA damage due to
etoposide treatment compared to MCF10A cells alone, and concurrent overexpression of
Bmi-l and H-Ras led to a further statistically significant decrease in apoptosis (p < 0.01)
at 12hr. At 24hr, overexpression of Bmi-l alone or in combination with H-Ras resulted in
a statisitically significant decrease in apoptosis (p < 0.05), in comparison to MCF10A or
MCF10+H-Ras cells (Figure 15B). The ability of the cell lines to degrade and migrate
through extracellular matrix was measured using the CytoSelectTM 24-well Cell Invasion
Assay (Cell Biolabs Inc, San Deigo CA). While MCF10A cells have negligible invasive
capability in Matrigel (measured as the number of cells that migrated through the
Matrigel barrier), MCF10A cells overexpressing Bmi-l (+Bmi-l) or H-Ras (+H-Ras) had
an 11% and 12% increase in invasion compared to MCF10A cells, and concurrent
overexpression of Bmi-l and H-Ras in MCF10A cells (+H-Ras+Bmi-l) caused a further

three-fold increase in invasion (p < 0.05) (Figure 15C).

Reduction in Bmi-l results in decreased proliferation and increased apoptosis in

vitro. In order to confirm that overexpression of Bmi-l in the MCF10A-derived cells was

directly related to proliferation and apoptosis, short hairpin RNA (shRNA) for Bmi-l was

84

used to reduce the expression of Bmi-l in four cell lines that overexpress Bmi-l:
MCF10A-Bmi-1+H-Ras, MCF7, MDA231, and ZR75-l. MCF10A-Bm-1+H-Ras cells
were used as proof of principle as these cells show increased proliferation and decreased
apoptotic response in vitro in comparison to MCF10A, MCF10A-Bmi-1 and MCF10A-
H-Ras cells alone. Utilizing the Cell Titer 960 Aqueous MTS-formazan proliferation
assay (Promega, Madison WI), proliferation of cells transduced with LV-pLL3.7+Bmi-li
was statistically significantly decreased in comparison to cells transduced with empty
vector alone (LV-pLL3.7) (Figure 16A-D). Decreased proliferative indices in cells
transduced with siRNA for Bmi-l suggests that Bmi-l is partially driving the
proliferation of these human established breast cancer cell lines in vitro.

To assess the apoptotic response to DNA damaging agents, each cell line was
treated with 100 uM etoposide for 24 hours, after which apoptosis was measured using
the Cell Death Detection ELISAPLUS (Roche, Indianapolis, IN). For the MDA231, ZR75-
1, and MCF7 cell lines, there was a trend towards an increase-in the apoptotic response to
etoposide in cells transduced with siRNA for Bmi-l (Figure 17A), suggesting that the
expression of Bmi-l in these established breast cancer cell lines confers a resistance to
apoptosis induced by DNA damaging agents, consistent with our previous findings [118].
Finally, the CytoSelectTM 24-well Cell Invasion Assay (Cell Biolabs Inc, San Deigo CA)
was used to assess the effect of Bmi-l knockdown on invasion. In three of four cell lines
tested, depression of Bmi-l led to a decrease in invasion (Figure 17B). Together, these
data show that Bmi-l plays a role in proliferation, invasion, and resistance to apoptosis in

breast cancer cell lines in vitro.

85

Bmi-l increases metastatic propensity and incidence. We examined the ability of each
of the four above cell lines to develop spontaneous metastases from a primary tumor. As
in our previous data, whereas MCF10A and MCF10A+Bmi-1 cells did not form primary
tumors, MCF10A+H-Ras and MCF10A+H-Ras+Bmi-l cells produced primary tumors in
100% of xenografted animals [118]. In MCF10A+H-Ras xenografted animals, micro- and
macro-metastases were only observed in the livers in 5/10 (50%) and in the spleens of
4/10 (40%) of the mice. In contrast, metastasis to both the spleens and livers were
observed in 10/ 10 (100%) of MCF 10A+H-Ras+Bmi-1 xenografted animals (Figure 18).
Interestingly, spontaneous brain metastasis was found in 3/10 (30%) of MCF10A+H-
Ras+Bmi-l xenografts, whereas no spontaneous metastases to the brain were found in
any of the MCF10A+H-Ras xenografted mice. The study of brain metastasis has been
limited by the lack of cell lines and mouse models that reliably reproduce the metastatic
disease seen in humans [180]. These data show a striking prevalence for brain metastasis
as a result of the overexpression of Bmi-l in collaboration with H-Ras overexpression,
making it a very important target in the study of metastatic disease and aggressive breast
cancer. These data support an increased propensity towards spontaneous metastasis to the
spleen and liver, and novel metastasis to the brain, in cells overexpressing both H-Ras

and Bmi-l, suggesting a role in increased metastasis due to Bmi-l overexpression.

Tail vein injection of MCF10A+H-Ras+Bmi-l resulted in fulminant pulmonary
tumors, whereas injection of other cell lines did not. To further characterize the ability
of these cell lines to escape anoikis, embolize, and develop metastatic foci at distant sites,

we utilized the tail vein injection technique to introduce the cells directly into the venous

86

circulation. A significant decrease in weight was observed over a five week period
following the tail vein injections in the MCF10A+H-Ras+Bmi-1 cohort. Collectively,
these animals lost approximately 30% of their body weight during this period, whereas
the body weight of all other cohorts remained stable (Figure 19A) and the mice remained
clinically normal. The MCF10A+H-Ras+Bmi-l-injected mice developed a roughened
haircoat, hunched posture, and elevated respiratory rate. At necropsy, fulminant gross
metastases were evident involving large portions of all lung lobes of MCF10A+H-
Ras+Bmi-l injected mice at 4-5 weeks (Figure 193). In contrast, lungs of MCF10A+H-
Ras injected mice were grossly unremarkable at 4-5 weeks. Histologically, at 14-16
weeks, MCF10A+H-Ras injected mice had small (l-2mm), expansile, non-infiltrative,
well-differentiated adenomatous lesions. In contrast, at 4 weeks, pulmonary metastases in
the MCF10A+H-Ras+Bmi-l injected mice were composed of sheets and bundles of
poorly-differentiated spindle-shaped cells with ,large pleomorphic nuclei and a high
mitotic rate (2-4/hpf), and effaced large portions of the normal pulmonary architecture
(Figure 19C). These data support the above data that Bmi-l as an important mediator in

the development and progression of metastatic disease in H-Ras transformed cells.

Bmi-l knockdown in MCF10A+H—Ras+Bmi—1 mammary fat pad xenografts slows
tumor progression, but does not alter phenotype. To investigate whether Bmi-l
. depression in MCF10A+H-Ras+Bmi-l cells would slow progression of cell growth in
vivo, we used the previously engineered MCF10A+H-Ras+Bmi-l cells expressing empty
vector (pLL3.7) or shRNA for Bmi-l (Bmi-li) for mammary fat pad xenograft

experiments. While the histologic phenotype of neither mammary fat pad nor intravenous

87

MCF10A+H-Ras+Bmi-l tumors was altered by knockdown of Bmi-l (Figure 20A), at
approximately 20-25 days post-injection, there was a statistically significant decrease in
tumor volume in the xenograft group expressing shRNA for Bmi-l (Figure 20B), that
persisted until the end of the study (p < 0.05, Two-Way ANOVA with Bonferroni
correction). Persistent knock-down of Bmi-l in these tumors was confirmed by western
blot (Figure 22). These data confirm the results of our previous xenograft experiments,
and suggests that the collaboration of Bmi-l and H-Ras may be important during the

early development of some tumors.

Bmi-l knockdown in MCF10A+H-Ras+Bmi-l intravenous xenografts does not alter
metastatic disease or phenotype. There was no significant difference in onset or
progression of clinical signs (weight loss, roughened haircoat, hunched posture, rapid
respiration) or in ultimate metastatic burden between tail vein xenografts expressing
empty vector and those with Bmi-l knockdown. Additionally, there was no significant

difference in histologic phenotype between the groups (Figure 20A).

Bmi-l knockdown in MDA231 xenografts does not alter metastatic incidence or
tumor phenotype, but slows onset of primary tumor formation. The mammary fat pad
and intravenous xenograft experiments were repeated with MDA231 cells expressing
empty vector or shRNA for Bmi-l. In the mammary fat pad xenografts, there was no
significant difference in primary tumor growth or spontaneous metastasis between
xenografts expressing empty vector and those with knockdown of Bmi-l. Additionally,

there was no significant difference between the histologic phenotype of mammary fat pad

88

xenografts expressing siRNA for Bmi-l and those expressing empty vector (Figure 21A).
However, there is a significant delay in the onset of primary tumor formation in tumors
with Bmi-l knockdown (Figure 218). By days 49 and 56 post-injection, 60% (p < 0.05)
and 70% (p < 0.01) of animals with MDA-MB-231 xenografts transfected with empty
vector, respectively, developed palpable tumors, whereas only 10% of animals with
xenografts transduced with shRNA for Bmi-l had tumors at either time point (Two-way
ANOVA). In the tail vein xenografts, there was no significant difference in metastatic
disease between MDA231 tail vein xenografts expressing empty vector and those with
knockdown of Bmi-l, and the histologic phenotype remained the same between groups
' (Figure 21A). Mammary fat pad tumors were harvested for protein analysis, and

knockdown was confirmed by western blot (Figure 22).

Discussion

Our previous work has demonstrated that Bmi-l — cooperates with H-Ras to
transform human mammary epithelial cells through the dysregulation of multiple growth
regulatory pathways, including the AKT and MAPK/ERK pathways, and cell cycle
mediators CDK4 and cyclin D [118]. This is consistent with our findings that
overexpression of Bmi-l and H-Ras leads to increased proliferative rates, and that co-
overexpression of both oncogenes results in a further significant increase in proliferation.
Furthermore, we have shown that overexpression of Bmi-l or H-Ras in MCF10A cells
confers a protective advantage against apoptosis in response to the DNA-damaging agent

etoposide. We have previously shown that MCF10A+Bmi-l+H-Ras have the highest

migratory potential in vitro when compared to MCF10A, MCF10A+Bmi-l, and

89

MCF10A+H-Ras using a wound-healing assay [118]. Given this increased migratory
potential, and a spindle shaped morphology characteristic of an EMT phenotype which is
closely linked to migration and invasion, we performed a Matrigel invasion assay to
determine the invasive capacity of these cells. Compared to MCF10A cells, which have
no invasive capability, our data show that H-Ras and Bmi-l overexpression causes a
significant increase in invasion of MCF10A cells, and that combined overexpression
causes a further significant increase in this invasive capability. Therefore, our data
suggest that overexpression of Bmi-l and H-Ras causes MCF10A cells to acquire an
increased invasion potential consistent with highly transformed HMECs.

Using intramammary fat pad injections in a SCID xenograft model, we showed
that Bmi-l overexpression alone in MCF10A cells was not sufficient to transform
HMECS, whereas overexpression of H-Ras alone and in combination with Bmi-l was
[118]. Interestingly, the gross and histologic morphology of tumors from each group were
strikingly different; tumors arising fi'om MCF10A cells overexpressing H-Ras were
grossly variably cystic and hemorrhagic, composed of vascular, smooth muscle, and mast
cell elements, while tumors arising from MCF10A cells overexpressing both H-Ras and
Bmi-l were grossly solid and homogeneous on cut surface, and histologically were
composed of locally infiltrative streams and bundles of poorly differentiated spindle-
shaped cells with atypical features and a high mitotic rate. Thus, overexpression of Bmi-l
in the face of H-Ras overexpression leads to a more poorly differentiated, more
aggressive phenotype.

The current data show that the overexpression of Bmi-l concurrently with H-Ras

overexpression leads to an increased incidence of spontaneous metastasis to the liver and

90

spleen, and novel metastases to the brain, in comparison to H-Ras overexpression alone
in severe combined immunodeficient (SCID) mice. Overexpression of Bmi-l has been
associated with increased incidence of spontaneous regional lymph node metastasis in
human gastric cancer [181], breast cancer [47, 182], and melanoma [183]. In addition,
Bmi-l has recently been associated with an ll-gene microarray signature indicative of
short interval to disease recurrence, metastasis, and poor response to therapy in prostate,
breast, lung, ovarian, and bladder tumors, as well as lymphoma, mesothelioma,
medulloblastoma, glioma, and acute myeloid leukemia [184]. This is consistent with our
xenograft data showing increased propensity of metastasis to liver, spleen, and brain.
However, increased metastasis to regional lymph nodes was not seen in our model. This
may be due to certain limitations with xenograft models, including impaired immune
responses in immunocompromised mice, the lack of a humanized stromal component in
the murine mammary gland, or because human cells are not firlly adapted to growth in
mice [185]. However, in our xenograft model, metastasis to the brain is of particular
interest, because there are few reliable models of spontaneous brain metastasis, and this is
a very important route of metastasis in human breast cancer. The incidence of clinical
brain metastasis in human breast cancer is approximately 10% [186], and nearly 30% of
patients with metastatic breast cancer have post-mortem evidence of metastasis to the
brain [187, 188]. Furthermore, using tail vein injections as a model for direct
hematogenous spread, we have shown that pulmonary tumors developed rapidly and
early (4-5 weeks) in the MCF10A+H-Ras+Bmi-l xenografts, but were still not grossly
evident until 14-15 weeks in MCF10A+H-Ras xenograft mice. This shows that

overexpression of Bmi-l in conjunction with H-Ras overexpression leads to the rapid

91

development of fulminant pulmonary metastasis leading to clinical illness, whereas
overexpression of H-Ras alone is not capable of inducing significant embolic pulmonary
metastasis. This is consistent with behavior of the MCF10A derived lines of cells
developed by Miller et al. Miller et al. (1993) derived the MCF10A immortalized normal
(non-transformed) breast epithelial cell line from a patient with fibrocystic disease.
Similar to the MCF10A+H-Ras cell line used in our experiments, MCFlOAtlk cells were
subsequently created by stable transfection with the neomycin-resistance gene and the
mutated T-24 Ha-ras gene [99, 189]. These cells form premalignant lesions that rarely
progress to carcinomas when injected into the mammary fat pad, and do not populate the
lung on tail vein injection [179]. Similar to Miller’s MCFlOAtlk cell line, our
MCF10A+H-Ras cell line does not result in significant pulmonary metastasis. This
indicates that phenotype of severe pulmonary metastasis is related solely to the
combination of overexpression of Bmi-l and H-Ras, since cells overexpressing Bmi-l
alone do not form tumors on tail vein injection. These data demonstrate that the
collaboration of Bmi-l with H-Ras plays an essential role in the development of distant
metastasis from a primary tumor, and that this collaboration is essential in the metastatic
survival and grth of MCF10A HMECs in vivo. Additionally, these data suggest that
breast cancers that overexpress both H-Ras and Bmi-l may be more metastatic. ’

Lastly, we wanted to determine the importance of Bmi-l in the behavior of
established human breast cancer cell lines overexpressing Bmi-l. It has been shown that
several breast cancer cell lines overexpress Bmi-l, and that overexpression of Bmi-l
induces telomerase activity and extends the replicative lifespan of mammary epithelial

cells [29]. This would suggest that Bmi-l overexpression may exert its effects on

92

telomerase in parental breast cancer cell lines that overexpress the protein. Thus,
knockdown of Bmi-l expression would potentially have a negative impact on the
proliferative capacity of Bmi-l-overexpressing breast cancer cell lines. To address this
question, we used small interfering RNA (siRNA) specific for Bmi-l, and achieved
substantial knockdown of Bmi-l expression at the protein level in the MCF10A-Bmi-
1+H-Ras cells, as well as three established breast cancer cell lines (MCF 7, ZR75-1, and
MDA-MB-231) based upon their relatively high levels of protein expression. In vitro,
through the use of MTS-formazan proliferation assays, we have shown that by decreasing
the expression of Bmi-l in these cell lines, rates of proliferation decreased significantly
(p < 0.05) in comparison with empty vector controls. This is consistent with our previous
findings that overexpression of Bmi-l in collaboration with H-Ras overexpression
dysregulates multiple growth pathways, including the ERK/MAPK and Akt pathways,
and upregulates cell cycle mediators such as CDK4 and cyclin D [118].

Conversely, using an anti-histone sandwich ELISA to measure apoptosis in
response to etoposide, we have shown that cell lines transduced with siRNA for Bmi-l
are more sensitive to apoptosis as a result of DNA damage, thus suggesting that Bmi-l
confers resistance to ap0ptosis, either as a result of cell cycle checkpoint regulation
through p53 dependent pathways, or by influencing DNA repair. Indeed, it has been
shown that Bmi-l suppresses the Ink4a/Arf tumor suppressor locus, which encodes for
p16 and p19, and that suppression of the locus leads to inhibition of retinoblastoma
phosphorylation and p53 degradation, respectively, leading to inhibition of cell cycle

arrest and resultant apoptosis or cellular senescence [18, 27, 30]. Additionally, we have

93

shown that Bmi-l confers a protective effect against oncogene-induced senescence and
apoptosis [1 18].

In human patients, overexpression of Bmi-l has been correlated with metastatic
disease in several types of cancer [184], and has‘been associated with regional lymph
node metastasis in breast cancer [47]. Consistent with this, our data show that by
depressing Bmi-l expression in certain breast cancer cell lines, the ability of these cells to
degrade and migrate through Matrigel is compromised in vitro. Taken together, our
results suggest that by knocking down Bmi-l expression, rates of proliferation, resistance
to apoptosis, and invasive capability decrease in established breast cancer cell lines.

Next we sought to determine if decreasing the expression of Bmi-l in established
breast cancer cell lines would lead to altered biologic behavior. The MDA-MB-23l
breast cancer cell line is an aggressive established cell line that forms tumors in vivo and
has metastatic behavior in both infra-mammary fat pad and tail vein xenograft animal
models [190]. Since we hypothesized that depressing Bmi-l expression in xenografts
would lead to a less aggressive phenotype, we used this cell line and the MCF10A+H-
Ras+Bmi-l cell line to evaluate for any decrease in tumor burden or metastatic disease in
mammary fat pad and tail vein xenograft models.

In the MCF10A+H-Ras+Bmi-l xenograft experiments, there was no difference in
tumor incidence or phenotype in mammary fat pad or intravenous xenografts. The lack of
difference in tumor incidence may be in part an effect of continual Bmi-l expression in
the face of H-Ras overexpression. The overexpression of Bmi-l in these cells may be too
great to overcome with the applied shRNA for Bmi-l. Likewise, the phenotype was

unable to be reverted to the previously shown MCF10A+H-Ras phenotype [118], likely

94

due to the effects of continued Bmi-l expression, although at lower levels, suggesting
that Bmi-l can act as a potent oncogene even when depressed to some degree. However,
there was a significant effect on primary tumor progression in the mammary fat pad
xenograft model, where animals with tumors expressing shRNA for Bmi-l had a
statistically significant decrease in tumor volume over time that persisted until the end of
study. This is an important biological indication that, when overexpressed in vivo, one of
the main fiinctions of Bmi-l is to exert a powerful proliferative effect on mammary
epithelial cells. The fact that all animals ultimately still developed mammary tumors may
be due to the possibility that the overexpression of Bmi-l is still too great in the face of
shRNA expression, or that the collaborative effect of H-Ras induced proliferation is
affecting the tumor cell biological behavior.

In the MDA-MB-23l cell line, shRNA knockdown of Bmi-l had no significant
affect on tumor grth over time, incidence of spontaneous or tail vein-induced
metastasis, or histologic phenotype. This may be due to several factors. First, it may be
that that Bmi-l alone does not have a prominent effect on MDA-MB-231 cell behavior in
the tumor microenvironment. Despite the in vitro findings of decreased proliferation,
increased apoptosis, and decreased invasion with shRNA knockdown of Bmi-l, it is not
unusual that the in vivo cell behavior should differ substantially from the in vitro cell
behavior. The MDA-MB-231 cell line has been shown to differ significantly in growth
rates and behavior in vitro compared to in vivo in SCID mouse models [190]. It is also
likely that several factors play a role in the overall aggressive and metastatic behavior of
these cells in vivo. Interestingly however, is the finding that the time to onset of

tumorigenesis in our mammary fat pad SCID xenograft model is significantly increased

95

in MDA-MB-231 cells that have shRNA induced knockdown of Bmi-l. Although this
effect does not persist throughout the length of the experiment, it does show a significant
outcome of Bmi-l depression in these cells, suggesting that Bmi-l may be involved in
the early stages of tumorigenesis. However, we cannot rule out the possibility that
expression of the shRNA for Bmi-l was lost during a selection event in vivo that selected
for more proliferative cells not expressing the shRNA. This would lead to an outgrowth
of more highly proliferative cells that eventually would lead to increased tumor growth.

In conclusion, we show that the collaboration of Bmi-l and H-Ras overexpression
in human mammary epithelial cells leads to a more aggressive phenotype in vitro and in
vivo, and induces a more highly metastatic phenotype. We also show that depression of
Bmi-l in several breast cancer cell lines leads to a less aggressive phenotype in vitro,
including decreased proliferation and invasion, and increased apoptosis as a result of
DNA damage. In vivo, we have shown that depression of Bmi-l in certain breast cancer
cell lines leads to a change in biologic behavior related to tumor progression or time to
tumor onset in xenograft models. These data support our hypothesis that Bmi-l plays a
significant role in the biologic behavior of certain established breast cancer cell lines in
vitro and in vivo. Should Bmi-l play an important role in the early growth of tumors or in
the sustainability of tumor progression over time, this would serve as an important target

of study in breast cancer disease, especially related to tumor recurrence and dormancy.

96

APPENDIX C

Figure 13: Morphology of MCF10A-derived cell lines in two dimensional culture.
MCF10A cells (A) grow in typical epithelioid fashion, with a polygonal shape, clear cell-
to-cell junctions, abundant cytoplasm, and round centralized nuclei; MCF10A+Bmi-l
cells (B) acquire a mesenchymal appearing spindle cell morphology; MCF10A+H-Ras
cells (C) are rounded, tear-drop shaped to triangular; MCF10A+H-Ras+Bmi-1 cells (D)
have an exaggerated spindle shaped morphology, often radiating outward from central
clusters; spheroid structures (E) arising from MCF10A+H-Ras+Bmi-1 cells after several
passages in culture.

 

 

 

 

 

 

 

 

 

97

Figure 14: Morphology of MCF10A-derived cell lines in three-dimensional Matrigel
culture. Hematoxylin and eosin staining (left column) and phalloidin
immunohistochemistry under confocal microscopy (right column): MCF10A cells show a
typical differentiation pattern with central lumen formation and polarization of cells
around this central lumen; MCF10A+Bmi-l (B) and MCF10A+H-Ras cells (C) remain
small disorganized clusters without lumen formation or polarization; MCF10A+H-
Ras+Bmi-1 cells (D) form disorganized highly proliferative clusters of cells with
somewhat spindle-shaped morphology and high mitotic rate.

 

 

 

 

 

 

 

 

98

Figure 15: Proliferative indices, apoptotic response to DNA damage, and invasive
properties of MCF10A-derived cell lines in vitro. (A) MTS-formazan proliferation assay.
MCF10A cells overexpressing H—Ras have significantly increased proliferation over
MCF10A cells, and overexpression of both Bmi-l and H-Ras causes further increase in
proliferation at days 1, 2, and 3 (p < 0.01) ns = not significant. (B) Anti-histone cell death
detection ELISA. MCF10A cells overexpressing Bmi-l or H-Ras have a decreased
' apoptotic response to DNA damage, and overexpression of both induces a firrther
significant decrease in the apoptotic response at 12 hr (p < 0.01). At 24 hr, Bmi-l
overexpression or Bmi—l+H-Ras overexpression exerts a protective effect against
apoptosis (p < 0.05) ns = not significant. (C) Matrigel Invasion Assay. MCF10A cells
overexpressing Bmi-l and H-Ras have a 12% and 11% increase in invasiveness, and
MCF10A cells overexpressing both Bmi-l and H-Ras .have a 31% increase in
invasiveness over MCF10A cells alone.

    

A' 2.0 PMCHOA T
A ‘EIMCF10A-Brri-1
E 1.6 LDMCF1OA-H-Ras l
§ llMCF1OA—H-Ras+Bmi-1
V1.2 % *

8

5 ns
9 0.8

O

(D

.D

< 0.4

0 I 1 2 3

Time (days)

99

Figure 15 (continued)

B.

Absorbance (490nm)

.0

Percent invasion

1.2
1.0
0.8
0.6
0.4
0.2

0

35
30
25
20
15

01

j "W ‘ “7 ns
iIMCI=10A E {—— n8
1 1:1 Bmi-1
in H-Ras

L

 

I H-Ras+Bmi-1

 

 

Time (hours)

31%

12%

11%

+Bmi-1+H-Ras +H-Ras
+Bmi-1

100

Figure 16: MTS-formazan proliferation assay in siRNA-Bmi-li induced established
breast cancer cell lines. In MCF10A+H-Ras+Bmi-l (A), MCF7 (B), MDA231 (C), and
ZR75-1 (D) cell lines, induction of siRNA for Bmi-l induces a decrease in proliferation
over time (*p < 0.05, **p < 0.01) in comparison to cell lines transduced with empty
vector alone.

A- 2-5 IMCF10A+Briri+Ras+pLL37
2.0 DMCF10A+Bmi+>Ras+Bmi-1i
1 5

0.5: .
.11 I] l, , -
1 3 4 5 6

 

 

Time (days)
B, A ‘ leCFlipLLIil
E 12, DMCF7+Bmi-1i n
E ‘ ' ’"TTA' 1
c . firth *‘I‘
g 0.8 , {,
§ 0.4]
< 1
0 . . . . . ,,,.
1 2 3 4
Time (days)
C. 16 ; lMDA231+pLL3.7 ..
E ' ‘ DMDA231+Bmi-1i ,
5 1.2 M .. *1
§ 1.0 . . 1
3
g 0, , ,, . , , ..
1 2 3 4 5 6
Time (days)
D. A 1-2 i'zmatanaii
g 1-0 nZR7s1+ami-1i ..
V 0.8 “fi’ ..
i ~, 2 i
g 0.6
o 0.4
3 .
< 0.2l
0 1 2" 3 ' 4
Time (days)

101

Figure 17: Anti-histone cell death detection ELISA and Matrigel Invasion Assay in
siRNA-Bmi-li induced established breast cancer cell lines. (A) In each cell line
transduced with siRNA for Bmi-l, there is a significant increase in apoptosis in response
to DNA damage compared to cells transduced with empty vector alone. (B) Matrigel
Invasion Assay. There is a significant decrease in invasiveness in the MCF7,
MCF10A+H-Ras+Bmi-l, and MDA231 cell lines transduced with siRNA for Bmi-l in
comparison to cells transduced with empty vector alone (p < 0.05).

A.

(D

 

l

E] pLL3.7
I Bmi-1i

Ll“

Ras+Bmi MCF7 ZR75—1 MDA231

l

 

 

 

l

l

l

l

Apoptosis (fold change)

O—‘NCDhUICDNCD

l

 

[j pLL3.7
I Bm‘-1i

 

 

0.5

Absorbance
.0 O .O
N 00 A

3(-

.0
d

 

 

MCF7 Ras+Bmi ZR75-1 MDA231

102

Figure 18: Spontaneous metastasis from intra-mammary fat pad MCF10A+H-Ras and
MCF10A+H-Ras+Bmi-l xenografis. In the MCF10A+H-Ras+Bmi-l xenograft model,
spontaneous metastasis occurred to the liver and spleen in 100% of animals, and brain
metastasis occurred in 30%. In contrast, in the MCF10A+H-Ras xenograft model,
metastasis to the liver occurred in 50%, and to the spleen in 40% of animals, while no
brain metastases were observed (Bar = 100 pm).

H-Ras+Bmi-1

    

Liver

  

103

Figure 19: MCF10A+H-Ras and MCF10A+H-Ras+Bmi-l tail vein xenografi model. (A)
Graph of body weight over time. By 4-5 weeks post-injection, mice injected with
MCF10A+H-Ras+Bmi-l cells lost ~30% of their body weight (*) in comparison with
other xenografts. (B) Grossly, mice with MCF10A+H-Ras+Bmi-1 tail vein xenografts
had fulminant pulmonary metastases involving all lung lobes, while MCF10A,
MCF10A+Bmi-l, and MCF10A+H-Ras tail vein xenografts had grossly unremarkable
lung fields. (C) Histologically, mice with MCF10A+H-Ras tail vein xenografts had
small, random, expansile lesions resembling adenomas or hyperplastic foci, while
MCF10A+H-Ras+Bmi—l xenografts had aggressive lesions that effaced large portions of
the lung (Bar = 100 um).

 

25
A.
’«T 20 -.
E
9
g» 15 MCF10A
E :1:
‘9 - -MCF10A+Bmi-1
g 10 +
>
8 MCF10A+H-Ras
m 5
+MCF1OA+H-Ras+Bmi-1
o
1 2 3 4 5

Post-injection time (weeks)

 

104

Figure 20: Mammary fat pad and tail vein model using MCF10A+H—Ras+Bmi-
1+pLL3.7 and MCF10A+H-Ras+Bmi-l+Bmi-1i cells. Histologically, there is no
difference in morphology between xenografts transduced with empty vector and those
transduced with siRNA for Bmi-l in mammary fat pad (SQ) (Bar = 100 um, inset 40X
objective) or intravenous (IV) models (A) (Bar = 200 um, inset 40X). Mammary fat pad
xenografts with shRNA knockdown of Bmi-l show a statistically significant decrease in
tumor volume over time (B) (Two-way ANOVA with Bonferroni correction*p < 0.05,
”p < 0.01, ***p < 0.001).

A. Ras+Bmi+

. |

pLL3.7 Ras+Bmi+Bmi-1

     
  
   
 
  
 
 
 

B. 2500' ***
p -E- Ras+Bm' + pLL3.7
E 2000' +Ras+3mi + Brri-1i
a:
g 1500- M”
15
a 1000-
E
.3 500-

 

 

0 10 20 30 40
Post-injection time (days)

105

Figure 21: Mammary fat pad and tail vein model using MDA231+pLL3.7 and
MDA231+Bmi—li cells. Histologically, there is no difference in morphology between
xenografts transduced with empty vector and those transduced with siRNA for Bmi-l in
mammary fat pad (SQ) (Bar = 100 um, inset 20X) or intravenous (IV) models (A) (Bar =
200 pm). Mammary fat pad xenografis with shRNA knockdown of Bmi-l show a
statistically significant lag in tumor formation at 60-65 days post-injection (B) (Two-way
ANOVA with Bonferroni correction, *p < 0.05, "p < 0.01).

A. MDA231+pLL3.7 MDA231+Bmi-1i

’-

       
 
  
 

A

 

 

   

8' 5 33: -B-MDA231+pLL3.7 **
E +MDA231+Bmi-1i
a 701
5 60-
E 50-
$ 40-
E 30-
g 20-
° 10- - _
°\ 0. A.

 

010 20 30 40 50 60 70
Post-injection time (days)

106

Figure 22: Western blot illustrating stable siRNA knock down of Bmi-l expression in
MCF10A+H-Ras+Bmi-l and MDA231 xenografts.

Bmi-1 C '

B—tubulin

 

107

CHAPTER FOUR

Bmi-l contributes to the outgrowth of embolic human breast cancer cells in the

lungs of SCID mice through increased proliferation and prevention of apoptosis

Abstract: Evasion of anoikis (apoptosis in response to lack of adhesion or inappropriate
adhesion) is an important mechanism by which tumor cells persist in the vasculature and
at distant metastatic sites. Bmi-l has been reported to be associated with anoikis
resistance, and we have previously shown that MCF10A cells overexpressing H-Ras and
Bmi-l (MCF 10A+H-Ras+Bmi-1) are able to survive and form aggressive distant
metastases in the lungs of tail vein-injected SCID mice in a xenograft model of direct
hematogenous spread, in contrast to MCF10A cells overexpressing Bmi-l
(MCF lOA+Bmi-1) or H-Ras (MCF10A+H-Ras) alone. In this study, by using an ex-vivo
imaging system, we show that increased proliferation and resistance to apoptosis, or
anoikis, are two primary mechanisms enabling MCF10A+H-Ras+Bmi-l cells to survive
and proliferate in the pulmonary vasculature ‘to form fulminant metastases leading to

clinical morbidity.

Introduction
Metastasis and dormancy are particularly important areas of study in breast cancer
research, because metastasis remains the major cause of mortality in patients with breast
cancer [191]. In addition, although clinicians have seen an overall improvement in long-
term survival over the years due to improved treatments [192], many patients with

minimal disease (small primary tumor, no evidence of regional lymph node involvement,

108.

no detectable metastatic disease) can have recurrence rates up to 30% after being
followed for 10-15 years [193-195]. The major cause of mortality in patients with breast
cancer is metastasis [191] Although the molecular mechanisms leading to metastasis of
breast cancer are poorly understood, the cellular mechanism of metastasis generally
follows several steps in its pathogenesis: invasion of surrounding tissue, intravasation
into the circulation, avoidance of adherence-independent apoptosis (anoikis), arrest in the
circulation at distant sites, extravasation, and growth within a new microenvironment
[191].

We have previously shown (Chapter 3) that human mammary epithelial cells
engineered to overexpress H-Ras concurrently with Bmi-l (MCF10A+H-Ras+Bmi-l)
exhibit aggressive tumorigenesis and metastatic disease when injected into SCID mice
when compared with MCF10A cells that overexpress H—Ras alone (MCF10A+H-Ras).
We have also shown that there is a significant increase in tumorigenesis, aggressiveness,
and metastasis of xenograft tumors produced by MCF10A+H-Ras+Bmi-l cells in both
mammary fat pad and tail vein-induced models. MCF 10+H-Ras+Bmi-1 tumors are
poorly differentiated and locally infiltrative, and in the tail vein model cause severe and
rapid pulmonary metastasis that effaces and destroys large portions of all lung lobes in
contrast to MCF10A+H-Ras cells, which develop small adenomatous or hyperplastic
lesions only after extended periods of time. Consistent with these data, overexpression of
Bmi-l has been associated with regional lymph node metastasis in patients with
infiltrative ductal carcinoma [47], and has also been associated with an 11 gene
microarray signature that is a powerful predictor of a short interval to disease recurrence,

marked propensity towards metastatic dissemination and death following therapy in

109

patients with multiple different types of cancer [48]. Additionally, it has been shown that
Bmi-l overexpression is associated with anoikis-resistance of circulating human prostate
carcinoma precursor cells, and that human prostatic carcinoma cells with siRNA
knockdown of Bmi-l are less resistant to anoikis,'possibly due to a reduction in Survivin
[196]. These data suggest that Bmi-l is an important regulator of metastatic spread,
survival, and growth in distant sites.

To determine the mechanisms leading to the striking differences in pulmonary
phenotypes between MCF10A+H—Ras and MCF10A+H-Ras+Bmi-l cells, we used a
SCID mouse tail vein xenograft model to determine the ability of disseminated cells to
intravasate and grow in the lungs. Each cell line was transduced with a lentivirus
expressing green fluorescent protein (GF P), and injected into the tail vein of SCH) mice.
Using an ex-vivo imaging system, cells were followed in the lungs over several weeks,
and sections of lung were examined for changes in proliferation and apoptosis by Ki67
immunolabeling and TUNEL staining, respectively. We report here mechanisms leading
to the genesis of an aggressive metastatic phenotype due to Bmi-l expression in

collaboration with H—Ras.

Materials and Methods
Cells, constructs, and animals: MCF10A-H-Ras and MCF10A-Bmi-1+H-Ras cells
were labeled with a GFP-tagged lentivirus and sorted by Fluorescence Activated Cell
Sorting (FACS), gating for high GFP expression. Two cohorts of 10 six-week-old female -
mice each were used, and each cohort received 1><106 MCF10A-H-Ras+GFP or

MCF10A-Bmi-l+H-Ras+GFP cells via tail-vein injection. Mice were euthanized at two

110

days, one week, and four weeks after injection. Lungs were harvested at the time of
necropsy, infused with 1 ml 1>< PBS, and visualized under a fluorescent microscope to
detect GFP-labeled cells (Zeiss, Thomwood NY). GFP signal was measured at 4X
magnification and quantified using ImagePr'o image analysis sofiware (Media
Cybernetics, Bethesda MD) and expressed as average pixels per field. Statisitical analysis
was performed using GraphPad Prism 4.02, and Student’s T-test was performed on all

samples.

Immunohistochemistry and TUNEL assay: Paraffin-embedded and frozen sections of
primary tumors and lungs with metastatic lesions were stained with a rabbit polyclonal
antibody for Ki67 (Vector Laboratories) and stained with a TUNEL ApopTag in situ kit
(Chemicon, Temecula CA). Formalin-fixed, paraffin-embedded sections were
deparaffinized in xylene and rehydrated in graded alcohol. Frozen sections were thawed
and fixed in 4% paraformaldehyde for 10 min. Antigen retrieval was performed by
microwaving in a pressure cooker for 4 min in citrate buffer (pH 6.0), followed by
cooling for 25 min. Endogenous peroxidases were blocked with 1.5% H202 in methanol
and protein block was performed with 10% goat serum for 30 min. Sections were
incubated in primary antibody (1210,000) at RT overnight in 10% blocking serum, and
biotinylated goat anti-rabbit secondary antibody (1:1000) in 10% blocking serum for 30
min. Following washing, the labeled avidin-biotinylated enzyme complex (ABC)
technique was employed for amplification of primary antibody binding. For visualization,
3,3-diaminobenzidine (DAB) was applied and counterstained with Gill’s hematoxylin.

For Apoptag (TUNEL) staining, proteolytic digestion was done with proteinase K (20

111

jig/ml) for 15 min at RT, and quenched in 2% H202 in PBS for 5 min at RT. l><
Equilibration Buffer was applied and sections were incubated for 15 min at RT. Sections
were then incubated with TdT enzyme for 30 min at 37 °C, washed in stop buffer for 30
min, and incubated with anti-digoxigenin-pe’roxidase for 30 min at RT. 3,3-
diaminobenzidine (DAB) was applied and counterstained with lithium carbonate. The
sections were dehydrated through graded alcohols, immersed in xylene, and mounted
with coverslips. Ki67 and Apoptag staining was assessed at 10X magnification and signal
was quantified either as labeled cells per field, or by using ImagePro image analysis
software (Media Cybernetics, Bethesda MD) as labeled pixels per field. Statisitical
analysis was performed using GraphPad Prism 4.02 (La Jolla CA), and Student’s T-test

was performed on all samples.

Results
Bmi-l contributes to the outgrowth of embolic breast cancer cells in the lungs. To
determine whether the mechanism leading to the fulminant metastatic phenotype in
MCF10A+Bmi-l+H-Ras is due to the effect of Bmi-l on proliferation and prevention of
apoptosis, MCF10A-Bmi-l+H-Ras and MCF10A+H-Ras cells were lentivirally tagged
with GFP and flow sorted for GF P expression. 1><106 cells were administered via tail vein
injection to two cohorts of ten female SCID mice each. Mice were sacrificed two hours,
two days, one week, and four weeks following tail vein injection, and lungs were
examined by ex-vivo imaging to detect the presence of GFP-expressing metastatic foci
using a Zeiss Axiovert 200 inverted microscope, Hamamatsu imaging camera, and

ImagePro sofiware to quantify the GF P signal (Figure 24A-B). By day two post injection,

112

there was a significant decrease in the number of cells in the lungs of MCF10A+H-Ras
injected mice, indicating massive die-off (p < 6X10'6) compared to MCF10A+H-
Ras+Bmi-l injected mice. At one and four weeks following injection, MCF10A+H-
Ras+Bmi-1 cells continued to proliferate, ultimately culminating in large tumor masses,
in contrast to MCF10A+H-Ras injected mice, in which few cells were detectable. These
data suggest that the mechanism of action of Bmi-l overexpression on MCF10A cells
overexpressing H-Ras is a combination of decreased cell death and increased
proliferation, leading to the ability to escape anoikis, implant in the pulmonary

vasculature, and result in severe metastatic pulmonary disease.

MCF10A+H—Ras+Bmi—l xenografts have increased proliferation and decreased
apoptosis compared to MCF10A+H-Ras xenografts at end-stage disease. To answer
the question of whether the difference in phenotype in the mammary fat pad xenografts
and tail vein model of metastasis (Chapter 3) is due to Bmi-l overexpression leading to
an increase in proliferation, increased resistance to apoptosis, or both, we examined Ki67
immunohistochemistry and TUNEL staining in pulmonary tumors at disease endpoint (5
weeks post-injection for IV xenografts and approximately 60 days after injection for
mammary fat pad xenografis). There was significantly greater staining for Ki67 in
MCF10A+H-Ras+Bmi-l fat pad (p < 1.3X10'7) and intravenous (p < 0.0002) xenografts
compared to MCF10A+H-Ras xenografts (Figure 25A-B), indicating that one mechanism
of collaboration between H-Ras and Bmi-l in vivo is through induction of proliferation.
In contrast, TUNEL staining was significantly increased (p <.0.0002) in mammary fat

pad MCF10A+H-Ras xenografts in comparison to MCF10A+H-Ras+Bmi-l xenografis.

113

The increased cellularity of tumors and decreased TUNEL staining suggests that
prevention of apoptosis is a major mechanism of tumor promotion in MCF10A+H-
Ras+Bmi-l mammary fat pad xenografts. There was a lack of significant TUNEL
staining in either intravenous model (Figure 26A—B), suggesting that these cells may have
been selected over time in vivo forthe ability to escape anoikis within the pulmonary
vasculature. Taken together, growth of a primary tumor within the mammary fat pad as
well as within the pulmonary vasculature is significantly impacted by Bmi-l
overexpression, suggesting that this collaboration between H-Ras and Bmi-li is essential

for the survival and growth of metastatic cells in vivo.

MCF10A+H-Ras+Bmi-1 xenografts have increased proliferation and decreased
apoptosis compared to MCF10A+H-Ras xenografts early in the process of
hematogenous spread. To determine the difference in kinetics of intravascular growth
over time leading to the phenotype observed at disease endpoint, frozen OCT-embedded
lungs collected fi'om tail vein-injected mice at two days and one week after injection were
stained for Ki67 and TUNEL. There was a statistically significant increase in Ki67
labeling at two days (p < 0.01) and one week (p < 8X10"2) in intravascular tumor cells in
the MCF10A+H-Ras+Bmi-l group as compared to the MCF10A+H-Ras group,
indicating that Bmi-l overexpression imparts a proliferative advantage early in the course
of hematogenous spread (Figure 27A). TUNEL staining demonstrated that there was a
statistically significant decrease in apoptosis in intravascular tumor cells (p < 0.05) in the
MCF10A+H-Ras+Bmi-l group as compared to the MCF10A+H—Ras group (Figure 278),

suggesting that Bmi-l overexpression imparts a protective effect against apoptosis

114

resulting from anchorage dependence, or anoikis, an important mechanism in metastatic
spread. Taken together, these results show that Bmi-l overexpression exerts a
proliferative and anti-apoptotic effect which, in the presence of I-I-Ras overexpression,
leads to an aggressive and fatal pulmonary phenotype. This indicates that the
collaboration of Bmi-l and H-Ras in mammary epithelial cells is essential for metastatic

survival and growth.

Discussion

We have shown that, in this tail vein xenograft model, Bmi-l collaborates with H-
Ras to produce an aggressive metastatic phenotype. While MCF10A cells overexpressing
H-Ras at most form micrometastases resembling adenomas and hyperplastic foci, cells
overexpressing both H-Ras and Bmi-l rapidly produce poorly differentiated pulmonary
metastases. Animals receiving cells overexpressing H-Ras alone survive for long periods
of time and do not develop clinical illness, whereas animals receiving cells
overexpressing H-Ras and Bmi-l concurrently develop clinical morbidity early in the
course of disease, characterized by rapid weight loss, hunched posture, and rapid
respiration.

One mechanism leading to the difference in this pulmonary phenotype is the
marked increase in proliferative ability of MCF10A+H-Ras+Bmi—l cells. The results of
ex-vivo imaging show that these cells continue to proliferate in an exponential manner
over the course of the disease. We have previously shown that these cells have an
increased proliferative capacity in vitro in comparison to MCF10A cells overexpressing

Bmi-l or H-Ras alone [118], and the results of Ki67 immunohistochemistry support this

115

biologic behavior in vivo. It is known that Ras activation leads to activation of multiple
growth and differentiation pathways as a result of signaling through multiple effectors
such as the Raf kinase and MAPK/ERK signaling cascade [197]. In our study, we
document Ki67 immunolabeling in the tail vein injected MCF10A cells overexpressing
H-Ras both early in the disease and at disease endpoint. However, there is a marked
increase in Ki67 immunolabeling in MCF10A+H-Ras+Bmi-l cells within the pulmonary
vasculature both early in the disease and at end-stage in comparison with MCF10A+H-
Ras cells. It is known that Bmi-l can act as an oncogene when overexpressed or
dysregulated, through its suppression of the Ink4a-Arf tumor suppressor locus [29, 118].
This locus encodes for pl4/pl9ARF and pl6INK4A, which doWnregulate MDM2 and
CDK4-6/cyclin D binding, respectively (Figure 28). MDM2 is a major suppressor of p53
by binding directly to it and preventing its transcription [198]. Inhibition of CDK4-
6/cyclin D binding results in prevention of Rb phosphorylation, and together with
inhibition of p53, causes cell cycle arrest. Thus, by inhibiting this locus constitutively,
Bmi-l overexpression results in a highly proliferative state with the cell remaining in the
cell cycle. We have previously shown that Bmi-l overexpression in vitro increases
proliferation of MCF10A cells and that Bmi-l collaborates with H-Ras through
dysregulation of multiple growth pathways such as PI3K, Akt, and MAPK/ERK, and cell
cycle regulators such as cyclin D and p16 [118]. Ries et al. (2002) have shown that
MDM2 is also driven by the Raf/MAPK/ERK pathway downstream from Ras signaling.
They show that activated Raf induces MDM2 to degrade p53 in the absence of
p14/19ARF. This results in inhibition of cell cycle arrest through a p53-dependent

mechanism. Taken together, the collaboration of H-Ras and Bmi-l on both p53 function

116

as well as through dysregulation of several growth pathways results in fulminant and
destructive metastases in the tail vein model of direct hematogenous spread.

Another mechanism leading to the aggressive metastatic phenotype is the ability
of MCF10A+H-Ras+Bmi-l cells to evade anoikis. Anoikis is apoptosis in response to
lack of adhesion or inappropriate adhesion [199]. Other than providing a physical
scaffold on which cells reside, signaling between components of the extracellular matrix
(ECM) and epithelial cells is important in the prevention of apoptosis in normal
biological processes such as proliferation and differentiation, as well as in metastasis.
Important mediators in cell-matrix signaling in prevention of apoptosis are the focal
adhesion kinases (FAKs), which mediate the fiinction of several downstream pathways
associated with growth and prevention of apoptosis, such as MAPK/ERK, PI3K, and Src
[199, 200]. Ras also exerts its proliferative effects through a shared downstream pathway,
the MAPK/ERK cascade. However, in our study, we have demonstrated that MCF10A
cells overexpressing H-Ras undergo massive cell death following tail vein injection. This
may result from the conflicting proapoptotic and antiapoptotic functions of Ras when
overexpressed. The proliferative effects of Ras signaling through the MAPK/ERK and
PI3K pathways are well known [197], and p53-independent anti-apoptotic effects have
been shown through Raf induced upregulation of MDM2, leading to p53 degradation
[198]. However, multiple pathways exist for Ras induction of apoptosis. For example,
Raf also activates the MDM2 inhibitor, p14/19ARF, leading to MDM2 inhibition and
accumulation of p53, resulting in cell cycle arrest [198]. In addition, overexpression of
constitutively active Ras in established cell lines results in cyclin-dependent kinase

inhibitor induction of cell cycle arrest, or in apoptosis [167, 201 , 202]. Overexpression of

117

Ras may lead to apoptosis through p53 dependent pathways via p14ARF activation and
subsequent MDM2 inhibition, or through p53 independent pathways [203, 204].
Nevertheless, our data support that in the balance of pro-apoptotic and anti-apoptotic
signaling through Ras, in the in vivo environment of the pulmonary vasculature, H-Ras
overexpressing cells are unable to escape anoikis and undergo massive apoptosis early in
the course of the disease. In contrast, by the endpoint of study, H-Ras overexpressing
cells are able to establish a few micrometastases within the pulmonary parenchyma.
However, at endpoint of disease, there was no detectable TUNEL staining within these
micrometastases, suggesting that there was eventually proliferation of non-apoptotic H-
Ras overexpressing outgrowths within the lung.

In contrast, MCF10A cells overexpressing H—Ras and Bmi-l have a strikingly
different behavior during the course of disease. We have previously shown that Bmi-l
exerts a protective effect against apoptosis due to DNA damage in vitro. However, these .
experiments were performed in two-dimensional adherent culture, and therefore may not
be translatable to the in vivo environment. It is known that Bmi-l acts to prevent cell
cycle arrest through inhibition of the Ink4A-ARF tumor suppressor locus, which encodes
the cyclin-dependent kinase inhibitors pl6mK"A and pl4ARF [29, 118]. Therefore, it is not
surprising that overexpression of Bmi-l should be protective against apoptosis due to
DNA damage and cell cycle arrest through p53-dependent mechanisms. However, little is
known about the effect of Bmi-l on the prevention of apoptosis due to anchorage
independence. The results of TUNEL staining in the current tail vein model suggest that
Bmi—l exerts a protective effect against anoikis in the pulmonary vasculature. While. there

is some apoptosis occurring early in the disease, the amount of apoptosis is significantly

118

reduced in comparison with MCF10A cells overexpressing H-Ras alone. Similar to H-
Ras overexpressing cells, there was no detectable TUNEL staining at disease endpoint,
suggesting outgrowth of non-apoptotic cell populations.

In conclusion, we show that overexpression of H-Ras and Bmi-l together in
MCF10A mammary epithelial cells in the tail vein xenograft model results in fiilminant
and aggressive pulmonary metastasis through massive induction of cell proliferation and
inhibition of apoptosis. Whether this phenomenon is primarily due to the collaboration
between Bmi-l and H-Ras, due to stem cell properties induced by Bmi-l overexpression
in the face of H-Ras overexpression, or both, needs further investigation. The results of
this work are in agreement with previous studies by Glinsky et al. that show the Bmi-l
pathway is associated with an 11 gene microarray signature that predicts a marked
propensity towards metastatic dissemination in patients with multiple different types of
cancer [48], as well as resistance to anoikis [196]. Should Bmi-l play a major role in the
pathophysiology of anoikis resistance and metastasis, it would be a critical molecular
target in patients with cancer and metastatic disease. If Bmi—l expression could be
detected early in the course of disease, it may be valuable both diagnostically and
prognostically as a predictor of metastatic propensity. Furthermore, if it does indeed
function in this capacity, the Bmi-l pathway would be a promising target for molecular

therapy.

119

APPENDIX D

Figure 23: MCF10A+H-Ras and MCF10A+H-Ras+Bmi-l GFP expressing cells in the
lungs of mice at 2 hours, 2 days, one week, and four weeks after injection. (A) At two
hours, there are similar numbers of cells present in the lungs of mice injected with
MCF10A+H-Ras (H-Ras) and MCFl0A+H-Ras+Bmi-l(H-Ras+Bmi-l) (4X
magnification). There is progressive and rapid loss of GFP expressing cells in the
MCF10A+H-Ras group, while GF P expressing cells persist in the MCF10A+H-
Ras+Bmi-l group. By one and four weeks, there are scant GFP expressing MCF10A+H-
Ras cells; by four weeks there are fulminant GFP expressing metastatic lesions (4X
magnification). (B) Graphic representation of data from 24A. There is significant loss of
GFP-expressing MCF10A+H—Ras cells, and a steady increase of GFP expressing
MCF10A+H-Ras+Bmi-1 cells in the lungs over time (*p < 6X10'6, Student’s T-test).

H-Ras H—Ras+Bmi-l

A.
2hr

2d

1wk

4wk

 

120

Figure 23 (continued)

a; 9.0- . r*—|
F 8.0« -Ras+Bmi

g 7.0-
g 6.0-

DRas

 

 

0.8 2.0 7.0 28.0
Post-injection time (days)

121

Figure 24: Ki67 immunohistochemistry in MCF10A+H-Ras (Ras) and MCF10A+H-
Ras+Bmi-1 (Ras+Bmi) mammary fat pad (SQ) and intravenous xenograft models at end-
stage disease. (A) There is a significant increase in Ki67 immunolabeling in the
MCF10A+H-Ras+Bmi-1 mammary fat pad and intravenous xenografts (right) compared
to MCF10A+H-Ras xenografis (left). (B) Graphic representation of data from 25A
(pixels per 10X field). There is a significant increase in proliferation in MCF10A+H-
Ras+Bmi-l compared to MCF10A+H-Ras xenografts (*p < 1.3X10'7, **p < 0.0002,
Student’s T-test).

A Ras Ras+Bmi
.
. - - , "VI?- . '_,__-g.‘- . ..- 1.! .~ . . ».
K“- ‘ 3“- \' T.‘ ' 5.” f 'v . . .
... - - . s _-
"...-:5"; ' L .-‘ ‘ ». ‘ - - .13, .
., , . , f
I; . ya; -“ ‘ 0“ n .. . .2;‘
.r.‘ 5 ... 7.. ‘ .' ...: . a
. T. . ‘ \»‘~ ,"_‘\': T :- - -, 5:? v] _._. .
SQ ‘ . ‘ x .. 1. t <~£-.-‘ ~ ~.\ . “Maw-61:3. -
. .~ \ \ - 9."; .
. :‘ ‘~n~\‘ :‘ ‘ .'. ‘ ‘ ‘— ~\-'.
".‘f -‘ , \ :‘ ‘ . s’ ‘3‘
‘ .‘_ ~31, ' ‘
'J § '7 ,
‘ .. ~ \ ~ .
. 5‘_ g \ 0 ~ .
‘ '1'“! \‘ ‘ ~ . . ' ‘7' '
, ' '. {Alf 1"“;{3' "it": 4" ,g .0", ”KS" 5'
r - . ' I, ‘ .
' '1 ’ ‘ H- . 3 . , 'J 'P M.
g: .‘ ’ ( .1 ‘ In:c .fi‘g’afifln ‘3‘"?
" r .' ~ fl»: ~11 .. .3"? :25! v
...: ~ , mu ..wt-ux a
,.. 1.: ...1‘1‘. w ‘ e.» -
1‘ ,1- .'. H. ‘ , purify? .2, w s
," -. '- ' ' .-+. ..
IV :1. . N I a ‘ '1‘3“1’1’ '1'1’1'0 1". I. 'I"./ .
l l.
. '0 l v a L.
. . '4 sf " ("I 3‘ '. ’ i};".'
.4. .w. v . ~
:, m .u ,_ g I , a}, , .
' ' l 1" .3 "d": .... ’19., '
q” ,. , .
-4« f4: 415v?!» m.» ~
.w ‘ :4.» v a— -
0 I “up"! 1.51", : 5" ’ “in" ’-
4 . .'o I'M: #u‘ ‘ "" 4 ‘~"

3.5-
3.04
2.5-
g 2.0-
" 1.5-
3.1.0-
% 0.5-
0.0-

field (1 03)

l'

 

Pix

 

122

Figure 25: TUNEL staining in MCF10A+H-Ras (Ras) and MCF10A+H-Ras+Bmi-l
(Ras+Bmi) mammary fat pad and intravenous xenograft models at end-stage disease. (A)
There is a significant increase in TUNEL staining in MCF10A+H-Ras mammary fat pad,
but not intravenous xenografts, compared to MCF10A+H-Ras+Bmi—l xenografts. (B)
Graphic representation of data from 26A (pixels per 10X field). There is a significant
increase in apoptosis in MCF10A+H-Ras mammary fat pad xenografts compared to
MCF10A+H-Ras+Bmi-l xenografts (*p < 0.0002, Student’s T—test). There is no
statistically significant difference in apoptosis in intravenous xenograft models at end-
stage disease.

A' Ras Ras+Bmi
_ . -'-f.;~" 1.... v.
( . ._ ‘ ' A‘
| ~ -V ' ‘
\ ".
.' . V. . . s
- .J '
SQ r.- . . .
u . I | > .
U
‘ ‘ l t . .
‘ Q
\
’ ‘.
‘ I ‘ r: n, ‘
. r: - “ 53..
IV i
~Q
N.
‘ Vb-

-Ras+ Bmi
3. DRas

Pixels per 10x field (102)
A M

h

 

 

 

 

123

Figure 26: Ki67 immunolabeling and TUNEL staining in MCF10A+H-Ras (Ras) and
MCF10A+H-Ras+Bmi-1 (Ras+Bmi) intravenous xenografts during early progression of
disease. (A) At 2 days and 8 days, there is a significant increase in Ki67 immunolabeling
in MCF10A+H-Ras+Bmi-l xenografts compared to MCF10A+H-Ras xenografts (*p <
0.01, **p < 8X10”, average cell number per field at 20X magnification, Student’s T-
test). (B) At 2 days and 8 days, there is a significant decrease in TUNEL staining in
MCF 10A+H-Ras+Bmi-1 xenografts compared to MCF10A+H-Ras xenografts (*p <
0.05, average cell number per field at 20X magnification, Student’s T-test).

 

 

 

 

 

 

 

 

 

   

 

 

A. h **
3. :3. -Ras+Bmi r——1
h o . R
.3 17.5« E: as
g 15.04 *
C 2 12.54 ' '
E’ 8 10.0«
3: 7.5.
a 6.04
3 2.6-1 l l [:1
,1 0.0-
2 f
Post-injection time (days)
* .
B- 2 s. r—fi * -Ras+Bmi
-. ”"7 ' .=' 1:1Ras
o 5-1
2
4.
2 1:
= T) 3.
8 C
o 2"
a
E 1‘
2
0d
“ 2 'i

Post-Inlection time (days)

124

Figure 27 : Bmi-1 prevents cellular senescence, cell cycle arrest and apoptosis by
inhibiting the Ink4a-Arf tumor suppressor locus. Normally, p19 suppresses RB
phosphorylation and p16 prevents degradation of p53, leading to cell cycle arrest, with
resultant apoptosis or senescence. Bmi-l suppresses the actions of p16 and p19 through
inhibition of the Ink4A-Arf tumor suppressor locus, leading to degradation of p53 and
suppression of RB phosphorylation, thus preventing cell cycle arrest, senescence, and
apoptosis. '

Bmi-1

— Prevention of cell cycle arrest

 

Ink4a-Ari

P14/19Ar/\ p16lnk4a Normal cell cycle

i _L checkpoint
regulation

Milli-:12 Cyclin liCDK4/6
p53 pRB

\ /

Cell cycle arrest
senescence
apoptosis

 

125

CHAPTER FIVE

Significance and Future Directions

1. Conclusions and Significance

We have shown that Bmi-1 plays a major role in promoting many cellular
functions related to transformation and metastasis. I have addressed the effects of Bmi-1
overexpression in normal immortalized human mammary epithelial cells (MCF10A)
alone and in collaboration with H-Ras overexpression in vitro and in vivo. To elucidate
the effects of Bmi-1 overexpression on MCF10A cells alone and in combination with H-
Ras overexpression in vitro, I examined 1) changes in cellular morphology in two- and
three-dimensional culture, 2) proliferation, 3) invasion, and 4) apoptosis due to DNA
damage in MCF10A cells overexpressing Bmi-1, H-Ras, and Bmi-1 concurrently with H-
Ras. Conversely, to determine what effect depressing Bmi-l has in established breast
cancer cell lines that overexpress it in vitro, shRNA was employed to knock down the
expression of Bmi-1 in several breast cancer cell lines, and changes in proliferation,

invasion, and apoptosis were similarly examined.

First, we found that the cellular phenotype in vitro changes dramatically based
upon the oncogene or combination of oncogenes overexpressed. While MCF10A cells
grow in a typical epithelioid fashion in two dimensional culture, with clear cell-to-cell
junctions, abundant cytoplasm and centralized nuclei, with Bmi-1 overexpression
MCF10A cells take on a spindle shaped, mesenchymal phenotype. With H-Ras

overexpression, cells become more rounded to triangular to stellate, and with the

126

combination of Bmi-1 and H-Ras overexpression, cells become markedly spindloid, ofien
radiating out from central clusters. Interestingly, after several passages in culture, these
cells form spheroid structures that are similar to mammospheres grown in three
dimensional suspension culture, which have been shown to be enriched for stem cell
markers. Further analysis of these spheroids for stem cell markers and evidence of stem

cell properties is an important future direction, and is forthcoming.

In three-dimensional culture in Matrigel basement membrane extract, we found
that there is a block in differentiation with either Bmi-l or H-Ras overexpression. While
normal MCF10A cells differentiate normally towards generation of luminal structures
through central apoptosis of cell clusters and well-organized cellular polarity around this
central lumen, with Bmi-1 or H-Ras overexpression, no such luminal formation occurred
and cells remained in disorganized clusters, indicating a block in differentiation.
Furthermore, with Bmi-l and H-Ras overexpression together, cells became more
spindloid and highly proliferative, forming large disorganized clumps of cells with a high
mitotic rate, illustrating a very proliferative and poorly differentiated phenotype as a
result of overexpression of both oncogenes. This provides evidence of the transforming
ability of these oncogenes in concert in a three dimensional in vitro context that is
designed to mimic the in vivo environment, giving significance and purpose to the

subsequent in vivo experiments to demonstrate tumorigenesis.

Given the effects on differentiation and proliferation that were observed in
culture, we sought to quantify these changes in vitro. We found that with Bmi-l or H-Ras

overexpression, proliferation of mammary epithelial cells increased significantly in an

127

MTS-formazan proliferation assay. With overexpression of both oncogenes, proliferation
then increased significantly farther. Conversely, we have shown that H-Ras and Bmi-l
overexpression alone have a protective effect against apoptosis due to DNA damage
using an anti-histone sandwich ELISA assay, and when overexpressed together, a
significant further inhibition of the apoptotic response is seen. Lastly, to evaluate effects
on invasive capability of cells in a three-dimensional in vitro context, cellular invasion
was measured using a Matrigel invasion assay. Bmi-1 or H-Ras overexpression led to a
significant increase in cellular invasion in this assay, and overexpression of both
oncogenes led to a statistically significant increase in proliferation over MCF10A cells
with overexpression of either oncogene alone. It was further determined through western
blotting that overexpression of H-Ras in combination with Bmi-l led to dysregulation of
multiple growth pathways, including the MAPK/ERK and AKT pathways, as well as cell
cycle regulators such as p53, pRB, and Cyclin D. This gives further confirmation of the

induction of an aggressive phenotype with Bmi-l and H-Ras overexpression in vitro.

Next we sought to determine the effects of Bmi—l and H-Ras overexpression,
alone and in combination, on MCF10A cells in the in vivo context to ascertain the true
biologic effects within the host microenvironment. Importantly, MCF10A cells
overexpressing Bmi-l did not form tumors in the intramammary fat pad xenograft model,
suggesting that overexpression of Bmi-1 alone is insufficient to transform human
mammary epithelial cells. H-Ras overexpression induced hemorrhagic and cystic lesions
composed of smooth muscle, hemangiomatous, and mast cell components that progressed
more slowly over time than those overexpressing both Bmi-1 and H-Ras. Importantly, the

xenografts overexpressing both Bmi-1 and H-Ras developed poorly differentiated tumors

128

resembling and epithelial to mesenchymal transition phenotype (EMT) that were locally
infiltrative and more highly metastatic than those tumors resulting from H-Ras
overexpression alone. Xenografts overexpressing both oncogenes consistently
metastasized to the spleen and liver at a greater rate than those overexpressing H-Ras
alone, and only those xenografts overexpressing both oncogenes metastasized to the
brain. This is particularly significant because it confirms the collaborative role of Bmi-l
and H-Ras in transformation of the MCF10A cell line in vivo, and recapitulates an
important route of metastasis in the human disease, the brain, that is particularly difficult
to consistently repeat in animal models of the human disease. Additionally, the marked
phenotypic difference between the tumors suggests significant differences in biologic
pathways in the development of each tumor phenotype. Further work to characterize the

mechanisms of the differences in phenotype is necessary, and is forthcoming.

The tail vein xenograft model of direct hematogenous spread to the lungs was
used to determine the metastatic capability of MCF10A cells overexpressing Bmi-l or H-
Ras or both oncogenes together. This experiment showed the marked capacity of
MCF10A cells overexpressing both oncogenes to escape anoikis, implant in the
pulmonary vasculature, and rapidly proliferate in comparison to those cells
overexpressing either oncogene alone. This further corroborates our hypothesis that Bmi-
l collaborates with H-Ras to induce a highly aggressive phenotype, consistent with its
known role in lymphomagenesis with c-myc. This phenotype was determined to be a
result of marked proliferation and a substantial resistance to apoptosis of MCF10A cells

overexpressing both Bmi-1 and H-Ras in comparison to H-Ras alone. Again, this

129

illustrates important differences in biologic pathways leading to the ability of these cells

to cause significant metastatic disease and clinical morbidity.

Given the results we obtained in the [first set of experiments, we sought to
determine whether reduction of Bmi-l had a direct effect on the biology of established
human mammary breast cancer cell lines that overexpress Bmi-1. Using shRNA
technology and lentiviral delivery, we established several cell lines with Bmi—1
knockdown. These tumor cell lines exhibited decreased proliferation by MTS-formazan
proliferation assay, decreased invasion in Matrigel, and increased apoptosis in response
to DNA damage. This work is significant because it confirms our previous findings in the
MCF10A cell line overexpressing both Bmi-l and H-Ras and illustrates the direct
biologic effects of Bmi-l non-engineered, established breast cancer cell lines derived
from human tumor samples. Thus, this reflects a true biologic role in the development of

some forms of breast cancer.

Next we evaluated two of the most aggressive cell lines for changes in
tumorigenesis and metastasis in vivo by using the xenograft mouse models illustrated
above. While the overall incidence of tumorigenesis and metastasis did not change, there
was a significant delay in tumor onset in one cell line, and overall tumor progression in
the other. This may suggest that Bmi-l is important in the early stages of the progression,
or in the continued growth and survival, including apoptosis resistance, of some forms of

breast cancer.

Because of these new findings, it is important to better define the mechanism

leading to the differences in the observed phenotype and biologic behavior of cells

130

overexpressing H-Ras, Bmi-l, or both oncogenes in vitro, and tumors induced by H-Ras
alone or in combination with Bmi-l in vivo. In order to better define the mechanisms
related to the differences in behavior and phenotype, microarray functional analysis is
underway and validation of important pathWays is forthcoming. In addition, it is
important to study the effects of Bmi-1 within a model of true genetic expression to
effectively determine the biologic effects of Bmi-1 in the host in vivo environment.
Therefore, we have developed a novel in vivo system to study the effects of Bmi-1
overexpression within the mammary gland on development and within the context of
expression of other oncogenes: an inducible, mammary gland specific transgenic mouse

model of Bmi-1 overexpression.
II. Microarray analysis of MCF10A-derived cell lines

Microarray analysis is an efficient means to evaluate differences in gene
expression between biological samples in a high-throughput manner. Several thousand
. genes can be assayed in a short period of time in a number of patient or research samples
to determine up- and down-regulated genes in each set [205]. This technology provides a
wealth of information on disease pathways and mechanisms that can be related to cancer
phenotype, biologic behavior, prognosis, and response to treatment [206-208]. Assaying
gene expression of thousands of genes in different types of cancer has led to the
identification of several reproducible portraits of several cancer types [208, 209], and has
been used to group breast cancer subtypes based upon hormone receptor status,
basal/luminal phenotype, and prognosis [208]. Further, microarray technology has been

used to determine gene signatures associated with tumor progression [210], treatment

131

failure, and death-from-cancer phenotypes[48]. Bmi-1 has been shown to be associated
with gene signatures predictive for poor prognosis and metastasis associated with a stem
cell phenotype in multiple types of cancer [48]. To further determine the mechanisms
associated with the Bmi-1 — H-Ras collaboration in our system, we used microarray
technology to identify differentially expressed genes between each of the cell lines
engineered, including the spontaneously arising spheroids generated from the

MCF10A+H-Ras+Bmi-l cells.

Preparation of samples for microarray analysis: Each of the five cell lines (MCF10A,
MCF10A+Bmi-l, MCF10A+H-Ras, MCF10A+H-Ras+Bmi-l, and MCF10A+H-Ras-
Bmi-1 spheroids) were grown in T75 flasks under previously described culture conditions
[118]. Cells were lysed and RNA was extracted using 1 ml Trizol reagent. RNA was
isolated using phenol-chloroform extraction, and cleaned with the RNeasy mini kit
(Qiagen, Valencia, CA). Approximately 15 pg RNA was used for each cell line for

microarray analysis.

Data normalization and analysis. RNA samples were sent to the microarray core
facility (Frederick, MD) for analysis. All mRNA chips were normalized using the G-C
Robust Multi-Array (GCRMA), Robust Multi-Array (RMA), or Microarray Suite 5

(MASS) procedure (www.bioconductororg) or using GCRMA in Partek Genomic Suite

 

(www.partck.com). For each contrast of classes, probesets were filtered based on the

 

detection calls derived from MASS procedure according to majority rule. Then the data
either from RMA or MAS normalization for those filtered probes were subjected to

Significance Analysis of Microarrays (SAM) procedure to determine the significant gene

132

lists based on intended false discovery rate (FDR), and then t-test p-values for those
genes were used for firrther filtering for final significant gene lists. Alternatively, one
way ANOVA model was used to derive the differentiated genes from different contrasts
of different treatment and phenotypes in Partek Genomic Suite (wwwpartekcom).

The analysis of pathway/network and enrichment analysis was performed using
in-house software WPS [211]. Briefly, Fisher’s exact test was performed based on 2 x 2
contingency tables, to determine whether a gene is in a given list or not versus whether
this gene is associated with a pathway (gene set, term) or not. A one-sided Fisher’s exact
test was used to measure whether a particular Biocarta pathway (wwwbiocartacom),
GSEA gene set term (www.broad.mit.edu/gsea/), or GO term (wwwgeneontologvorgf)
was enriched in a given gene list. The terms were ranked based on their Fisher’s exact
test P-values with the most enriched term listed at the top. To compare biological themes
at the pathway, gene set, and GO term level across multiple gene lists of different
contrasts, these gene lists were also subjected to a pathway-level pattern extraction
pipeline (Yi and Stephens, unpublished work). Briefly, after batch computation of
Fisher’s exact test for the gene lists, the log-transformed p-values were retrieved and
combined into an enrichment score matrix for clustering analysis or pathway pattern
extraction. The terms (pathways, or GO terms) of selected clusters with interests were
further used to retrieve the associated genes from the original gene list. Interested

pathways were displayed along with the data in WPS program.

133

Results

Comparison of MCF10A cell lines overexpressing Bmi-l yields a distinct Bmi-l
signature. By hybridizing the labeled cDNA. from each cell line to the Affymetrix
Human Genome US 133 Array, the expression of 13,000 genes could be analyzed. We set
cut-offs for the gene analysis at p < 0.05 and 1.5 fold change. Comparing the gene
expression signatures of MCF10A with MCF10A+Bmi-1 and MCF10A+H-Ras with
MCF10A+H-Ras+Bmi-l resulted in a unique set of differentially expressed genes. Figure
29 shows the differentially expressed gene expression pathways associated with enriched
BioCarta pathways unique to this Bmi-1 signature in a dendrogram heatrnap. Several
genes associated with the ATM growth pathway, and G2/M cell cycle checkpoint,
including p53 and Rb pathways, were most differentially expressed within this analysis.
Some important target genes that were found to be downregulated in this analysis
included p53, p21 (CDKNIA), and Rad51, while c-Abl was upregulated (Figure 30),
consistent with the role of Bmi-1 in suppression of the p53, cell cycle checkpoint. p21 is a
universal inhibitor of cyclin-dependent kinases [212], suggesting that inhibition of p21
would result in increased proliferation and possibly transformation. Rad51 complexes
with BRCA2 to repair DNA double strand breaks, and dysfunction of this complex leads
to increased chromosomal instability and susceptibility to breast cancer [213].
Interestingly, c-Abl is a protein tyrosine kinase that is activated by DNA damage and in
turn induces cell cycle arrest and apoptosis, which requires MAPK6 activation [214],
consistent with our findings of a collaborative role of H-Ras activation and Bmi-1
function. Gene-Ontology Biological Processes associated with this Bmi-1 signature

included cell cycle control, cell proliferation, and mitosis (Figure 31). These data are

134

consistent with our in vitro experiments that show Bmi-1 overexpression is associated

with increased proliferation of MCF10A cells.

Future Directions

The preliminary microarray analysis that has been performed has confirmed our
previous in vitro data and has shed some new light on other interesting pathways that
may influence the phenotype of MCF10A cells transformed by Bmi-l and H-Ras. It will
be critical to analyze the array data further for differential gene expression specific to
each cell line to identify pathways that would explain the differences in the in vivo tumor
phenotype as well as the dramatic difference in metastatic phenotype. In addition, we
plan to determine the critical differences between the MCF10A+H-Ras+Brni-l adherent
cells and the spontaneously derived spheroid structures originating from the same cell
line. It would be important to determine if there are stem cell pathways upregulated such
as the Notch, Wnt, and Shh pathways that would provide evidence of enrichment for stem
cells and even support for Bmi-l in the tumor stem cell hypothesis in this model. Overall,
several promising pathways for study have been discovered through the use of microarray
technology that may further aid in the determination of the mechanisms of Bmi-1
overexpression in tumorigenesis and metastasis. Validation of targets identified by gene
microarray will be performed, and interesting genes will undergo functional analysis to
ascertain their biologic relevance to oncogenesis and metastasis. These data sets hold
great potential in identifying important targets in the pathogenesis of breast cancer, and

may yield insight into targets for translational research of clinical relevance.

135

111. Development of a tissue-specific conditional transgenic mouse model of Bmi-l

overexpression in the mammary gland

To determine the role of Bmi-l in mammary gland development and
tumorigenesis, a tissue-specific conditional mouse model of Bmi-l overexpression
restricted to the mammary gland was developed. Using the tetracycline responsive
element to drive the expression of Bmi-1 under the control of the CMV promoter,
transgenic mice overexpressing the Bmi-1 gene in the presence of doxycycline were
developed. These mice were bred to the MMTV-rtTA mouse model to obtain double
positive transgenic mice expressing both the pTRE-CMV-Bmi-l and MMTV-rtTA
construct. The MMTV-rtTA construct employs the MMTV promoter, a mammary gland-
specific promoter, to drive the expression of the reverse tetracycline transactivator (rtTA)
protein specifically in the mammary gland. The rtTA protein associates with the
tetracycline responsive element (pTRE) in the presence of doxycycline, and activates the
CMV promoter to drive Bmi-l expression. Thus, animals harboring both transgenes
overexpress Bmi-l within the mammary gland upon the addition of doxycycline in the
diet. This model provides a powerful tool in the study of Bmi-l in mammary gland
development and tumorigenesis because of its tissue specific and inducible nature,
allowing study of the biologic effects of Bmi-l overexpression at specific stages of
development. Additionally, overexpression can be terminated by removal of doxycycline
from the diet, to evaluate for regression of any phenotypic changes that may occur once
Bmi-l expression is turned off. Given the collaborative role Bmi-l plays with other

oncogenes, this model will be usefirl in experiments where oncogenes are co-expressed in

136

other mouse models to evaluate in vivo collaborative effects of Bmi-1 on oncogene

firnction.

Construction of pTRE-CMV-thi-l Plasmid: Wild type human Bmi-l (thi-l)
cDNA is 3.2kb in size, located on chromosome 10p13, and contains ten exons. The ORF
encodes 326 amino acids and the protein product is 44-46 kD. The human gene shares
92% nucleotide sequence and 98% amino acid sequence identity with the gene in the
mouse (mBmi-l).

The pTRE-CMV-thi-l construct was created by subcloning the human Bmi-l
cDNA into the pTRE-tight expression vector. The pBabepuro-thi-l plasmid was
obtained from Dr. Goberhan Dimri from the Feinberg School of Medicine at
Northwestern University (Figure 32A). The pBabepuro vector is 5169 bp and the internal
human Bmi-l (thi-l) fragment is 981 bp in size. The plasmid contains an ampicillin
resistance gene (Amp) for propagation in bacteria and a puromycin resistance gene (Puro)
for selection in mammalian cells. The human Bmi-l cDNA fragment was excised from
the pBabe-puro expression vector with BamHI and Sal]. The 981 bp thi-l cDNA was
inserted into the BamHI and Sal] sites of the pTRE-tight expression vector (Figure 328).
The pTRE-tight expression vector contains a modified Tet response element (pTRE),
which consists of seven repeats of a 36-bp sequence that contains the l9-bp tet operator
sequence (tetO). The pTRE is upstream of a CMV promoter (PminCMV), which lacks a
CMV enhancer. Therefore, pTRE-tight is silent in the absence of binding of TetR or
rTetR to the tetO sequences. When in the presence of doxycycline and the reverse
tetracycline transactivator (rtTA), the pTRE is activated, resulting in transcription of the

gene of interest.

137

Following successful cloning of the thi-l fragment into the pTRE—tight vector,
the 800 bp injection fragment containing the pTRE promoter, inserted thi-l, and the
SV40 polyA was excised from the pTRE-tight-thi-l construct (Figure 32C) by
restriction digestion with Xhol. The plasmid construct structure was confirmed by
restriction digestion. The pBabe-Bmi-l vector was digested with BamHI and Sal] to
obtain the 981 bp thi-l fiagment, and digested with A7201, which gave the expected ~2
kb fragment. The thi-l fragment did not possess a restriction digest site for X1101
enzyme that would interfere with the injection fiagment to be excised in the final

construct.

Generation, Maintenance and Identification of pTRE-Bmi-l Transgenic Mice:
Since the pTRE-tight system had been used successfirlly in vivo previously, the newly
constructed pTRE-tight/Bmi-l plasmid was sent to the transgenic core facility (NCI-
Frederick) for injection of the appropriate DNA fiagment in blastocysts in order to create
a pTRE/Bmi-l transgenic mouse. The injection fragment was microinjected into the male
pronucleus of a one day old blastocyst, then implanted in the oviduct of a pseudopregnant
female FVB mouse [215]. The resulting pups were screened by Southern blot analysis. 1
pg of pTRE-Bmi-l injection fragment DNA and 10-20 pg of tail genomic DNA was
digested with BamHI and separated on an agarose gel. The DNA was transferred to
Hybond-N Nylon membranes (Amersham) and hybridized in QuikHyb Solution
(Stratagene) with a 32P-random primed-labeled probe.

Ten founder mice were determined to carry the transgene. The founders were

determined by southern blot following digestion of genomic DNA with BamHI and XhoI.

138

Since the injection fragment is inserted randomly into the genome during microinjection
of the blastocyst, insertion in varying orientations had to be accounted for, such as head-
to-head, head-to-tail, and tail-to-tail insertions. The injection fragment was ~2 kb in size
when digested with XhoI, with a BamHI restriction site at 350 bp. When the genomic
DNA of founders inheriting the pTRE-thi-l injection fragment is digested with
BamHI, a number of fragments can result based upon the orientation of the inserted
injection fragment (Figure 33). If the fragment was inserted head-to-head, a ~700 bp
fragment could result. If the fragment was inserted tail-to—tail, a ~3.2 kb fragment could
be seen. Finally, if the fragment was inserted head-to-tail, a ~2 kb fragment could occur.
These fragments were present in the control injection fragment (C) digested with BamHI,
and the founders were identified by the most common insertion orientations yielding a
3.2 kb or 2 kb fragment (Figure 34). Relative copy number was estimated fi'om signal
intensity on southern blot, and ranged from one to 80 copies. Following founder
identification, the highest expressing lines (A1, K9, and E4) were propagated by breeding
to FVB mice and screened by PCR. The E4 line ultimately was propagated because of its
willingness to breed and its large litters. For PCR, 50-200 ng of genomic tail DNA was
added to a 25 pl reaction volume [10x PCR buffer, 50 mM MgC12, 2.5 mM dNTP’s, l U
Taq Polymerase, and 10 pM each of pTRE forward 5'- TGT CGA GTT TAC TCC CTA
TCA GTG -3‘ and thi-l reverse 5‘-CAT TTT TGA AAA GCC CTG GA -3‘ primers.
Cycling conditions consisted of 95 °C for 3 min, followed by 33 cycles of 94 °C for 45
sec, 56 °C for 45 sec, and 72 °C for 45 sec, then an extension at 72 0C for 10 min,
resulting in a PCR product 576 bp. Mice harboring the genotype were then bred to

MMTV-rtta mice to generate double positive mice, screened by using MMTV-rtta

139

specific primers (rtTA.497F 5‘-TGC CGC CAT TAT TAC GAC AAG C-3‘ and
rtTA.1002R 5‘-ACC GTA CTC GTC AAT TCC AAG GG-3‘). All methods used to
breed animals were in compliance with the National Institutes of Health Animal Care and

Use Committee regulations.

Induction of Transgene Expression:

Once double positive animals were successfully identified, the inducible function
of the construct was tested in a set of mice. Animals harboring both transgenes (pTRE-
Bmi-l X MMTV-rtTA) possessed the inducible pTRE construct and the mammary gland
specific MMTV promoter driving the rtTA gene. In the presence of doxycycline, the
rtTA fimctionally binds to the pTRE, and drives Bmi-l expression in the mammary
gland. A dose of 2 mg/ml doxycycline [216] was administered in the drinking water for
14 days, and animals were sacrificed and mammary glands from either mammary chain
were harvested for protein and RNA analysis. Glands were snap frozen in liquid nitrogen,
pulverized with a mortar and pestel, and lysates were prepared in a tissue macerator with
1 ml RIPA lysis buffer containing anti-proteases. Similarly, glands collected for RNA
analysis were snap frozen, pulverized, and lysed in Trizol reagent, followed by phenol-
chloroform isolation and purification. Additionally, glands were collected for
histopathologic examination. Glands were fixed in 4% paraformaldehyde and routinely
processed into paraffin blocks from which 4 pm sections were prepared and stained with
hematoxylin and eosin (H&E).

RNA samples from wildtype (WT), Bmi-l+/rtTA-, and Bmi-l+/rtTA+ mice were

tested for Bmi-1 mRNA expression using RT-PCR. The RT reaction (1 pl oligoDT, 1 pl

140

10 deNTP, 10 pl H2O) was incubated for 5 min at 65 °C, and added to 4 pl 5X buffer,
1 pl 0.1M DTT, 1 pl RNAse out, and 1 pl 8811 RT enzyme per sample. The RT reaction
mixture was incubated at 50 °C for 60 min, and the 70 °C mixture was incubated for 15
min, then chilled to 4 °C to obtain cDNA. The 220 bp human Bmi-1 cDNA fragment was
amplified with human specific Bmi-1 primers (thi-lF 5‘ TCA TCC TTC TGC TGA
TGC TG-3‘ and thi-lR 5‘-GCA TCA CAG TCA TTG CTG CT/-3‘). The amplified

cDNA was run on a 1% agarose gel and photographed (Figure 35).

Results

Generation of pTRE-Bmi-l X MMTV-rtTA Mice

Transgenic mice overexpressing the human Bmi-l gene under the inducible
control of the pTRE-CMV promoter were created by microinjection of the linearized
pTRE-CMV promoter, thi-l, and the SV40 polyA tail into the male pronucleus of a
blastocyst. Founder pups were screened by Southern blot analysis using tail snip DNA,
and offspring were screened by PCR using specific primers recognizing the pTRE and
thi-l. Positive offspring were bred to the MMTV-rtTA colony and animals were

screened for the presence of both transgenes by PCR.

Induction of transgene expression

All heterozygous pTRE-Bmi-l animals exhibited normal behavior and were
healthy throughout the experiment. Males and females were fertile, and females were
able to lactate and nurse their pups. A subset of animals possessing both the pTRE-Bmi-l
and MMTV-rtTA transgenes were given doxycycline (2 mg/ml) in their drinking water

for at least 14 days. After 14 days of doxycycline administration, animals were sacrificed

141

in diestrus, and mammary glands were harvested for histopathology and protein and RNA
analysis. Since we have shown by immunohistochemical analysis that Bmi-l expression
in the mouse mammary gland is lowest during diestrus, mice were examined at this stage
of the estrus cycle. The stage of estrus was determined by using an estrus monitor to
measure the electrical impedence of the vaginal epithelial cell layer. The electrical
impedence of the vaginal epithelium differs based upon the thickness of the epithelium at
different stages of the estrus cycle. Histopathologic examination of glands of animals
with Bmi-1 overexpression for two weeks showed no significant changes in mammary
gland structure or function. There was no evidence of hyperplasia, inflammation, or
atypical cellular changes. PCR examination following 14 days of doxycycline
administration determined that the 200 bp human Bmi-1 mRNA was overexpressed in
Bmi-l+/rtTA+ animals, but not in Bmi-1+/rtTA- or wildtype (WT) animals. There was
minimal expression of Bmi-1 mRNA in Bmi-l+/rtTA- animals, indicating a small
amount of leakiness present in the model. Human beta-actin (500 bp) was used as a
control.
Future directions and utility of the pTRE-Bmi X MMTV-rtta model

To adequately examine the role of Bmi-1 expression in mammary gland biology,
three major experiments will be designed: 1) exposure to Bmi-1 overexpression
chronically and throughout different stages of development, 2) breeding to other GEM
models of human breast cancer, and 3) overexpression of Bmi-l in chemically induced
models of breast cancer. Chronic overexpression of Bmi-1 will determine if
overexpression of Bmi-1 will impact mammary gland biology over extended periods of

time. Acute, short term exposure to Bmi-1 overexpression in the mammary gland may

142

not be sufficient to alter or induce a phenotype, and this may require longer exposure
periods. Examinations will be based on histopathologic and immunohistochemical
analysis of the mammary glands following 1, 4, 6, and 12 month exposures to
doxycycline. I

Since Bmi-1 has been shown to be important in morphogenesis and embryonic
development [217], exposure to high levels of Bmi-1 throughout the in-utero time period
into the pre—weaning ages is important to address the biologic effects of Bmi-1
overexpression in the mammary gland at early stages of development. The mammary
gland of the mouse develops in four stages: 1) in utero, 2) during puberty, 3) during each
estrus cycle, and 4) during pregnancy [218]. In utero, rudimentary ducts are initially
formed, and during puberty the ducts elongate and branch to fill the mammary fat pad.
The density of ductal side branching as well as the number of alveolar buds increases
with each estrus cycle, and during pregnancy large lobuloalveolar structures producing
milk arise from these alveolar buds. The mammary gland is thus an incredibly
proliferative organ with incredible regenerative capacity. This can also be observed
during involution of the gland following estrus or pregnancy, wherein the gland loses a
large majority of the epithelial cell component.

Given its role in the maintenance of stem cell populations, overexpression of
Bmi-1 in utero when the rudimentary ducts are formed may cause any number of
deleterious phenotypes. It has been thought that hormone exposure in utero or during the
early postnatal period may influence the proliferative potential of stem cells [219-221]. If
Bmi-1 overexpression has a positive influence on stem cell number within the mammary

gland, this may increase the number of cells that are exposed to hormones later in life

143

during puberty. In addition, since some breast cancers are thought to arise from
transformation of mammary stem cells [220, 222] increased numbers of stem cells within
the mammary gland may translate to an increased risk for the development of mammary
neoplasia.

We have previously shown that when Bmi-1 is dysregulated or overexpressed, it
can act as an oncogene [118]. During puberty, when the ducts are branching to fill the fat
pad, Bmi-l overexpression may alter the branching morphology or increase the number
or density of branching, leading to an abnormal phenotype. It has been shown that
overexpression of genes responsible for. stem cell maintenance such as Notchl and
Notch3 result in a decreased proliferation of ductal and alveolar epithelial cells during
puberty, resulting in impaired ductal and lobuloalveolar development and inability to
form certain milk proteins [223]. Bmi-l overexpression during this time may have a
similar effect on mammary gland morphology, or it may act to increase the proliferation
of ductal and lobuloalveolar structures, making the gland more prone to tumorigenesis.

It is well known that the development of many mammary cancers is influenced by
hormonal factors, particularly estrogens and their receptors [220, 224]. Exposure to
estrogen over an extended period 'of time may increase a woman’s risk of breast cancer.
Increased exposure occurs in women who enter early menarche, since this increases the
number of years the breast tissue is exposed to estrogen. In addition, the more years a
woman menstruates, the longer the breast tissue is exposed to estrogen, and so late
menopause also influences a woman’s risk of breast cancer [225]. Finally, late
pregnancy or never having been pregnant are risk factors because estrogen levels are low

during pregnancy and breast tissue is exposed to more estrogen in women who become

144

pregnant for the first time after the age of 35, or in those who never become pregnant
[226]. Should Bmi-1 alter or increase the density of ductal or alveolar cells during estrus,
this would have a significant effect on the likelihood of mammary epithelial
transformation over several estrous cycles.

Given its collaborative role with other oncogenes in the induction of leukemias
and lymphomas [28, 53], and the transformation of mammary epithelial cells in vitro and
in vivo [29, 118], breeding pTRE-Bmi X MMTV-rtTA transgenic mice to other GEM
models of breast cancer would yield important insights on the effects of Bmi-1
overexpression in the context of other oncogenes. It has been shown that overexpression
of more than one oncogene in a GEM model can significantly alter the resulting
phenotype, whether it is by morphology or by increased incidence or acceleration of the
disease process. Overexpression of Bmi-1 in the Ep-myc transgenic mouse model of
lymphoma resulted in a dramatic increase in incidence as well as acceleration of pre—B
cell lymphomagenesis [28, 53]. We have shown that overexpression of Bmi-1 in the
context of H-Ras overexpression in normal human mammary epithelial cells results in
acceleration of tumor formation as well as a more poorly-differentiated, aggressive
phenotype in xenograft models [118]. However, these models lack an endogenous
gerrnline mutation that is characteristic of natural breast cancer and development of
neoplasias in GEM models. Therefore, the development of a bitransgenic, inducible
model of oncogene collaboration between Bmi-1 and other oncogenes involved in human
breast cancer would be a very versatile and powerful tool to study molecular mechanisms

involved in oncogene collaboration.

145

The pTRE-Bmi-l transgenic mouse would be particularly usefirl in determining if
Bmi-1 overexpression affects the response to carcinogens. Several models of carcinogen
driven mammary tumorigenesis exist, most notably the 7,12-dimethylbenz[a]anthracene
(DMBA) model. The mechanism of action of DMBA is through metabolism into a diol
epoxide which incorporates into DNA and causes mutations [227], as well as by blocking
transport of glutathione to the liver [228]. Mice treated with DMBA generally acquire
mammary tumors by 90 days following administration characterized as adenocarcinoma,
squamous cell carcinoma, or myoepithelial carcinoma [229]. Through its role in
suppression of the lnk4A-Arf tumor suppressor locus, Bmi—l overexpression may result
in increased incidence of mammary tumor formation, acceleration of tumorigenesis, or
modification of the histologic phenotype. This would lend further support to the role of

Bmi-l as a collaborating oncogene.

IV. Future Directions

Future studies arising from this research should include 1) investigation of the
stem cell phenotype of MCF10A+H-Ras+Bmi-l spheroids, 2) characterization of the
mechanisms of apoptosis inhibition and metastatic propensity by Bmi-1, and 3) exploring
the collaborative relationship of Bmi-l with other oncogenes in breast cancer.

To investigate the stem cell phenotype of MCF10A+H-Ras+Bmi-l spheroids, it
will be important to identify stem cell markers present or upregulated in these cells
compared to adherent MCF10A+H-Ras+Bmi-l cells. In vitro, this would be achieved
through western blot analysis of spheroids for markers such as CD44, CD24, Oct3/4, and

aldehyde dehydrogenase (ALDH). One of the important characteristics of stem cells is

146

their ability to escape apoptosis induced by anchorage-dependent cells detaching from the
surrounding extracellular matrix (anoikis). This could be tested in vitro by culturing
MCF10A+H-Ras+Bmi-l and MCF10A+H-Ras cells in suspension culture, and assaying
the cells for apoptosis and proliferation. We Would hypothesize that the MCF10A+H-
Ras+Bmi-l cells would be induced to form spheres due to anchorage independence, and
continue to proliferate, whereas the MCF10A+H-Ras cells would undergo anoikis. To
test the stem cell properties of these spheroids in vivo, an athymic nude mouse xenograft
model would be used to test the self-renewal capability of these transformed cells. The
definitive test of the tumor stem cell nature of a transformed cell is the in vivo
recapitulation of a tumor from a very small number of tumor cells that possess a stem cell
phenotype. MCF10A+H-Ras+Bmi-l spheroids would be disassociated by trypsinizatiOn
and counted, followed by implantation of cells into the mammary fat pad in a dilution
series ranging fiom 1X106 cells to 100 cells. A similar dilution series would be
performed with MCF10A+H-Ras+Bmi-l adherent cells- and MCF10A+H-Ras cells as
controls. If animals with a relatively small number of cells derived from spheroids form
tumors and other groups do not, this would be further evidence of the stem cell nature of
these transformed cells.

The mechanism of apoptosis inhibition in MCF10A+H-Ras+Bmi-l certamly
deserves further investigation. Western blot analysis of all MCF10A derived cell lines for
mediators of apoptosis such as the caspases or the BCL2 family will potentially illustrate
the mechanism behind this apoptosis inhibition, and provide targets for functional
analysis in vitro. In addition, the data obtained fiom microarray analysis can be used to

identify pathways that Bmi-l is signaling through to prevent apoptosis, and these

147

pathways can be targeted with specific drugs, antibodies, or shRNA to knock out these
mediators to provide additional functional analysis and physiologic relevance to Bmi-l
induced apoptosis inhibition.

The mechanism leading to the striking difference in the metastatic phenotype
between the MCF10A+H-Ras and MCF10A+H-Ras+Bmi-l cells is perhaps the most
interesting and obvious further pursuit. First, an analysis by western blot of the
pulmonary metastases from MCF10A+H-Ras (obtained by laser microdissection) and
MCF10A+H-Ras+Bmi-l tumors for mediators involved in the metastatic phenotype such
as the matrix metalloproteinases (MMPs) or angiogenic mediators (V EGF , Angl, Tie2)
would be integral. Secondly, identification of such mediators in the microarray data
would provide targets for functional analysis in vivo. An important functional experiment
would be to knock down the expression of one or more such genes identified on the
microarray or by western blot in MCF10A+H-Ras+Bmi-l cells and test their invasive
capabilities in a Matrigel invasion assay compared to cells without knock down.
Ultimately, it would be important to use these cells in a mammary fat pad xenograft
model to analyze for changes in spontaneous metastasis, or in a tail vein model to
measure differences in embolization and establishment of pulmonary metastases.

Since H-Ras is not the only important oncogene involved in the genesis of breast
cancer, collaborative studies with Bmi-1 and other oncogenes will be important. Similar
experiments as those that have been performed in this dissertation research with H-Ras
can be performed with other oncogenes important in breast cancer, such as c-erb—BZ
(Her2neu), c-myc, TGFa, Akt, p53, BRCAl, Notch, and Wnt. Some of these mediators

have been shown to be involved in the Bmi-l pathway (e.g., p53), and it would be

148

important to determine the functional consequences of Bmi-l overexpression in the
presence of each mutation. Other mediators have been associated with stem cell pathways
(BRCAl, Notch, Wnt), and the determination of their association with Bmi-l in any
mechanistic sense would be important to understanding the role Bmi-1 plays in
influencing tumorigenesis, and establishing that the effects of Bmi-1 are not only
oncogene specific, but are also global throughout the genome.

Finally, since we have shown that Bmi-1 has important effects on cell growth and
transformation in vitro and tumorigenesis and metastasis in vivo, it is important to
investigate the clinical relevance of the role of Bmi-1 in breast cancer through
translational research. Since Bmi-l overexpression is associated with a more aggressive
phenotype, targeting Bmi-l as a tumor marker would be potentially important in the
diagnosis and treatment of aggressive breast cancer. Development of drugs to inhibit the
Bmi-l pathway would be an important potential means to treat or slow the progression of

disease, including distant metastasis.

149

APPENDIX E

Figure 28: All enriched BioCarta pathways shared by Bmi-l overexpressing cells.
Heatmap dendrogram comparing gene expression pathways in MCF10A vs.
MCF10A+Bmi-l and MCF10A+H-Ras vs. MCF10A+H-Ras+Bmi-l, revealing a Bmi-l
specific gene expression signature; several genes associated with the ATM growth
pathway, and G2/M cell cycle checkpoint, including p53 and Rb pathways, are most
differentially expressed. Each column represents a comparison between two MCF10A-
derived cell lines, and the red signal indicates a significant change in ATM and G2/M
checkpoint signaling related to Bmi-l expression.

-D.0 1:]. 3.0
‘6 U ‘U
-1 -¢ --1
o o o
‘U‘Ufilla UII-‘UlI-‘U
—r-r-cin HID—CID—t
O O O I 0 IO '0
hhhdUh-Illu—(ll.
win-n I-run Inn in
I I IV! o In in, r
Hl—‘I-‘las-‘lm-‘la-‘l
mun-he IwEmEm
BZBZOZ‘J-‘Im' ”IE‘IE
e.e.e.° a 91.:- e.° e.
-1 -r
.1 "unbeaten—Ik- I)
O QIU OIIU ¢
~0le I—«nnln‘.
II. I l I~hu l l I l
U n 01 02:9 0 0 0 0
I: d d can: u u u u
Infiniti- I lo 0 o o
n-a—u-ne-a n ”£85:
git: Q, G 9.9.9.9.
unnatural/1min
, , ,_

ATM signaling
GZ/M checkpoint

 

150

Figure 29: Genes up— and down-regulated in the ATM and G2/M cell cycle Checkpoint
pathways. BioCarta pathways differentially expressed in MCF10A vs. MCF10A+Bmi-l
and MCF10A+H-Ras vs. MCF10A+H-Ras+Bmi-l reveals a unique Bmi-l expression
profile of up- and down-regulated pathways. Green represents downregulation and red
represents upregulation of genes in each pathway.

MRE11A

uni-:- ABL ALO36776
III-II) .Z-I-

RAD51
ATM signaling pathway

m
£15357 \RADso
CDKN‘lA
IIIIIIED
-IIID

GAD45 NFKB1A
- EIIAIEUJ was an 7
RPSGKA1 [:23- CHK‘l
IIIIIIJ III-

YWHAI-l \

W GZIM checkfiint
coc2//””
II-

CCN 1
‘11-

EB:
43%}. JAE, PRKDC

 

151

Figure 30: Gene-Ontology Biological Processes associated with Bmi-1 signature.
Pathways associated with the unique Bmi-1 gene profile include cell cycle control, cell
growth, and mitosis. Each column represents a comparison between two MCF10A-
derived cell lines, and the red signal indicates a significant change in ATM and G2/M
checkpoint signaling related to Bmi-1 expression.

1: 1: tr
... _. ..
o o o
v v v n. 1: k- 'U to. 1:
”c: 2 3 “’. ‘3 “’. 'o‘ "’. 3'
In. In. In. —< '5 I.- —c In. —( h.
V) In W I v-G In I In I V,
I I I v! o I v; I “'3 I
“I «I ..I D‘ a“; “I °' HI 9. HI
1:: a; a: ‘l '4' n; ”I n. ‘I a;
5.521%: a ‘3: Pas: E3.
r‘ n .— ll. e .< p-e ... lg. n
o q I: u o l: I: u v
-c a. an I: I ... In In I: a:
In I l l >—< In. I l l I
I: c 6 m I: u u u u
I a: a; n; ”I ”I E E 2: £2-
... ... .. ...
I: I: I: I: q q a. n. a. 8'
an in an an a. a. VI in in

cell cycle
cell growth
M phase
mitosis

 

Figure 31: Vectors and expression plasmids used in development of inducible
mammary-specific transgenic mouse model of Bmi-l overexpression. (A) The ~5.2 Kb
pBabepuro-thi-l retroviral plasmid contains the 981 bp human Bmi-l gene flanked by
the GAG and pSV40 genes, and an ampicillin resistance gene for propagation in bacteria
and a puromycin resistance gene for selection in mammalian cells. (B) The 2.6Kb pTRE-
tight expression vector contains a modified Tet response element (pTRE), containing the
tet operator sequence (tetO) upstream from the CMV promoter (PminCMV). (C) The
final construct containing the pTRE-CMV promoter, thi-l gene, SV40 polyA tail, and
ampicillin resistance gene (Amp').

A_ Baml Sall

   

 

Puro H LTR

 

152

Figure 31 (continued)

B.

   
  

 
   

Xhol (2)
M63 (323-41 1 )
Pvul
(1 985)
Xhol (602)
C. Xhol (2)
EcoRI (324)
BamHI (345)
EcoRI (752)

Xhol (1928)

Sall (1326)

153

Figure 32: Possible orientation of integrations of the transgene and their respective
fragment sizes following restriction enzyme digestion. Possible orientations of the
transgene following random integration include head-to-head (resulting in a ~700 bp
fragment), tail-to-tail (resulting in a ~3.2 kb fragment), and head to tail (resulting in a ~2
kb fragment).

Head-to—head:

' ctio >1.6 ' '
jun "l ) 1011(1ng (>16)

 

 

WW—HH—I’VV

350bp 350bp
~700bp

 

Tail-to-tail:

junction (>350bp) ju ction >350bp)
,—— —\ e ,3 f—‘nx 5“
13d of —> /\ (— f1?

350bp

¥ ,4
f

~3.2kb

Head-to-tail:

 

Figure 33: Southern blot analysis of 10 pg of genomic DNA from pTRE-Bmi-l
transgenic founders. Ten founder lines were identified with relative copy numbers
ranging from one (H8) to eighty (A1) copies.

 

 

 

A
~32”) C A1 '6 P1111
~2.0kb .- an
~7OODp '
GS GB
“ “ — sag—3.}
K9

H8 H10

EL- 54.

 

155

Figure 34: Tetracycline induced mammary-specific mRN A expression of Bmi-1. mRNA
expression in mammary glands of single positive (pTRE-BmiVMMTV-rtta'), double
positive (pTRE-BmiVMMTV-rttf), and wildtype F VB mice following 14 days of
doxycycline administration.

 

156

10.

ll.

12.

13.

14.

15.

REFERENCES
Sparmann, A. and M. van Lohuizen, Polycomb silencers control cell fate,
development and cancer. Nat Rev Cancer, 2006. 6(11): p. 846-56.
Holliday, R., Epigenetics: an overview. Dev Genet, 1994. 15(6): p. 453-7.

Baylin, SB. and J .E. Ohm, Epigenetic gene silencing in cancer - a mechanism for
early oncogenic pathway addiction? Nat Rev Cancer, 2006. 6(2): p. 107-l6.

Henikoff, S. and K. Ahmad, Assembly of variant histones into chromatin. Annu
Rev Cell Dev Biol, 2005. 21: p. 133-53.

Jones, P. A. and S. B. Baylin, T he fundamental role of epigenetic events in cancer.
Nat Rev Genet, 2002. 3(6). p. 415 2.8

Smith, CL. and CL. Peterson, AT P—dependent chromatin remodeling. Curr Top
Dev Biol, 2005. 65: p. 115-48.

Strahl, B.D. and CD. Allis, The language of covalent histone modifications.
Nature, 2000. 403(6765): p. 41-5.

Barcellos-Hoff, M.H., It takes a tissue to make a tumor: epigenetics, cancer and
the microenvironment. J Mammary Gland Biol Neoplasia, 2001. 6(2): p. 213-21.

Barcellos-Hoff, M.H. and D. Medina, New highlights on stroma-epithelial
interactions in breast cancer. Breast Cancer Res, 2005. 7(1): p. 33-6.

Cavalli, G., Chromatin and epigenetics in development: blending cellular memory
with cell fate plasticity. Development, 2006. 133(11): p. 2089-94.

Mello, CC. and D. Conte, J r., Revealing the world of RNA interference. Nature,
2004. 431(7006): p. 338-42.

Jones, PL. and AP Wolffe, Relationships between chromatin organization and
DNA methylation in determining gene expression. Semin Cancer Biol, 1999. 9(5):
p. 339-47.

Qiu, J ., Epigenetics: unfinished symphony. Nature, 2006. 441(7090): p. 143-5.

Wolffe, AP. and MA. Matzke, Epigenetics: regulation through repression.
Science, 1999. 286(5439): p. 481-6.

Jones, RA. and D. Takai, The role of DNA methylation in mammalian
epigenetics. Science, 2001. 293(5532): p. 1068-70.

157

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

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

Wolffe, A.P., Packaging principle: how DNA methylation and histone acetylation
control the transcriptional activity of chromatin. J Exp Zool, 1998. 282(1-2): p.
239-44. '

Itahana, K., et al., Control of the replicative life span of human fibroblasts by p16
and the polycomb protein Bmi-I. Mol Cell Biol, 2003. 23(1): p. 389-401.

Guney, 1., S. Wu, and J .M. Sedivy, Reduced c-Myc signaling triggers telomere-
independent senescence by regulating Bmi—I and p1 6(INK4a). Proc Natl Acad Sci
U S A, 2006. 103(10): p. 3645-50.

Hernandez-Munoz, 1., et al., Association of BM] with polycomb bodies is
dynamic and requires PR C2/EZH2 and the maintenance DNA methyltransferase
DNMT 1. Mol Cell Biol, 2005. 25(24): p. 11047-58.

Voncken, J .W., et al., Chromatin-association of the Polycomb group protein
BM] is cell cycle-regulated and correlates with its phosphorylation status. J Cell
Sci, 1999. 112 (Pt 24): p. 4627-39.

Bracken, A.P., et al., The Polycomb group proteins bind throughout the INK4A-
ARF locus and are disassociated in senescent cells. Genes Dev, 2007. 21(5): p.
525-30.

Levine, S.S., I._F. King, and RE. Kingston, Division of labor in polycomb group
repression. Trends Biochem Sci, 2004. 29(9): p. 478-85.

Bracken, A.P., et al., Genome-wide mapping of Polycomb target genes unravels
their roles in cell fate transitions. Genes Dev, 2006. 20(9): p. 1123-36.

van Kemenade, F .J., et al., Coexpression of BM—I and EZH2 polycomb-group
proteins is associated with cycling cells and degree of malignancy in B-cell non-
Hodgkin lymphoma. Blood, 2001. 97(12): p. 3896-901.

Simon, LA. and J.W. Tamkun, Programming off and on states in chromatin:
mechanisms of Polycomb and trithorax group complexes. Curr Opin Genet Dev,

2002. 12(2): p. 210-8.

Al-Hajj, M., et al., Therapeutic implications of cancer stem cells. Curr Opin
Genet Dev, 2004. 14(1): p. 43-7.

Alkema, M.J., et al., Characterization and chromosomal localization of the
human proto-oncogene BM-I. Hum Mol Genet, 1993. 2(10): p. 1597-603.

158

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

Dimri, G.P., et al., The Bmi-1 oncogene induces telomerase activity and
immortalizes human mammary epithelial cells. Cancer Res, 2002. 62(16): p.
4736-45.

Jacobs, J .J ., et al., Bmi-1 collaborates with c—Myc in tumorigenesis by inhibiting
c-Myc-induced apoptosis via INK4a/ARF. Genes Dev, 1999. 13(20): p. 2678-90.

Lessard, J. and G. Sauvageau, Bmi—I determines the proliferative capacity of
normal and leukaemic stem cells. Nature, 2003. 423(6937): p. 255-60.

Molofsky, A.V., et al., Bmi-1 promotes neural stem cell self-renewal and neural
development but not mouse growth and survival by repressing the p161nk4a and

p1 9Arf senescence pathways. Genes Dev, 2005. 19(12): p. 1432-7.

Cui, H., et al., Bmi-1 is essential for the tumorigenicity of neuroblastoma cells.
Am J Pathol, 2007. 170(4): p. 1370-8.

Cui, H., et al., Bmi-1 regulates the differentiation and clonogenic self-renewal of
I-type neuroblastoma cells in a concentration-dependent manner. J Biol Chem,

2006. 281(45): p. 34696-704.

Liu, S., et al., Hedgehog signaling and Bmi-I regulate self-renewal of normal and
malignant human mammary stem cells. Cancer Res, 2006. 66(12): p. 6063-71.

Haga, K., et al., Eflicient immortalization of primary human cells by pI6INK4a-
specific short hairpin RNA or Bmi-1, combined with introduction of h T ERT.
Cancer Sci, 2007. 98(2): p. 147-54.

Song, L.B., et al., Bmi-I is a novel molecular marker of nasopharyngeal

carcinoma progression and immortalizes primary human nasopharyngeal
epithelial cells. Cancer Res, 2006. 66(12): p. 6225-32.

Kang, M.K., et al., Elevated Bmi-1 expression is associated with dysplastic cell

transformation during oral carcinogenesis and is required for cancer cell
replication and survival. Br J Cancer, 2007. 96(1): p. 126-33.

Nowak, K., et al., BM] is a target gene of E2F -I and is strongly expressed in
primary neuroblastomas. Nucleic Acids Res, 2006. 34(6): p. 1745-54.

Leung, C., et al., BmiI is essential for cerebellar development and is
overexpressed in human medulloblastomas. Nature, 2004. 428(6980): p. 337-41.

Vonlanthen, S., et al., The bmi-I oncoprotein is differentially expressed in non-
small cell lung cancer and correlates with INK4A -ARF locus expression. Br J
Cancer, 2001. 84(10): p. 1372-6.

159

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

ReinisCh, C.M., et al., Expression of BM—I in normal skin and inflammatory and
neoplastic skin lesions. J Cutan Pathol, 2007. 34(2): p. 174-80.

Neo, S.Y., et al., Identification of discriminators of hepatoma by gene expression
profiling using a minimal dataset approach. Hepatology, 2004. 39(4): p. 944-53.

Kim, J .H., et al., The Bmi-I oncoprotein is overexpressed in human colorectal
cancer and correlates with the reduced p1 6INK4a/pI4ARF proteins. Cancer Lett,
2004. 203(2): p. 217-24.

Reinisch, C., et al., BM-I : a protein expressed in stem cells, specialized cells and
tumors of the gastrointestinal tract. Histol Histopathol, 2006. 21(11): p. 1143-9.

Tateishi, K., et al., Dysregulated expression of stem cell factor Bmi] in
precancerous lesions of the gastrointestinal tract. Clin Cancer Res, 2006. 12(23):
p. 6960-6.

Kim, J .H., et al., Overexpression of Bmi-1 oncoprotein correlates with axillary
lymph node metastases in invasive ductal breast cancer. Breast, 2004. 13(5): p.
383-8.

Glinsky, G.V., O. Berezovska, and AB. Glinskii, Microarray analysis identifies a
death-from-cancer signature predicting therapy failure in patients with multiple
types of cancer. J Clin Invest, 2005. 115(6): p. 1503-21.

Steele, J .C., et al., The polycomb group proteins, BM—I and EZH2, are tumour-
associated antigens. Br J Cancer, 2006. 95(9): p. 1202-11.

Silva, J ., et al., Implication of polycomb members Bmi-I, Mel-18, and Hpc-2 in
the regulation of p1 6INK4a, p1 4ARF, h-T ER T, and c-Myc expression in primary
breast carcinomas. Clin Cancer Res, 2006. 12(23): p. 6929-36.

Berns, A., et al., Identification and characterization of collaborating oncogenes in
compound mutant mice. Cancer Res, 1999. 59(7 Suppl): p. 17738-17773.

Haupt, Y., et al., Novel zinc finger gene implicated as myc collaborator by
retrovirally accelerated lymphomagenesis in E mu-myc transgenic mice. Cell,

1991. 65(5): p. 753-63.

van Lohuizen, M., et al., Identification of cooperating oncogenes in E mu-myc
transgenic mice by provirus tagging. Cell, 1991. 65(5): p. 737-52.

Krimpenfort, P., et al., Loss of pl 6Ink4a confers susceptibility to metastatic
melanoma in mice. Nature, 2001. 413(6851): p. 83-6.

160

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

Sharpless, N.E., et al., Loss of p] 6Ink4a with retention of p1 9Arf predisposes mice
to tumorigenesis. Nature, 2001. 413(6851): p. 86-91.

Sherr, OJ. and F. McCormick, The RB and p53 pathways in cancer. Cancer Cell,
2002. 2(2): p. 103-12.

Kamijo, T., et al., Tumor spectrum in ARF-deficient mice. Cancer Res, 1999.
59(9): p. 2217-22.

Hirakawa, T. and HE. Ruley, Rescue of cells from ras oncogene-induced growth
arrest by a second, complementing, oncogene. Proc Natl Acad Sci U S A, 1988.
85(5): p. 1519-23.

J onkers, J ., et al., Activation of a novel proto-oncogene, F rat] , contributes to
progression of mouse T -cell lymphomas. Embo J, 1997. 16(3): p. 441-50.

Lloyd, A.C., et al., Cooperating oncogenes converge to regulate cyclin/cdk
complexes. Genes Dev, 1997. 11(5): p. 663-77.

Pitot, H.C., Fundamentals of Oncology. 1986, New York: Marcel Dekker, Inc.

Ragozzino, M.M., et al., Mechanisms of oncogene cooperation: activation and
inactivation of a growth antagonist. Environ Health Perspect, 1991. 93: p. 97-103.

Vogelstein, B. and K.W. Kinzlef, The multistep nature of cancer. Trends Genet,
1993. 9(4): p. 138-41.

Adams, J.M. and S. Cory, Oncogene co-operation in leukaemogenesis. Cancer
Surv, 1992. 15: p. 119-41.

Cardiff, RD. and W.J. Muller, Transgenic mouse models of mammary
tumorigenesis. Cancer Surv, 1993. 16: p. 97-113.

Hunter, T., Cooperation between oncogenes. Cell, 1991. 64(2): p. 249-70.

J onkers, J. and A. Bems, Retroviral insertional mutagenesis as a strategr to
identify cancer genes. Biochim Biophys Acta, 1996. 1287(1): p. 29-57.

Land, H., L.F. Parada, and RA. Weinberg, Cellular oncogenes and multistep
carcinogenesis. Science, 1983. 222(4625): p. 771-8.

Sinn, E., et al., Coexpression of MMT V/v-Ha-ras and MMTV/c-myc genes in
transgenic mice: synergistic action of oncogenes in vivo. Cell, 1987. 49(4): p.
465-75.

161

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

Palmieri, S., Oncogene requirements for tumorigenicity: cooperative effects
between retroviral oncogenes. Curr Top Microbiol Irnmunol, 1989. 148: p. 43-91.

Ruley, H.E., Transforming collaborations between ras and nuclear oncogenes.
Cancer Cells, 1990. 2(8-9): p. 258-68.

Weinberg, R.A., Oncogenes, antioncogenes, and the molecular bases of multistep
carcinogenesis. Cancer Res, 1989. 49(14): p. 3713-21.

Schutte, J., J.D. Minna, and M.J. Birrer, Deregulated expression of human c-jun
transforms primary rat embryo cells in cooperation with an activated c-Ha—ras
gene and transforms rat-1a cells as a single gene. Proc Natl Acad Sci U S A,
1989. 86(7): p. 2257-61.

Jochemsen, A.G., et al., Diflerent activities of the adenovirus types 5 and 12 E IA

regions in transformation with the El Ha-ras oncogene. J Virol, 1986. 59(3): p.
684-91.

Fitzgerald, K., A. Harrington, and P. Leder, Ras pathway signals are required for
notch-mediated oncogenesis. Oncogene, 2000. 19(37): p. 4191-8.

Land, H., L.F. Parada, and RA. Weinberg, T umorigenic conversion of primary
embryo fibroblasts requires at least two cooperating oncogenes. Nature, 1983.

304(5927): p. 596-602.

Newbold, RF. and R.W. Overell, Fibroblast immortality is a prerequisite for
transformation by EJ c-Ha-ras oncogene. Nature, 1983. 304(5927): p. 648-5].

Ridley, A.J., et al., Ras-mediated cell cycle arrest is altered by nuclear oncogenes
to induce Schwann cell transformation. Embo J, 1988. 7(6): p. 1635-45.

Hulit, J ., D. Di Vizio, and R.G. Pestell, Inducible transgenics. New lessons on

events governing the induction and commitment in mammary tumorigenesis.
Breast Cancer Res, 2001. 3(4): p. 209-12.

Jacobs, J .J . and M. van Lohuizen, Cellular memory of transcriptional states by
Polycomb-group proteins. Semin Cell Dev Biol, 1999. 10(2): p. 227-35.

van Lohuizen, M., The trithorax-group and polycomb-group chromatin modifiers:
implications for disease. Curr Opin Genet Dev, 1999. 9(3): p. 355-61.

Alkema, M.J., et al., Pertubation of B and T cell development and predisposition
to lymphomagenesis in Emu Bmi] transgenic mice require the Bmi] RING finger.
Oncogene, 1997. 15(8): p. 899-910.

162

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

Haupt, Y., et al., bmi-I transgene induces lymphomas and collaborates with myc
in tumorigenesis. Oncogene, 1993. 8(11): p. 3161-4.

Nesbit, C.E., J .M. Tersak, and E.V. Prochownik, MYC oncogenes and human
neoplastic disease. Oncogene, 1999. 18(19): p. 3004-16.

F acchini, L.M. and L.Z. Penn, The molecular role of Myc in growth and

transformation: recent discoveries lead to new insights. Faseb J, 1998. 12(9): p.
633-51.

Henriksson, M. and B. Luscher, Proteins of the Myc network: essential regulators
of cell growth and differentiation. Adv Cancer Res, 1996. 68: p. 109-82.

Langdon, W.Y., et al., The c-myc oncogene perturbs B lymphocyte development
in E-mu-myc transgenic mice. Cell, 1986. 47(1): p. 11-8.

Rochlitz, C.F., et al., Incidence of activating ras oncogene mutations associated
with primary and metastatic human breast cancer. Cancer Res, 1989. 49(2): p.
357-60.

Sparano, J .A., et al., Targeted inhibition of farnesyltransferase in locally
advanced breast cancer: a phase I and II trial of tipifamib plus dose-dense
doxorubicin and cyclophosphamide. J Clin Oncol, 2006. 24(19): p. 3013-8.

Thor, A., et al., ras gene alterations and enhanced levels of ras p21 expression in
a spectrum of benign and malignant human mammary tissues. Lab Invest, 1986.
55(6): p. 603-15. .

Bunone, G., et al., Activation of the unliganded estrogen receptor by E GF
involves the MAP kinase pathway and direct phosphorylation. Embo J, 1996.
15(9): p. 2174-83.

Kato, S., et al., Molecular mechanism of a cross-talk between oestrogen and
growth factor signalling pathways. Genes Cells, 2000. 5(8): p. 593-601 .

Smith, C.A., et al., Correlations among p53, Her-Z/neu, and ras overexpression
and aneuploidy by multiparameterflow cytometry in human breast cancer:
evidence for a common phenotypic evolutionary pattern in infiltrating ductal
carcinomas. Clin Cancer Res, 2000. 6(1): p. 112-26.

Janocko, L.E., et al., Assessing sequential oncogene amplification in human
breast cancer. Cytometry, 1995. 21(1): p. 18-22.

Theillet, C ., et al., Loss of a c-H-ras-I allele and aggressive human primary
breast carcinomas. Cancer Res, 1986. 46(9): p. 4776-81.

163

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

Kleer, C.G., et al., Characterization of RhoC expression in benign and malignant
breast disease: a potential new marker for small breast carcinomas with
metastatic ability. Am J Pathol, 2002. 160(2): p. 579-84.

van Golen, K.L., et al., Reversion of RhoC GT Pose-induced inflammatory breast
cancer phenotype by treatment with a farnesyl transferase inhibitor. Mol Cancer
Ther, 2002. 1(8): p. 575-83. ' '

Voncken, J .W., et al., MAPKAP kinase 3pK phosphorylates and regulates

chromatin association of the polycomb group protein Bmi]. J Biol Chem, 2005.
280(7): p. 5178-87.

Miller, F .R., et al., Xenograft model of progressive human proliferative breast
disease. J Natl Cancer Inst, 1993. 85(21): p. 1725-32.

Al-Hajj, M., et al., Prospective identification of tumorigenic breast cancer cells.
Proc Natl Acad Sci U S A, 2003. 100(7): p. 3983-8.

Dontu, G., eta1., Stem cells in normal breast development and breast cancer. Cell
Prolif, 2003. 36 Suppl 1: p. 59-72.

Dontu, G. and MS. Wicha, Survival of mammary stem cells in suspension
culture: implications for stem cell biology and neoplasia. J Mammary Gland Biol
Neoplasia, 2005. 10(1): p. 75-86.

Liu, S., G. Dontu, and MS. Wicha, Mammary stem cells, self-renewal pathways,
and carcinogenesis. Breast Cancer Res, 2005. 7(3): p. 86-95.

Pathak, S., Organ- and tissue-specific stem cells and carcinogenesis. Anticancer
Res, 2002. 22(3): p. 1353-6.

Reya, T., et al., Stem cells, cancer, and cancer stem cells. Nature, 2001.
414(6859): p. 105-11.

Sell, S. and GB. Pierce, Maturation arrest of stem cell diflerentiation is a
common pathway for the cellular origin of teratocarcinomas and epithelial

cancers. Lab Invest, 1994. 70(1): p. 6-22.

Tu, S.M., S.H. Lin, and CJ. Logothetis, Stem-cell origin of metastasis and
heterogeneity in solid tumours. Lancet Oncol, 2002. 3(8): p. 508-13.

Soltysova, A., V. Altanerova, and C. Altaner, Cancer stem cells. Neoplasma,
2005. 52(6): p. 435-40.

Singh, S.K., et al., Identification of human brain tumour initiating cells. Nature,
2004. 432(7015): p. 396-401.

164

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

121.

122.

123.

Taipale, J. and RA. Beachy, The Hedgehog and Wnt signalling pathways in
cancer. Nature, 2001. 411(6835): p. 349-54.

Hemmati, H.D., et al., Cancerous stem cells can arise fiom pediatric brain
tumors. Proc Natl Acad Sci U S A, 2003. 100(25): p. 15178-83.

Smith, K.S., eta1., Bmi-1 regulation of INK4A—ARF is a downstream requirement
for transformation of hematopoietic progenitors by E2a-Pbx1. Mol Cell, 2003.
12(2): p. 393-400.

Valk-Lingbeek, M.E., S.W. Bruggeman, and M. van Lohuizen, Stem cells and
cancer; the polycomb connection. Cell, 2004. 118(4): p. 409-18.

Park, [K., et al., Bmi-1 is required for maintenance of adult self-renewing
haematopoietic stem cells. Nature, 2003. 423(6937): p. 302-5.

lwama, A., et al., Enhanced self-renewal of hematopoietic stem cells mediated by
the polycomb gene product Bmi-1. Immunity, 2004. 21(6): p. 843-51.

Molofsky, A.V., et al., Bmi-l dependence distinguishes neural stem cell self-
renewal fiom progenitor proliferation. Nature, 2003. 425(6961): p. 962-7.

Qiao, C., et al., Human mesenchymal stem cells isolated from the umbilical cord.
Cell Biol Int, 2007.

Datta, S., et al., Bmi-1 Cooperates with H-Ras to Transform Human Mammary
Epithelial Cells via Dysregulation of Multiple Growth-Regulatory Pathways.
Cancer Research, 2007. 67(21): p. 10286-10295.

Dahmane, N. and A. Ruiz i Altaba, Sonic hedgehog regulates the growth and
patterning of the cerebellum. Development, 1999. 126(14): p. 3089-100.

Wechsler-Reya, R]. and MP. Scott, Control of neuronal precursor proliferation
in the cerebellum by Sonic Hedgehog. Neuron, 1999. 22(1): p. 103-14.

Wang, V.Y. and H.Y. Zoghbi, Genetic regulation of cerebellar development. Nat
Rev Neurosci, 2001. 2(7): p. 484-91.

Sawa, M., et al., BM—I is highly expressed in MO-subtype acute myeloid
leukemia. Int J Hematol, 2005. 82(1): p. 42-7.

Kennedy, C ., et al., Cell proliferation in the normal mouse mammary gland and
inhibition by phenylbutyrate. Mol Cancer Ther, 2002. 1( 12): p. 1025-33.

165

124.

125.

126.

127.

128.

129.

130.

131.

132.

133.

134.

135.

136.

Longacre, TA. and SA. Bartow, A correlative morphologic study of human
breast and endometrium in the menstrual cycle. Am J Surg Pathol, 1986. 10(6): p.
382-93. '

Green, J .E., et al., The C3 ( I )/S V40 T -antigen transgenic mouse model of
mammary cancer: ductal epithelial cell targeting with multistage progression to
carcinoma. Oncogene, 2000. 19(8): p. 1020-7.

Kavanaugh, C. and J .E. Green, The use of genetically altered mice for breast
cancer prevention studies. J Nutr, 2003. 133(7 Suppl): p. 2404S-2409S.

Shibata, M.A., et al., The C3 (1)/S V40 T antigen transgenic mouse model of
prostate and mammary cancer. Toxicol Pathol, 1998. 26(1): p. 177-82.

Maroulakou, I.G., et al., Prostate and mammary adenocarcinoma in transgenic

mice carrying a rat C3 (1) simian virus 40 large tumor antigen fusion gene. Proc
Natl Acad Sci U S A, 1994. 91(23): p. 11236-40.

Maroulakou, I.G., et al., Reduced p53 dosage associated with mammary tumor
metastases in C3(1)/TAG transgenic mice. Mol Carcinog, 1997. 20(2): p. 168-74.

Abd El-Rehim, D.M., et al., Expression of luminal and basal cytokeratins in
human breast carcinoma. J Pathol, 2004. 203(2): p. 661-71.

Nielsen, T.O., et al., Immunohistochemical and clinical characterization of the
basal-like subtype of invasive breast carcinoma. Clin Cancer Res, 2004. 10(16):
p. 5367-74. '

Honrado, E., J. Benitez, and J. Palacios, Histopathology of BRCA1 - and BRCA2-
associated breast cancer. Crit Rev Oncol Hematol, 2006. 59(1): p. 27-39.

Ford, D., et al., Genetic heterogeneity and penetrance analysis of the BRCA1 and
BRCA2 genes in breast cancer families. The Breast Cancer Linkage Consortium.
Am J Hum Genet, 1998. 62(3): p. 676-89.

Lakhani, S.R., The pathology of hereditary breast cancer. Dis Markers, 1999.
15(1-3): p. 113-4.

Honrado, E., et al., Pathology and gene expression of hereditary breast tumors
associated with BRCA], BRCA2 and CHEKZ gene mutations. Oncogene, 2006.
25(43): p. 5837-45.

Lakhani, S.R., The pathology of familial breast cancer: Morphological aspects.
Breast Cancer Res, 1999. 1(1): p. 31-5.

166

137.

138.

139.

140.

141.

142.

143.

144.

145.

146.

147.

148.

149.

Bertucci, F ., et al., Gene expression profiling shows medullary breast cancer is a
subgroup of basal breast cancers. Cancer Res, 2006. 66(9): p. 4636-44.

Qiu, T.H., et al., Global expression profiling identifies signatures of tumor
virulence in MMTV-PyMT—transgenic mice: correlation to human disease. Cancer
Res, 2004. 64(17): p. 5973-81.

Guy, C.T., R.D. Cardiff, and W.J. Muller, Induction of mammary tumors by

expression of polyomavirus middle T oncogene: a transgenic mouse model for
metastatic disease. Mol Cell Biol, 1992. 12(3): p. 954-61.

Lin, E.Y., et al., Progression to malignancy in the polyoma middle T oncoprotein
mouse breast cancer model provides a reliable model for human diseases. Am J
Pathol, 2003. 163(5): p. 2113-26.

van de Vijver, M.J., The pathology of familial breast cancer: The pre-
BRCA I/BRCAZ era: historical perspectives. Breast Cancer Res, 1999. 1(1): p. 27-
30.

Ferrini, F .S., et al., Schirrous invasive ductal carcinoma of the breast overexpress
p53 oncoprotein. Sao Paulo Med J, 2001. 119(1): p. 4-6.

Aouni, N.E., et al., Medullary breast carcinoma: a case report with cytological
features and histological confirmation. Diagn Cytopathol, 2006. 34(10): p. 701-3.

Ridolfi, R.L., et al., Medullary carcinoma of the breast: a clinicopathologic study
with 1 0 year follow-up. Cancer, 1977. 40(4): p. 1365-85.

Liu, S., et al., BRCA] regulates human mammary stem/progenitor cell fate. Proc
Natl Acad Sci U S A, 2008. 105(5): p. 1680-5.

Wright, M.H., et al., BrcaI breast tumors contain distinct CD44 +/CD24- and
CD133+ cells with cancer stem cell characteristics. Breast Cancer Res, 2008.
10(1): p. R10.

Ringrose, L. and R. Paro, Epigenetic regulation of cellular memory by the
Polycomb and T rithorax group proteins. Annu Rev Genet, 2004. 38: p. 413-43.

Raaphorst, F .M., Deregulated expression of Polycomb-group oncogenes in human
malignant lymphomas and epithelial tumors. Hum Mol Genet, 2005. 14 Spec No
1: p. R93-R100.

Bea, S., et al., BM-I gene amplification and overexpression in hematological
malignancies occur mainly in mantle cell lymphomas. Cancer Res, 2001. 61(6): p.
2409-12.

167

150.

151.

152.

153.

154.

155.

156.

157.

158.

159.

1.60.
161.

162.

Jacobs, J.J., et al., The oncogene and Polycomb-group gene bmi-I regulates cell
proliferation and senescence through the ink4a locus. Nature, 1999. 397(6715): p.
164-8.

Dimri, G.P., What has senescence got to do with cancer? Cancer Cell, 2005. 7(6):
p. 505-12.

Kim, R.H., et al., Bmi-1 cooperates with human papillomavirus type 16 E6 to
immortalize normal human oral keratinocytes. Exp Cell Res, 2007. 313(3): p.
462-72.

Guo, W.J., et al., Mel-18 acts as a tumor suppressor by repressing Bmi-1
expression and down-regulating Akt activity in breast cancer cells. Cancer Res,
2007. 67(11): p. 5083-9.

Berman, H., et al., Genetic and epigenetic changes in mammary epithelial cells
identify a subpopulation of cells involved in early carcinogenesis. Cold Spring
Harb Symp Quant Biol, 2005. 70: p. 317-27.

Cowell, J .K., et al., Molecular characterization of the t(3,'9) associated with
immortalization in the MCF10A cell line. Cancer Genet Cytogenet, 2005. 163(1):
p. 23-9.

Band, V., et al., Human papilloma virus DNAs immortalize normal human
mammary epithelial cells and reduce their growth factor requirements. Proc Natl
Acad Sci U S A, 1990. 87(1): p. 463-7.

Elenbaas, B., et al., Human breast cancer cells generated by oncogenic
transformation of primary mammary epithelial cells. Genes Dev, 2001. 15(1): p.
50-65.

Seeburg, P.H., et al., Biological properties of human c-Ha-rasl genes mutated at
codon 12. Nature, 1984. 312(5989): p. 71-5.

Debnath, J ., et al., The role of apoptosis in creating and maintaining luminal
space within normal and oncogene-expressing mammary acini. Cell, 2002.

111(1): p. 29-40.

Landis, M.W., et al., Cyclin DI -dependent kinase activity in murine development
and mammary tumorigenesis. Cancer Cell, 2006. 9(1): p. 13-22.

Yu, Q., et al., Requirement for CDK4 kinasefunction in breast cancer. Cancer
Cell, 2006. 9(1): p. 23-32.

Dawson, P.J., et al., MCF10A T: a model for the evolution of cancer from
proliferative breast disease. Am J Pathol, 1996. 148(1): p. 313-9.

168

163.

164.

165.

166.

167.

168.

169.

170.

171.

172.

173.

174.

175.

Giunciuglio, D., et al., Invasive phenotype ofMCF10A cells overexpressing c-Ha-
ras and c-erbB-2 oncogenes. Int J Cancer, 1995. 63(6): p. 815-22.

Moon, A., et al., H-ras, but not N-ras, induces an invasive phenotype in human
breast epithelial cells: a role for MAP-2 in the H-ras-induced invasive
phenotype. Int J Cancer, 2000. 85(2): p. 176-81.

Santner, S.J., et al., Malignant MCF1 0CA .1 cell lines derived from premalignant
human breast epithelial MCF1 OAT cells. Breast Cancer Res Treat, 2001. 65(2): p.
101-10.

Shin, I., et al., H-Ras-specific activation of Rac-AKK3/6-p38 pathway: its critical
role in invasion and migration of breast epithelial cells. J Biol Chem, 2005.
280(15): p. 14675-83.

Serrano, M., et al., Oncogenic ras provokes premature cell senescence associated
with accumulation of p53 and p1 6INK4a. Cell, 1997. 88(5): p. 593-602.

Sun, P., et al., PRAK is essential for ras-induced senescence and tumor
suppression. Cell, 2007. 128(2): p. 295-308.

Wang, W., et al., Sequential activation of the [lax-extracellular signal-regulated .
kinase and W3/6-p38 mitogen-activated protein kinase pathways mediates
oncogenic ras-induced premature senescence. Mol Cell Biol, 2002. 22(10): p.
3389-403.

Chen, G., et al., The p38 pathway provides negative feedback for Ras proliferative
signaling. J Biol Chem, 2000. 275(50): p. 38973-80.

el-Deiry, W.S., Regulation of p53 downstream genes. Semin Cancer Biol, 1998.
8(5): p. 345-57.

Nakano, K. and K.H. Vousden, PUMA, a novel proapoptotic gene, is induced by
p53. Mol Cell, 2001. 7(3): p. 683-94.

Flatt, P.M., et al., p53-dependent expression of PIG3 during proliferation,
genotoxic stress, and reversible growth arrest. Cancer Lett, 2000. 156(1): p. 63-
72.

Boehm, J .S. and WC. Hahn, Understanding transformation: progress and gaps.
Curr Opin Genet Dev, 2005. 15(1): p. 13-7.

Clark, G.J. and OJ. Der, Aberrant function of the Ras signal transduction

pathway in human breast cancer. Breast Cancer Res Treat, 1995. 35(1): p. 133-
44.

169

176.

177.

178.

179.

180.

181.

182.

183.

184.

185.

186.

187.

188.

von Lintig, F .C., et al., Ras activation in human breast cancer. Breast Cancer Res
Treat, 2000. 62(1): p. 51-62.

Sarkisian, C.J., et al., Dose-dependent oncogene-induced senescence in vivo and
its evasion during mammary tumorigenesis. Nat Cell Biol, 2007. 9(5): p. 493-505.

Dahl, C., et al., Human mast cells express receptors for IL-3, IL-5 and GM-CSF;
a partial map of receptors on human mast cells cultured in vitro. Allergy, 2004.
59(10): p. 1087-96.

Tang, B., et al., T GF -beta switches from tumor suppressor to prometastatic factor
in a model of breast cancer progression. J Clin Invest, 2003. 112(7): p. 1116-24.

Kim, L.S., et al., Vascular endothelial growth factor expression promotes the
growth of breast cancer brain metastases in nude mice. Clin Exp Metastasis,
2004. 21(2): p. 107-18.

Huang, K.H., et al., [Association of Bmi-I mRNA expression with diflerentiation,
metastasis and prognosis of gastric carcinoma]. Nan Fang Yi Ke Da Xue Xue
Bao, 2007. 27(7): p. 973-5.

F eng, Y., et al., [Expression and significance of Bmi-I in breast cancer]. Ai
Zheng, 2007. 26(2): p. 154-7.

Mihic-Probst, D., et al., Consistent expression of the stem cell renewal factor
BM-I in primary and metastatic melanoma. Int J Cancer, 2007. 121(8): p. 1764-
70.

Glinsky, G.V., Death-fi'om-cancer signatures and stem cell contribution to
metastatic cancer. Cell Cycle, 2005. 4(9): p. 1171-5.

F antozzi, A. and G. Christofori, Mouse models of breast cancer metastasis. Breast
Cancer Res, 2006. 8(4): p. 212.

Clayton, A.J., et al., Incidence of cerebral metastases in patients treated with
trastuzumab for metastatic breast cancer. Br J Cancer, 2004. 91(4): p. 639-43.

Bendell, J .C., et al., Central nervous system metastases in women who receive
trastuzumab-based therapy for metastatic breast carcinoma. Cancer, 2003.
97(12): p. 2972-7.

Tsukada, Y., et al., Central nervous system metastasis from breast carcinoma.
Autopsy study. Cancer, 1983. 52(12): p. 2349-54.

170

189.

190.

191.

192.

193.

194.

195.

196.

197.

198.

199.

200.

201.

Strickland, L.B., et al., Progression of premalignant MCF1 OAT generates
heterogeneous malignant variants with characteristic histologic types and
immunohistochemical markers. Breast Cancer Res Treat, 2000. 64(3): p. 235-40.

J essani, N., et al., Breast cancer cell lines grown in vivo: what goes in isn't always
the same as what comes out. Cell Cycle, 2005. 4(2): p. 253-5.

Vernon, A.E., S.J. Bakewell, and LA. Chodosh, Deciphering the molecular basis
of breast cancer metastasis with mouse models. Rev Endocr Metab Disord, 2007.
8(3): p. 199-213.

Clarke M., et al., Eflects of chemotherapy and hormonal therapy for early breast
cancer on recurrence and I5-year survival: an overview of the randomised trials.
Lancet, 2005. 365(9472): p. 1687-717.

Brackstone, M., J .L. Townson, and A.F. Chambers, Tumour dormancy in breast
cancer: an update. Breast Cancer Res, 2007. 9(3): p. 208.

Fisher, 3., et al., Findings from recent National Surgical Adjuvant Breast and
Bowel Project adjuvant studies in stage I breast cancer. J Natl Cancer Inst
Monogr, 2001(30): p. 62-6.

Wallgren, A., et al., Risk factors for locoregional recurrence among breast cancer
patients: results fi'om International Breast Cancer Study Group Trials I through
VII. J Clin Oncol, 2003. 21(7): p. 1205-13.

Glinsky, G.V., Genomic models of metastatic cancer: functional analysis of
death-from-cancer signature genes reveals aneuploid, anoikis-resistant,
metastasis-enabling phenotype with altered cell cycle control and activated
Polycomb Group (PcG) protein chromatin silencing pathway. Cell Cycle, 2006.
5(11): p. 1208-16.

Avruch, J ., et al., Ras activation of the Raf kinase: tyrosine kinase recruitment of
the MAP kinase cascade. Recent Prog Horm Res, 2001. 56: p. 127-55.

Ries, S., et al., Opposing eflects of Ras on p53: transcriptional activation of
mdm2 and induction of p1 9ARF . Cell, 2000. 103(2): p. 321-30.

Valentijn, A.J., N. Zouq, and AP. Gilmore, Anoikis. Biochem Soc Trans, 2004.
32(Pt3): p. 421-5.

Parsons, J .T., Focal adhesion kinase: the first ten years. J Cell Sci, 2003. 116(Pt
8): p. 1409-16.

Downward, J ., Ras signalling and apoptosis. Curr Opin Genet Dev, 1998. 8(1): p.
49-54.

171

202.

203.

204.

205.

206.

207.

208.

209.

210.

211.

212.

213.

214.

Frame, S. and A. Balmain, Integration of positive and negative growth signals
during ras pathway activation in vivo. Curr Opin Genet Dev, 2000. 10(1): p. 106-
13.

J oneson, T. and D. Bar-Sagi, Ras eflectors and their role in mitogenesis and
oncogenesis. J Mol Med, 1997. 75(8): p. 587-93.

Mayo, M.W., et al., Requirement of NF -kappaB activation to suppress p53-
independent apoptosis induced by oncogenic Ras. Science, 1997. 278(5344): p.
l 8 12-5.

Sotiriou, C., et al., Breast cancer classification and prognosis based on gene
expression profiles fiom a population-based study. Proc Natl Acad Sci U S A,
2003. 100(18): p. 10393-8.

Brennan, D.J., eta1., Application of DNA microarray technology in determining
breast cancer prognosis and therapeutic response. Expert Opin Biol Ther, 2005.
5(8): p. 1069-83.

Kaklamani, V.G. and W.J. Gradishar, Gene expression in breast cancer. Curr
Treat Options Oncol, 2006. 7(2): p. 123-8.

Mullan, PB. and RC. Millikan, Molecular subtyping of breast cancer:

opportunities for new therapeutic approaches. Cell Mol Life Sci, 2007. 64(24): p.
3219-32.

Perou, C.M., et al., Molecular portraits of human breast tumours. Nature, 2000.
406(6797): p. 747-52.

Gebauer, G., et al., Identification of genes associated with tumor progression
using microarrays. Anticancer Res, 2005. 25(3A): p. 1477-82.

Yi, M., et al., WholePathwayScope: a comprehensive pathway-based analysis tool
for high-throughput data. BMC Bioinformatics, 2006. 7: p. 30.

Xiong, Y., eta1., p21 is a universal inhibitor of cyclin kinases. Nature, 1993.
366(6456): p. 701-4.

Pellegrini, L., et al., Insights into DNA recombination fi'om the structure of a
RAD5I-BRCA2 complex. Nature, 2002. 420(6913): p. 287-93.

Cong, F. and SF. Goff, c-A bl-induced apoptosis, but not cell cycle arrest,
requires mitogen-activated protein kinase kinase 6 activation. Proc Natl Acad Sci
U S A, 1999. 96(24): p. 13819-24.

172

215.

216.

217.

218.

219.

220.

221.

222.

223.

224.

225.

226.

227.

Hogan, B., F. Costantini, and E. Lacy, Manipulating the Mouse Embryo. A
Laboratory Manual. 1986, Cold Spring Harbor: Cold Spring Harbor Laboratory.
332.

Gunther, E.J., et al., A novel doxycycline-inducible system for the transgenic
analysis of mammary gland biology. F aseb J, 2002. 16(3): p. 283-92.

van der Lugt, N.M., et al., Posterior transformation, neurological abnormalities,
and severe hematopoietic defects in mice with a targeted deletion of the bmi-I
proto-oncogene. Genes Dev, 1994. 8(7): p. 757-69.

Orrnandy, C.J., et al., Investigation of the transcriptional changes underlying
fitnctional defects in the mammary glands of prolactin receptor knockout mice.
Recent Prog Horm Res, 2003. 58: p. 297-323.

Baik, 1., et al., Association of fetal hormone levels with stem cell potential:
evidence for early life roots of human cancer. Cancer Res, 2005. 65(1): p. 358-63.

Thomsen, A.R., et al., Estrogenic effect of soy isofiavones on mammary gland
morphogenesis and gene expression profile. Toxicol Sci, 2006. 93(2): p. 357-68.

Trichopoulos, D., Hypothesis: does breast cancer originate in utero? Lancet,
1990. 335(8695): p. 939-40.

Petersen, O.W., et al., Epithelial progenitor cell lines as models of normal breast
morphogenesis and neoplasia. Cell Prolif, 2003. 36 Suppl 1: p. 33—44.

Hu, C., et al., Overexpression of activated murine Notchl and Notch3 in

transgenic mice blocks mammary gland development and induces mammary
tumors. Am J Pathol, 2006. 168(3): p. 973-90.

Mori, T., H. Nagasawa, and HA. Bern, Long-term effects of perinatal exposure to
hormones on normal and neoplastic mammary growth in rodents: a review. J
Environ Pathol Toxicol, 1979. 3(1-2): p. 191-205.

Chlebowski, R.T., et al., Influence of estrogen plus progestin on breast cancer

and mammography in healthy postmenopausal women: the Women 's Health
Initiative Randomized Trial. J ama, 2003. 289(24): p. 3243-53.

Bilimoria, M.M. and M. Morrow, The woman at increased risk for breast cancer:
evaluation and management strategies. CA Cancer J Clin, 1995. 45(5): p. 263-78.

Singletary, K.W. and J .A. Milner, DNA binding and adduct formation of 7, 12-

dimethylbenz[ a] -anthracene by rat mammary epithelial cell aggregates in vitro.
Carcinogenesis, 1986. 7(1): p. 95-8.

173

228. Cao, Y., et al., Eflect of 7, I2-dimethylbenz[a]anthracene (DWA) on gut
glutathione metabolism. J Surg Res, 2001. 100(1): p. 135-40.

229. Yin, Y., et al., Characterization of medroxyprogesterone and DWA -induced
multilineage mammary tumors by gene expression profiling. Mol Carcinog, 2005.
44(1): p. 42-50.

174

LIBRARIES

In

S
R
E
V
M
U

MCH‘GAN sms

null 1111;

1111293

1|