Cholecalciferol Protects Against Deoxycholic Acid- Induced Loss of Ephb2 in Human Colorectal Cancer Cells

Cholecalciferol Protects Against Deoxycholic Acid- Induced Loss of EphB2 in Human Colorectal Cancer Cells

Item Type text; Electronic Thesis

Authors Comer, Shawna Beth

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 08/10/2021 19:10:52

Link to Item http://hdl.handle.net/10150/193312 CHOLECALCIFEROL PROTECTS AGAINST DEOXYCHOLIC ACID -INDUCED LOSS OF EPHB2 IN HUMAN COLORECTAL CANCER CELLS

Shawna Beth Comer

A Thesis Submitted to the Faculty of the

DEPARTMENT OF NUTRITIONAL SCIENCES

In Partial Fulfillment of the Requirement s For the Degree of

MASTER OF SCIENCE

In the Graduate College

THE UNIVERSITY OF ARIZONA

2007 2

STATEMENT BY AUTHOR

This thesis has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is depo sited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission fo r extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scho larship. In all other instances, however, permission must be obtained from the author.

SIGNED: ______Shawna Comer______

APPROVAL BY THESIS DIRECTOR

This thesis has been approved on the date shown below:

______May 3, 2007______Emmanuelle J. Meuillet Date Assistant Profe ssor of Nutritional Sciences 3

ACKNOWLEDGMENTS

I would like to thank my committee members for their guidan ce and incomparable research expertise: Dr. Ronald L. Heimark, Dr. William S. Garver, and advisor Dr. Emmanuelle J. Meuillet.

I would like to thank the following members of the lab for their patience, technical support, and, above all, perpetual encourage ment: Sherif Morgan, Dr. Samira Jean -Louis, Lee Wisner, Margareta Berggren, Song Zuohe, Ben George, and Sylvestor Moses.

Finally, I would like to thank my family, friends, loved ones, and role models for keeping me sane and laughing. I n particular , I wou ld like to thank my boyfriend Tyler Matthews , parents Craig and Lori Comer , sisters Shelby and Shannon Comer and niece Kadyn , grandmother Marlene Bisek , and friends Jamie Critelli , Caitlin Gade , Janae Renteria, Joelle Biggers , and Erin Mendoza . My greates t wish is that someday you will find someone to believe in you the w ay you all have believed in me. 4

DEDICATION

I would like to dedicate this manuscript to my grandmother, Dolores Hall, who succumbed to lung cancer in July of 2004. Until we meet again, “ home again, home again, jiggety -jog” 5

TABLE OF CONTENTS

LIST OF FIGURES ...... 6 ABSTRACT ...... 8 INTRODUCTION ...... 9 I. Summary of Problem ...... 9 II. Contribution to Project D evelopment ...... 10 III. Literature R eview ...... 11 A. Backgrou nd on colon cancer ...... 11 B. Colon tumorigenesis ...... 1 3 C. Tumor promotion by secondary bile acids ...... 1 8 D. Regulation of cell -cell adhesion ...... 2 2 E. Cell migratory regulation by Eph receptors and ephrins ...... 30 F. Therapeutic potential of vitamin D ...... 37 PRESENT STUDY ...... 45 REFERENCES ...... 63 APPENDIX A : DEOXYCHOLIC ACID DOWNREGULAT ES EPHB2 EXPRESSION IN HUMAN COLORECTAL CANCER CELLS: PROTECTIVE ROLE OF CHOLECALCIFEROL ...... 6 9 I. Abstract ...... 70 II. Introduction ...... 71 III. Materials and Methods ...... 74 IV. Results ...... 78 V. Discussion ...... 83 VI. References ...... 91 VII. Figures ...... 95 6

LIST OF FIGURES

Supplemental Figure 1. Accumulat ion of genetic changes i n colorectal cancer progression ...... 17 Supplemental Figure 2. Structure of cholic acid (A) and deoxycholic acid (B) ...... 21 Supplemental Figure 3. β-catenin regulation and signaling in the absence of Wnt signaling ...... 28 Supplemental Figure 4. β-catenin regulation and signaling with active cano nical Wnt signaling ...... 29 Supplemental Figure 5. Silencing of EphB2 expression by transformed colorectal cancer cells ...... 36 Supplemental Figure 6. Structure of cholecalciferol ...... 44 Supplemental Figure 7. Summary of cell line variations in genoty pe and response to treatments ...... 60 Figure 1. Morphology of colorectal cancer cells is altered in response to deoxycholic acid and cholecalciferol ...... 48 Figure 2 . Wound migration is impaired with cholecalciferol pre -treatment as compared to DCA ...... 49 Figure 3. DCA -induced loss of Ephb2 expression is abrogated with cholecalciferol pre - treatment ...... 50 Figure 4 . Cholecalciferol protects against DCA -mediated loss of EphB2 protein expression in a dose -dependent manner ...... 51 Figure 5A . Cholecalciferol protects against DCA -induced loss of EphB2 RNA expre ssion in a dose -dependent manner ...... 53 Figure 5B. Cholecalciferol protects against DCA -induced loss of EphB2 RNA expression in a dose -dependent manner ...... 5 4 Figure 6 . DCA decreases expression of E -cadherin in a time -dependent manner in HT29 colorectal cancer cells ...... 5 5 Figure 7 . DCA treatment decreases expression of E -cadherin with subsequent induction of β-catenin in a time -dependent manner ...... 5 6 7

LIST OF FIGURES – Continued

Figure 8. DCA increases expression of β-cateni n in a dose -dependent manner ...... 5 8 Figure 9. Cholecalciferol protects against DCA -induced disrupti on of lipid rafts...... 59 8

ABSTRACT

Research has identified a linear relationship between saturated fat intake and colon cancer, and has de monstrated that high fat diets enhance tumorigenesis through elevation of secondary bile acids such as deoxycholic acid (DCA) . We and others have shown that DCA can manipulate cell adhesion by decreasing expression of E -cadherin and increasing expression o f β-catenin. We have also shown that DCA significantly reduces EphB2 expression, which regulates cell positioning and segregation . Importantly , vitamin D can reinstate membranous E-cadherin/ β-caten in interactions and increase E -cadherin expression . In the present study, we sought to analyze the effects of DCA and vitamin D (cholecalciferol) treatment on EphB2 in colorectal cancer cells . Pre -treatment with cholecalciferol restored EphB2 expression in a dose -dependent manner , even with combined DCA treatment . This observation may be EGFR -dependent , suggest ing that cholecalciferol may antagonize the effects of DCA. Taken together, these results suggest that cholecalciferol may represent an adjuvant therapy for colorectal cancer patients . 9

INTRODUCTION

I. Su mmary of Problem

Colorectal cancer is a growing health issue, particularly in developed countries. As part of the digestive tract, the colon is considerably affected by individual components of digestion and nutrient status. In particular, bile acids, suc h as deoxycholic acid (DCA), which aid in lipid emulsification, have been shown to have a potent tumor -promoting effect on colon tissue. Interestingly, similar biological properties appear to be influenced by DCA and by vitamin D . It stands to reason, then , that these two compounds, both cholesterol derivatives, may influence protein expression and cell signaling. Of particular importance are the molecular mechanisms, including cell -cell adhesion and cell compartmentalization, affected by DCA and vitamin D. Though several anti -tumor effects of vitamin D have already been established at the transcriptional level, our research indicates an additional role for vitamin D in manipulation of membrane dynamics, particularly as a protective agent against DCA -induced changes in cell adhesion and segregation . Ultimately, these results may further emphasize the need for vitamin D supplementation as an adjuvant therapy for individuals with advanced colorectal carcinoma, even in the absence of nuclear vitamin D receptor. 10

II. C ontribution to Project Development

I, Shawna Comer, as a graduate student in the Department of Nutritional Sciences at the University of Arizona, assert that I performed all experiments included in this thesis, including all cell culture, immunoflu orescence, wound migration assays, clonogenic assays, and Western blot analysis. The exception to this statement is data obtained from RT -PCR techniques: while I prepared the samples for the experiments, including culture and subsequent addition of drugs, Zuohe Song performed the extraction and amplification of RNA, and ran the RNA through electrophoresis for detection. I conducted the background literature search for this project, developed the project under the direction of Dr. Emmanuelle Meuillet, as wel l as authored the appended paper for intended submission to Carcinogenesis in May 2007, with revision and guidance from Dr. Meuillet, the primary investigator and mentor for this study. 11

III. Literature Review

A. Background on colon cancer :

According to the American Cancer Society (ACS) [1] , c olorectal cancer is the third most common type of cancer, as well as the second leading cause of cancer death s in the United States, second only to lung cancer for both men and women. Based on statistics from the Colon Cancer Alliance (CCA) and the ACS , one out of every eighteen people in the Unit ed States will develop color ectal cancer in their lifetime , and 55,170 people died of this disease in 2006 [1] . Typically , this disease is discovered in its early stages as polyps via colonoscopy. However, in the absence of early detection , polyps may progress into adenoma then carcinoma in situ (CIS). Only when the tumor has developed int o a large enough mass will the patient notice the symptoms associated with advanced colon cancer, such as bloody stool and abdominal cramping.

When diagnosed early, colorectal cancer is readily treatable through a combination of therapies including surge ry, chemotherapy, radiation, and adjuvant therapy. Despite the availability of the above listed treatments , according to the CCA, only 37% of colorectal cancers are detected at the local stage, when the cancer has n ot spread to surrounding tissue. Most can cers, however , are identified after the cancer has spread to regional tissue, and approximately 20% of cancers are identified after the disease has spread to distant organs. Normally, t he five -year survival rate decreases significantly as the cancer progre sses, from 90% at the local stage, to 66% at the regional stage, to 8.5% five -year survival rate when cancer ha s progressed to distant organs [1] .

While yearly fecal occult blood tests, flexible sigmoidoscopies every five years, and ten - year colonoscopies are recommended for healthy individuals over the age of 50 , this recommendation oft en does not take into consideration potential risk factors for developing colorectal cancer. In order to recognize high -risk individuals , epidemiological 12

data have identified several markers for colorectal cancer , including genetics and lifestyle [1] . According to the ACS and CCA, lifestyle risk factors include low physical activity, overw eight /obese , smoking, and high alcohol consumption. Racial correlations also exist: African American or Ashkenazi Jewish descent is associated with increased risk of colorectal cancer. Finally, individuals over the age of 50 are at a higher risk, as are individuals with inflammatory bowel disease . Several genetic mutations have also been identified to predispose individuals to develop colorectal cancer (discussed in the following section ). In addition, a s the large intestine is a crucial component of the digestive tract, several dietary components have also been identified in relation to colorectal cancer risk. Specifically, t he ACS has identified h igh consumption of red meat and processed foods, as well as low consumption of fruits, vegetables, and whole g rains with incre ased risk of colorectal cancer . In addition , more recent epidemiological data obtained by the ACS indicate a correlation between consumption of a diet high in saturated fat and increased risk of developing colorectal cancer. 13

B. Colon tum origenesis :

B.1 . Normal colon epithelium morphology and characteristics:

Normal colonic epithelium is characterized by a thin layer of epithelial cells facing the lumen of the large intestine. Furthermore, these cells are arranged in crypts roughly 50 ce lls deep that function to increase surface area for water resorption [2] . Proliferation of undifferentiated stem cells occurs in the lower portion of the crypt, and these immature colonocytes migrate upwards towards the top of the crypt [2] . This migration is characterized by differentiation of cells as well as loss of the ability to divide [2] . Eventually, differentiated cells reach the lumen, undergo death via apoptosis, and are shed in fecal matter. Generally, renewal of surface epithelium occurs constantly, and is replaced in its entirety once every six days [2] . However, in colorectal cancer, normal differentiat ion and turnover of cells is impaired. Instead of undergoing differentiation and apoptosis, cells retain a stem cell -like phenotype and continue to divide [2] , leading to formation of a benign lesion or polyp. Further accumulation of additional mutations confers growth advantages to these cells, ultimately aiding in cancer progression and metasta sis [3] .

B.2. Cancerous colon epithelium:

In 1990, Fearon and Vogelstein proposed a model for development of most colorectal tumors, comprised of four components [3] . First, they suggested that mutational activation of o ncogenes and suppression of tumor suppressor gen es leads to formation of tumors. Second, for a tumor to achieve malignancy , four to five gen etic mutations are typically needed. Third, the total accumulation of changes, and not the order in which those chan ges are acquired, is responsible for the pr operties exhibited by the tumor. Finally, Fearon and Vogelstein suggested that tumor suppressor genes may not be recessive at the cellular level, since effects are seen even with heterozygous expression of 14

these genes [3] . Over a decade later, this theory still dominates the aren a of colorectal carcinogenesis, though more recent research has made important advances in delineation of the exact mutations involved in development of this disease. ( See Supplemental Figure 1, page 17)

Colorectal cancer can be divided into three categories : 1) sporadic (60 -70% of cases); 2) familial (20 -30% of cases); and 3) hereditary (<10% of cases). Contributing to familial and hereditary colorectal cancer, several dis eases have also been demonstrated as predisposing to colorectal cancer, including familial adenomatous polyposis (FAP), hereditary non -polyposis colorectal cancer (HNPCC), familial ulcerative colitis, and Crohn’s disease [2] , among others. However, most cases of colorectal cancer are sporadic, and arise from a combination of spontaneous gene mut ations.

Normal colon tissue is polyclonal, having arisen from several s tem cells of different lineages. However, ne oplastic tissue is typically monoclonal, as it usually arises from one mutated cell or cluster of mutated cells due to clonal expansion [3] . A cell can undergo nearly infinite mutations in its lifetime; however, several mutations have been identified that frequently occur in colorectal cancer. Furthermore, it has been suggested that each mutation confers an additional growth ad vantage to the transformed cells, leading ultimately to cancer progression and metastasis [2] . Though mutational events including deletions, substitutions, and insertions occur on a small scale, these may result in dramatic alterations in both protein expression and cell phenotype. Furthermore, these mutations may result in silencing or amplifica tion of specific proteins [2] , which can be termed tumor suppressors or oncogenes , depending on effects seen with mutation of these genes . Fearon and Vogelstein demonstrated that accumulation of mutations occurs prog ressively in colorectal cancer. Furthermore, they showed that mutations in at least one tumor suppressor gene and one oncogene were typically required for development of colorectal cancer [4] . 15

B.3. Oncogene activation:

Another common mutation involves the small GTPase ras. Ras mutations are seen in at least half of colorectal carcinomas, as well as in the majority of adenomas greater than 1 cm in size [3] . However, ras mutation is usually only found in one cell per early tumor, indicating that it is not responsible for ini tiating neoplastic growth, but rather may be responsible for transformation of a small adenoma into a large malignant carcinoma [3] . Mutations in the K -ras oncogene in colorectal cancer reduce GTPase activity and ability to interact with GT Pase activating proteins (GAPs) and lead to an accumulation of activated ras protein in tumor cells [5] . Furthermore, one study identified mutated K -ras in 58 -73% of aberrant crypt f oci (ACF) [5] , early neoplastic lesions in colorectal cancer

B.4. Inactivation of t umor suppressor genes:

Inactivation of tumor suppressor genes represents a critical step in color ectal cancer development. One gene commonly mutated in colorectal cancer is the tumor suppressor gene, p53, which is responsible for check -point regulation of the cell cycle. In nearly 75% of colorectal ca rcinomas, a large portion of the p17 gene, which co ntains p53, is lost due to chromosome loss or mitotic recombination [3] . Interestingly, this is seen in very few adenomas, suggesting that loss of p53 may be responsible for acquisition of malignancy in cancer cells [3] . Cou pled with loss of one allele, several colorectal carcinomas also display amino acid substitution mutations that result in inactivation of the other p53 allele, leading to loss of the wild -type function [3] . However, even if one allele of p5 3 remains wild -type , research has shown that the mutated p53 allele may interfere with the function of the wild -type allele, prohibiting its binding to normal cellular constituents and thus decreasing its function as a tumor suppressor gene [3] .

Anoth er common mutation involves chromosome 18q, which codes for tumor suppressor genes DCC (deleted in colorectal cancer) gene, SMAD2, and SMAD4/ DPC4 [5] . 16

Deletions in th is chromosome have been suggested to be present in up to 70% of colorectal carcinomas [5] . Mutations in DCC , which encodes a protein similar in function to cell adhesion proteins, re sult in a two -fold decrease in membranous adhesion protein expression, which is associated with a 30 -fold reduction in cell adhesion [3] . Loss of heterozygosity and decreased expression of DCC in neoplastic tissue are seen in colorectal can cer [3] , and restoration of DCC expression suppresses tumor formation [5] . Furthermore, recent data has shown that DCC can exert pro -apoptotic effects via its role as a caspase substrate [5] . Specifically, if the binding site for caspase -3 is mutated, the pro -apoptotic effects of DCC are completely abrogated [5] , providing strong support for the role of DCC as a tumor suppressor. SMAD (small Mad -related protein) -2 and SMAD4 are key components of transforming growth factor (TGF) -β signaling, and, when activated, induce cell cycle arrest [5] .

While several gene mutations confer growth advantages to transformed cells, mutation of the adenomatous polyposis co li (APC) gene has been suggested to be particularly important to development of colorectal cancer, as this gene is considered to be the ‘gatekeeper’ of cell proliferation within the colon [2] . In support of this theory, the APC gene has been shown to be mutated in up to 85% of non -familial cases of colorectal cancer [2,6] . Mutation of this gene t ypically produces inactive truncated proteins that are unable to perform normal cellular duties, which include cell cycle block, regulation of β- catenin signaling as a component of the Wnt signaling cascade, and regulation of cell adhesion and cell migrati on [2] , likely due to APC interaction with components of the cytoskeleton [6] . Mutations to the APC gene frequently occur in the mutation cluster region (MCR), which contains a 20 -amino acid sequence respo nsible for β-catenin binding [6] . Regulation of β-catenin will be discussed later in this review. Importantly, lo ss of control of cell adhesion and migration represents a key step in cancer progression [6] . 17

Chromosome 5q 12p 18q 17p Alteration: Mutation/ Loss Mutation Loss Loss Gene: APC K-ras DCC p53

Normal Hyperproliferative Early Intermediate Late Carcinoma Metastasis epithelium epithelium adenoma adenoma adenoma

CANCER PROGRESSI ON

Supplemental Figure 1. Accumulation of genetic changes in colorectal cancer progression. The progression from normal epithelium to carcinoma requires a series of mutations, which usually occur in the order depicted: Mutation or loss of APC oft en results in uncontrolled proliferation of cells, leading to preneoplastic tissue. Next, mutation of K -ras activates this oncogene, and contributes to progression of early adenoma to intermediate adenoma. Next, DCC (deleted in colorectal cancer) is mutate d and results in silencing of this tumor -suppressive gene, resulting in progression to late adenoma. Finally, p53, a powerful tumor -suppressive gene, is mutated and often silenced, resulting in unchecked cancer progression from adenoma to carcinoma, then u ltimately to distal metastasis . (Adapted from: Fearon, E.R. and Vogelstein, B. (1990) “A genetic model for colorectal tumorigenesis.” Cell , 61 , 759 -767. ) 18

C. Tumor promotion by secondary bile acids :

High intake of saturated fat is correlated with increas ed production and secretion of bile acids (reviewed in [7] ). These amphipathic cholesterol derivatives are crucia l to digestion, aiding in the emulsification, digestion and absorption of lipids [8,9] . Though most bile acids are reabsorbed, a small portion, roughly 5% per cycle, escapes the small intes tine and passes into the colon [7] . Colonic microflora metabolize a small portion of primary bile acids, cholic acid and chenodeoxycholic acid, into secondary bile acids, the most intensively studied of which is deoxycholic acid (DCA - a tumor promoter ). Due to its amphipathic nature, DCA has been found to partially absorb in to the colon. (See Supple mental Figure 2, page 20 ) One study found that individuals witho ut colorectal cancer, serum DCA concentration is roughly 0.32 uM, while patients with colorectal adenom a had nearly a 3 -fold increase, up to 0.89 uM , in serum concentration of DCA as compared to controls without colorectal cancer [10] . Additionally, animal studies have demonstrated that DCA can cause tissue damage. Because a subsequent increase i n cell proliferation is seen immediately following DCA -mediated tissue damage, it is suggested that this response may be the basis for the tumor -promoting capacity of DCA (reviewed in [7] ).

In addition to stimulating cell proliferation, DCA has been demonstrated to induce several tumor -promoting effects at the molecular level, including manipulation of membrane com position and decreased membrane fluidity [11] , induction of oxidative stress and mitochondrial c ytochrome c release [9] , activation of protein kinase C (PKC), induction of cyclooxygenase -2 (COX -2) promoter activity, and stimulation of activator protein -1 (AP -1) and nuclear factor kB (NF -kB) expression [12] . Interestingly, DCA seems to exert two op posite effects on cells. W hile DCA can stimulate cell proliferation at low concentrations [14], it can also induce both caspase -dependent and caspase - independent apoptosis at high concentrations [7,9] . Though the apoptotic effect may be largely due to cellular cytotoxicity, studies have indicated that these seemingly 19

contradictory effects may select for a population of cells th at is both resistant to apoptosis and rapidly proliferative. In support of the aforementioned tumorigenic characteristics, animal studies have also confirmed the tumor -promoting potential of DCA on initiated rodents [7,13] . For example, studies performed by Flynn et al. demonstrated that mice initiated with azoxymethane developed more aberrant crypt foci, which are preneoplastic lesions, when fed a diet containing 0.2 5% DCA as compa red to mice fed a control diet [14] .

In addition to inducing apoptosis in cells, DCA has also been shown to influence adhesion, invasion, and migration in colorectal cancer cells, as evidenc ed by the correlation between Dukes’ stage of colon tumors and abnormal fecal concentrations of DCA [15] . Colorectal cancer cell lines HCT8/E11, LoVo, and SW480 (among others), when treated with DCA exhibited a dose -dependent stimulation of invasion into collagen and increased clonogenicity [15,16] , as well as increased expression of cyclin D1, uPA, and uPAR, all of which are associated with increased tumorigenicity [16] . In addition, DCA -induced invasion was associated with activation of Rac1 and RhoA GTPases [16] .

In regards to cell -cell adhesion protein interactions, DCA has been shown to have a particularly potent effect . In addition to decreasing total protein expression of E -cadherin, DCA can decrease the binding of β-catenin to E -cadherin while increasing β-catenin tyrosine phosphorylation and subsequent nuclear localization [16] . One possible mechanism for this may be through the activation of EGFR and Src due to the perturbation of the plasma membrane by DCA [11] . Specifically, DCA treatment was shown to decrease fluidity of the plasma membrane by increasing cholesterol content, which resulted in activation and downstream signal transduction of EGFR [11] . As a result, β-catenin dissociates from the membrane and subsequent ly translocates to the nucleus, which can lead to i ncreased invasion, proliferation, and transcription of β- catenin/TCF target genes. Generally, β-catenin membrane localization and binding to E - 20

cadherin is associated with decreased invasion, increased differentiation, and inhibition of β-catenin signaling [16 -18] . 21

Supplemental Figure 2. Structure of cholic acid (A) and deoxycholic acid (B). (Adapted fr om Chemistry WikiProject, Wikipedia Online Encyclopedia, en.wikipedia.org) 22

D. Regulation of cell -cell adhesion:

D. 1. Epithelial -to -mesenchymal transition:

The epithelial -to -mesenchymal transition (EMT) is characterized by the following properties: loss of an epithelial phenotype, down -regulation of cell -cell adhesive proteins, loss of polarity, increased invasive capacity, and gain of mesenchymal characteristics [19] . Though normally a crucial component of processes such as embryonic development and tissue remodeling, these mesenchymal characteristics of the EMT are often associated with abnormal growth and migration such as that seen in cancer progression [19] . For example, once a cell d etaches from the extracellular matrix (ECM), it normally undergoes programmed cell death, termed anoikis [20] . Only small populations of cells that detach from the ECM survive, and even smaller populations of these cells can metastasize. This su rvival may be due in part to cells undergoing the EMT , as well as anti -apoptotic mutations in genes such as Bak, Bax, focal adhesion kinase (FAK), and integrin -linked kinase (ILK) [20,21] . Thus, l oss of cell -cell and cell -ECM connections favor invasion through the b asement membrane and metastasis to different areas of the body [8,19] .

In the case of cancer, EMT confers an advantage to transformed cells by allowing them to acquire a more aggressive phenoty pe . For example , mem brane -bound adherens junctions (AJ) (favoring an epithelial phenotype) are composed primarily of the epithelial cadherin (E-cadherin )/β-catenin /α-catenin complex , and seem to play a key role in tumor suppression by maintaining organization and adhesion within the epithelium [15] . Individual AJ protein expression is altered in transformed cells, possibly due to a switch from an epithelial differentiated phenotype to a mesenchymal phenotype [15] . Since i n poorly differentiat ed and highly invasive cells, decreas ed expression of E -cadherin and increased expression of β-catenin is observed, these proteins are now implicated either as 23

a tumor -suppressor in the case of E -cadherin or tumor -promoter in the case of β-catenin [22] .

D.2. E -cadherin:

E-cadherin is a transmembrane protein directly i nvolved in cell adhesion ; in fact, functional inactivation of this protein is considered a critical step in acquisition of EMT [19] . E -cadherin protein s of adjacent cells can dimerize at the plasma membrane , facilitating formation of an adherens junction between the two cells. In normal colon tissue, E -cadherin stains strongly at the membrane in a ‘honeycomb’ pattern [15] . However, in colorectal cancer, expression of E -cadherin is significantly reduced and even lost complete ly in some advanced carcinomas [15,22] . The observed decrease of E - cadherin may occur by several different mechanisms, including 1) silencing of the promoter by hypermethylation , or 2) trans criptional down -regulation by the transcription factor Snail or Snail2 [15,19,23] . Snail has been shown to rep ress E-cadherin expression by binding to E -boxes in the CDH1 gene , which encodes for E -cadherin and was among the first to be considered an invasion -suppressing gene [15,19,23] . Though Snail is typically absent from normal colon tissue , it is found to be upregulated in colorectal cancer , particularly in metastatic colon cancer [15] . The CDH1 gene has also been demonstrated to undergo loss -of -function mutations in specific cancers, such as lobular breast cancer s and endometrial cancer s [23] . Immunohistochemistry performed on cytokeratin -stained colorectal tumor tissue slides revealed decreased E -cadherin staining at the invasive, de -differentiated edge of tumor s; however, the central, highly differentiated region of tumor s tend to exhibit prominent E -cadherin staining [22] , providing support for the theory that loss of E -cadherin is associated with increased invasion and metastasis [23] . Interestingly, the present study suggests a novel mechanism for downregulation of E -cadherin, involving DCA -mediated modulation of the plasma membrane. Though the precise mechanism of downregulation remains unclear , low 24

expression of E -cadherin is now associated with poor prognosis, increased invasion and metastasis, and decr eased differentiation of cells [1 9,24] .

D.3. β-catenin:

The other essential component of cell -cell adhesion is β-catenin , which is found in three forms in the cell: membrano us, cytosolic, and nuclear [15,24] . At the membrane, β- catenin is bound to E -cadherin and α-catenin in a phosphorylation -dependent manner . Tyrosine (Y) phosphorylation of β-catenin is accomplished by several kinases, including tyrosine phosphorylation site 654 (Y654 ) by Src and epidermal growth factor recept or (EGFR) , that of Y489 by Abl, and that of Y142 by Fyn, Fer, and cMet [25] . In general, all sites of tyrosine phosphorylation lead to β-catenin dissociation from the adherens complex [26,27] . Its d issociation from E -cadherin results in cyt osolic translocation and in loss of suppression of β-catenin signaling [24] . Once in its soluble form , β-catenin is either degraded or stabilized for nuclear translocation, which depends on several factors, including the prese nce of Wnt signaling, mutations in the components of the destruction complex, which target β-catenin for ubiquitination and subsequen t degradation , or stabilizing mutations in β-catenin itself, which favor stabilization and nuclear translocation [26,27] .

The destruction complex for β-catenin is composed of glycogen synthase kinase (GSK )−3β, Axin, and adenomatous polyposis coli (APC) protein [28] . In the absence of Wnt signaling, GSK3 β phosphorylates β-catenin on serine 33/37 and threonine 41 , marking it for recognition and ubiquitination by the F -box protein β-TrCP and subsequen t degradation in the proteasome [23,27,28] . APC and Axin can also enhance the phosphorylation of β-catenin by GSK3 β, favoring its degradation in the proteasome . Though an alternative pathway independent of GSK3 β, involving Siah -1 and APC, has been suggested, the destruction complex composed of GSK3 β, Axin, and APC is 25

generally recognized as the key re gulator of β-catenin levels [28] . In colo rectal cancer, however, several compo nents of this destruction complex are compromised , such as mutations in APC and Axin that interfere with β-catenin binding and subsequent phosphorylation [28] . ( See Supplemental Figur e 3, page 27 )

D.4. Mutations affecting the destruction complex:

The destruction complex functions to maintain low levels of cytoplasmic β-catenin within the cell. Within this complex, APC plays a key role in targeting β-catenin for degradation. T he effect of APC on β-catenin phosphorylation is dependent on binding to β-catenin via two specific regions. In fact, m utations in the armadillo r epeats of APC are among the earliest geneti c mutations that occur in colorectal cancer , and are found in 80% of sporadic colorectal cancers [23,27] . These mutations consequently prevent binding of APC to β-catenin, thereby preventing ubiquitinati on and proteasomal degradation of β-catenin [27,28] . Interesting ly, APC also plays a critical function in nuclear export of β-catenin , and truncated APC protein produced in response to mutations has been shown to lack several nuclear export sequences [29] , which further highl ights the tumor -promoting potential of APC mutations. In addition, APC is a component of the Wnt signaling pathway, mutations in which are seen in nearly all colorectal cancers [30,31] . During normal Wnt signaling , the Wnt prot ein is an extracellular secreted glycoprotein that binds to Frizzled (Frz) , a transmembrane receptor, which recruits Disheveled (Dvl), an intracellular phosphoprotein, to the membrane for activation [27,30,32] . Dvl activation inhibits GSK3 β-mediated phosphorylation of β- catenin, thus preventing β-catenin from degradation [27,28] . Though the precise mechanism of Dvl -mediated inhibition of GSK3 β is unknown, recent research suggests that FRAT1/GBP may play a role, as this protein competes with Axin for binding to GSK3 β, possibly preventing association of GSK3 β with Axin and APC, as well as access to β-catenin, which is required for kinase activit y of GSK3 β [reviewed in [33] . 26

Constitutive activation of Wnt signaling is seen in the majo rity of colorectal cancers, and through increased nuclear accumulation of β-catenin, is associated with poor prognosis and increased invasion and metastasis [27,28] . (See Supplement al Figure 3, page 28 )

Axin functions, in combination with APC, to enhance the phosphorylation of β-catenin by GSK3 β, favoring proteasomal degradation [28] . However, in colorec tal cancer, Axin is frequently mutated: nonsense mutations have been shown to produce truncated proteins that are unable to bind β-catenin [28] . Furthermore, these mutations were associated with increased nu clear localization of β-catenin and increased transcriptional activity of β- catenin/TCF [28] . In add ition to nonfunctional proteins created during mutational events, Axin mutants have been identified that interfere with β-catenin degradation. For example, a leucine -to -methionine substitution on residue 396 of Axin produces a protein that interferes with binding of GSK3 β to β-catenin, prohibiting ubiquitin -mediated proteasomal degradation of β-catenin [28,30] .

In addition to alterations in Wnt signaling , β-catenin itself is mutated in about 10% of sporadic colon cancer cases, and these mutations are associated with increased stability of the protein and ac tivation of Wnt signaling [27,30] .

D.5. β-catenin signaling in colorectal cancer:

Upon activation of Wnt signaling, such as that seen in the majority of sporadic colorectal cancer s, cytoplasmic β-catenin is stabilized and translocates to the nucleus where it associates with a member of the T -ce ll factor (TCF)/ lymphoid enhancement factor (LEF) families, of which the most relevant to colorectal cancer are TCF -1, TCF -4, and LEF -1 [28] . Upon binding to the promoter region of target genes, the β-catenin/TCF/LEF complex can modulate transcription of several genes by recruiting co -activators such as p300/cAMP -response -element b inding protein (CBP) and SWI/SNF complexes (chromatin -remodeling transcriptional unit) [27] . Examples of t arget genes include those 27

encoding for E-cadherin, c -myc , cyclin D1 , cyclooxygenase (COX) -2, matrix metalloproteinase (MMP) -7/matrilysin , urok inase plasminogen activator (uPA), urokinase plasminogen activator receptor (uPAR) , and ephrin receptor B2 ( EphB2 ) [16,28,32,34] , all of which are important in cell proliferation , adhesion , and segregation . Interestingly, EphB2 , a receptor tyrosine kinase responsible for cell -cell repulsion and, in some cases, cell -cell adhesion, is also a transcriptional target of β-catenin/TCF signaling [34] . EphB2 has been demonstrated to have a tumor -suppressor role in colorectal carcinogenesis, suggesting that not all genes activated by β-catenin/TCF signaling are pro -tumorigenic [35] .

Research has indicated that progenitor c ells in the colonic crypt display nuclear localization of β-catenin, indicating that these cells are exposed to Wnt signaling as part of normal physiological signaling [35,36] . Interestingly, early colorectal cancer lesion s also display nuclear accumulation of β-catenin, suggesting at least two possibilities : first, that the loss of compartmentalization of differentiated and progenitor cells may directly enhance tumorigenesis [35,36] , or second, in the presence of Wnt signaling, cells may remain undifferentiated an d adopt a progenitor phenotype [35] . Further , blocking β- catenin/TCF -mediated transcription in colorectal cancer cell lines induces cell cycle arrest and differentiation, even in the presence of other tumor -promoting alt erations, such as APC mutation s [34] , ultimately resulting in the inhibition of cancer progression . 28

E-cad 1 E-cad Src CELL MEMBRANE β-cat P

2 GSK3 β 3 P 5 β-cat APC Ub Proteasome Axin 4

NUCLEUS E-cadherin 6 TCF Groucho LEF EphB2 HDAC Cyclin D1

Supplemental Figure 3. β-catenin regulation and signaling in the absence of Wnt signaling. 1) In the absence of Wnt signaling, β-catenin is complexed with E -cadherin at the plasma membrane. However, β-catenin contains several phosphorylation sites, and has been shown to be targeted by Src, EGFR, and Abl, among others. 2) Phosphorylation of β-catenin results in dissociation from E -cadherin and the plasma membrane, and β- catenin translocates into the cytoplasm o f the cell. 3) Members of the destruction complex, APC, Axin, and GSK3 β, function to target β-catenin for proteasomal degradation. APC and Axin bind to β-catenin and promote phosphorylation of β-catenin by GSK3 β. 4) GSK3 β-mediated phosphorylation of β-cate nin targets this molecule for polyubiquitination. 5) After ubiquitination, β-catenin is degraded in the proteasome, thereby maintaining low levels of cellular β-catenin in the absence of Wnt signaling. 6) When β-catenin is degraded, TCF/LEF is bound to Gro ucho and histone deacetylase (HDAC), and is transcriptionally inactive. (Adapted from Wong and Pignatelli (2002) “Beta -catenin -- a linchpin in colorectal carcinogenesis?” Am J Pathol , 160 , 389 -401; Polakis (2000) “Wnt signaling and cancer.” Genes Dev , 14 , 1 837 -51; Lustig and Behrens (2003) “The Wnt signaling pathway and its role in tumor development.” J Cancer Res Clin Oncol , 129 , 199 -221.) 29

E-cad Wnt 1 E-cad Frz Src CELL MEMBRANE β-cat P

2 3 Dsh GSK3 β

P β-cat APC Ub Proteasome Axin

NUCLEUS E-cadherin 5 β-cat TCF LEF EphB2

Cyclin D1

Supplemental Figure 4. β-catenin regulation and signaling with active canonical Wnt signaling. 1) β-catenin remains complexed with E -cadherin at the plasma membrane in a de -phosphorylated state. 2) Phosphorylation of β-catenin results in dissociation from E -cadherin and the plas ma membrane, and β-catenin translocates into the cytoplasm of the cell. 3) In the absence of Wnt signaling, the destruction complex would phosphorylate β-catenin, targeting it for proteasomal degradation. However, with active Wnt signaling, extracellular W nt proteins signal to membrane -bound Frizzled (Frz) receptors, which activate intracellular Disheveled (Dsh) proteins. 4) Dsh proteins interfere with GSK3 β-mediated phosphorylation of β-catenin , possibly through competition with Axin for binding to GSK3 β. This functions to stabilize β-catenin and promoting nuclear translocation of β-catenin. 5) Upon entering the nucleus, β-catenin dissociates Groucho and HDAC from TCF and forms a transcriptional complex with TCF, which can affect transcription of several ge nes, including E -cadherin, EphB2, and cyclin D1, among others. Specifically, β-catenin/TCF signaling has been shown to decrease expression of E -cadherin and increase expression of EphB2. (Adapted from Wong and Pignatelli (2002) “Beta -catenin -- a linchpin in colorectal carcinogenesis?” Am J Pathol , 160 , 389 -401; Polakis (2000) “Wnt signaling and cancer.” Genes Dev , 14 , 1837 -51; Lustig and Behrens (2003) “The Wnt signaling pathway and its role in tumor development.” J Cancer Res Clin Oncol , 129 , 199 -221.) 30

E. Cell migratory regulation by Eph receptors and ephrins:

E.1. Introduction to EphRs and ephrins:

In vertebrates, 8 ephrin ligands and 14 Eph receptors (EphRs) have be en identified, making Eph receptors the largest family of receptor tyrosine kinases in the human genome [37] . Two classes of ephrins are known to exist: class A, which is bound to the membrane by a GPI linkage, and class B, which pass es through the plasma membrane and possess es a cytoplasmic domain [38] . Eph receptors are transmembrane proteins that tend to cluster, part icularly at adherens junctions [37] , and are classified according to which class of ephrin ligands they bind; for example, EphA receptors bind ephrinAs, and EphB recep tors bind only ephrinB ligands [38,39] .

Eph receptors and ephrins function primarily in cell -cell repulsion. The ephrin/Eph complex exists primarily as a cyclic heterot etramer with 2:2 stoichiometry [38] , suggesting that clustering of both receptor and ligand is necessary for downstream signal transduction [40] . When an ephrin binds to an Eph receptor of a different cell , activation of the Eph receptor by tyrosine phosphory lation creates docking sites for secondary proteins that bind to the receptor via their Src -homology -2 (SH2) and phospho -tyrosine - binding (PTB) sites, both of which bind ph osphorylated tyrosine residues [37] . Several docking proteins interact with activated Eph receptors, including Fyn, Src, Grb2, and Nck [37] ; based on their role in cell proliferation and maintenance of the actin cytoskeleton, the effects of Eph activation on small GTPases Rac, Rho, and cdc42 may explain several aspects of Eph signaling, particularly that of repulsion [37] . In particular, p120RasGAP can bind to activated Eph receptors via its SH2 domain, which complexes with p190RhoGAP to regulate the activity of RhoGAP, a negati ve inhibitor of the GTPase Rho [37] . Activation of Rho has been demonstrated to induce neu rite retraction, providing a mechanism f or Eph -mediated cell repulsion [41] . In addition, interaction with an Eph receptor can also stimulate tyrosine p hosphoryla tion of ephrin ligands [38] . 31

Ultimately, t his bidirectional signaling le ads to collapse of filopodia and subsequent repul sion o f cells. In fact, t his role in repulsion is of particular importance in embryonic development, where Eph receptors play a role in ‘path -finding’ during development (reviewed i n [42] ). Additionally , migrating growth cones and neural crest cells have also been found to rely on Eph/ ephrin signaling for proper localization to tissues [42] . Besides embryonic development , cell -cell repulsion is also critical for tissue patterning and segre gation and vascular remodeling [40] . Due to this contact -mediated repulsion, Eph receptors and ephrins are rarel y found in the same tissue type, rather exhibiting complementary expression such as that seen in EphB receptor -expressing neuronal avoidance of e phrinB -expressing caudal cells [39] .

In addition to regulation of cell adhesion, Eph receptors have been demonstrated to have several anti -tumorigenic effects, including inhibition of cell proliferation as evidenced by down -regulation of mitogen -activated protein kinase (MAPK) and Abl tyrosine kinase activity [43] .

E.2. EphRs, ephrins, and carcinogenesis :

While the primary function of the EphR/ephrin complex has been delineated in cell -cell repulsion, research has shown that EphRs and ephrins may function in cell -cell adhesion when co -localized , in addition to maintaining correct compartmentalizat ion of cells through repulsion [35,44] . In particular, EphB2 has been demonstrated to phos phorylate Ng -CAM, a member of the L1 fam ily of cell adhesion molecules ( reviewed in [37] ), which is associated with activation of the neural/glia cell adhesion molecule ( Ng -CAM ) signaling cascade [45] . Coating the surface of a tissue culture flask with EphB2 results in adherence of ephrinB -expressing cells; similarly, coating a surface with ephrinB1 results in adher ence of cells expressing E phB2 , further supporting the notion that ephrin/EphR interaction s result in weak cell -cell adhesion [39] . 32

In the colon , EphB2, EphB3, EphB4, and EphA2 have been identified as the main Eph Rs . Interestingly, while EphB3, EphB4, and especially EphB2 have been demonstrated to be down -regulated in advanced colorectal cancer [35,46] , recent studies have demonstrated a significant positive correlation between clinical tumor stage, lymphatic vessel invasion, and liver m etastasis and EphA2 expression [47] .

E.3. EphB2 in colorectal cancer:

As the majority of my research has focused on expression of EphB2 in col orectal cancer, the remainder of this section will focus on this particular receptor . Physiologically, the loss of EphB2 protein expression in colorectal cancer is quite beneficial for transformed cells. Normally, once a cell expressing an EphB receptor encounters a cell expressing a membranous ephrin ligand, the interaction caus es bidirectional signaling, resulting in confinement of the EphB -expressing cell to one area, while the ephrin -expressing cell localizes to a different area [39] . However, if a cell silences EphB2 expression , interaction with its corresponding ligand is consequently impaired , and the cell is free to migrate to other areas. In fact, studies con ducted by Batlle et al. support this theory, where they demonstrated that complementary expression of E phR and ligand restrict cell positioning along the crypt -villus axis in adult mice [44] . In addition, by using a dominant -negative form of the EphB2 receptor that lacke d the intracellular domain, they were able to demonstrate random mispositioning of cells along the axis, with intermingling of differen tiated and proliferative cells [35,44] .

As stated previously, i n a normal colonic villus, undifferentiated proli ferative cells occupy the crypt while the precursor cells migrate up towards the tip of the villus , undergoing cell cycle arrest and differentiation upon reaching the crypt -villus junction , and are shed roughly every six days [35,44] . However, i n EphB2/EphB2 null mice, the proliferative cells of the crypt and the differentiated cells of the upper villus tend to intermingle, demonstrating evasion of normal migrat ory regulations in the colon [44,48] . 33

Additionally, p remalignant lesions in the colon have been found to express high levels of EphB2; however, as the tumor progre sses, EphB2 expression was found to be completely lost [34,44,48] . To explain this loss, Clevers et al . have recently proposed an excellent theory as to the early activation and eventual lo ss of EphB2 protein expression [35] . They suggested that founder tumor cells acquire a mutatio n in APC or other component of the regulatory complex which may lead to constitutive activation of β-catenin/TCF signaling. These founder cells remain undifferentiated and continue to populate the crypt with their descendants, all of which strongly express membranous EphB receptor due to the active β-catenin/TCF signaling. Once these founder -like cells reach the surface epithelium, where other cells express the ephrinB ligands, the receptor and ligand interact, resulting in compartmentalization of the tumor and inhibition of its ability to proliferate. However, tumor cells are conditioned eventually to silence EphB expression. When this occurs, compartmentalization of the tumor is lost and tumor cells resume their rapid proliferation and expansion, leading t o tumor progression. (See Supplemental Figure 5, page 3 5)

Recent research by Guo et al. provides subst antial evidence for this theory. The authors found that EphB2 was expressed in 100% of normal colon crypt cells, 78% of adenomas, 55.4% of primary colore ctal cancers, 37.8% of lymph node metastases and 32.9% of liver metastases [48] . Other studies conducted by Batlle et al. further demonstrated that virtually all adenocarcinomas lost expression of EphB2, with more than 25% of tumors completely lack ing expression of EphB2 [44] . Overall, these studies support Clevers’ theory and demonstrated that decreased EphB2 expression was associated with poor differentiation, increased colony form ation, advanced Dukes’ tumor stage, and poor overall survival as well as disease -free survival [48] .

Additionally, animal studies using genetically engineered mice with the APC min /+ genetic background (leads to development of multiple benign intestinal lesions) crossed mice containing a mutation resulting in very low EphB expression demonstrated accelerated tumor progression in the large intestine as compared to control [35] . Overall , it is now 34

widely accepted that loss of EphB2 expression is correlated with poor prognosis and lower patient survival [46] . In the context of cancer , the inverse relationship between EphB2 and tumor stage makes logical sense: as the tumor progresses from adenoma to carcinom a to metastasis, EphB2 expression and thus control of cell adhesion is progressively lost [3 5] .

E.4. Effect of β-catenin/TCF signaling on EphB2 expression:

As discussed previously, constitutive activation of β-catenin/TCF signaling is often seen in colorectal cancer. The negative effect of β-catenin/TCF signaling on cell -cell adhesion is well characterized, especially noted for decreasing expression of E -cadherin. However, recent studies also suggest that EphB2 expression can also be affected by β-catenin/TCF signaling. A study conducted by Batlle et al. used a dominant -negative (dn) form of TC F to elucidate genes affecte d by this transcription factor [36] . Use of dn -TCF inhibited β- cat enin/TCF signaling, and among the genes affected, EphB2 expression decreased significantly, suggesting that EphB2 is a direct transcriptional target of active β- catenin/TCF signaling [36] . Though a dditional studies further support this observation, contrary to expectations, several studies have demonstrated that , in colorectal cancer tissue samples , the majority of tumors had more than 50% EphB2 -negative cells despite extensive nuclear localization of β-catenin [e.g. [35] .

Though initially up -regulated by β-catenin/TCF, EphB2 expression is typically down - regulated in advanced carcinom as through an unknown mechanism. More recently, however, s everal theories have attempt ed to explain the mechanism responsible for this decrease , including evidence for promoter or gene hypermethylation, point mutatio ns, or loss of heterozygosity [35] . In particular, one study has identified a CpG island spanning the EphB2 promoter and first exon as a site of hypermethylation in 53% of colon tumors and 25% of colorec tal cancer cell lines examined [49] . In addition, the same group identified mutations in exon 17 of EphB2 in 41.1% microsatellite unstable colon tumors 35

and 35.7% o f colorectal cancer cell lines [49] . These results suggest that hypermethylation of the EphB2 gene is likely to contribute to decreased expression of EphB2 protein. In support of this theory, SW620, a colorectal cancer cell line expressing aberrant methylation of the EphB2 promote r, seemed to express a significantly higher level of EphB2 protein when treated with 5 -aza -2’ -deoxycytidi ne (a DNA methylase inhibitor) [49] .

In support of the mutation theory, the chromosomal location of EphB2 is a frequent site of loss of heter ozygosity in colorectal tumors [48] . However, none of the mechanisms previously suggested completely account for the loss of EphB2, EphB3, and EphB4 detected in the majority of advan ced colorectal cancers [35] . This suggests that another, more universal mechanism may be responsible for this effect. I ndeed, Clevers et al. and Batlle et al. demonstrated that the majority of EphB2 down -regul ation occurs at the mRNA level [35,44] . In addition to transcriptional regulation and post -translational modification of EphB2, alternative splicing has also been shown to generate a trun cated form of EphB2 that can bind ligands but do es not t ransduce extracellular signals (reviewed in [39] ). Based on the above listed mechanisms, it is suggested that the expression and functionality of EphB2 protein may be altered in several ways.

Taken together, the inform ation presented above describes a phenomenon in colorectal cancer progression: near -constant loss of EphB2 expression in advanced colorectal carcinomas despite the presence of active β-catenin/TCF signaling . Yet if EphB2 is a transcriptional target of β-catenin/TCF [34] , and extensive nuclear localization of β- catenin is seen in colorectal cancer [28] , EphB2 expression should increase as cancer progresses. This discrepancy clearly implicates the involvement of additional, though currently unknown, pathways in the regulation of EphB2 expression. 36

EphrinB +

Crypt - villus junction

EPH EPH EPH B2 + + B2 + B2 (+++) (+) (+++)

1. A founder cell ( ) 2. Due to constitutive β- 3. Upon reaching the acquires an initial mutation catenin/TCF signaling, founder surface epithelium, leading to constitutive activity cell adopts a progenitor EphB2 + cells encounter of β-catenin/TCF, and phenotype and repopulates the ephrinB + cells. therefore expresses high crypt with its descendants, all levels of EphB2. of which are EphB2 +.

Growth halted

DCA

EPH EPH EPH B2 + B2 + B2 + (+++) (-) (-)

6. Tumor cells exhibit rapid 5. Tumor cells ‘learn’ to 4. Interaction of ephrinB ligand proliferation and expansion silence EphB2 and resume with EphB2 receptor within the colonic crypt, proliferation. compartmentalizes the tumor intermingling with ephrinB + and inhibits proliferation by cells. tumor cells.

Adapted from Clevers et al. (2006) Cancer Res., 66, 2 -5. Supplemental Figure 5. Silencing of EphB2 Expression by Tra nsformed Colorectal Cancer Cells . 37

F. Therapeutic potential of vitamin D:

F.1. Vitamin D and its receptor:

Vitamin D is consumed in th e diet as cholecalciferol (CCF). (Se e Supplemental Figure 6, page 43 ) Although relatively few natural dietary sources of this vitamin exist, milk fortification is the primary source of CCF in the diet [50] . Even t hough dietary sources are relatively scarce, deficiency of this vitamin is uncommon in healthy individuals, as vitamin D can be synthesized endogenously in the skin. 7 -dehydrocholesterol in the epidermis is converted to cholecalciferol by UVB radiation. Next, h ydroxylation to 25 - hydroxycholecalcifero l (calcidiol) occurs primarily in the liver, and subsequent hydroxylation to 1,25 -dihydroxycholecalciferol (calcitriol), the most active form of the vitamin, occurs primarily in the kidneys [50] .

One of the most important roles of vitamin D in the body is maintenance of blood calcium and phosphorus levels. However, vitamin D , particularly calcitriol, has also been demonstrated to decrease ce ll proliferation and increase differentiation [51,52] . This effect is likely mediated through the vitamin D re ceptor (VDR), a nuclear receptor th at, upon activation by calcitriol , serves as a transcription factor for several genes . VDR, when dimerized with retinoid X receptor (RXR), serves as a powerful transcription factor. Vitamin D response elements ( VDREs ) hav e been identified in p21/Waf and calcium -sensing receptor, and vitamin D has also been demonstrated to regulate the expression of several other genes, including p27/Kip1, c -myc , laminin, tenascin, fibronectin, cyclin C, c -fos , c -jun , PLC γ, ornithine decarb oxylase, and members of the TGF β family ( reviewed in [18] ). Even thoug h VDREs have not been identified in all of these genes, vitamin D has been demonstrated to regulate their expression , possibly in an indirect manner , such as demethylation . 38

However, VDR expression is often decreased or lost in colorectal cancer. This ph enomenon is demonstrated in a study by Shabahang et al. in 1993: in human colorectal adenoma and carcinoma cell lines, more highly differentiated cell lines show higher levels of V DR mRNA and functional protein [53] . Also, HT29 cells showed dose - dependent inhibition of growth (measured by thymidine incorporation and clonogenic assay ) with increasing concentrations of calcitriol , but SW620 , a de -differentiated cell lin e, did not display this effect [53] . Importantly, i n support of the cell line data, in twelve tissue samples from patients, nine of the twelve have lower VDR express ion in tumor tissue as compared to adjacent normal tissue [53] .

In addition to decreased expression of VDR in colorectal cancer, studies have demonstrated that VDR may be mutate d or nonfunctional . For example, Ingles et al. showed that presence of VDR start codon (Fok1) “f” allele was associated with a decreased risk of large colorectal adenoma, specifically, 70% decrease was detected for the genotype “ff ” and 20% de crease for the genotype “Ff ” [54] . This obser vation becomes exceptionally interesting when considering the work of Makishima et al. , who demonstrated that lithocholic acid (LCA) can bind directly to the VDR, which in turn activates CYP3A, the enzyme re sponsible for catabolizing LCA [55] . This may be a potential mechanism for detoxification of excess secondary bile acids. Thus, if VDR is mutated or nonfunctional, as seen in many cases o f colorectal cancer, catabolism of secondary bile acids may be compromised, allowing concentration of secondary bile acids to remain high.

F.2. Vitamin D properties and metabolism:

Colonocytes are capable of synthesizing calcitriol, as evidenced by the expression of hydroxylases simila r to those found in the kidney [51,56] . Interestingly, hydroxylases CYP27B1, responsible for synthesizing calc itriol, and CYP24, responsible for catabolizing calcitriol, are both upregulated in early adenomas [56] . Thus, while synthesis 39

of calcitriol increases early on in colorectal cancer, degradation also increases, resulting in no overall change in its expression. In addition, colon cells have been shown to express 1α-hydroxy lase as well as 25 -hydroxylase [51,57] , both of which a re required for the conversion of CCF to the biologically active form of vitamin D, calcitriol. The expression of these hydroxylases in the colon is still somewhat controversial. Matusiak et al. and others provide evidence for consistent expression of 1α-hydroxylase in colorectal tissue, regardless of the degree of differentiati on or tumor stage of the cells [57,58] . However, Cross et al. argues that t hese enzymes , as well as VDR, increase in protein expression as well as mRNA in early adenomas, but de crease in advanced de -differentiated tumors [51] . In spite of the discrepant observations of hydroxyla se expression, both studies demonstrate a decrease in VDR expression with increased tumor stage and loss of differentiatio n [51,57] . Overall, t hese data suggest that the initial up -regulation of hydroxylase and VDR expression may be an attempt to halt cell proliferation and tumor progression, but that upon advancement of tumor stage, tumors may become conditioned to silence VDR and hydroxylase activity, similar to the decrease in expression of EphB2 observed in the colon , as discussed previously.

F.3. Clinical demonstrations of the t umor -protective ability of vitamin D :

Epidemiological data has indicated that individuals living in areas with low exposure to natural sunlight, such as large cities and high altitude locations, were at increased risk o f developing colorectal cancer ( reviewed in [59] ). In addition, residents of states with the highest mean solar radiation levels, such as Arizona and New Mexico, exhibited significa ntly lower incidence of colorectal cancer as compared to states with the lo west levels of solar radiation ( reviewed in [59] ). In general, s erum calcidiol and calcitriol levels were inversely correlated with risk of colorectal cancer, particularly when advanced adenomas were present ( reviewed in [60] ). In particular, increased serum calcidiol levels were significantly associated with the size of the proliferative crypt compartment [61] . A case -control clinical trial of individuals with colorectal adenoma 40

demonstrated an inverse relationship between vitamin D and colorectal cancer; specifically, every 10 ng/ml increase in serum calcidiol was associated with a 26% decrease in risk of recurren t colorectal adenoma [62] . In addition to these data, Harris and Go present ed a thorough rev iew of clinical trial outcomes of vitamin D supplementation on the risk s o f developing colorectal cancer [re viewed in [18] .

F.4. Molecular and cellular tumor -protective effects of vitamin D:

On a molecular level, vitamin D exhibited several powerful anti -tumor effects. For example, rats initiated with azoxymethane follow ed by prolonged injection of vitamin D (either 1 α-hydroxycholecalciferol or calcitriol) , showed reduced incidence of colon tumors, suggestin g an anti -proliferative effect for vitamin D [63] . Interestingly, vitamin D treatment also decreased expression of vascular endothelial growth factor (VEGF) and microvessel counts in tumors, implying an add itional anti -angiogenic effect [63] . Additional studies by Kallay et al. using vitamin D receptor (VDR) knock -out mice to measu re colonocyte proliferation found that VDR expression was inversely correlated with cyclin D1, 8 -hydr oxy -2’ -deoxyguanosine, 8-OHdG - a proliferation marker , and proliferating cell nu clear antigen (PCNA) [64] , all of which are cell cycle markers . Finally, vitamin D has also been shown to induce apoptosis, likely due to its upregulation of Bak and Bax [reviewed in [18] .

Furthermore, d iet studies on rodents have provided insight into tumor -protective effects of vitamin D. Mice fed a nutritional stress diet (high fat and phosphate content, low calcium and vitamin D content) developed hyperplasia and increased [3H] -thymidine labeling of epithelial cells in bo th the sigmoid and the ascending colon [65] . While several components of this diet may h ave negatively affected the outcome, more recent work by Mokady et al. supports a potent role of vitamin D in protecting against hyperplasia. This study showed that 1,2 -dimethyl hydrazine (DMH) -induced rats fed a stress diet (mimicking a typical Western di et with 20% fat by weight) had a four -fold 41

increase in colon tumor multiplicity, as well as increased bromodeoxyuridine incorporation into crypt cell DNA and thymidine ki nase activity [66] . However, when the same rats fed the stress diet were supplemented with vitamin D, the tumor -promoting effects of the stress diet were almost entirely abrogated [66] .

In vitro studies using differentiated cell lines showed higher levels of VDR mRNA and functional VDR protein. Studies conducted by Shabahang et al. using HT29 cells, an adenocarcinoma cell line wi th a high degree of differentiation, showed dose -dependent inhibition of growth (measured by thymidine incorporation) with increasing concentrations of calcitriol, whereas SW620 cells, a metastatic, de -differentiated cell line, did not display this effect, likely because SW620 cells do not express VDR [53] . In the same study, using human tissues, out of the twelve tissue samples from patients, nine had lower VDR expre ssion in tumor tissue as comp ared to adjacent normal tissue [53] .

Of particular interest to our study , research conducted by Palmer et al. demonstrated that treatm ent with calcitriol caused a proportion of SW480 colorectal cancer cells to change phenotype to a more adhesive, epithelial phenotype [18] . Also, calcitriol treatment increased expression of E -cadherin , ZO -1, ZO -2, occludin, and vinculin (all components of intercellular junctions) as well as promoted reformation of both tigh t junctions and adhe rens junctions at the membrane [18] . Calcitriol tr eatment also increased membrane localization of β-catenin and subsequent colocalization of β-catenin and E -cadherin [18 ]. Immunoprecipitation and luciferase reporter techniques also strongly suggested that the VDR, when ligand -activated by calcitriol, competes with TCF -4 for binding to β-catenin; this interaction may account for the down -regulation of target genes seen with concomitant calcitriol treatment as well as activation of genes containing VDREs within the promoter region [18] . Target genes of β-catenin/TCF demonstrated to be downregulated by calcitriol treatment include c -myc , PPAR -delta , Tcf -1, and CD44 [41], among others . 42

Though the exact mechanism behind the observation s by Palmer et al. is unknown, sev eral theories have been suggested. Sp ecifically , the gene responsible for coding for E - cadherin, cdh1 , has been shown to be hypermethylated in both microsatellite stable (MSS) and microsatellite instable (MSI) colon carcinomas [67] . Interestingly, vitamin D treatment has been found to be associated with demethylation of the osteocalcin gene in both human osteosarcoma cells and rat osteosarcoma tumors, leading to transcriptional activation of this gene [68] . Vitamin D analogs appear to fu nction in demethylation as well. For example, EB1089, a well -established vitamin D analog, has been demonstrated to demethylate and increase transcription of human breast cancer gene ka llikrein 6 [69] . This may therefore represent a plausib le mechanism whereby vitamin D can activate transcription of E -cadherin.

Finally, and in relation to our present study, in addition to its nuclear effects, vitamin D has also been shown to affect components of the plasma membrane. Recent research has iden tified membrane -associated rapid -response steroid (MARRS) binding proteins for calcitriol (1,25 -(OH) 2D3) and 24,25 -(OH) 2D3 (a less active vitamin D metabolite) on the plasma membrane [70] . Binding to MARRS of its vitamin D metabolite induces nuclear translocation of the MARRS protein, which may be another mechanism, in addition to genomic effects, whereby vitamin D regulates serum calcium levels and exerts anti -tumor effects, such as promotion of differentiation [70] . In relation to these observations, epidermal growth factor (E GF) stimulation of EGF receptor (EGFR) has been shown to stimulate growth and promote proliferation, particularly in cancerous cells [71] , which frequently overexpress both EGF and EGFR in neoplastic tissu e as compared to surrounding normal tissue [72] . This enhancement of growth, coupled with the anti - apoptotic mechanisms present in the majority of tumors, provides an ideal environment for tumor progression and expansion. Treatment of Caco -2 colorectal cancer cells with calcitriol led to a sign ificant reduction in EGFR numbers on both basolateral and apical sides [71] . Of great importance, the same study found that, while EGF treatment stimulated cell growth, concomitant treatment with calcitrio l halted growth and decreased 43

cell numbers as compared to control [71] . Interestingly, stimulation of EGFR by EGF induced a decrease in mRNA expression of VDR [71] , providing another plausible mechanism behind the loss of VDR expression in advanced colorectal carcinomas.

F.5. Non -hypercalcemic analogs of vitamin D:

Despite its anti -tumor effects, vitamin D can modulate calcium levels in the blood, which can result in hypercalcemia in the presence of excess vitamin D. In order to achieve the same effects on differentiation and proliferation and avoid hypercalcemia , several vitamin D analogs have been developed. Animal studies u sing the APC min mouse given

thrice -weekly injections of either calcitriol or 1 α,25 -(OH) 2-16 -ene -19 -nor -24 -oxo -D3 (a common vitamin D analog) yielded decreased tumor load as compared to control with minimal hypercalcemic effects [73] . Another synthetic analog, 22 -ox a-calcitriol (OCT), was used to determine effects on colon carcinogenesis. Otoshi et al. demonstrated that, in mice initiated with DMH, eight weeks of injections (six per week) of OCT decreased the formation of aberrant crypt foci (ACF) [74] . In the same study, non -initiated mice were given four weeks of injections (six per week) of OCT, and demonstrated decreased PCNA expression in colonic epithelial cells as a prol iferative marker [74] .

Of importance to our studies, vitamin D may be able to compete with bile acids for incorporation into the membrane and regulate transcripti onal effects , either through activation of nuclear VDR or interference with DCA -induced changes in cell signaling, such as promotion of β-catenin nuclear localization and signaling . In fact, studies by Jiang et al. demonstrated that calcitriol can decrease expression of CYP7 α, the rate - limiting enzyme for bile acid synthesis, by interfering with transcriptional activity of LXR α [75] . This suggests a potential therapeutic mechanism by which vitamin D could alter the tumor -promoting effects induced by secondary bile acids. 44

Sup plemental Figure 6. Structure of cholecalciferol. (Adapted from Chemistry WikiProject, Wikipedia Online Encyclopedia, en.wikipedia.org) 45

PRESENT STUDY

The purpose of this research is to accomplish the following specific aims:

I. To identify the effect deoxy cholic acid (DCA) has on adherens junctions and ultimately cell -cell adhesion. This relationship will be examined by determining E - cadherin, β-catenin , and EphB2 expression and localization in relation to treatment with DCA.

II. To characterize the effect of cholecalciferol on E-cadherin, β-catenin , and EphB2 expression , particularly in combination with DCA treatment.

III. To visualize the molecular interactions at the plasma membrane , including E - cadherin/ β-catenin complexes and EphB2 localization, in response t o combined treatment with cholecalciferol and DCA.

We hypothesized that cholecalciferol, a form of vitamin D that does not enter the nucleus, would counteract negative effects of deoxycholic acid on expression of EphB2 and E - cadherin, key components of c ell adhesion in human colorectal cancer cells. Furthermore, we hypothesized that, in advanced colorectal carcinoma cell lines SW620 and Colo 201, which express very low levels of VDR, this interaction would occur primarily at the cell membrane. 46

Summary o f Results

The following summary attempts to highlight key methods, results, and conclusions obtained from the present study, and is intended as a guide to the appended paper. The article intended for submission, titled “Deoxycholic Acid Downregulates Eph B2 Expression in Human Colorectal Cancer Cells: Protective Role of Cholecalciferol ” is attached in its entirety in Appendix A, beginning on page 70 .

As both DCA and vitamin D have been shown to affect the plasma membrane , either through modulation of mem brane -bound protein activity or manipulation of membrane composition itself [11,70,71] , we sought to identify morphological change s in colorectal cancer cells in response to these treatments. DCA treatment induced pronounced rounding and detachment of cells, whereas pre -treatment with cholecalciferol abrogated these changes and yielded a phenotype more similar to that seen in control cells . To determine capacity for migration, we used a wound migration assay to determine the effects of DCA and cholecalciferol. DCA treatment was associated with increased migration; however, cholecalciferol pre -treatment inhibited migration capacity of cells as compared to DCA -treated cells. Finally, immunofluorescence revealed that DCA treatment reduced membrane localization of EphB2, while cholecalciferol in combination with DCA was able to protect against changes in EphB2 localization , comparable to basal levels. (Figures 1 -3).

Based on the morphological changes seen in Figure 1, we used Western blot analysis and RT -PCR to determine protein and RNA expression of EphB2, and Western blot analysis to analyze protein expression of E -cadherin and β-catenin . Importantly, assessment of EphB2 protein expression by Western blot analysis in SW620 cells revealed that cholecalciferol can protect against DCA -induced loss of EphB2, which is a hallmark of cancer progression [34,44,48] . Furthermore, RT -PCR demons trated that although DCA strongly decreased expression of EphB2, cholecalciferol treatment was able to protect 47

against loss of EphB2 in a dose -dependent manner in cell lines HCT116, HT29, SW620, and Colo 201. In addition, we decided to examine the effects of a specific EGFR inhibitor, ZD1839, which has been shown to effectively inhibit the kinase domain of EGFR and therefore inhibit the pro -tumorigenic effects of overactive EGFR signaling, including stimulation of proliferation and evasion of apoptosis [rev iewed in [72] . In our studies, through use of a specific EGFR inhibitor, ZD1839, we demonstrated that downregulation of EphB2 by DC A may be EGFR -dependent, based on increased expression of EphB2 in response to cholecalciferol, DCA, and ZD1839 treatment (Figures 4 -5) .

In addition to decreasing expression of EphB2, DCA induced a time - and dose -dependent decrease in expression of E -cadherin in HCT116, HT29, and SW620 cells. Furthermore, the decrease in expression of EphB2 and E -cadherin in response to DCA was often accompanied by a simultaneous increase in expression of β-catenin, suggesting that DCA may increase total expression of β-catenin in addition to the previously described effect of increasing phosphorylation and nuclear translocation [16] (Figures 6 -8) .

Finally, we sought to analyze modulation of the plasma membrane in response to cholecalciferol and DCA treatment using fluorescently tagged cholera toxin, which associates with lipid rafts. DCA treatment strongly decreased total membrane staining of GM1 as compared to control , and noticeably disrupted punctated patterns of GM1 , suggesting possible disruption of lipid rafts. Pre -treatment with cholecalcifero l was able to restore expression of GM1, as well as promote reformation of lipid rafts, thus interfering with the effects of DCA. These potential effects of cholecalciferol and DCA on lipid rafts may be of extreme importance to cell adhesion, as EphB2 and E -cadherin have both been demonstrated to be capable of localizing to lipid rafts [83,84] (Figure 9) .

The individual characteristics of cell lines HT29, SW620, HCT116, and Colo 201 used in this study are summarized in Supplemental Figure 7 (page 59). 48

Figure 1. Morphology of Colorectal Cancer Cells is Altered in Response to Deoxycholic Acid and Cholecalci ferol.

Ctrl CCF

DCA CCF+DCA

SW620 colorectal cancer c ells were serum -starved for 24 hours prior to treatment. Treatments were as follows: Control; 0.1 uM cholecalciferol (CCF) for 24 hours; 250 uM DCA for 10 hours; or 0.1 uM CCF pre -treatment for 24 hours with addition of 250 uM DCA for 10 hours. Control and CCF -treated cells display morphology common to epithelial cells, though CCF treatment induced slight rounding of cells. DCA induced dramatic rounding of cells as well as possible formation of filopodia. CCF pre -treatment protected against chang es induced by DCA and favored a phenotype similar to that seen in CCF -treated cells. 49

Figure 2 . Wo und Migration is Impaired with Cholecalciferol Pre - Treatment as Compared to DCA .

Control CCF

DCA DCA + CCF

SW620 colorectal cancer c ells were serum -starved for 24 hours prior to treatment. Treatments were as follows: Control; 0.1 uM CCF for 24 hours; 250 uM DCA for 10 hours; 0.1 uM CCF pre -treatment for 24 hours with addition of 250 uM DCA for 10 hours ; or 250 uM cholic acid (CA) for 24 hours, which a biologically inert primary bile acid . A wound was made with a spatula in the plane of conflue nt cells at 0h, and subsequent pictures were obtained after 10 hours of treatment. DCA induced pronounced migration from the initial wound site as compared to control, CCF -treated, and CA -treated cells. However, pre - treatment with CCF was able to impair th e migration of cells induced by DCA treatment. 50

Figure 3. DCA -Induced Loss of EphB2 Local ization is Abrogated with Cholecalciferol Pre -Treatment.

Ctrl CCF

DCA CCF+DCA

SW620 colorectal cancer c ells were serum -starved for 24 hours prior to treatment. Treatments were as follows: Control; 0.1 uM CCF for 24 hours; 250 uM DCA for 10 hours; or 0.1 uM CCF pre -treatment for 24 hours with addition of 250 uM DCA for 10 hours. Immunofluorescence for EphB2 protein was conducted, and images were obtained following 10 hours treatment. DCA induced a loss of EphB2 localization at the membrane as compared to both control and CCF -treated cells, while CCF pre -treatment protected against DCA -induced loss of EphB2 membranous localization. 51

Figure 4 . Cholecalciferol protects against DCA -mediated loss of EphB2 protein expression in a dose -dependent manner.

SW620

0h 10h A - - + + - - + + DCA (250µM) - + - + - + - + CCF (0.1µM)

EphB2

β-actin

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2

GAPDH

Cells were serum -starved for 24 hours prior to treatment. In part A, treatments were as follow s: Control; 0.1 uM CCF for 24 hours; 250 uM DCA for 10 hours; or 0.1 uM CCF pre -treatment for 24 hours with addition of 250 uM DCA for 0 or 10 hours. In part B, treatments were as follows: Control; 250 uM DCA for 10 hours; 0.1 uM CCF for 24 hours; isopr opanol pre -treatment followed by 10 hours of 250 uM DCA; 0.05 uM CCF for 24 hours followed by 250 uM DCA for 10 hours; 0.1 uM CCF for 24 hours followed by 250 uM DCA for 10 hours; 1 uM CCF for 24 hours followed by 250 uM DCA for 10 hours; or 0.1 uM CCF for 24 hours followed by 250 uM DCA treatment for 10 hours, with 10 uM ZD1839 added two hours before end of DCA treatmen t. Antibodies were directed against EphB2 or loading control , and protein expression was determined through Western blot analysis. In part A, 10 hours treatment with 250 uM DCA induced loss of EphB2 protein expression, which was protected against by 24 hours pre -treatment with CCF. In part B, the protective effect of CCF was demonstrated to be dose -dependent, with maximal effects 52

seen at 1 uM CCF treatment. Interestingly, use of ZD1839 increased expression of EphB2 as compared to control, suggesting possible involvement of EGFR. 53

Figure 5A. Cholecalciferol protects against DCA -induced loss of EphB2 RNA expression in a dose -dependent manner.

A HT29

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2 β-actin

SW620

- + + + + + + DCA (250 µM) - - 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2 β-actin

Cells were serum -starved for 24 hours prior to treatment. Treatments for HT29 were as follows: Control; 250 uM DCA for 10 hours; 0.1 uM cholecalciferol for 24 hours; isopropanol pre -treatment followed by 10 hours of 250 uM DCA ; 0.05 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; 0.1 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; 1 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; or 0.1 uM cholecalciferol for 24 hou rs followed by 250 uM DCA treatment for 10 hours, with 10 uM ZD1839 added two hours before end of DCA treatment. Treatments for SW620 were the same as used in HT29 cells, with the exclusion of 0.1 uM cholecalciferol for 24 hours. RT -PCR was conducted using primers designed for EphB2 and β-actin. . In SW620 cells, the protective effect of CCF was demonstrated to be dose -dependent, with maximal effects seen at 1 uM CCF treatment. Interestingly, use of ZD1839 increased expression of EphB2 as compared to contro l, suggesting possible involvement of EGFR. 54

Figure 5B. Cholecalciferol protects against DCA -induced loss of EphB2 RNA expression in a dose -dependent manner.

B Colo 201

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2 β-actin

HCT116

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2 β-actin

Colo 201 and HCT116 c ells were serum -starved for 24 hours prior to treatment. Treatments for HT29 were as follows: Control; 250 uM DCA for 10 hours; 0.1 uM cholecalciferol for 24 hours; isopropanol pre - treatment followed by 10 hours of 250 uM DCA; 0.05 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; 0 .1 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; 1 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours; or 0.1 uM cholecalciferol for 24 hours followed by 250 uM DCA treatment for 10 hours, with 10 uM ZD1839 added two hours before end of DCA treatment. RT -PCR was conducted using primers designed for EphB2 and β-actin. 55

Figure 6 . DCA Decreases Membrane Localization of E -cadherin in a Time -Dependent Manner in HT29 Colorectal Cancer Cells.

Ctrl DCA, 8h

DCA, 24h CA, 24h

HT29 colorectal cancer c ells were serum -starved for 24 hours prior to treatment. Treatments we re as follows: Control; 250 uM DCA for 8 hours, 250 uM DCA for 24 hours, or 250 uM CA for 24 hours. 250 uM DCA began to decrease membrane localization of E -cadherin at 8 hours, with a maximal effect seen at 24 hours treatment with DCA. Use of CA did not al ter membrane localization of E -cadherin, and displayed similar staining to that seen in control cells. 56

Figure 7. DCA treatment decreases expression of E -cadherin with subsequent induction of β-catenin in a time -dependent manner.

HT29

250 µM DCA l A o r C t n h h h m o 4 4 h h h h 6 0 2 2 1 2 4 8 1 3 111 C kDa E-cadherin

β−catenin

β-actin

SW620

250 µM DCA l A o r C t n h h h m o 4 4 h h h h 6 0 2 2 1 2 4 8 1 3 C

E-cadherin

β−catenin

GAPDH

HCT116

l 250 µM DCA A o r C t n m h h h o 0 h h h h 6 4 4 C 3 1 2 4 8 1 2 2

E-cadherin

β−catenin

GAPDH 57

HT29, SW620, and HCT116 cells were serum -starved for 24 hours prior to treatment. Treatments in all cell lines are as f ollows: Control; 250 uM DCA for 30 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 16 hours, 24 hours; or 24 hours CA. DCA treatment decreased expression of E -cadherin in all three cell lines with maximal effects seen between 8 and 24 hours treatment with DCA. In HT29 and SW620 cells, 8 and 16 hours treatment with DCA increased expression of β-catenin as compared to control. This increase in expression of β-catenin was not observed in HCT116 cells, possibly due to high susceptibility of this cell line to cytotoxicity from DCA treatment. 58

Figure 8. DCA increases expression of β-catenin in a dose -dependent manner.

HT29

- 50 100 250 500 - DCA (µM) ---- -500 CA (µM)

Anti -β -catenin

Anti -β -actin

SW620

- 50 100 250 500 - DCA (µM) ---- -500 CA (µM)

Anti -β -catenin

Anti -β -actin

HCT116

- 50 100 250 500 - DCA (µM) ---- -500 CA (µM)

Anti -β -catenin

Anti -β -actin

HT29, SW620, and HCT116 colore ctal cancer cells were treated with increasing concentrations of DCA for 16 hours. Treatments were as follows: Control; 50 uM DCA; 100 uM DCA; 250 uM DCA; 500 uM DCA; or 500 uM CA for 16 hours. In both HT29 and SW620 cells, an increase in expression of β-catenin was seen with 250 uM and 500 uM treatment with DCA for 16 hours. Again, this effect was not observed in HCT116 cells, possibly due to cytotoxic effects of DCA on this cell line. 59

Figure 9. Cholecalciferol protects against DCA -induced disruption of GM1 localization .

Ctrl CCF

DCA CCF+DCA

Cells were serum -starved for 24 hours prior to treatment. Treatments were as follows: Control; 0.1 uM cholecalciferol for 24 hours; 250 uM DCA for 10 hours; or 0.1 uM cholecalciferol pre -treatment for 24 hours with addition of 250 uM DCA for 10 hours. Immunofluorescence for fluorescently -tagged cholera toxin was conducted, and images were obtained following 10 hours treatment. Both control and CCF - treated cells display strong membranous staining of GM1. DC A treatment decreased membrane localization of GM1, possibly indicating disruption of lipid rafts. CCF pre -treatment was able to protect cells against DCA -mediated loss of GM1 at the membrane, and cells displayed similar GM1 staining to that seen in contro l cells. 60

HT29 SW620 HCT116 Colo 201

44 year -old 51 year -old Adult male, 70 year -old Origin female, primary male, lymph primary male, ascites tumor node metastasis tumor metastasis

Adenoc arcin Type of Cancer Adenocarcinoma Adenocarcinoma Carcinoma oma Fibroblast - Epithelial; Epithelial; Epithelial; Morphology like; grow in adherent adherent adherent suspension myc+; ras+ ; myc+; ras+; myc+; myb+ ; myb+ ; fos+; myb+; fos+; ras+; fos+; sis+; Mutations ras+ sis+; p53+; sis+; p53+; abl -; p53+; abl -; ros -; abl -; ros -; ros -; src - src - src - VDR Expression Moderate Low/ absent Moderate Low/ absent EphB2 Low Moderate Low High Expression DCA increases Yes Yes Unknown Unknown migration? CCF+DCA reduces Unknown Yes Unknown Unknown migration? DCA decreases N/A Yes N/A Yes EphB2? CCF+DCA increases N/A Yes N/A Yes EphB2? DCA decreases Yes Yes Yes Yes E-cadherin? DCA increases β- Yes Yes No Unknown catenin? Very susceptible Stabilizing to DCA Additional Info mutation in β- treatment catenin due to cytotoxicity

Supplemental Figure 7. Summary of cell line variations in genotype and response to treatments . This chart identifie s the morphology and mutations associated with each cell line used in this study. Furthermore, this chart attempts to summarize the results obtained from each cell line used in the present study. 61

Additional Experiments

Prior to submission of the app ended paper, the following experiments will be conducted:

Separation of nuclear and cytosolic fractions:

Separation of nuclear and cytosolic fractions will be conducted in cell lines HT29, HCT116, SW620, and Colo 201. The purpose of this experiment is t o determine localization of β-catenin in each cell line under normal conditions, then in response to cholecalciferol, DCA, and combination treatments. We expect to see decreased nuclear localization with cholecalciferol treatment, increased nuclear localiz ation with DCA treatment, and decreased nuclear localization with combination treatment.

Immunofluorescence of β-catenin: Immunofluorescence of β-catenin will be conducted in cell lines SW620 and HT29. The purpose of this experiment is to visualize local ization of β-catenin in response to cholecalciferol, DCA, and combination treatments. As stated before, we expect to see decreased nuclear localization with cholecalciferol treatment, increased nuclear localization with DCA treatment, and decreased nuclear localization with combination treatment. In addition, we expect to see increased expression of β-catenin with DCA treatment as compared to control, cholecalciferol, and combination treatments, in support of results obtained from previous Western blot anal ysis experiments.

Biotinylation of E-cadherin at the cell membrane: Biotinylation of E-cadherin will be conducted in the cell line HCT116. The purpose of this experiment is to visualize membrane dynamics, particularly localization of E- cadherin , in resp onse to cholecalciferol, DCA, and combination treatments. We expect to see perturbation of the membrane and loss of expression of E-cadherin at the membrane in response to DCA treatment. However, we expect to see normal membrane dynamics restored in cells treated with either cholecalciferol or combination treatments. 62

Conclusion

Colorectal cancer is a growing threat to national health, particularly in developed countries. In the present study, t reatment of colorectal cancer cells with cholecalciferol and D CA exerted several effects on cell adhesion and segregation , two key tumor - protective systems in colorectal cancer . Three findings in this study have important implications for colorectal cancer . First, DCA treatment drastically decreased expression of Eph B2 RNA and protein . As loss of EphB2 is associated with accelerated cancer progression, this finding ma y represent yet another pro -tumorigenic effect of DCA. Second, cholecalciferol was able to restore both protein and RNA expression of EphB2 when combined with DCA treatment. Interestingly, cholecalciferol treatment alone did not increase levels of EphB2 as compared to control , suggesting that cholecalciferol functions best as a protective agent against pro -tumorigenic changes induced by DCA treatment , rath er than as an anti -tumor agent in itself . Finally , these protective effects of cholecalciferol appear to be exerted at the membrane, and at least in part through modulation of EGFR activity, as was demonstrated using a specific EGFR inhibitor, ZD1839. Use of ZD1839 increased expression of EphB2 as compared to control and to DCA -only treatment, suggesting that DCA may signal through EGFR to exert its observed effects on EphB2. However, ZD1839 treatment did not drastically increase EphB2 expression above that seen with combined cholecalciferol and DCA treatment, in contrary to expectations, suggesting that another mechanism in addition to EGFR may be responsible for loss of EphB2 expression. In conclusion, these results provide evidence for use of cholecalcife rol as an adjuvant therapy for individuals with advanced colorectal cancer. Furthermore, as a lifetime of high saturated fat and low vitamin D intake increases risk of developing colorectal cancer, one treatment of cholecalciferol will not exert the same p rotective effects in the body as demonstrated in the cell. Rather, additional emphasis should be placed on meeting vitamin D requirements throughout life , not only to maintain calcium levels, but, as demonstrated in this study and others, to protect agains t development and progression of colorectal cancer. 63

REFERENCES

1. American Cancer Society, I. (2006) SOS - Colorectal Cancer. 2. Cruz -Bustillo Clarens, D. (2004) Molecular genetics of colorectal cancer. Rev Esp Enferm Dig (Madrid) , 96 , 48 -59. 3. Fearon, E.R. and Vogelstein, B. (1990) A genetic model for colorectal tumorigenesis. Cell , 61 , 759 -67. 4. Glinghammar, B. and Rafter, J. (1999) Carcinogenesis in the colon: interaction between luminal factors and genetic factors. Eur J Cancer P rev , 8 Suppl 1 , S87 - 94. 5. Laurent -Puig, P., Blons, H. and Cugnenc, P.H. (1999) Sequence of molecular genetic events in colorectal tumorigenesis. Eur J Cancer Prev , 8 Suppl 1 , S39 -47. 6. Buda, A. and Pignatelli, M. (2002) Genetics -- cellular basis. Br Med B ull , 64 , 45 - 58. 7. Nagengast, F.M., Grubben, M.J. and van Munster, I.P. (1995) Role of bile acids in colorectal carcinogenesis. Eur J Cancer , 31A , 1067 -70. 8. Debruyne, P.R., Bruyneel, E.A., Karaguni, I.M., Li, X., Flatau, G., Muller, O., Zimber, A., Gespa ch, C. and Mareel, M.M. (2002) Bile acids stimulate invasion and haptotaxis in human colorectal cancer cells through activation of multiple oncogenic signaling pathways. Oncogene , 21 , 6740 -50. 9. Yui, S., Saeki, T., Kanamoto, R. and Iwami, K. (2005) Charac teristics of apoptosis in HCT116 colon cancer cells induced by deoxycholic acid. J Biochem (Tokyo) , 138 , 151 -7. 10. Jean -Louis, S., Akare, S., Ali, M.A., Mash, E.A., Jr., Meuillet, E. and Martinez, J.D. (2006) Deoxycholic acid induces intracellular signali ng through membrane perturbations. J Biol Chem , 281 , 14948 -60. 11. Glinghammar, B., Inoue, H. and Rafter, J.J. (2002) Deoxycholic acid causes DNA damage in colonic cells with subsequent induction of caspases, COX -2 promoter activity and the transcription f actors NF -kB and AP -1. Carcinogenesis , 23 , 839 - 45. 12. Narisawa, T., Magadia, N.E., Weisburger, J.H. and Wynder, E.L. (1974) Promoting effect of bile acids on colon carcinogenesis after intrarectal instillation of N -methyl -N' -nitro -N-nitrosoguanidine in ra ts. J Natl Cancer Inst , 53 , 1093 -7. 13. Flynn, C., Montrose, D.C., Swank, D.L., Nakanishi, M., Ilsley, J.N. and Rosenberg, D.W. (2007) Deoxycholic acid promotes the growth of colonic aberrant crypt foci. Mol Carcinog , 46 , 60 -70. 14. Debruyne, D., Oliveira, M.J., Bracke, M., Mareel, M. and Leroy, A. (2006) Colon cancer cells: pro -invasive signalling. Int J Biochem Cell Biol , 38 , 1231 -6. 15. Pai, R., Tarnawski, A.S. and Tran, T. (2004) Deoxycholic acid activates beta - catenin signaling pathway and increases co lon cell cancer growth and invasiveness. Mol Biol Cell , 15 , 2156 -63. 16. Coluccia, A.M., Benati, D., Dekhil, H., De Filippo, A., Lan, C. and Gambacorti - Passerini, C. (2006) SKI -606 decreases growth and motility of colorectal cancer 64

cells by preventing pp60 (c -Src) -dependent tyrosine phosphorylation of beta - catenin and its nuclear signaling. Cancer Res , 66 , 2279 -86. 17. Palmer, H.G., Gonzalez -Sancho, J.M., Espada, J., Berciano, M.T., Puig, I., Baulida, J., Quintanilla, M., Cano, A., de Herreros, A.G., Lafarga , M. and Munoz, A. (2001) Vitamin D(3) promotes the differentiation of colon carcinoma cells by the induction of E -cadherin and the inhibition of beta -catenin signaling. J Cell Biol , 154 , 369 -87. 18. Bates, R.C. and Mercurio, A.M. (2005) The epithelial -mes enchymal transition (EMT) and colorectal cancer progression. Cancer Biol Ther , 4, 365 -70. 19. Reddig, P.J. and Juliano, R.L. (2005) Clinging to life: cell to matrix adhesion and cell survival. Cancer Metastasis Rev , 24 , 425 -39. 20. Gilmore, A.P. (2005) Ano ikis. Cell Death Differ , 12 Suppl 2 , 1473 -7. 21. Brabletz, T., Jung, A., Reu, S., Porzner, M., Hlubek, F., Kunz -Schughart, L.A., Knuechel, R. and Kirchner, T. (2001) Variable beta -catenin expression in colorectal cancers indicates tumor progression driven by the tumor environment. Proc Natl Acad Sci U S A , 98 , 10356 -61. 22. Mareel, M. and Leroy, A. (2003) Clinical, cellular, and molecular aspects of cancer invasion. Physiol Rev , 83 , 337 -76. 23. Buda, A. and Pignatelli, M. (2004) Cytoskeletal network in colo n cancer: from genes to clinical application. Int J Biochem Cell Biol , 36 , 759 -65. 24. Lilien, J. and Balsamo, J. (2005) The regulation of cadherin -mediated adhesion by tyrosine phosphorylation/dephosphorylation of beta -catenin. Curr Opin Cell Biol , 17 , 45 9-65. 25. Haraguchi, K., Nishida, A., Ishidate, T. and Akiyama, T. (2004) Activation of beta -catenin -TCF -mediated transcription by non -receptor tyrosine kinase v -Src. Biochem Biophys Res Commun , 313 , 841 -4. 26. Behrens, J. (2005) The role of the Wnt signal ling pathway in colorectal tumorigenesis. Biochem Soc Trans , 33 , 672 -5. 27. Wong, N.A. and Pignatelli, M. (2002) Beta -catenin -- a linchpin in colorectal carcinogenesis? Am J Pathol , 160 , 389 -401. 28. Polakis, P. (2000) Wnt signaling and cancer. Genes Dev , 14, 1837 -51. 29. Lustig, B. and Behrens, J. (2003) The Wnt signaling pathway and its role in tumor development. J Cancer Res Clin Oncol , 129 , 199 -221. 30. Mulholland, D.J., Dedhar, S., Coetzee, G.A. and Nelson, C.C. (2005) Interaction of nuclear receptors w ith the Wnt/beta -catenin/Tcf signaling axis: Wnt you like to know? Endocr Rev , 26 , 898 -915. 31. van de Wetering, M., Sancho, E., Verweij, C., de Lau, W., Oving, I., Hurlstone, A., van der Horn, K., Batlle, E., Coudreuse, D., Haramis, A.P., Tjon -Pon -Fong, M., Moerer, P., van den Born, M., Soete, G., Pals, S., Eilers, M., Medema, R. and Clevers, H. (2002) The beta -catenin/TCF -4 complex imposes a crypt progenitor phenotype on colorectal cancer cells. Cell , 111 , 241 -50. 32. Clevers, H. and Batlle, E. (2006) Eph B/EphrinB receptors and Wnt signaling in colorectal cancer. Cancer Res , 66 , 2 -5. 65

33. Batlle, E., Henderson, J.T., Beghtel, H., van den Born, M.M., Sancho, E., Huls, G., Meeldijk, J., Robertson, J., van de Wetering, M., Pawson, T. and Clevers, H. (2002) Bet a-catenin and TCF mediate cell positioning in the intestinal epithelium by controlling the expression of EphB/ephrinB. Cell , 111 , 251 -63. 34. Bruckner, K. and Klein, R. (1998) Signaling by Eph receptors and their ephrin ligands. Curr Opin Neurobiol , 8, 375 -82. 35. Cutforth, T. and Harrison, C.J. (2002) Ephs and ephrins close ranks. Trends Neurosci , 25 , 332 -4. 36. Zisch, A.H. and Pasquale, E.B. (1997) The Eph family: a multitude of receptors that mediate cell recognition signals. Cell Tissue Res , 290 , 217 -26 . 37. Mellitzer, G., Xu, Q. and Wilkinson, D.G. (2000) Control of cell behaviour by signalling through Eph receptors and ephrins. Curr Opin Neurobiol , 10 , 400 -8. 38. Hall, A. (1998) Rho GTPases and the actin cytoskeleton. Science , 279 , 509 -14. 39. Robinson , V., Smith, A., Flenniken, A.M. and Wilkinson, D.G. (1997) Roles of Eph receptors and ephrins in neural crest pathfinding. Cell Tissue Res , 290 , 265 - 74. 40. Yu, H.H., Zisch, A.H., Dodelet, V.C. and Pasquale, E.B. (2001) Multiple signaling interactions of Abl and Arg kinases with the EphB2 receptor. Oncogene , 20 , 3995 -4006. 41. Batlle, E., Bacani, J., Begthel, H., Jonkheer, S., Gregorieff, A., van de Born, M., Malats, N., Sancho, E., Boon, E., Pawson, T., Gallinger, S., Pals, S. and Clevers, H. (2005) EphB receptor activity suppresses colorectal cancer progression. Nature , 435 , 1126 -30. 42. Zisch, A.H., Stallcup, W.B., Chong, L.D., Dahlin -Huppe, K., Voshol, J., Schachner, M. and Pasquale, E.B. (1997) Tyrosine phosphorylation of L1 family adhesion molecules: implication of the Eph kinase Cek5. J Neurosci Res , 47 , 655 - 65. 43. Jubb, A.M., Zhong, F., Bheddah, S., Grabsch, H.I., Frantz, G.D., Mueller, W., Kavi, V., Quirke, P., Polakis, P. and Koeppen, H. (2005) EphB2 is a prognostic factor in colorectal cancer. Cl in Cancer Res , 11 , 5181 -7. 44. Saito, T., Masuda, N., Miyazaki, T., Kanoh, K., Suzuki, H., Shimura, T., Asao, T. and Kuwano, H. (2004) Expression of EphA2 and E -cadherin in colorectal cancer: correlation with cancer metastasis. Oncol Rep , 11 , 605 -11. 45. Guo, D.L., Zhang, J., Yuen, S.T., Tsui, W.Y., Chan, A.S., Ho, C., Ji, J., Leung, S.Y. and Chen, X. (2006) Reduced expression of EphB2 that parallels invasion and metastasis in colorectal tumours. Carcinogenesis , 27 , 454 -64. 46. Alazzouzi, H., Davalos, V., K okko, A., Domingo, E., Woerner, S.M., Wilson, A.J., Konrad, L., Laiho, P., Espin, E., Armengol, M., Imai, K., Yamamoto, H., Mariadason, J.M., Gebert, J.F., Aaltonen, L.A., Schwartz, S., Jr. and Arango, D. (2005) Mechanisms of inactivation of the receptor t yrosine kinase EPHB2 in colorectal tumors. Cancer Res , 65 , 10170 -3. 47. Harris, D.M. and Go, V.L. (2004) Vitamin D and colon carcinogenesis. J Nutr , 134 , 3463S -3471S. 66

48. Cross, H.S., Bareis, P., Hofer, H., Bischof, M.G., Bajna, E., Kriwanek, S., Bonner, E . and Peterlik, M. (2001) 25 -Hydroxyvitamin D(3) -1alpha -hydroxylase and vitamin D receptor gene expression in human colonic mucosa is elevated during early cancerogenesis. Steroids , 66 , 287 -92. 49. Greenwald, P., Kelloff, G.J., Boone, C.W. and McDonald, S. S. (1995) Genetic and cellular changes in colorectal cancer: proposed targets of chemopreventive agents. Cancer Epidemiol Biomarkers Prev , 4, 691 -702. 50. Shabahang, M., Buras, R.R., Davoodi, F., Schumaker, L.M., Nauta, R.J. and Evans, S.R. (1993) 1,25 -Dih ydroxyvitamin D3 receptor as a marker of human colon carcinoma cell line differentiation and growth inhibition. Cancer Res , 53 , 3712 -8. 51. Ingles, S.A., Wang, J., Coetzee, G.A., Lee, E.R., Frankl, H.D. and Haile, R.W. (2001) Vitamin D receptor polymorphis ms and risk of colorectal adenomas (United States). Cancer Causes Control , 12 , 607 -14. 52. Makishima, M., Lu, T.T., Xie, W., Whitfield, G.K., Domoto, H., Evans, R.M., Haussler, M.R. and Mangelsdorf, D.J. (2002) Vitamin D receptor as an intestinal bile acid sensor. Science , 296 , 1313 -6. 53. Cross, H.S., Bises, G., Lechner, D., Manhardt, T. and Kallay, E. (2005) The Vitamin D endocrine system of the gut -- its possible role in colorectal cancer prevention. J Steroid Biochem Mol Biol , 97 , 121 -8. 54. Matusiak, D. , Murillo, G., Carroll, R.E., Mehta, R.G. and Benya, R.V. (2005) Expression of vitamin D receptor and 25 -hydroxyvitamin D3 -1{alpha} - hydroxylase in normal and malignant human colon. Cancer Epidemiol Biomarkers Prev , 14 , 2370 -6. 55. Tangpricha, V., Flanagan, J.N., Whitlatch, L.W., Tseng, C.C., Chen, T.C., Holt, P.R., Lipkin, M.S. and Holick, M.F. (2001) 25 -hydroxyvitamin D -1alpha - hydroxylase in normal and malignant colon tissue. Lancet , 357 , 1673 -4. 56. Garland, C.F. and Garland, F.C. (1980) Do sunlight and v itamin D reduce the likelihood of colon cancer? Int J Epidemiol , 9, 227 -31. 57. Giovannucci, E. (2005) The epidemiology of vitamin D and cancer incidence and mortality: a review (United States). Cancer Causes Control , 16 , 83 -95. 58. Holt, P.R., Arber, N., Halmos, B., Forde, K., Kissileff, H., McGlynn, K.A., Moss, S.F., Kurihara, N., Fan, K., Yang, K. and Lipkin, M. (2002) Colonic epithelial cell proliferation decreases with increasing levels of serum 25 -hydroxy vitamin D. Cancer Epidemiol Biomarkers Prev , 11, 113 -9. 59. Peters, U., McGlynn, K.A., Chatterjee, N., Gunter, E., Garcia -Closas, M., Rothman, N. and Sinha, R. (2001) Vitamin D, calcium, and vitamin D receptor polymorphism in colorectal adenomas. Cancer Epidemiol Biomarkers Prev , 10 , 1267 -74. 60. Isek i, K., Tatsuta, M., Uehara, H., Iishi, H., Yano, H., Sakai, N. and Ishiguro, S. (1999) Inhibition of angiogenesis as a mechanism for inhibition by 1alpha - hydroxyvitamin D3 and 1,25 -dihydroxyvitamin D3 of colon carcinogenesis induced by azoxymethane in Wist ar rats. Int J Cancer , 81 , 730 -3. 67

61. Kallay, E., Pietschmann, P., Toyokuni, S., Bajna, E., Hahn, P., Mazzucco, K., Bieglmayer, C., Kato, S. and Cross, H.S. (2001) Characterization of a vitamin D receptor knockout mouse as a model of colorectal hyperprolif eration and DNA damage. Carcinogenesis , 22 , 1429 -35. 62. Newmark, H.L., Lipkin, M. and Maheshwari, N. (1990) Colonic hyperplasia and hyperproliferation induced by a nutritional stress diet with four components of Western -style diet. J Natl Cancer Inst , 82 , 491 -6. 63. Mokady, E., Schwartz, B., Shany, S. and Lamprecht, S.A. (2000) A protective role of dietary vitamin D3 in rat colon carcinogenesis. Nutr Cancer , 38 , 65 -73. 64. Thorstensen, L., Lind, G.E., Lovig, T., Diep, C.B., Meling, G.I., Rognum, T.O. and L othe, R.A. (2005) Genetic and epigenetic changes of components affecting the WNT pathway in colorectal carcinomas stratified by microsatellite instability. Neoplasia , 7, 99 -108. 65. Haslberger, A., Varga, F. and Karlic, H. (2006) Recursive causality in evo lution: a model for epigenetic mechanisms in cancer development. Med Hypotheses , 67 , 1448 -54. 66. Pampalakis, G. and Sotiropoulou, G. (2006) Multiple mechanisms underlie the aberrant expression of the human kallikrein 6 gene in breast cancer. Biol Chem , 38 7, 773 -82. 67. Farach -Carson, M.C. and Nemere, I. (2003) Membrane receptors for vitamin D steroid hormones: potential new drug targets. Curr Drug Targets , 4, 67 -76. 68. Tong, W.M., Kallay, E., Hofer, H., Hulla, W., Manhardt, T., Peterlik, M. and Cross, H.S . (1998) Growth regulation of human colon cancer cells by epidermal growth factor and 1,25 -dihydroxyvitamin D3 is mediated by mutual modulation of receptor expression. Eur J Cancer , 34 , 2119 -25. 69. Messa, C., Russo, F., Caruso, M.G. and Di Leo, A. (1998) EGF, TGF -alpha, and EGF -R in human colorectal adenocarcinoma. Acta Oncol , 37 , 285 -9. 70. Huerta, S., Irwin, R.W., Heber, D., Go, V.L., Koeffler, H.P., Uskokovic, M.R. and Harris, D.M. (2002) 1alpha,25 -(OH)(2) -D(3) and its synthetic analogue decrease tumor load in the Apc(min) Mouse. Cancer Res , 62 , 741 -6. 71. Otoshi, T., Iwata, H., Kitano, M., Nishizawa, Y., Morii, H., Yano, Y., Otani, S. and Fukushima, S. (1995) Inhibition of intestinal tumor development in rat multi - organ carcinogenesis and aberrant crypt foci in rat colon carcinogenesis by 22 - oxa -calcitriol, a synthetic analogue of 1 alpha, 25 -dihydroxyvitamin D3. Carcinogenesis , 16 , 2091 -7. 72. Jiang, W., Miyamoto, T., Kakizawa, T., Nishio, S.I., Oiwa, A., Takeda, T., Suzuki, S. and Hashizume, K. (2006) Inhibition of LXRalpha signaling by vitamin D receptor: possible role of VDR in bile acid synthesis. Biochem Biophys Res Commun , 351 , 176 -84. 73. Himanen, J.P., Rajashankar, K.R., Lackmann, M., Cowan, C.A., Henkemeyer, M. and Nikolov, D.B. (2001) Crystal s tructure of an Eph receptor -ephrin complex. Nature , 414 , 933 -8. 74. Mendelsohn, J. and Baselga, J. (2006) Epidermal growth factor receptor targeting in cancer. Semin Oncol , 33 , 369 -85. 68

75. Hahm, E.R., Gho, Y.S., Park, S., Park, C., Kim, K.W. and Yang, C.H. (2004) Synthetic curcumin analogs inhibit activator protein -1 transcription and tumor - induced angiogenesis. Biochem Biophys Res Commun , 321 , 337 -44. 76. Brambilla, R. and Klein, R. (1995) Telling axons where to grow: a role for Eph receptor tyrosine kinas es in guidance. Mol Cell Neurosci , 6, 487 -95. 77. Smit, L., Baas, A., Kuipers, J., Korswagen, H., van de Wetering, M. and Clevers, H. (2004) Wnt activates the Tak1/Nemo -like kinase pathway. J Biol Chem , 279 , 17232 -40. 78. Ishitani, T., Ninomiya -Tsuji, J. a nd Matsumoto, K. (2003) Regulation of lymphoid enhancer factor 1/T -cell factor by mitogen -activated protein kinase - related Nemo -like kinase -dependent phosphorylation in Wnt/beta -catenin signaling. Mol Cell Biol , 23 , 1379 -89. 79. Wang, C., Deng, L., Hong, M ., Akkaraju, G.R., Inoue, J. and Chen, Z.J. (2001) TAK1 is a ubiquitin -dependent kinase of MKK and IKK. Nature , 412 , 346 -51. 80. Stewart, D.B., Barth, A.I. and Nelson, W.J. (2000) Differential regulation of endogenous cadherin expression in Madin -Darby can ine kidney cells by cell -cell adhesion and activation of beta -catenin signaling. J Biol Chem , 275 , 20707 -16. 81. Seveau, S., Bierne, H., Giroux, S., Prevost, M.C. and Cossart, P. (2004) Role of lipid rafts in E -cadherin -- and HGF -R/Met -- mediated entry of Listeria monocytogenes into host cells. J Cell Biol , 166 , 743 -53. 82. Gauthier, L.R. and Robbins, S.M. (2003) Ephrin signaling: One raft to rule them all? One raft to sort them? One raft to spread their call and in signaling bind them? Life Sci , 74 , 207 -16 . 69

APPENDIX A

DEOXYCHOLIC ACID DOWNREGULATES EPHB2 EXPRESSION IN HUMAN COLORECTAL CANCER CELLS: PROTECTIVE ROLE OF CHOLECALCIFEROL

Shawna Comer 1, Zuohe Song 1, Samira Jean -Louis 1, Margareta Berggren 1, Sherif Morgan 1, Ben George 1, and Emmanuelle J. Meui llet 1,2,3*

1. Department of Nutritional Sciences, College of Agriculture and Life Sciences, at the University of Arizona, Tucson, AZ. 85721, USA. 2. Arizona Cancer Center, Tucson AZ 3. Department of Molecular and Cellular Biology at the University of Ari zona, Tucson, AZ. 85721, USA.

* To whom correspondence should be addressed: Emmanuelle J. Meuillet The University of Arizona Department of Nutritional Sciences, College of Agriculture and Life Sciences 1177 E. 4 th Street PO BOX 210038 Tucson, AZ 85721 -00 38 Phone: (520) 626 -5794 Fax: (520) 626 -4455 Email: [email protected]

Running title: EphB2 expression is restored in colon cancer by vitamin D treatment.

Keywords: EphB2, vitamin D, bile acids, colon cancer, β-catenin, E -cadherin 70

I. Abstract A diet high in fat has long been implicated in the etiology, promotion, and progression of colon cancer. Though the mechanisms by which dietary fat increases colon carcinogenesis are not fully understood, it has been sug gested that high fat diets enhance tumorigenesis through elevation of secondary bile acids such as deoxycholic acid (DCA) that act as promoters of tumor development. We and others have shown that many effects of this specific cholesterol derivative origin ate at the plasma membrane and are further propagated within the cell. In this study, we show that DCA decreases the expression of E -cadherin and increases the expression of β-catenin. Interestingly, we also show that DCA significantly reduces EphB2 expression in colorectal cancer cells, a downstream target of β-catenin and powerful regulator of both cell positioning and cell adhesion. Hence, given that vitamin D can regulat e the interaction between E -cadherin and β-catenin as well as modulate EGFR, we sought to analyze the effects of DCA in vitamin D pre -treated colon cancer cells. Interestingly, we found that the effects of DCA are EGFR -dependent since treatment with ZD183 9, a specific EGFR inhibitor, increases EphB2 expression even with concomitant DCA treatment. However, pre -treatment of the cells with vitamin D prevented the effects of bile acids. These results strongly suggest that vitamin D may antagonize the effects of DCA, and restores basal levels of EphB2 in colorectal cancer cells. Taken together, our study suggests that vitamin D may be useful as an adjuvant therapy for individuals with advanced colorectal carcinoma. 71

II. Introduction

Though many factors contri bute to colorectal carcinogenesis, epidemiological data obtained by the American Cancer Society indicate a distinct positive correlation between consumption of a diet high in saturated fat and increased risk of developing colorectal cancer. High intake of saturated fat is correlated with increased production and secretion of bile acids, which are amphipathic cholesterol derivatives aiding in the emulsification of lipids, enhancing digestion and absorption of these macronutrients [8,9] . Importantly, secondary bile acids have been demonstrated to induce nuclear translocation of β-catenin and to induce transcription of β-catenin/TCF/LEF target genes. In addition to stimulating cell proliferation, Deoxycholic Acid (DCA) a secondary bile aci d has been demonstrated to induce several tumor -promoting effects at the molecular level, including, induction of oxidative stress and mitochondrial cytochrome c release [9] , activation of protein kinase C (PKC), induction of cycl ooxygenase -2 (COX -2) promoter activity, stimulation of activator protein -1 (AP -1) and nuclear factor kB (NF -kB) expression [12] , and manipulation of membrane composition and decreased membrane fluidity [11] . DCA has been shown to have a particularly potent effect on adhesion protein interactions. In addition to decreasing total protein expression of E -cadherin, DCA decrea ses the binding of β-catenin to E -cadherin while increasing β-catenin tyrosine phosphorylation and subsequent nuclear localization [16] . One possible mechanism for this is the activation of EGFR and Src by perturbation of the plasma membrane by DCA [11] . β-catenin dissociation from the membrane and subsequent nuclear localization is associated with increased invasion, proliferation, and transcription of β-catenin/TCF target genes, while β-catenin membrane localization and binding with E -cadherin is associated with decreased invasion, increased differentiation, and inhibition of β-catenin signaling [16 - 18] .

Another cholesterol derivative, vitamin D, has been demonstrated to have powerful anti - tumor effects, both in vitro and in vivo. Epidemiological data has indicated that 72

individuals living in areas with low exposure to natural sunlight, such as larg e cities and high altitude locations, were at increased risk of developing colorectal cancer [reviewed in [59] . In addition, residents of states with the highest mean solar radiation levels, such as Arizona and New Mexico, exhibited significantly lower incidence of colorectal cancer as compared to states with the lowest levels of solar radiation [reviewed in [59] . On a molecular level, vitamin D has exhibited several powerful anti -tumor effects. In rats initiated with azoxymethane, prolonged injection of vitamin D (either 1 α- hydroxycholecalciferol or calcitriol) reduced incidence of colon tumors, suggesting an anti -proliferative effect [63] . Interestingly, vitamin D treatment also decreased expression of vascular endothelial growth factor (VEGF) and microvessel counts in tumors, implying an additional anti -angiogenic effect [63] .

In addition to the nuclear actions of vitamin D, that involves the Vitamin D Receptor (VDR), this speci fic vitamin has also been shown to affect components of the plasma membrane. Treatment of Caco -2 colorectal cancer cells with calcitriol led to a significant reduction in EGFR numbers on both basolateral and apical sides [71] . In the same study, while EGF treatment stimulated cell growth, concomitant treatment with calcitriol halted growth and decreased cell numbers as compared to control [71] . Interesti ngly, stimulation of EGFR by EGF induced a decrease in mRNA expression of VDR [71] , which may provide yet another insight into the mechanism behind the loss of VDR expression in advanced colorectal carcino mas. Importantly, VDR expression is often decreased or lost in colorectal cancer. In addition to decreased expression of VDR in colorectal cancer, research has demonstrated that VDR may be mutated or nonfunctional in some cases. For example, Ingles et al. showed that presence of VDR start codon (Fok1) f allele was associated with a decreased risk of large colorectal adenoma, specifically, 70% decrease for genotype ff and 20% decrease for genotype Ff [54] . Based on this information, we hypothesized that cholecalciferol, a form of vitamin D that does no t enter the nucleus, would prevent the effects of DCA on EphB2 and E -cadherin expression, key components of cell adhesion. Furthermore, we hypothesized that, in 73

advanced colorectal carcinoma cell lines SW620 and Colo 201, which express very low levels of V DR, this interaction would occur primarily at the cell membrane. 74

III. Materials and Methods Cell Culture Human colon cancer lines HCT116, HT29, Colo 201, and SW620 (American Type Culture Collection (ATCC)) were grown in McCoy’s 5A Modified Medium (fo r HCT116 and HT29), RPMI 1640 Medium, and Leibovitz’s L15 Medium, respectively. All cell culture as well as experiments pertaining to cell culture used media supplemented with 10% fetal bovine serum and 1X Penicillin -Streptamycin -Glutamine (Pen/Strep) (Gib co), unless stated otherwise. All cell lines were incubated at 37ºC under 5% CO 2 except

SW620, which was incubated at 37ºC under 0% CO 2.

Detection of Proteins by Western Blotting Cell lines HCT116, HT29, Colo 201, and SW620 were grown to 60 -70% confluenc y in sterile 6 -well tissue culture plates and allowed to attach for 48 hours. Cells were serum - starved for 24 hours, and were incubated with indicated dosage of cholecalciferol (Sigma -Aldrich, St. Louis, MO) (dissolved in isopropanol) for 24 hours prior to indicated dosage of deoxycholic acid (DCA) (Sigma -Aldrich, St. Louis, MO), dissolved in double -

distilled water (ddH 2O), for specified time points. Upon completion of treatments, cells were lysed in RIPA lysis buffer (0.05 M Tris, pH 7.5, 0.15 M NaCl, 1% T riton X -100, 1% SDS, 1% NP -40, and 5% phenylmethylsulfonyl fluoride (PMSF)) containing protease inhibitors (protease inhibitor cocktail) (Sigma -Aldrich, St. Louis, MO) and phosphatase inhibitors (10 mM NaF, 10 mM sodium pyrophosphate, and 10 mM NaVO 4). Who le cell lysate was then centrifuged at 13,000 rpm for 5 minutes at 4ºC. Equal amounts of proteins (30 µg) were separated by 7% or 10% SDS -PAGE for 90 minutes at 135 V under reducing conditions and transferred onto PVDF membranes (Invitrogen, Carlsbad, CA) for 2 ½ hours. After blocking with 5% non -fat dry milk in 1X PBS -T for one hour at room temperature, membranes were incubated with primary antibodies directed against the protein of interest. Primary antibodies (rabbit anti -β-catenin (1:1000), rabbit anti - phosph o-β-catenin (Ser33/37/Thr41) (1:1000), rabbit anti -phospho -EGF receptor 75

(Tyr845) (1:1000), mouse anti -β-actin (1:5000) [Cell Signaling Technology, Danvers, MA]; mouse anti -E-cadherin (1:2500) [BD Biosciences, San Jose, CA]; mouse anti - phospho -β-catenin (Ty r654) (1:100) [Abcam, Cambridge, MA]; rabbit anti -EphB2 (N - term) (1:100) [Abgent, San Diego, CA] were diluted as recommended in company protocol and incubated at 4ºC overnight with gentle agitation. This incubation was followed by washing 3X with 1X PBS -T for 10 minutes each. Membranes were incubated with anti -mouse or anti -rabbit secondary antibody conjugated with horseradish peroxidase (GE Healthcare, Piscataway, NJ) for one hour at room temperature with gentle agitation. This incubation was followed by w ashing 3X with 1X PBS -T for 10 minutes each. Chemiluminescence kits were obtained from PerkinElmer Life Sciences, Inc., Waltham, MA. These experiments were independently conducted three times.

Wound Migration Assay Colon cancer cell migration was assesse d using a wound migration assay as described previously [76] , with slight modifications. Briefly, cell lines HT29, and SW620 w ere seeded at 200,000 cells per well in media containing 10% FBS and 1X Pen/Strep in sterile 6 -well tissue culture plates and grown to 60 -70% confluency as described previously. After 48 hours incubation, cell culture media was aspirated from wells and ser um -free media was added to each well. Cells were serum -starved for 24 hours, and then were incubated with cholecalciferol (Sigma -Aldrich, St. Louis, MO) for 24 hours. Media was aspirated from wells, and wells were washed once with 1X PBS -T. After using a s patula to wound the confluent cells, each well was washed once with 1X PBS -T to remove any detached cells. Serum -free media was added to each well, and cholecalciferol, sodium deoxycholate (DCA), and sodium cholate (CA) (Sigma -Aldrich, St. Louis, MO) treat ments were added simultaneously to designated wells. Images were obtained at 0, 10, and 24 hours after initial wound using a MiniVid Camera (LW Scientific, Inc., Lawrenceville, GA) under a microscope at 10X magnification. These experiments were independent ly conducted three times. 76

Immunofluorescence for E -cadherin and EphB2 Colon cancer cell lines HT29 and SW620 were seeded onto cover slips in sterile 6 -well tissue culture plates. Wells were incubated with pre -warm 3% para -formaldehyde 1X PBS -T (Sigma -Al drich, St. Louis, MO) for 12 minutes on an orbital shaker. Para - formaldehyde was aspirated, and wells were washed three times with 1X PBS for 5 minutes each time. For detection of E -cadherin, wells were incubated with 0.5% Triton X-100 (Sigma -Aldrich, St. Louis, MO) in 1X PBS -T for 10 minutes at room temperature with gentle agitation, then were washed twice with 1X PBS -T for 5 minutes each time. For detection of EphB2, incubation with Triton X -100 was omitted. After blocking with 5% BSA for 30 minutes at ro om temperature, cells were washed twice with 1X PBS for 5 minutes each time. Cells were incubated with primary antibodies (mouse anti -E-cadherin (1:100) [BD Biosciences, San Jose, CA]; rabbit anti -EphB2 (1:100) [Abgent, San Diego, CA]) with gentle agitati on at room temperature for 1 hour, followed by washing 5 times with 1X PBS. Cells were incubated with secondary antibodies (goat anti -mouse Alexa Fluor 488 (3 µl/1 ml); goat anti -rabbit Alexa Fluor 594 (3 µl/1 ml) [Invitrogen, Carlsbad, CA]) for 1 hour at room temperature, followed by washing 5 times with 1X PBS -T. Nuclei were stained and slides mounted in DAPI, and images were obtained with a Nikon fluorescent microscope under 40X magnification.

RT -PCR for human EphB2 and EphA2 Total RNA was extracted fr om confluent T75 tissue culture flasks for cell lines HT29, SW620, and Colo 201 using FastRNA Pro Green Kit (Qbiogene, Inc.) according to the manufacturer's instructions. The RNA was then further purified with RNAeasy kit (Qiagen). The SuperScript III Firs t-Strand Synthesis System (Invitrogen) was used to synthesize first -strand cDNA from 1.5 µg purified total RNA. PCRs were run for 25 cycles. The primers for cDNA PCR were sense 5' - TCCATCTGGGACTTTCAAGG -3’ and antisense 5' - GACCACGACAGGGTAATGCT – 3’ which targeted a 554 bp product of human EphB2 gene. Human β-actin -- sense 5' -GCTCGTCGTCGACAACGGCT C -3', antisense 5' -CAAACATGATCTGGG TCATCTTCTC -3' which targeted a 353 bp 77

product, were also amplified at the same condition as the internal control during the PCR procedure. PCR products were analyzed on 1.5% agarose gels stained with ethidium bromide.

Cytoplasmic and Nuclear Extracts Cell lines SW620, HT29, and Colo 201 were grown to confluency in sterile T75 tissue culture flasks in the conditions described abov e. Samples were trypsinized from the flask and collected in 15 mL centrifuge tubes, then were spun at 800 rpm for 3 minutes at 4ºC to repellet the cells. After washing cells once with cold 1X PBS -T, cells were centrifuged again to repellet the sample. Pell et was resuspended in 1 mL cold 1X PBS -T and transferred to an Eppendorf tube. Samples were centrifuged at 2,000 rpm for 5 minutes at 4ºC to repellet, then were resuspended by vigorous pipetting in 200 µL sucrose buffer

(ddH20, 0.5 M sucrose, 3.1 mM CaCl 2, 2 mM MgAc, 0.1 mM EDTA, 1 mM DTT, 0.5 mM PMSF [Sigma -Aldrich, St. Louis, MO]) + NP -40 (sucrose buffer + 0.5% NP -40 [Sigma -Aldrich, St. Louis, MO]). After incubation on ice for 5 minutes, nuclei were pelleted by centrifugation at 5,000 rpm for 5 minutes at 4ºC. Supernatant, considered to be the cytoplasmic fraction, was transferred to a new tube and stored at -20ºC until use. Pelleted nuclei were resuspended in 1 mL sucrose buffer, and were then centrifuged at 5,000 rpm for 5 minutes. Supernatant of the sam ple was discarded, removing leftover cytoplasmic contaminants. Pelleted nuclei were resuspended in 50 µL low -salt buffer

(ddH20, 20 mM HEPES (pH 7.9), 25% glycerol, 1.5 mM MgCl 2, 20 mM KCl, 0.2 µM EDTA, 1 mM DTT, and 0.5 mM PMSF). To each sample, 0.2X volu me high -salt buffer

(ddH20, 20 mM HEPES (pH 7.9), 25% glycerol, 1.5 mM MgCl 2, 0.8 mM KCl, 0.2 mM EDTA, 1% NP -40, 1 mM DTT, and 0.5 mM PMSF) was added and tube was gently flicked. Continual addition of 0.2X volume and flicking of tube occurred until 1X volu me (50 µL) was added and nuclei began to shrink and viscosity increased. Samples were incubated on a rotary platform at 4ºC for 20 minutes, then were centrifuged at 13,000 rpm for 15 minutes at 4ºC. The supernatant was considered to be the nuclear fraction , and samples were stored at -20ºC until use. 78

IV. Results Deoxycholic acid and cholecalciferol induce morphological changes of colorectal cancer cells. It has been shown that deoxycholic acid (DCA) modulates the lipid and protein composition of the p lasma membrane [7,11,12] . Figure 1 describes the effects of DCA on the morphology of SW620 colon cancer cells. Cells were fixed with paraformaldehyde and examined under a microscope at 10X magnification. Control cells exhibited a healthy, fibroblast -like phenotype with distinct attachments foci to the s lide. Treatment with 0.1 µM cholecalciferol (CCF) for 24 hours induced a slight rounding. Upon treatment with 250 µM DCA for 10 hours, cells displayed pronounced rounding and shrinkage. Interestingly, DCA -treated cells also displayed short, spiky membra ne protrusions, possibly indicating the formation of filopodia. When cells were pre -treated with CCF then exposed to DCA, cells exhibited a phenotype similar to that of the cholecalciferol -only treated cells, regaining an epithelial phenotype. This data suggests that cholecalciferol treatment helps maintain an epithelial phenotype, even in the presence of DCA.

Cholecalciferol protects DCA -induced migration of colorectal cancer cells As loss of an epithelial phenotype and gain of mesenchymal characte ristics is known to enhance migration and metastasis cell morphology [8,19] , and since cell morphol ogy was altered in response to DCA and cholecalciferol treatment, we next sought to identify whether these morphological changes were associated with changes in migration. Figure 2 describes migrational changes of SW620 colorectal cancer cells in response to cholecalciferol and DCA. Using a wound migration assay, we observed very little migration from the initial wound in control cells after 10 hours, as well as in cells treated with 250 µM of a control non -tumorigenic bile acid, cholic acid (CA). However, upon treatment of cells with 250 µM DCA, we observed a pronounced increase in cell migration from the initial wound after 10h. Pre -treatment of cells with 0.1 µM cholecalciferol restored migration to that seen in control cells, suggesting that 79

cholecalci ferol protects against DCA -induced increase in migration of colorectal cancer cells .

Cholecalciferol restores DCA -induced loss of EphB2 protein expression in colon cancer cells. To determine localization and expression of EphB2 protein, we used immunoflu orescence to observe changes in this protein in response to cholecalciferol and DCA, as depicted in Figure 3. In SW620 cells, control cells and cells pre -treated with cholecalciferol displayed strong membranous EphB2 expression. However, treatment with 25 0 µM DCA for 10 hours dramatically decreased expression of EphB2 at the membrane, while expression was restored to basal levels when cells were pre -treated with cholecalciferol. In addition, control cells exhibited a punctated pattern of EphB2 expression a t the membrane, while cells treated with DCA for 10 hours lost this punctated pattern, in addition to loss of overall expression, whereas cells pre -treated with cholecalciferol favored the original punctated pattern of expression. This suggests that in add ition to restoring expression of EphB2, pre -treatment with cholecalciferol favors clustering of EphB2 receptors , possibly in lipid rafts , which is associate d with increased activity [77] .

Cholecalciferol restores DCA -induced loss of EphB2 protein expression in colon cancer cells. As HT29 and HCT116 cells express very low levels of EphB2 protein, SW620 cells were treated with DCA, cholecalciferol, or the combinat ion for 0 and 10 hours, and were probed for EphB2 ( Figure 4). Control cells demonstrated strong expression of EphB2, with no change in expression with cholecalciferol treatment. DCA strongly reduced expression of EphB2 at 10h, and combination treatment o f DCA and cholecalciferol restored EphB2 expression to normal at 10h. In order to better elucidate the response of cells pre -treated with cholecalciferol to DCA, we conducted a dose -response experiment with increasing concentrations of cholecalciferol in c ombination with DCA. We observed that cholecalciferol was able to restore EphB2 expression to basal lev els in a dose - 80

dependent manner. Furthermore, use of a specific EGFR inhibitor, ZD1839, increased expression of EphB2 as compared to control and DCA -only treatment. Together, these results suggest that the protective effect of cholecalciferol on loss of EphB2 expression in response to DCA may occur via manipulation of plasma membrane -associated signaling proteins, such as EGFR.

Cholecalciferol protects ag ainst DCA -induced loss of RNA expression of EphB2 in a dose -dependent manner. To further confirm the results of the previous experiment, we used RT -PCR to evaluate RNA expression of EphB2 in four cell lines: HCT116, HT29, SW620, and Colo 201, as depicted i n Figure 5A and Figure 5B. Control cells express high levels of EphB2, but EphB2 expression decreases with 250 uM DCA treatment. Cholecalciferol treatment at 250 uM for 24 hours displayed RNA expression of EphB2 similar to control cells. Interestingly, ch olecalciferol pre -treatment of cells for 24 hours at 0.1 uM in combination with DCA was able to protect against DCA -induced loss of EphB2 RNA. This effect was observed best in SW620 colorectal cancer cells. This data confirms previous observations, that ch olecalciferol is able to protect against DCA -induced loss of EphB2 expression.

DCA decreases membranous E -cadherin staining in a time -dependent manner. As DCA is known to decrease cell -cell adhesion [16] and vitamin D is known to promote cell -cell adhesion [18] , we used immunofluorescence to determine relative protein expression of E -cadherin in HT29 cells ( Figure 6 ). Upon treatment of HT29 cells with 250 µM DCA, cells expressed a time -depe ndent decrease in expression of E -cadherin, with a slight decrease in expression at 8 hours, and near complete loss of expression of EphB2 at 24 hours treatment with DCA. Upon treatment with cholic acid, a primary bile acid, cells displayed normal express ion of E -cadherin in cells treated with 250 µM CA. 81

DCA decreases expression of E -cadherin with subsequent induction of β-catenin expression in a time - and dose -dependent manner. To determine if overall expression of adhesion proteins was altered in the presence of DCA and cholecalciferol, we used Western blot analysis to determine expression of E - cadherin ( Figure 7). Based on this data, we exposed cells to a time -course of treatment with 250 µM DCA. At 2h, expression of E -cadherin starts to decline, with expression almost completely absent at 16h and 24h DCA treatment. However, treatment with cholic acid does not exert this effect of E -cadherin expression.

Batlle et al. and van de Wetering et al. identified EphB2 as a direct transcriptional target of β-catenin/TCF signaling [34,36] . In addition, loss of E -cadherin expression is suggested to be due at least in part to β-catenin/TCF signaling. Based on this observation, in order to understand the loss of EphB2 expression observed in response to DCA treatment, we assessed β-catenin expression and its localization within colon cancer cells. Assessment of total cell lysat e revealed an increase in expression of β-catenin at 250 and 500 µM DCA treatment for 16h ( Figure 8). At the membrane, if β-catenin is phosphorylated, it dissociates from E -cadherin and translocates to the cytoplasm [24 -27] . One kinase de monstrated to be responsible for phosphorylation of β-catenin is epidermal growth factor receptor (EGFR), which phosphorylates β-catenin on tyrosine residue 654 (Tyr 654 ) [25] . Using a specific EGFR inhibitor, ZD1839 [78] , we demonstrated that treatment of HT29 cells with ZD1839 strongly decreased Tyr 654 -phosphorylation of β- catenin even in the presence of DCA, which is known to activate EGFR activity [11] and was demonstrated to increase Tyr 654 -phosphorylation of β-catenin (data not shown) . In order to determine the time required for up -regulation of total β-catenin expression, we treated cells with a time -course of 250 µM DCA, and observed that the increase in β- catenin expression occurred at around 2h treatment with DCA ( Figure 7 ).

Chole calciferol exerts protective effects against DCA by modulation of lipid rafts in the plasma membrane. 82

As DCA has been shown to perturb the plasma membrane, and as vitamin D has been shown to interact with the membrane as well, we sought to identify the effects of these compounds on plasma membrane dynamics. Specifically, as both EphB2 and E -cadherin can localize to lipid rafts [83,84] , we used cholera toxin, which binds to the lipid raft - associated ganglioside GM1, to analyze effects of DCA and cholecalciferol on lipid r aft formation in SW620 colorectal cancer cells. Control cells expressed strong membranous GM1 staining, as well as a distinct punctated pattern of expression ( Figure 9). Upon treatment of 0.1 uM cholecalciferol for 24 hours, staining of GM1 remained consta nt, while punctated pattern became more pronounced, possibly indicating increased lipid raft formation . However, treatment with DCA strongly decreased GM1 expression, as well as disrupted the punctated pattern of expression seen in control and cholecalcife rol -treated cells. Upon pre -treatment with 0.1 uM cholecalciferol for 24 hours followed by 250 uM DCA for 10 hours, cells expressed similar GM1 staining as compared to control. Furthermore, cholecalciferol protected cells from DCA -induced loss of GM1 punct ated pattern. 83

V. Discussion Colorectal cancer is the second leading cause of cancer death for both men and women in the United States [1] . The progression of cancer from original transformed cell to benign adenoma to malignant carcinoma is hallmarked with many changes in cell genotype and phenotype, due to progressive accumulation of mutations and influence of the colon environment. Recent data strongly implicates the loss of cell -cell adhesion and participation in the epithelial -to -mesenchymal transition as a critical step in the acquisition of malignancy by transformed cells [8,19] . Importantly, expression of proteins involved in cell adhesion, including E -cadherin, EphB2, and β-catenin, have been shown to be dramatically influenced by luminal factors such as bile acids and nutrients [15,22] . Specifically, deoxycholic acid, a cholesterol derivative produced by microbial met abolism of bile acids, has been demonstrated to decrease expression of E - cadherin, induce loss of cell -cell adhesion, and increase invasion and metastasis in colorectal cancer cells [16 -18] . Conversely, vitamin D has been shown to increase expression of E -cadherin and promote reformation of adherens junctions at the plasma membrane [18] . Of interest, EphB2, a receptor tyrosine kinase involved in cell adhesion and cell segregation, is silenced in advanced col orectal carcinoma [48] . Furthermore, loss of EphB2 expression is associated with increased invasion and metastasis, acceleration of cancer progression, and poor patient prognosis [48] . However, th e effects of DCA and vitamin D on EphB2 expression are currently unknown.

For the present study, as DCA and vitamin D appear to exert opposite effects on cell -cell adhesion, we sought to identify three components of this interaction. First, we sought to analyze the effect of combined DCA and vitamin D (cholecalciferol) treatment on markers of invasion and metastasis, including EphB2 and E -cadherin, in human colorectal cancer cells. Furthermore, we sought to identify the location of this interaction, postu lating that the plasma membrane was likely to be affected, as both DCA and cholecalciferol are cholesterol derivatives and are easily incorporated into the membrane. 84

Finally, we sought to identify the role of EGFR in these interactions, as DCA and vitamin D have been shown to increase or decrease activity of this receptor, respectively [11] .

Our re sults demonstrate a powerful protective effect of cholecalciferol treatment on EphB2 expression with combined DCA treatment, repeatedly demonstrating its ability to restore EphB2 to basal expression at both RNA and protein levels in multiple cell lines. In terestingly, this effect was not seen with cholecalciferol treatment alone, suggesting that cholecalciferol functions best as a protective agent at the membrane, rather than at the transcriptional level in the nucleus. Together, these results highlight thr ee critical findings of this study. First, DCA treatment alone strongly decreased expression of EphB2; however, 24 -hour pre -treatment with cholecalciferol in combination with DCA restored EphB2 expression as compared to control. This effect was demonstrate d to be dose -dependent in SW620 cells, showing a gradual increase in both RNA and protein expression in response to increasing concentrations of cholecalciferol. Second, the interactions between cholecalciferol and DCA occur, at least in part, at the plasm a membrane. Using cholera toxin to identify manipulation of the plasma membrane, we showed that DCA treatment strongly disrupted GM1 localization at the plasma membrane, possibly indicating disruption of lipid raft formation, though further studies are req uired to confirm this effect . However, with cholecalciferol pre -treatment, DCA failed to induce perturbation of the membrane as previously seen. Finally, using a specific EGFR inhibitor, ZD1839, we demonstrated that the effects of cholecalciferol may be me diated, at least in part, through manipulation of EGFR. Combined treatment of DCA, cholecalciferol, and ZD1839 induced higher expression than seen in control or in cells treated with only DCA.

DCA treatment has been demonstrated to increase nuclear local ization of β-catenin, as well as increased transcriptional effects by β-catenin/TCF [16] . As vitamin D possesses a nuclear receptor (VDR) and dimerizes with RXR to form its own transcription -regulating complex [51,52] , it is therefore possibl e that vitamin D treatment modulates transcription 85

within the nucleus. Furthermore, vitamin D has previously been demonstrated to compete with TCF for binding to β-catenin [18] , which may represent a mechanism whereby vitamin D exerts its well -known effect of inducing cell differentiation and inhibiting proliferation [18,53,63,66] , since inhibiting transcription of downstream targets of β- catenin/TCF, many of which are involved in cell cycle [16,28,32] , would theoretically lead to cell cycle arrest.

However, based on our results, this is not likely to be the mechanism responsible for the effects of cholecalciferol seen in this study, for several reasons. First, the primary cell line used in this study, SW620, is a metastatic derivative of the SW480 adenocarcinoma cell line, and does not express VDR [18,53] . Second, the form of vitamin D used in this study was cholecalciferol, not calcitriol, th e most active form of vitamin D in the body, which is also the most active VDR ligand [51,56] . Instead, we propose a mechanism whereby cholecal ciferol is incorporated into the plasma membrane, thereby interfering with the tumor -promoting effects of DCA, such as increased nuclear localization of β- catenin and increased transcriptional activity of β-catenin/TCF , which was confirmed in our study usi ng cholera toxin and biotinylation studies . This may be accomplished in several ways. First, cholecalciferol, as a cholesterol derivative, may simply compete with DCA for incorporation into the plasma membrane, creating a physical barrier. This theory is s upported by the fact that only pre -treatment with cholecalciferol, and not simultaneous addition of cholecalciferol with DCA, protects against DCA -induced loss of EphB2 (unpublished data), suggesting that pre -treatment may saturate the membrane with choles terol and prevent DCA from exerting an effect. Second, vitamin D has been shown to decrease both expression and activity of EGFR at the plasma membrane [71] . Furthermore, in our study, using a specific EGF R inhibitor, ZD1839, in combination with DCA and cholecalciferol treatment, we observe an increase in expression of EphB2 above that seen in response to DCA and cholecalciferol treatment only, suggesting that inhibition of EGFR favors restoration of EphB2. Finally, the presence of membrane - associated rapid -response steroid (MARRS) proteins has recently been identified, and 86

represents a possible mechanism for membrane -mediated effects of vitamin D [70] .

MARRS proteins have been identified for both calcitriol and 24,25 -(OH) 2D3 at the membrane; however, no MARRS proteins have been identified as responsive to cholecalciferol. Though vitamin D hydroxylase levels typically decrease during colorectal cancer [51] , research has shown that colonocytes are capable of making calcitriol, the active form of vitamin D [51,57] . Therefore, MARRS proteins may represent an alternate mechanism for vit amin D -mediated effects, in addition to those mediated through the VDR [70] .

Of interest, a recent study conducted by van de Wetering et al. identified EphB2 as a target of β-catenin/TCF -mediated transcription using a dominant -negative form of TCF (dn -TCF ), which was unable to bind DNA U se of dn -TCF resulted in decreased expression of EphB2, suggesting that active β-catenin/TCF signaling results in increased transcription of EphB2 [34] . However, this study stands in stark contrast to our results. DCA is known to increase nuclear β-catenin as well as β-catenin/TCF -mediated transcription [16] ; thus, if β-catenin/TCF increases transcription of EphB2, as indicated by van de Wetering et al [34] ., DCA treatment should increase expression of EphB2. However, we have repeatedly demonstrated that DCA significantly decreases EphB2 at both the protein and RNA levels through multiple techniques. The Tak1/Tab/Nlk signaling pathway within the cell may prov ide a possible mechanism for both observations.

Briefly, Tak1 is related to MAPKKK proteins, and i s activated in response to ligands such as transforming growth factor (TGF) -β, interleukin (IL) -1, lipopolysaccharides (LPS), ephrin B1, and Wnt proteins [79,80] , among others. Interaction with Tab1, a membrane -bound protein, results in activat ion of Tak1, as does ubiquitination through a Ubc13/Uev1A -mediated mechanism [81] . Upon activation, Tak1 is involved in several signal transduction pathways, including nuclear factor (NF) -kB, activation of c -Jun , p38 α, and, of interest to this study, Nemo -like kinase (Nlk) signaling [79,80] . Nlk, in 87

turn, has been found to negatively regulate β-catenin/TCF signaling by phosphorylat ion of two serine/threonine residues located in the central region of TCF [80] , preventing TCF from binding to DNA [79,80] . Therefore, activation of Nlk leads to decreased transcription of β-catenin/TCF target genes such as EphB2. In add ition, Wnt proteins can activate Tak1 and result in Nlk activation [80 ]. Of particular interest to the present study, DCA has been shown to enhance Wnt signaling in colorectal cancer, as DCA destabilizes β-catenin at the membrane, leading to cytosolic and ultimately nuclear translocation [16] . Theoretically, Wnt signaling can take one of two directions: in the canonical pathway, it can lead to stabilization and ultimately nuclear localization and signaling of β-catenin/TCF [79] . The Tak1/Nlk pathway, however, provides an alternative to the canonical Wnt signaling pathway, and may function to regulate the canonical signaling pathway [79] . This could provide a mechanism whereby DCA downregulates EphB2 expression. However, several problems exist with this theory.

First, other target genes of β-catenin/TCF signaling include uPA/ uPAR, MMP -7, cdc44, and cyclin D1 [16] . If DCA initiates the Tak1/Nlk pathway, transcription of all target genes, many of which are involved in cell proliferation and migration [16] , would theoretically be impaired. Furthermore, these results would suggest that DCA does, in fact, protect against cancer progression, which numerous cell culture, animal, and epid emiological studies firmly indicate is incorrect [7,9,11 -14] . However, the aforementioned study by v an de Wetering et al. identifying EphB2 as a transcriptional target of β-catenin/TCF was conducted in LS174T and DLD -1 cells, suggesting that this effect may not be universal among colorectal cell lines.

The authors of this study chose cholecalciferol be cause this is the form of vitamin D provided in the diet. In addition, as our primary cell line, SW620, does not express VDR [53] , calcitriol would be unable to init iate signaling associated with activation of VDR. Second, multiple cell lines, HCT116, HT29, SW620, and Colo 201, were used . Interestingly, while Colo 201 and SW620 cells display strong EphB2 expression, a 88

previous study by using immunohistochemical analys is found that both HT29 and HCT116 express very low levels of EphB2, with low or absent nuclear β-catenin expression [46] . This may be explained by tumor origin: a tumor that arose from a transformed, undifferentiated crypt cell would e xpress high EphB2, whereas a tumor derived from a transformed cell higher up on the villus would express very low levels of EphB2, and would rather express high levels of ephrin ligand [39,44] . Complementary patterns of expr ession of these proteins ensure , through bi -directional signaling and subsequent repulsion [39] , that proliferative cells and differentiated cells remain necessarily confined to their individual anatomical locations to prevent intermingling of these cell populations.

Our study has several strong implications in terms of prevention and treatment of colorectal cancer. Epidemiological data has repeatedly demonstrated that low serum vitamin D was associated with increased risk of colorectal cancer [62] , as was living in a location with low ultraviolet ray exposure from the sun, which impairs vitamin D synthesis [59] . However, clinical trials using either calcitriol or calcitriol analogs have had mixed success [53,63,64,66] . Based on our data, the varied outcomes of these trials may be due to the fact that VDR expression is lost in most advanced colorectal carcinomas [53] , and therefore the drugs may fail to have an effect, as they essentially have no target. In addition, we have demonstrated in this study that timing of treatment is critical to success. Specifically, 10 hours treatment with cholecalciferol yielded maximal positive effect on EphB2 expression, and time -course experiments with DCA tre atment indicate that E -cadherin expression starts to decrease dramatically between 5 and 10 hours.

Furthermore, wound migration assays show increased migration between 5 and 10 hours as compared to control and shorter time points. Though the reason behin d these observations may be multi -faceted, these effects are likely due at least in part to the fact that E -cadherin reaches steady state expression upon cell -cell contact, and therefore 89

recycles, between 5 to 10 hours [82] , providing a possible window for maximal effectiveness of treatment. Interestingly, both E -cadherin and EphB2 have been suggested to localize to lipid rafts in order to maximize signaling efficie ncy [83,84] ; therefore, perturbation of E -cadherin by DCA could theoretically disrupt EphB2 localization as well. Ideally, the entire population would meet their daily requirements for vitamin D, which has been correlated with a decreased incidence of colorectal cancer. However, in the incidence of colorectal cancer, our data provides strong evidence for use of vitamin D, and specifically cholecalciferol or similar analogs, as an adjuvant therapy for advanced colorectal carcinoma, even in the absence of nuclear VDR.

Clearly, elucidation of the sig naling pathways involved with DCA, cholecalciferol, and EphB2 represents the largest obstacle in understanding the effects observed in this study. In addition, further work remains to be done on the exact nature of the manipulation of the plasma membrane o bserved in response to cholecalciferol and vitamin D. Finally, confirmation of co -localization of EphB2 and E -cadherin in lipid rafts remains a desirable target, as this may provide an explanation to the similar effects on the two proteins in response to c holecalciferol and vitamin D.

In summary, pre -treatment of human colorectal cancer cells with cholecalciferol protected against DCA -induced loss of cell adhesion and migration proteins EphB2 and E-cadherin. These effects were seen on several levels, incl uding transcriptional modulation of RNA levels, translational changes in protein expression, and morphological responses to treatment, including decreased migration and increased adhesion protein expression. Interestingly, cholecalciferol treatment alone d id not increase levels of EphB2 as compared to control, suggesting that cholecalciferol functions best as a protective agent against pro -tumorigenic changes induced by DCA treatment, rather than as an anti -tumor agent in itself. Furthermore, these protecti ve effects of cholecalciferol appear to be exerted at the membrane, and at least in part through modulation of EGFR activity, as was demonstrated using a specific EGFR 90

inhibitor, ZD1839. These results suggest that colorectal cancer patients may benefit fro m adjuvant therapy with cholecalciferol in addition to standard chemotherapeutive and radiotherapeutive techniques. 91

VI. References 1. Debruyne, P.R., Bruyneel, E.A., Karaguni, I.M., Li, X., Flatau, G., Muller, O., Zimber, A., Gesp ach, C. and Mareel, M.M. (2002) Bile acids stimulate invasion and haptotaxis in human colorectal cancer cells through activation of multiple oncogenic signaling pathways. Oncogene , 21 , 6740 -50. 2. Yui, S., Saeki, T., Kanamoto, R. and Iwami, K. (2005) Chara cteristics of apoptosis in HCT116 colon cancer cells induced by deoxycholic acid. J Biochem (Tokyo) , 138 , 151 -7. 3. Glinghammar, B., Inoue, H. and Rafter, J.J. (2002) Deoxycholic acid causes DNA damage in colonic cells with subsequent induction of caspases , COX -2 promoter activity and the transcription factors NF -kB and AP -1. Carcinogenesis , 23 , 839 - 45. 4. Jean -Louis, S., Akare, S., Ali, M.A., Mash, E.A., Jr., Meuillet, E. and Martinez, J.D. (2006) Deoxycholic acid induces intracellular signaling through me mbrane perturbations. J Biol Chem , 281 , 14948 -60. 5. Pai, R., Tarnawski, A.S. and Tran, T. (2004) Deoxycholic acid activates beta - catenin signaling pathway and increases colon cell cancer growth and invasiveness. Mol Biol Cell , 15 , 2156 -63. 6. Coluccia, A. M., Benati, D., Dekhil, H., De Filippo, A., Lan, C. and Gambacorti - Passerini, C. (2006) SKI -606 decreases growth and motility of colorectal cancer cells by preventing pp60(c -Src) -dependent tyrosine phosphorylation of beta - catenin and its nuclear signaling. Cancer Res , 66 , 2279 -86. 7. Palmer, H.G., Gonzalez -Sancho, J.M., Espada, J., Berciano, M.T., Puig, I., Baulida, J., Quintanilla, M., Cano, A., de Herreros, A.G., Lafarga, M. and Munoz, A. (2001) Vitamin D(3) promotes the differentiation of colon carcinoma cells by the induction of E -cadherin and the inhibition of beta -catenin signaling. J Cell Biol , 154 , 369 -87. 8. Garland, C.F. and Garland, F.C. (1980) Do sunlight and vitamin D reduce the likelihood of colon cancer? Int J Epidemiol , 9, 227 -31. 9. Iseki, K ., Tatsuta, M., Uehara, H., Iishi, H., Yano, H., Sakai, N. and Ishiguro, S. (1999) Inhibition of angiogenesis as a mechanism for inhibition by 1alpha - hydroxyvitamin D3 and 1,25 -dihydroxyvitamin D3 of colon carcinogenesis induced by azoxymethane in Wistar r ats. Int J Cancer , 81 , 730 -3. 10. Tong, W.M., Kallay, E., Hofer, H., Hulla, W., Manhardt, T., Peterlik, M. and Cross, H.S. (1998) Growth regulation of human colon cancer cells by epidermal growth factor and 1,25 -dihydroxyvitamin D3 is mediated by mutual mo dulation of receptor expression. Eur J Cancer , 34 , 2119 -25. 11. Ingles, S.A., Wang, J., Coetzee, G.A., Lee, E.R., Frankl, H.D. and Haile, R.W. (2001) Vitamin D receptor polymorphisms and risk of colorectal adenomas (United States). Cancer Causes Control , 12, 607 -14. 12. Hahm, E.R., Gho, Y.S., Park, S., Park, C., Kim, K.W. and Yang, C.H. (2004) Synthetic curcumin analogs inhibit activator protein -1 transcription and tumor - induced angiogenesis. Biochem Biophys Res Commun , 321 , 337 -44. 92

13. Nagengast, F.M., Gru bben, M.J. and van Munster, I.P. (1995) Role of bile acids in colorectal carcinogenesis. Eur J Cancer , 31A , 1067 -70. 14. Bates, R.C. and Mercurio, A.M. (2005) The epithelial -mesenchymal transition (EMT) and colorectal cancer progression. Cancer Biol Ther , 4, 365 -70. 15. Himanen, J.P., Rajashankar, K.R., Lackmann, M., Cowan, C.A., Henkemeyer, M. and Nikolov, D.B. (2001) Crystal structure of an Eph receptor -ephrin complex. Nature , 414 , 933 -8. 16. van de Wetering, M., Sancho, E., Verweij, C., de Lau, W., Oving , I., Hurlstone, A., van der Horn, K., Batlle, E., Coudreuse, D., Haramis, A.P., Tjon -Pon -Fong, M., Moerer, P., van den Born, M., Soete, G., Pals, S., Eilers, M., Medema, R. and Clevers, H. (2002) The beta -catenin/TCF -4 complex imposes a crypt progenitor phenotype on colorectal cancer cells. Cell , 111 , 241 -50. 17. Batlle, E., Henderson, J.T., Beghtel, H., van den Born, M.M., Sancho, E., Huls, G., Meeldijk, J., Robertson, J., van de Wetering, M., Pawson, T. and Clevers, H. (2002) Beta -catenin and TCF mediate cell positioning in the intestinal epithelium by controlling the expression of EphB/ephrinB. Cell , 111 , 251 -63. 18. Buda, A. and Pignatelli, M. (2004) Cytoskeletal network in colon cancer: from genes to clinical application. Int J Biochem Cell Biol , 36 , 7 59 -65. 19. Lilien, J. and Balsamo, J. (2005) The regulation of cadherin -mediated adhesion by tyrosine phosphorylation/dephosphorylation of beta -catenin. Curr Opin Cell Biol , 17 , 459 -65. 20. Haraguchi, K., Nishida, A., Ishidate, T. and Akiyama, T. (2004) Ac tivation of beta -catenin -TCF -mediated transcription by non -receptor tyrosine kinase v -Src. Biochem Biophys Res Commun , 313 , 841 -4. 21. Behrens, J. (2005) The role of the Wnt signalling pathway in colorectal tumorigenesis. Biochem Soc Trans , 33 , 672 -5. 22. Mendelsohn, J. and Baselga, J. (2006) Epidermal growth factor receptor targeting in cancer. Semin Oncol , 33 , 369 -85. 23. American Cancer Society, I. (2006) SOS - Colorectal Cancer. 24. Debruyne, D., Oliveira, M.J., Bracke, M., Mareel, M. and Leroy, A. (2006 ) Colon cancer cells: pro -invasive signalling. Int J Biochem Cell Biol , 38 , 1231 -6. 25. Brabletz, T., Jung, A., Reu, S., Porzner, M., Hlubek, F., Kunz -Schughart, L.A., Knuechel, R. and Kirchner, T. (2001) Variable beta -catenin expression in colorectal canc ers indicates tumor progression driven by the tumor environment. Proc Natl Acad Sci U S A , 98 , 10356 -61. 26. Guo, D.L., Zhang, J., Yuen, S.T., Tsui, W.Y., Chan, A.S., Ho, C., Ji, J., Leung, S.Y. and Chen, X. (2006) Reduced expression of EphB2 that parallel s invasion and metastasis in colorectal tumours. Carcinogenesis , 27 , 454 -64. 27. Cross, H.S., Bareis, P., Hofer, H., Bischof, M.G., Bajna, E., Kriwanek, S., Bonner, E. and Peterlik, M. (2001) 25 -Hydroxyvitamin D(3) -1alpha -hydroxylase and vitamin D receptor gene expression in human colonic mucosa is elevated during early cancerogenesis. Steroids , 66 , 287 -92. 93

28. Greenwald, P., Kelloff, G.J., Boone, C.W. and McDonald, S.S. (1995) Genetic and cellular changes in colorectal cancer: proposed targets of chemoprev entive agents. Cancer Epidemiol Biomarkers Prev , 4, 691 -702. 29. Shabahang, M., Buras, R.R., Davoodi, F., Schumaker, L.M., Nauta, R.J. and Evans, S.R. (1993) 1,25 -Dihydroxyvitamin D3 receptor as a marker of human colon carcinoma cell line differentiation a nd growth inhibition. Cancer Res , 53 , 3712 -8. 30. Mokady, E., Schwartz, B., Shany, S. and Lamprecht, S.A. (2000) A protective role of dietary vitamin D3 in rat colon carcinogenesis. Nutr Cancer , 38 , 65 -73. 31. Wong, N.A. and Pignatelli, M. (2002) Beta -cate nin -- a linchpin in colorectal carcinogenesis? Am J Pathol , 160 , 389 -401. 32. Mulholland, D.J., Dedhar, S., Coetzee, G.A. and Nelson, C.C. (2005) Interaction of nuclear receptors with the Wnt/beta -catenin/Tcf signaling axis: Wnt you like to know? Endocr Rev , 26 , 898 -915. 33. Cross, H.S., Bises, G., Lechner, D., Manhardt, T. and Kallay, E. (2005) The Vitamin D endocrine system of the gut -- its possible role in colorectal cancer prevention. J Steroid Biochem Mol Biol , 97 , 121 -8. 34. Farach -Carson, M.C. and Neme re, I. (2003) Membrane receptors for vitamin D steroid hormones: potential new drug targets. Curr Drug Targets , 4, 67 -76. 35. Matusiak, D., Murillo, G., Carroll, R.E., Mehta, R.G. and Benya, R.V. (2005) Expression of vitamin D receptor and 25 -hydroxyvitami n D3 -1{alpha} - hydroxylase in normal and malignant human colon. Cancer Epidemiol Biomarkers Prev , 14 , 2370 -6. 36. Hall, A. (1998) Rho GTPases and the actin cytoskeleton. Science , 279 , 509 -14. 37. Robinson, V., Smith, A., Flenniken, A.M. and Wilkinson, D.G. (1997) Roles of Eph receptors and ephrins in neural crest pathfinding. Cell Tissue Res , 290 , 265 - 74. 38. Bruckner, K. and Klein, R. (1998) Signaling by Eph receptors and their ephrin ligands. Curr Opin Neurobiol , 8, 375 -82. 39. Brambilla, R. and Klein, R. (1995) Telling axons where to grow: a role for Eph receptor tyrosine kinases in guidance. Mol Cell Neurosci , 6, 487 -95. 40. Clevers, H. and Batlle, E. (2006) EphB/EphrinB receptors and Wnt signaling in colorectal cancer. Cancer Res , 66 , 2 -5. 41. Smit, L., Baas, A., Kuipers, J., Korswagen, H., van de Wetering, M. and Clevers, H. (2004) Wnt activates the Tak1/Nemo -like kinase pathway. J Biol Chem , 279 , 17232 -40. 42. Ishitani, T., Ninomiya -Tsuji, J. and Matsumoto, K. (2003) Regulation of lymphoid enhancer fact or 1/T -cell factor by mitogen -activated protein kinase - related Nemo -like kinase -dependent phosphorylation in Wnt/beta -catenin signaling. Mol Cell Biol , 23 , 1379 -89. 43. Wang, C., Deng, L., Hong, M., Akkaraju, G.R., Inoue, J. and Chen, Z.J. (2001) TAK1 is a ubiquitin -dependent kinase of MKK and IKK. Nature , 412 , 346 -51. 94

44. Narisawa, T., Magadia, N.E., Weisburger, J.H. and Wynder, E.L. (1974) Promoting effect of bile acids on colon carcinogenesis after intrarectal instillation of N -methyl -N' -nitro -N-nitrosog uanidine in rats. J Natl Cancer Inst , 53 , 1093 -7. 45. Flynn, C., Montrose, D.C., Swank, D.L., Nakanishi, M., Ilsley, J.N. and Rosenberg, D.W. (2007) Deoxycholic acid promotes the growth of colonic aberrant crypt foci. Mol Carcinog , 46 , 60 -70. 46. Jubb, A.M ., Zhong, F., Bheddah, S., Grabsch, H.I., Frantz, G.D., Mueller, W., Kavi, V., Quirke, P., Polakis, P. and Koeppen, H. (2005) EphB2 is a prognostic factor in colorectal cancer. Clin Cancer Res , 11 , 5181 -7. 47. Zisch, A.H. and Pasquale, E.B. (1997) The Eph family: a multitude of receptors that mediate cell recognition signals. Cell Tissue Res , 290 , 217 -26. 48. Batlle, E., Bacani, J., Begthel, H., Jonkheer, S., Gregorieff, A., van de Born, M., Malats, N., Sancho, E., Boon, E., Pawson, T., Gallinger, S., Pals, S. and Clevers, H. (2005) EphB receptor activity suppresses colorectal cancer progression. Nature , 435 , 1126 -30. 49. Peters, U., McGlynn, K.A., Chatterjee, N., Gunter, E., Garcia -Closas, M., Rothman, N. and Sinha, R. (2001) Vitamin D, calcium, and vitamin D receptor polymorphism in colorectal adenomas. Cancer Epidemiol Biomarkers Prev , 10 , 1267 -74. 50. Kallay, E., Pietschmann, P., Toyokuni, S., Bajna, E., Hahn, P., Mazzucco, K., Bieglmayer, C., Kato, S. and Cross, H.S. (2001) Characterization of a vitamin D receptor knockout mouse as a model of colorectal hyperproliferation and DNA damage. Carcinogenesis , 22 , 1429 -35. 51. Stewart, D.B., Barth, A.I. and Nelson, W.J. (2000) Differential regulation of endogenous cadherin expression in Madin -Darby canine kidney cells by cell -cell adhesion and activation of beta -catenin signaling. J Biol Chem , 275 , 20707 -16. 52. Seveau, S., Bierne, H., Giroux, S., Prevost, M.C. and Cossart, P. (2004) Role of lipid rafts in E -cadherin -- and HGF -R/Met -- mediated entry of Listeria monocytogenes into host cells. J Cell Biol , 166 , 743 -53. 53. Gauthier, L.R. and Robbins, S.M. (2003) Ephrin signaling: One raft to rule them all? One raft to sort them? One raft to spread their call and in signaling bind them? Life Sci , 74 , 207 -16. 95

VII. Figures

A A B

C D

B A B

C D

Figure 1 96

DCA, 8h A Ctrl DCA, 8 h

CA, 24h DCA, 24h CA, 24h

B A B

C D

Figure 2 97

0h 10h

EphB2

β-actin

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)

EphB2

GAPDH

Figure 3 98

- + - + + + + + DCA (250 µM) - - 0.1 0 0.05 0.1 1 0.1 CCF (µM) ------+ ZD1839 (10 µM)0

HT29

EphB2 β-actin

SW620

EphB2 β-actin

Colo 201

EphB2 β-actin

HCT116

EphB2 β-actin

Figure 3 99

HT29

250 µM DCA l A o r C t n m h h h o 0 h h h h 6 4 4 C 3 1 2 4 8 1 2 2

E-cadherin

β−catenin

β-actin

SW620

250 µM DCA l A o r C t n m h h h o 0 h h h h 6 4 4 C 3 1 2 4 8 1 2 2

E-cadherin

β−catenin

GAPDH

HCT116

250 µM DCA l A o r C t n m h h h o 0 h h h h 6 4 4 C 3 1 2 4 8 1 2 2

E-cadherin

β−catenin

GAPDH

Figure 4 100

A Ctrl CCF

DCA CCF+DCA

B DCA UDCA DCA+VitD Ctr 2h 4h 8h 12h 2h 4h 8h 12h Biotin E-cad β-actin

Figure 5 101

- - 25 0 250 500 500 - - DCA (µM) ------500 500 CA (µM) - + - + - + - + ZD1839 (10 µM)

Anti -phospho -β-catenin (Y654)

Anti -β-catenin

Figure 6