A dissertation presented to. the faculty of. In partial fulfillment. of the requirements for the degree. Doctor of Philosophy. Venktesh S.

Transcription

1 Molecular Mechanisms of Circulating Tumor Cell Adhesion in Breast Cancer Metastasis A dissertation presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Doctor of Philosophy Venktesh S. Shirure May Venktesh S. Shirure. All Rights Reserved.

2 2 This dissertation titled Molecular Mechanisms of Circulating Tumor Cell Adhesion in Breast Cancer Metastasis by VENKTESH S. SHIRURE has been approved for the Department of Chemical and Biomolecular Engineering and the Russ College of Engineering and Technology by Monica M. Burdick Assistant Professor of Chemical and Biomolecular Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

3 3 ABSTRACT SHIRURE, VENKTESH S., Ph.D., May 2013, Chemical Engineering Molecular Mechanisms of Circulating Tumor Cell Adhesion in Breast Cancer Metastasis Director of Dissertation: Monica M. Burdick During hematogenous metastasis, tumor cells dissociate from a primary tumor, migrate through the tissue space, and enter the circulatory system. The circulating tumor cells (CTCs), once bloodborne, travel to distant sites, where they adhere to the endothelial cells lining the vessel wall, potentially extravasate, and form secondary tumors if directed by niche factors. Determining the mechanisms of each of these steps can provide insights into novel diagnostics and therapeutics for cancer. Of particular interest is elucidating the molecular mechanisms by which metastatic cells adhere to the endothelium while resisting the disruptive shear exerted by the blood flow. We hypothesized that breast cancer cell adhesion is mediated by interaction of endothelial E- selectin with its counter-receptor(s) expressed on breast cancer cells. This hypothesis was tested by using a variety of specialized biochemical techniques and tumor cell/endothelial cell adhesion assays. It was found that breast cancer cells express gangliosides (sialylated lipids), a novel glycoprotein ligand known as Mac-2BP, and CD44 molecules that are functional E-selectin ligands under physiological flow conditions. Further efforts were made to find whether E-selectin ligand activity is related with breast cancer stemlike cells (BCSCs), which are the subset of tumor cells thought to possess properties necessary to maintain and grow tumor mass. For this purpose, breast cancer cell lines which were BCSCs and non-bcscs were analyzed for E-selectin ligand activity.

4 4 Interestingly, the non-bcsc cells expressed higher levels of E-selectin ligand activities than that of BCSCs. Epithelial to mesenchymal transition (EMT) is a process by which tumor cells are believed to gain metastatic potential and BCSC properties. The results indicated that E-selectin ligand activity of breast cancer cells may be regulated by EMT. These data suggesting close association of E-selectin ligands with breast cancer metastasis motivated us to develop methods to find E-selectin ligand activity of tumor tissues. First, the E-selectin ligand activity of cancer tissues was analyzed by immunohistochemistry (IHC). This study showed that E-selectin reactive molecules are abundantly expressed by some cancer tissues. However, mere presence of molecules reactive to E-selectin under static (no-flow) conditions is not sufficient for E-selectin ligand function. The E-selectin ligands need to possess certain biophysical properties to serve as adhesion molecules resisting the shear forces exerted by the flow of circulatory fluid (e. g., blood). To analyze the E-selectin ligand activity of tissue samples under physiological flow conditions, an assay termed dynamic biochemical tissue analysis (DBTA) was developed. Ultimately, the E-selectin ligands found in this study potentially reveal a new therapeutic target for breast cancer, and DBTA provides a tool for the development of novel diagnostics and prognostics for cancer based on E-selectin ligands.

5 5 DEDICATION To my beloved family

6 6 ACKNOWLEDGMENTS I would like to thank my advisor Dr. Monica M. Burdick for her invaluable guidance and encouragement throughout my dissertation. I am deeply indebted to Dr. Douglas J. Goetz for his suggestions and support. I would like to thank to Dr. Fabian Benencia, Dr. Ramiro Malgor, and Dr. David F. J. Tees for their advice in preparing my dissertation. My thanks to all current and former Burdick lab students, especially Luis F. Delgadillo, Nathan M. Reynolds, Tiantian Liu, Grady Carlson, Karissa A. Henson, and Daniel Marling. They were always available to discuss science as well as to assist in experiments. I would like to thank to students from Benencia and Goetz lab, especially China Kummitha, Sudhir Deosarkar, and Pooja Bhatt for helpful discussions. I would also like to thanks to my friends Santosh and Rita Vijapur, Yatin Kumbhar, Supradeep Vijaya Kumar, Aditya Kulkarni, Nilesh Khade, Abhijeet Chinchore, Baswaraj and Preeti Patil, and Chintan Patel.

7 7 TABLE OF CONTENTS Page Abstract... 3 Dedication... 5 Acknowledgments... 6 List of tables List of figures List of abbreviations Chapter 1: Introduction to cell adhesion in metastasis Need for studying breast cancer metastasis Mechanism of metastasis Cancer stem cells Seed and soil hypothesis Cell adhesion in metastasis Hematopoietic stem cell (HSC) homing pathway Molecular mediators of cell adhesion Selectins Selectin ligands Detection of selectin ligands Monoclonal antibodies Flow adhesion assay Hypothesis Specific aim 1: To identify and characterize breast cancer cell E-selectin ligands Specific aim 2: To investigate the relation of E-selectin ligand activity to BCSC and EMT Specific aim 3: To develop methods for investigation of E-selectin ligand activity of cancer tissues Chapter 2: Gangliosides expressed on breast cancer cells are E-selectin ligands Abstract Introduction... 39

8 2.3 Materials and methods Cell culture Antibodies and recombinant proteins Flow cytometry Extraction of gangliosides Thin layer chromatography (TLC) and immuno-overlay assay Cell treatments Shear-dependent cell adhesion assays Statistics Results BT-20 and MDA-MB-468 breast cancer cells express E-selectin ligands that mediate tethering on activated endothelium BT-20 but not MDA-MB-468 cells express sle X and sle A Protease-insensitive molecules possess E-selectin ligand activity Gangliosides are E-selectin ligands Gangliosides mediate tethering and rolling on endothelial E-selectin Discussion Chapter 3: Mac-2 binding protein is a novel E-selectin ligand expressed by breast cancer cells Abstract Introduction Materials and methods Cell culture Antibodies Enzyme treatments and cell surface biotinylation Flow cytometry RNA interference Parallel plate flow chamber adhesion assay Cell lysis and immunoprecipitation SDS-PAGE and Western blotting Mass spectrometry analysis of E-selectin reactive protein Immunofluorescence microscopy and image deconvolution

9 Statistics Results Adhesion of ZR-75-1 breast cancer cells to HUVECs is mediated by E-selectin ZR-75-1 cells do not express known E-selectin ligands Identification of a novel E-selectin ligand Mac-2BP expressed by ZR-75-1 cells is an E-selectin ligand Mac-2BP regulates E-selectin mediated rolling and adhesion of ZR Mac-2BP is expressed by several other breast cancer cell lines that also possess E-selectin ligand activity Expression of Mac-2bp correlates with E-selectin ligand activity in invasive breast cancer tissue Discussion Chapter 4: CD44v expressed by breast cancer cells are functional E-selectin ligands under flow conditions Abstract Introduction Materials and methods Cell culture Antibodies and chimera constructs Quantitative reverse transcriptase polymerase chain reaction (qrt-pcr) Cell lysis and immunoprecipitation RNA interference SDS-PAGE and Western blotting Flow cytometry Flow adhesion assay and antigen capture Fluorescence microscopy Statistics Results Breast cancer cell lines express CD44 isoforms CD44 expressed by BT-20 cells, but not by MDA-MB-231 cells, possess HCELL activity

10 4.4.3 CD44 on intact BT-20 cells possesses HCELL activity under physiological flow conditions BT-20 cell HCELLv isoforms are sufficient for shear-resistant adhesion of CHO-E cells BT-20 cell HCELL is HECA-452 negative FT-3 and FT-6 regulate E-selectin ligand activity in BT-20 cells Breast cancer cell expression of epithelial and mesenchymal cell markers Discussion Chapter 5: Breast cancer stem-like cells possess lower E-selectin ligand activity than nonstem-like cells Abstract Introduction Materials and methods Cell culture Antibodies and chimera constructs Flow cytometry Flow adhesion assays Quantitative reverse transcriptase polymerase chain reaction (qrt-pcr) Fluorescence microscopy TGF-β treatment for EMT Statistics Results Selectin ligand activity of stem-like and non-stem like breast cancer cell lines Selectin ligand activity of breast cancer cells under physiological flow conditions Breast cancer cell lines express carbohydrate epitopes for selectin ligand activities Levels of α-(1,3)- and α-(1,4)-fucosyltransferases are different in BCSCs and non-bcscs Expression of epithelial and mesenchymal markers by various breast cancer cells EMT induced by TGF-β increases may alter expression of E-selectin ligands

11 5.5 Discussion Chapter 6: Various levels of E-selectin ligand activities of cancer tissues and cell lines are detectable by HECA-452 and E-selectin chimera Abstract Introduction Materials and methods Cell culture Antibodies and chimera constructs Preparation of sle X microsphere Flow cytometry Tissue microarrays Immunohistochemistry and image analysis Results Affinities of human and murine E-Ig chimeras towards sle X are distinct E-Ig chimera and HECA-452 mab can detect different levels of activities on cancer cells Activity of colon carcinoma tissues is detectable by HECA-452 mab and E-Ig chimera A range of activities of breast cancer tissues is detectable by HECA-452 mab and E-Ig chimera Some tissue antigens are detected by the one but not by the other probe Discussion Chapter 7: Dynamic biochemical tissue analysis Abstract Introduction Methods Microsphere preparation Flow cytometry Tissue slide preparation Flow chamber set up Immunohistochemistry Results

12 7.5 Discussion Chapter 8: Mathematical modeling of microsphere adhesion to immobilized substrate in parallel plate flow chamber Introduction Experimental methods and description of mathematical model Preparation of E-selectin substrate and sle X microspheres for experimentation Model assumptions Description of symbols used in mathematical model Model equations Solution technique for mathematical equations Results Experimental results of adhesion of sle X microspheres Experimental determination of K D Validation of solution technique using the published data Determination of K A Effect of Peclet s number Discussion Chapter 9: Conclusion and recommendation for future work References Appendix Appendix Appendix

13 13 LIST OF TABLES Page Table 3.1: Diverse breast cancer cell lines possess E-selectin ligand activity Table 4.1: Primer sequences of CD44 and fucosyltransferases for qrt-pcr Table A1: Parameter values used for calculations Table A2: Dimensionless time calculated from perfusion time.206

14 14 LIST OF FIGURES Page Figure 1.1. Prevailing model of distant metastasis Figure 1.2. Subset of tumor cells possesses properties necessary for metastasis Figure 1.3. HSC make a series of interactions with endothelial cells which are mediated by specific molecular mediators Figure 1.4. Selectin reactive carbohydrate structures Figure 1.5. Experimental set up used in dynamic cell adhesion studies Figure 2.1. Breast cancer cell adhesion to IL-1β activated HUVECs is E-selectinmediated Figure 2.2. Protease treatment does not alter the E-selectin activity of breast cancer cells Figure 2.3. Breast cancer cell gangliosides are E-selectin ligands Figure 2.4. Effect of protease treatment on breast cancer cell adhesion to HUVECs Figure 2.5. Screening for molecules capable of carrying selectin reactive glycans are expressed by breast cancer cells Figure 3.1. ZR-75-1 cell adhesion to IL-1β activated HUVECs is E-selectin mediated.. 70 Figure 3.2. Identification of a novel protein, Mac-2BP, as an E-selectin ligand Figure 3.3. Immunostaining of ZR-75-1 cells shows co-localization of signals for Mac- 2BP and E-selectin ligand activity Figure 3.4. Mac-2BP silencing of ZR-75-1 cells reduces their adhesion to E-selectin Figure 3.5. Several breast cancer cell lines express Mac-2BP and Mac-2BP of BT-20 cells possess E-selectin ligand activity Figure 3.6. Immunofluorescence analysis shows co-localization of Mac-2BP and E- selectin ligand activity on breast cancer tissue Figure 3.7. Hypothesized model for the role of Mac-2BP in metastasis Figure 4.1. CD44s and CD44v are expressed by breast cancer cells Figure 4.2. Differential levels of CD44v isoforms are expressed on the surface of breast cancer cells Figure 4.3. CD44 expressed by BT-20 cell possesses functional E-selectin ligand activity Figure 4.4. CD44 knock down in BT-20 cells reduces their adhesion to E-selectin expressing cells

15 Figure 4.5. CD44v isoforms on BT-20 cells possess sufficient E-selectin ligand activity to support cell adhesion under hematogenous flow conditions Figure 4.6. HECA-452 negative glycans confer E-selectin ligand activity to BT-20 cell CD Figure 4.7. BT-20 are epithelial-like and MDA-MB-231 are mesenchymal-like cells Figure 5.1. BCSCs and non-bcscs express E-, P-, and L-selectin ligand activities Figure 5.2. Different levels of selectin ligand activity are expressed by BCSCs and non- BCSCs under physiological flow conditions Figure 5.3. Putative E-selectin reactive glycans are expressed by BCSCs and non- BCSCs Figure 5.4. α-(1,3)- and/or α-(1,4)-fucosyltransferases are expressed at higher levels in non-bcscs than BCSCs Figure 5.5. BCSCs are mesenchymal-like cells and non-bcscs are epithelial like cells Figure 5.6. TGF-β induces EMT in BT-20 cells Figure 5.7. EMT increases E-selectin mediated rolling velocity of BT-20 cells Figure 5.8. Hypothesized model to relate E-selectin mediated adhesion with BCSC and EMT models in metastasis Figure 6.1. Human and murine E-Ig chimeras reveal different reactivities with sle X coated microspheres Figure Breast and colon cancer cell lines show various levels of reactivities to murine E-Ig chimera and HECA-452 mab igure 6.3. Colon cancer tissues were positive for HECA-452 reactivity and E-Ig chimera Figure 6.4. Breast carcinoma tissues were positive for HECA-452 and E-Ig chimera Figure 6.5. The HECA-452 reactivity and E-Ig activity of non-cancerous tissue is restricted to certain areas Figure 6.6. Levels of activities of a same tissue detected by HECA-452 mab and murine E-Ig chimera differ Figure 7.1. Experimental setup used in DBTA Figure 7.2. E-selectin chimera construct conjugated polystyrene microspheres are specifically recognizable by E-selectin recognizing mab Figure 7.3. E-selectin microspheres specifically adhere to cancer tissues Figure 7.4. E-selectin microspheres exhibit rolling interactions with carcinoma tissue

16 Figure 7.5. Controlled modulations in particle size and shear stress result in discernible changes in adhesion Figure 7.6. DBTA can be applied to tissue microarrays, and the microarrays used in DBTA can be reused for immunohistochemistry Figure 8.1. Schematics of sle X coated microsphere interactions with E-selectin coated substrate in flow chamber Figure 8.2. Schematic diagram representing boundary conditions and FDM grids Figure 8.3. sle X microspheres were perfused over E-selectin substrate at a shear stress of 1.5 dyne/cm Figure 8.4. Comparison of normalized bound particles predicted by the model and found by experiments Figure 8.5. Effect of Peclet s number (P) on particle profile in the flow chamber Figure 9.1. Mac-2BP silencing of ZR-75-1 cells reduces Gal-1 ligand activity

17 17 LIST OF ABBREVIATIONS AC ALDH ANOVA BCSC BSA CC CEA CHO-E CSC CTB CTC DBTA DMEM DPBS EDTA EGF E-Ig chimera EMT EpCAM FBS FEM FFPE FITC FT Gal-1 GAPDH HCELL HECA-452 HSC HUVECs Adenocarcinoma Aldehyde dehydrogenase Analysis of variance Breast cancer stem-like cells Bovine serum albumin Carcinoid colon cancer Carcinoembryonic antigen E-selectin transfected Chinese hamster ovary cells Cancer stem-like cells Cholera toxin subunit B Circulating tumor cell Dynamic biochemical tissue analysis Dulbecco s modified eagle medium Dulbecco s phosphate buffered saline Ethylenediaminetetraacetic acid Epidermal growth factor Recombinant mouse E-Selectin/human immunoglobulin Fc Epithelial to mesenchymal transition Epithelial cell adhesion molecule Fetal bovine serum Finite element method Formalin fixed paraffin embedded Fluorescein isothiocynate Fucosyltransferase Galectin-1 Glyceraldehyde 3-phosphate dehydrogenase Hematopoietic cell E- and L-selectin ligand High endothelial cell antigen-452 Hematopoietic stem cell Human Umbilical Vein Endothelial Cells

18 18 IDC IgG IHC IL-1β LDC mab MAC Mac-2BP MALDI MC MCF-7 MDA-MB-231 MDA-MB-468 MdC MET mrna MS NC PBS PC PCLP PCR PE PSGL-1 qrt-pcr SE sle A sle X TGF-β TMA TOF Invasive ductal breast carcinoma Immunoglobulin G Immunohistochemistry Interleukin-1β Mixed lobular and duct breast carcinoma Monoclonal antibody Mucinous colon adenocarcinoma with necrosis Mac-2 binding protein Matrix-assisted laser desorption/ionization Mucinous breast carcinoma Michigan Cancer Foundation-7 M.D. Anderson-metastatic breast-231 M.D. Anderson-metastatic breast-468 Medullary breast carcinoma Mesenchymal to epithelial transition Messenger ribonucleic acid Mass spectrometry Neuroendocrine breast carcinoma Phosphate buffered saline Papillary colon adenocarcinoma Podocalaxyn-like protein Polymerase chain reaction Phycoerythrin P-selectin glycoprotein ligand-1 Quantitative reverse transcriptase polymerase chain reaction Standard error Sialyl Lewis A Sialyl Lewis X Transforming Growth Factor-β Tissue microarray Time of flight

19 19 CHAPTER 1: INTRODUCTION TO CELL ADHESION IN METASTASIS 1.1 Need for studying breast cancer metastasis The process of formation of new tumors distinct from the primary mass is called metastasis. Such a manifestation of cancer has devastating consequences on human health. For example, bone metastasis of the osteolytic type dissolves bones leading to collapse of the structural framework of body and eventually a painful death. Breast cancer metastasis, in particular, is important because it is one of the common causes of cancer deaths among women, with about 39,000 projected deaths for the year 2011 in America [1]. Notably, if breast cancer is detected in early stages, the survival rate is almost 100%, but once the tumor has metastasized the five-year survival rate drops precipitously to 20% [1]. These statistics indicate that the present treatment strategies are inadequate to fight against the destructive advancement of the breast cancer. Thus, there is a great necessity to better understand the molecular mechanism of breast cancer metastasis. 1.2 Mechanism of metastasis The journey of tumor cells from a primary tumor to distant metastatic sites involves very complex yet highly regulated events (Fig. 1.1). These events can be broadly categorized into the following sequential steps: local invasion, entry of cancer cells into systemic circulation (intravasation), transportation and arrest of the cancer cells in various organs, entry into tissue parenchyma (extravasation), and formation of metastatic mass [2-5]. Each of these steps poses unique challenges that cancer cells have to overcome to survive and ultimately form metastases at a distant site [4,5]. To provide a

20 20 broad picture of metastasis a simplified description of these events is presented here. More details can be obtained from articles written to address specific steps [2-7]. During local invasion, cancer cells migrate through basement membranes and fibrous architecture formed by connective tissues. To perform these activities, cancer cells acquire migratory properties and become motile by shedding certain cell adhesion molecules. Apart from such phenotypic alterations, cancer cells themselves release, or induce other non-cancerous cells to secrete, several cytochemicals, such as matrixmetaloprotineases, which cut matrix proteins, to facilitate the cell migration [5]. Further, cancer cells gain access to systemic circulation in a variety of ways, such as via newly formed vessels in the tumor by angiogenesis [5]. Once in the lumen of the vasculature, cancer cells have to endure a hostile and drastic environment of blood or lymph and hydrodynamic forces, which are high enough to tear cancer cells apart. These tumor cells in the systemic circulations are called as circulating tumor cells (CTCs). The CTCs travel to distantly located sites, primarily through well-connected hematogenous routes [2-5,7]. The arrest of CTCs in the vasculature of secondary organs is mediated by interactions between CTCs and endothelial cells lining the vascular lumen. The attached cancer cells then extravasate into the surrounding tissue parenchyma, where the survival capability of cancer cells is again tested by the new niche [2-5]. Thus, metastasis is an extremely intricate process that is a product of successful cross-talk between cancer cells and their surrounding environment. In this view, cancer cells and the potential metastatic tissue need certain traits to foster metastasis, which is discussed in the following sections.

21 21 Figure 1.1. Prevailing model of distant metastasis. Fidler Originally published in Nature Reviews Cancer, doi: /nrc1098. Reproduced with permission. (a) Primary tumor grows to certain size, and (b) form blood vessels by angiogenesis. (c) The tumor cells detach and (d) enter into circulatory system and transported to different organs. (e) The CTCs attach to lumen and extravasate to form new tumor mass [4]. 1.3 Cancer stem cells Cancer tumors are made up of a heterogeneous population of cells characterized by differences in cell morphology, surface markers, genetics, rate of cell proliferation, and response to therapy [5,8,9]. Two models have been proposed to explain such variations in tumor cells. According to the stochastic model, all cells of a tumor are

22 22 biologically similar; however, the differences in cell traits arise due to unknown intrinsic and extrinsic factors that include genetic epigenetic, niche, and environmental factors [8]. In this view, all cells of a tumor possess equal capacity to initiate new tumors. The other model is based on hierarchical organization, a concept borrowed from the knowledge of tissue regeneration [8]. According to this model, only a subset of tumor cells possesses functional traits to maintain and generate all cells of tumor mass. These cells are termed cancer stem cells (CSCs), as they are capable of unlimited self-renewal which can generate CSC and non-csc progenies [5,8,9]. It is important to note that CSCs are not necessarily cells of tumor origin, which are the cells first altered during the tumor formation [8,9]. Yet the cells of tumor origin can later acquire CSC properties, or tissue resident physiological stem cells upon genetic mutations can become CSCs [8,9]. A major distinction that separates the two models is that according to the stochastic model, any cell, if given appropriate cues, can acquire CSC properties, and according to the hierarchical model this capacity is limited to only a subset of the cells [8-10]. Which model better explains the tumor heterogeneity is still a partially unanswered question, as discussed in many excellent review papers [2-4,8,10]. Yet with the identification of a subset of tumor cells with stem cell properties [11-14] that are capable of generating heterogeneous tumor mass resembling the original tumor [11,12], the hierarchical model is getting wider acceptance, at least for some forms of cancers [5,6,8,11]. The CSCs, identified by expression of certain cell surface marker proteins, have been reported in a number of cancer types, including breast cancers [8,11,12,15]. In

23 23 a seminal paper, Al-Hajj et al. (2003) found that a small number of breast epithelial cancer cells with positive expression of CD44 and negative or low expression of CD24 (CD44 + /CD24 -/low ) are sufficient to form new tumors in mouse models [12]. Furthermore, these breast cancer stem-like cells (BCSCs) were able to generate secondary tumor with heterogeneous phenotypic cell populations resembling the original tumor mass [12]. With acquisition of additional functional traits, the CSCs are thought to lead to metastatic spread of tumors [5,8,9]. These properties include the capabilities to disseminate from the primary tumor, survive in the circulatory system, attach to endothelium, and grow a tumor in the new microenvironment (Fig. 1.2; [8]). Such functional traits may also be possessed by BCSCs characterized with cell surface marker expression CD44 + /CD24 -/low, as these cells have been shown to possess high invasive potential [16] and are suggested to favor distant metastasis in breast cancer patients [17]. On the other hand, it is also possible that only a subset of BCSCs with CD44 + /CD24 -/low expression have the potential for metastasis [18]. As a significant development of BCSC theory, recent reports suggest that BCSC phenotypes are obtainable from non-bcscs by epithelial to mesenchymal transition (EMT) [19], a process which is marked by loss of epithelial adhesion molecule E- cadherin and increase in expression of N-cadherin. Indeed, the putative stem cell signatures on BCSCs were induced by TGF-β cytokine, a well-known EMT inducer [19]. It is believed that cancer cells undergo EMT to achieve higher migratory properties necessary to pass through the maze of cellular and matrix proteins to invade surrounding tissue. Accumulating evidence also strongly suggests that EMT is germane to metastatic

24 invasion [5,9,19]. In support, circulating breast tumor cells have been found in the state of EMT [18]. 24 Figure 1.2. Subset of tumor cells possesses properties necessary for metastasis. Baccelli and Trumpp, Originally published in The Journal of Cell Biology. doi: /jcb Reproduced with permission. A tumor is composed of bulk of non-metastatic non-cscs, non-metastatic CSC clones and a small subset of metastatic CSCs. The metastatic CSC clones can be organ specific [8]. 1.4 Seed and soil hypothesis Based on the autopsies of 735 female breast cancer patients, Stephan Paget published the seed and soil hypothesis in 1889 [20]. According to his hypothesis, the formation of metastasis depends upon the compatibility of cancer cells (the seed) and the metastatic organ (the soil), analogous to a conventional wisdom that seeds grow only in

25 25 suitable soil. The close connection between tumor cells and metastatic microenvironment has been tested and validated by many others over years for a variety of cancers. For instance, a study performed by autopsy of breast cancer patients found that breast cancer preferentially metastasizes to certain organs, such as bone marrow and lungs, with much higher frequency than other organs, such as kidneys [2,4]. A counter argument to the seed and soil hypothesis is that metastatic pattern is associated with lymphatic and blood circulatory system. According to this point of view, the circulating tumor cells are physically trapped into the organs of first exposure and thus metastasis is mere a probabilistic event. Although some results can be explained by this model, several other issues remain unaddressed. Such as, the frequency of metastasis cannot be explained by the blood volume, blood flow rate, blood volume per unit weight of the organ, or organ weight [21]. Also, implanted melanoma cells in animal models fail to form tumors in renal tissues but generate tumors in pulmonary or ovarian tissues [22]. Moreover, the mechanical trapping model also fails to explain why a large number of micro-metastases in certain organs remain dormant for long periods of time [5]. Thus, organ preference of metastasis may not be because of mere mechanical trapping but is also driven by biological mechanisms [2-5]. 1.5 Cell adhesion in metastasis A prominent mechanism that partly explains the preferential organotropism of metastasis is that the luminal surface of the vasculature expresses tissue specific adhesion molecules to arrest tumor cells and promote the metastasis. In support of this notion, endothelial cells lining lumen of blood vessels obtained from various organs supported

26 26 the adhesion of cancer cells in a differential pattern correlating with in vivo metastatic incidences [23]. This concept is maintained by a plethora of other reports showing that cancer cells express specific counter receptors to endothelial adhesion molecules [24-28]. Also, the selective deletion or blocking of these adhesion molecules from cancer cells reduced the metastatic potential of cancer cells in animal models [29,30]. The preferential osteotropism, metastasis to bones, of breast cancer, in particular, can be explained by the cell adhesion process. As endothelium of human bone marrow constitutively expresses certain adhesion molecules such as E-selectin [31], cancer cells expressing counter receptors adhere to endothelium and perhaps extravasate to form metastasis. Although post adhesion events are undeniably important, the entry of CTCs into the metastatic niche is perhaps controlled by the adhesion. Thus, blocking the adhesive interactions between tumor cells and endothelial cells may be an attractive strategy to prevent metastasis. To develop this idea, it is important to understand the adhesion pathways and molecular mediators of cell adhesion Hematopoietic stem cell (HSC) homing pathway The molecular model for CTC adhesion to endothelial cells is derived from leukocyte cell adhesion pathways. Hematopoietic stem cells (HSC) use cell adhesion pathway for bone marrow homing, and other leukocyte use the pathway during tissue repair [32,33]. From the knowledge of these pathways, it is now abundantly clear that circulating cells make a series of interactions with progressively increasing affinity, prior to firmly attaching to endothelial cells (Fig. 1.3). Each of these interactions is mediated

27 27 by specific molecular mediators [32,33]. In order to understand cancer cell adhesion, a brief description of the well-reported HSC cell adhesion pathway is provided. HSC in blood circulation follow a series of adhesive interactions with bone marrow endothelium, which involves the following events: cells still in flow form loose bonds with endothelium (tethering), then they roll over the endothelium and activate molecules to finally form firm adhesion to endothelium. The tethering and rolling events are mediated by endothelial E-selectin and its counter receptors expressed on HSC. Chemokines, such as SDF-1, then stimulate HSC to express integrins which brings rolling cells to complete arrest (firm adhesion) by binding to endothelial VCAM-1. The firm adhesion of HSC to endothelium activates downstream pathways which allow extravasation of HSC across endothelial cells [32,33]. 1.6 Molecular mediators of cell adhesion As mentioned in the previous section, the adhesion of circulating cells to endothelium is initiated by tethering and rolling types of interactions. These interactions are mediated by endothelial E- or P-selectins and their respective ligands expressed on the circulating cells [24-26]. As the main focus of the present work is selectin mediated adhesion of breast cancer cells, a brief discussion of selectins and their ligands are included.

28 28 Figure 1.3. HSC make a series of interactions with endothelial cells which are mediated by specific molecular mediators. Flowing HSCs tether and roll over endothelium by forming bonds with E-selectin. The rolling cells are then firmly anchored on endothelial cells by other cell adhesion molecules to finally transmigrate into bone marrow micro-environment Selectins The selectin family of adhesion molecules consists of E-, P-, and L-selectin. All of the selectins are transmembrane proteins and share many structural features that includes an amino terminus, epidermal growth factor (EGF)-like domain, short consensus repeats, and transmembrane and cytoplasmic domains [24,29,33,34]. Despite these similarities, their distribution among cells and mechanisms of action are different, signifying their role in a variety of physiologic and pathologic processes [24,29,33,34]. E- and P-selectin are expressed by endothelial cells in response to inflammatory cytokines such as TNF-α or IL-1β, and L-selectin is expressed de novo by most types of leukocytes [24,29,33,34]. Also, E-selectin is constitutively expressed by bone marrow

29 29 endothelium, and P-selectin is expressed by activated platelets [24,29,31,33,34]. E- and P-selectin are believed to facilitate direct interactions between flowing tumor cells and endothelial cells, but L-selectin, which is a primary receptor for lymphocyte homing in peripheral lymph nodes, is thought to promote secondary interactions with endothelium by forming aggregates of tumor cell-leukocytes [24,29]. Particularly E-selectin is believed to be a major player in cancer metastasis since (a) many cancers frequently spread to bone marrow, endothelium of which constitutively expresses E-selectin [31], and (b) adhesion of cancer cells to endothelium is specifically mediated by E-selectin under physiological flow conditions [24-26,35] Selectin ligands Selectins primarily recognize glycan epitopes displayed on the appropriate carrier molecules, and such glycoconjugates are referred as selectin ligands. For binding of any of the selectins, the minimum motifs required are sialylated and fucosylated epitopes located at the terminal end of the glycans [36,37]. Examples of such glycans include sialyl lewis X (sle X ) and its stereo isomer sialyl Lewis A (sle A ), and closely associated structures (Fig. 1.4; [38]). Although many of these carbohydrates are also expressed by some healthy tissues, their levels in cancer tumors are aberrantly high [24,29,39]. Indeed, the level of expression of sle X has been shown to be higher in metastatic tumors than non-metastatic tumors [40]. These data underscore the importance selectin-ligand mediated cell adhesion in cancer metastasis. The biosynthesis of carbohydrates is directed by a unique set of glycosyltransferases in the Golgi apparatus. Each glycosyltransferase typically transfers a

30 30 sugar molecule, such as sialic-acid, fucose, galactose, N-acetylgalactosamine, N- acetylglucosamine, sequentially elongating the glycan chain. The terminal sialofucosylations of the glycans are catalyzed by sialyltransferases and fucosyltransferases. Six sialyltransferases of the ST3Gal family (named ST3Gal I to VI) form α-(2, 3) sialic acid linkages of sle X [41]. The final step of the sle X/A synthesis is transfer of the fucose group which is carried out by fucosyltransferases that form α-(1, 3)- and α-(1, 4)-glycosidic linkages of fucose. Of nine enzymes encoded by FUT-1-9 genes, FT-3-7 and FT-9 are α-(1, 3)-fucosyltransferases that catalyze the formation of sle X and FT-3 and FT-5 also act as α-(1, 4)-fucosyl transferase that direct synthesis of sle A [24]. Notably, sle X and related structures are only a terminal portion of a long polymer chain extending from a scaffold molecule. The core glycan chains, apart from the terminal end, appear to influence selectin binding. For instance, sle X extended on core-2 types of O-glycans provide high affinity selectin ligand activity to colon carcinoma cells [42,43]. Complicated mechanisms, such as competition among glycosyltransferases sharing similar acceptors, contribute to the heterogeneity of glycans [44,45]. Such structural variations of the glycans are believed to be a result of specific physiologic or pathologic conditions [24,39]. In addition to the glycans, the attachment site of a glycan to a core protein may influence the selectin binding and may be related to specific disease conditions. More specifically, the polymeric chains of glycans are either attached on asparagine-linked (N-linked glycans) and/or serine/threonine-linked (O-linked glycans) to core protein molecules. While the selectin ligand activity of leukemia cell CD44 is

31 associated with N-linked [46] and that of colon cancer cell CD44 is related to O-linked glycosylations [47]. 31 Figure 1.4. Selectin reactive carbohydrate structures. Glycans have to be displayed on appropriate scaffold molecules to form a functional E-selectin ligand. Several such core molecules have been identified including CD44, CD43, P-selectin glycoprotein ligand-1 (PSGL-1), podocalaxyn-like protein (PCLP), CEA, and gangliosides [24-26,48-51]. Additional features are necessary to serve as a ligands for the other two selectins. To be recognized by P-selectin, the glycans need

32 32 to be O-linked glycans attached to only certain proteins [24,38,52], as some portion of P- selectin also bind to core molecule of the ligand. The examples of P-selectin ligands are PSGL-1, CD44, or CD24 [24,52,53]. Functional L-selectin ligands need sulfated groups in addition to sialic acid and fucose moieties primarily on O-linked glycans. Such glycans carried by CD34, CD44, MadCAM-1, and PSGL-1 [24,26,54]. 1.7 Detection of selectin ligands Monoclonal antibodies To detect sle X or related glycan structures, a number of commercially prepared mabs with varying detection specificity is available. Hence, some discussion on mabs for detecting sialofucosylated structures is warranted. The two major mabs used for detecting sle X are CSLEX-1 and KM-93. While KM-93 mab detects a broad array of sle X decorated on proteins as well as lipids, CSLEX-1 mab recognizes more restricted domain of antigens, which mainly includes sle X decorated on protein backbones but the mab weakly bind to certain lipid sle X [38,55]. Additionally, CSLEX-1 mab needs Galβ1--->4 GlcNAc linkage for binding [38]. None of these mabs detect the almost similar structure of sle A or VIM-2 glycans [55]. Another sle X -recognizing mab is FH6 which is raised primarily to detect extended form of sialyl dimeric Le X and possesses limited cross-reactivity to the sle X [56]. FH6 mab requires sialic acid and fucose groups in addition to Galβ1--->4 GlcNAc linkage for binding [38]. Further, VIM-2 glycans on proteins and lipids are specifically detected by anti-cd65s mab [57,58]. Furthermore, sle A oligosaccharide decorated on protein as well as lipids can be recognized by KM231, and sulfated sialofucosylated antigens are

33 33 detectable by MECA-79 mab [59]. On the other hand, HECA-452 mab generally detects a broad array of sialofucosylated antigens and depends on sialic acid and fucose residues but is independent of Galβ1--->4 GlcNAc linkage [38]. Due to its large domain of sialofucosylated antigen recognition, HECA-452 mab is a mab of choice to detect potential selectin ligands on a variety of cells [26,60]. However, this mab does not recognize some sialofucosylated structures, such as internally fucosylated structures of VIM-2 glycans [38]. Alternative probes to explore selectin ligands are humanized forms of recombinant selectin chimeras. These constructs consist of lectin-like antigen binding region from murine, rat, or human E-selectin connected to the Fc domain of human IgG. To obtain optimal levels of detection signal, the choice of appropriate chimera construct appears to be critical, maybe because affinities of chimeras obtained from different species vary dramatically [61]. For instance, even with very sensitive analysis techniques such as flow cytometry, staining of highly E-selectin ligand positive leukocytes with human E-Ig chimera revealed very weak signal, in contrast to staining of these cells with murine E-Ig chimera, provided highly positive signal [61] Flow adhesion assay A critical aspect to consider in selectin research is that selectins bind to its ligands under fluid shear force exerted by blood flow. In this view, in addition to biochemical characteristics, a functional E-selectin needs to possess certain biophysical properties, such as high on and off intrinsic kinetic rates (k on and k off ) and high tensile strength of the bond that allows a cell to tether and roll over selectins under shear force [62]. Thus,

34 34 detection of selectin ligands under non-flow conditions is insufficient to find functional ligand activity required to support cell adhesion under physiological flow conditions. To circumvent these issues, functional selectin ligand activity is studied under flow conditions. Physiological flow conditions are created in vitro by a parallel plate flow chamber, which forms a leak-proof channel of defined cross section over the surface of a tissue culture dish (Fig. 1.5). The cross sectional area of the flow channel can be varied by using a gasket of different width or thickness. The fluid flow in the chamber is maintained by a high precision syringe pump. The cells under investigation, typically suspended in buffer or culture medium, are drawn over a tissue culture dish, over which endothelial cells are grown or purified molecules are immobilized. The flow rate of the cell suspension fluid is set to generate a physiological shear rate. The wall shear rate ( ) is calculated by the following equation [63]. µ = viscosity of fluid Q = flow rate b and h = channel breadth and height τ w = wall shear stress

35 35 Figure 1.5. Experimental set up used in dynamic cell adhesion studies. This set up consists of a vacuum sealed flow chamber (shown in the box) placed on a microscope equipped with a video recording system. The flow rate is controlled by a high precision syringe pump. The parallel plate flow chamber is placed on the stage of an inverted microscope equipped with a video camera connected to VHS tape recorder or computerized recorder. Adhesion events are recorded and later analyzed. The adhesion events are categorized as cell tethering, rolling, and firm attachment. Typically, tethering of flowing cells is determined by counting the number of cells attaching from the free stream, and cells remaining stationary for more than 5 sec are considered firmly adhering cells. Cell rolling

36 36 velocity is calculated by capturing stills from video over 5 sec and converting pixel distance into microns using Image J software [27,28,64]. 1.8 Hypothesis For the reasons detailed earlier in the chapter, the hypothesis of the present study was as follows. The adhesion of circulating breast tumor cells to endothelium is mediated by endothelial E-selectin and its ligands expressed by breast tumor cells. Thus, identifying mechanisms regulating expression of E-selectin ligands and relating cancer progression with E-selectin ligand activity in situ could lead to novel therapeutic/ prognostic/diagnostic strategies against breast cancer. This hypothesis was tested with the following specific aims Specific aim 1: To identify and characterize breast cancer cell E-selectin ligands As multiple ligands co-operate to mediate adhesion of circulating cells via the E- selectin pathway, independent studies were designed to investigate various E-selectin ligands, including glycolipid and glycoprotein ligands, on breast cancer cells (Chapters 2-4). The functional E-selectin ligand activity of breast cancer cell lines was investigated under physiological flow conditions Specific aim 2: To investigate the relation of E-selectin ligand activity to BCSC and EMT Breast cancer stem-like cells (BCSCs) are the subset of tumor cells that are thought to possess properties necessary to maintain and grow tumor mass, and epithelial to mesenchymal transition (EMT) is a process by which tumor cells are believed to gain metastatic potential and BCSC properties. Due to close association of BCSCs and EMT

37 37 to metastasis, efforts were made to find whether E-selectin ligand activity is related with BCSC and EMT phenotypes (Chapter 5). For this purpose, breast cancer cell lines which were BCSCs and non-bcscs were analyzed for E-selectin ligand activity Specific aim 3: To develop methods for investigation of E-selectin ligand activity of cancer tissues As the data from specific aim 1 and 2 suggested close association of E-selectin ligands with breast cancer metastasis, E-selectin ligand activity of breast cancer tissues may be used as marker of advanced stage cancers. The E-selectin ligand activity of cells has been classically explored by HECA-45 mab and E-selectin chimera constructs. The important questions to address to use these probes for tissue analysis are whether they can distinguish between cancer and non-cancer tissues and whether they can identify distinct levels of E-selectin ligand activities (Chapter 6). Mere presence of molecules reactive to E-selectin under static (no-flow) conditions is not sufficient for E-selectin ligand function. The ligands need to possess certain biophysical properties to serve as E-selectin ligand operational under the shear forces exerted by the flow of circulatory fluid (e. g. blood). Thus, the functional E- selectin ligand activity of tissues under physiological flow conditions may distinguish cancer tissues from non-cancer tissues. An assay termed as dynamic biochemical tissue analysis (DBTA) was developed (Chapter 7) to analyze the functional E-selectin ligand activity in situ. Also, a mathematical model was proposed to study effect of hydrodynamics on the microsphere adhesion in flow chamber (Chapter 8).

38 38 CHAPTER 2: GANGLIOSIDES EXPRESSED ON BREAST CANCER CELLS ARE E- SELECTIN LIGANDS Abstract Cancer cell adhesion to vascular endothelium is a critical process in hematogenous metastasis. We hypothesized that breast cancer cells express ligands that bind under blood flow conditions to E-selectin expressed by endothelial cells. At a hemodynamic wall shear rate, BT-20 and MDA-MB-468 breast cancer cells adhered to cytokine-activated human umbilical vein endothelial cells (HUVECs) but not to anti-eselectin monoclonal antibody treated HUVECs, demonstrating that adhesion was specifically mediated by E-selectin. Characterization of glycans expressed on breast cancer cells by a panel of antibodies revealed that BT-20 cells expressed sialyl Lewis X (sle X ) and sialyl Lewis A (sle A ) but MDA-MB-468 cells did not, suggesting that the former possess classical glycans involved in E-selectin mediated adhesion while the latter have novel binding epitopes. Protease treatment of the breast cancer cells failed to significantly alter the carbohydrate expression profiles, binding to soluble E-selectin-Ig chimera, or the ability of the cells to tether and roll on E-selectin expressed by HUVECs, indicating that glycosphingolipids are functional E-selectin ligands on these cells. Furthermore, extracted breast cancer cell gangliosides supported binding of E-selectin-Ig chimera and adhesion of E-selectin transfected cells under physiological flow conditions. 1 Previously published as: Gangliosides expressed on breast cancer cells are E-selectin ligands. Shirure VS, Henson KA, Schnaar RL, Nimrichter L, and Burdick MM. Biochemical and Biophysical Research Communications, 2011; Text and figures used with permission.

39 39 In summary, our results demonstrate that breast cancer cells express sialylated glycosphingolipids (gangliosides) as E-selectin ligands that may be targeted for prevention of metastasis. 2.2 Introduction Metastasis to distant organs involves complex yet systematic events, in which cancer cells disseminate from a primary tumor and journey through the vasculature to a secondary site, where they attach and extravasate into the tissue to form a new growth. Although a metastatic niche at the secondary site is undeniably important for successful colonization, a circulating tumor cell cannot invade without attaching to the vascular endothelium at the target site. Treatment strategies that prevent cancer cell adhesion to endothelium are thus attractive methods to inhibit metastasis, but first an understanding of the molecular mediators involved is necessary to develop such interventions. It is hypothesized that cancer cells adhere to endothelium by a process similar to that of leukocyte homing. In this model, cells in flow are captured on the endothelial surface (also known as tethering), slowed by transient adhesive interactions with endothelial selectins (rolling), and firmly anchored on endothelium (firm adhesion) to enable entry into the underlying tissue. The selectins, particularly E-selectin, are recognized to mediate adhesion and thus potentiate metastasis of certain cancers [24,29]. E-selectin is expressed in most vascular beds in response to inflammatory stimuli and is also constitutively expressed by endothelium of hematopoietic tissues such as bone marrow [31]. Notably, breast cancer frequently spreads to bone, and breast cancer cells

40 40 have been reported to specifically bind endothelial E-selectin under hematogenous shear conditions [65,66]. In complement, breast cancer lesions but not normal breast cells express sialofucosylated modifications (e.g., sialyl Lewis X; sle X ) in situ [40] that are recognized by E-selectin [24,29], with higher expression correlating with disease progression [40]. Despite persuasive evidence for the involvement of E-selectin and its ligands in breast cancer metastasis, breast cancer cell E-selectin ligands operative under physiological flow conditions have not been identified. In contrast, we have previously isolated and characterized numerous sialylated glycosphingolipid (ganglioside) E-selectin ligands expressed by normal human leukocytes [25,67]. Also, in studies of E-selectin ligands on colon cancer, prostate cancer, and leukemia, gangliosides were identified as significant mediators of cell adhesion [48,68,69]. The contribution of these molecules in breast cancer cell adhesion is currently unknown, which is surprising given their importance in various metastatic processes, including epithelial to mesenchymal transition (EMT) [70]. Therefore, the present study was designed to investigate the potential role of gangliosides as fluid shear-resistant E-selectin ligands expressed by invasive breast cancer cell lines. 2.3 Materials and methods Cell culture The BT-20 breast invasive ductal carcinoma cell line and MDA-MB-468 breast adenocarcinoma cell line [71] were obtained from the American Type Culture Collection (ATCC; Manassas, VA). BT-20 and MDA-MB-468 cells were cultured in MEM and

41 41 DMEM (Invitrogen, Carlsbad, CA), respectively, supplemented with 10% fetal bovine serum (FBS) and 1x penicillin-streptomycin. E-selectin transfected Chinese Hamster Ovary (CHO-E) cells were a generous gift from Dr. Robert Sackstein (Harvard Medical School, Boston, MA). CHO-E cells were cultured in MEM supplemented with 10% FBS, 0.1 mm nonessential amino acids, and 1x penicillin-streptomycin. Human umbilical vein endothelial cells (HUVECs) were purchased from Lonza (Allendale, NJ) and cultured as described previously [26,48] Antibodies and recombinant proteins Anti-human CD43 (1G10), CD44 (515), PSGL-1 (KPL-1), CD62E (68-5H11), CD66 (COL-1), HECA-452 (recognizing several sialofucosylated epitopes correlating with E-selectin ligand activity), sialyl Lewis X (sle X ; CSLEX-1) monoclonal antibodies (mabs), and all isotype controls were obtained from BD Biosciences (San Jose, CA). Anti-sialyl Lewis A (sle A ; KM-231) was from Calbiochem (San Diego, CA), and antihuman PCLP (3D3) and E-cadherin (67A4) mabs were from Santa Cruz Biotechnology (Santa Cruz, CA). Recombinant mouse E-Selectin/human immunoglobulin Fc chimera (E-Ig chimera) and biotin-conjugated cholera toxin subunit B (CTB) were obtained from R & D systems (Minneapolis, MN) and Invitrogen (Carlsbad, CA), respectively. Fluorescein isothiocyanate (FITC)-conjugated and phycoerythrin (PE)-conjugated polyclonal secondary antibodies were purchased from Southern Biotech (Birmingham, AL).

42 Flow cytometry The procedure for cell labeling has been described previously [26,48]. A FACSort flow cytometer (BD Biosciences) was used for analysis Extraction of gangliosides The Svennerholm method using 4:8:3 chloroform/methanol/water with phase partitioning was used to extract breast cancer cell gangliosides as described previously [67,72]. Following phase partitioning, samples were subjected to chromatography using Sep-Pak Plus C18 cartridges (Waters, Milford, MA) to obtain fractions enriched in gangliosides. Protein contamination was not detected when tested by BCA protein assay (Thermo Fisher Scientific, Rockford, IL) Thin layer chromatography (TLC) and immuno-overlay assay Ganglioside extracts from equivalent numbers of breast cancer cells were resolved on glass-backed silica gel TLC plates (Merck, Gibbstown, NJ) using chloroform/methanol/0.25% aqueous KCl (60:35:8 v/v/v) as the developing solvent. The ganglioside bands were visualized after spraying with resorcinol reagent (0.3% (w/v) resorcinol, 0.003% (w/v) cupric sulfate pentahydrate, and 30% (v/v) concentrated HCl in water) [73], and subsequently heating the plates for 20 min at 125 o C. To evaluate whether gangliosides from breast cancer cells are ligands for E- selectin, an immuno-overlay assay using E-Ig chimera was performed on gangliosides resolved by TLC. Dry TLC plates were immersed in a mixture of hexanes and then transferred to a solution of 1 mg/ml polyisobutylmethacrylate in hexanes [73]. The plates were dried, immersed in phosphate buffered saline (PBS) for 5 min and then transferred

43 43 to blocking buffer (0.1% bovine serum albumin (BSA)/ 0.05% Tween 20 in PBS) for 30 min at room temperature (RT). The plates were overlaid with E-Ig chimera for 1 hr at RT, followed by alkaline phosphatase conjugated anti-human IgG under the same conditions. After extensive washes with PBS and one wash with water, the TLC plates were immersed in nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution (Sigma-Aldrich, St. Louis, MO). A standard mixture of gangliosides (GM3, GM1, GD1a, GD1b, and GT1b), lacking glycans that support E-selectin binding, was used as a negative control. Brain gangliosides were purchased from Matreya (Pleasant Gap, PA) or purified from bovine brain extract [72], and GM3 was from Sigma-Aldrich. After development the TLC plates were washed with water and dried. Replicate plates were spotted and developed with sulfuric-resorcinol solution as described above Cell treatments Breast cancer cells were treated with 0.1 U/ml Vibrio cholerae sialidase (Roche Biochemicals, Indianapolis, IN) for 60 min at 37 C to remove terminal sialic acids. To cleave cell-surface proteins, breast cancer cells were treated with a broadly active protease, bromelain (Sigma-Aldrich), at 1% (w/v) for 60 min at 37 C. After enzyme treatments, cells were washed and resuspended in 0.1% BSA/Dulbecco s phosphate buffered saline (DPBS) Shear-dependent cell adhesion assays Cancer cells, with or without enzyme treatment, were perfused over HUVECs at a bone marrow microvasculature wall shear rate of 80 sec -1 [74] using a parallel plate flow chamber (Glycotech, Rockville, MD). E-selectin expression was induced in HUVECs

44 44 with 50 U/ml interleukin-1β (IL-1β; Calbiochem, San Diego, CA) at 37 C for 6 hr [26,48], and E-selectin contribution was assessed by pretreating activated HUVECs with a function-blocking anti-cd62e mab [26]. To observe adhesion in real-time, the flow chamber was placed on a Nikon TE300 inverted microscope equipped with a video camera. Adhesion events were recorded for 2 min and analyzed offline. Initial tethering was determined as the number of cancer cells attaching from the laminar flow stream, while cancer cells that remained stationary on HUVECs for more than 5 sec were considered firmly adherent. Cell rolling velocity was calculated by determining the distance traveled by a cell in 5 sec, using Image J software [48]. To determine whether gangliosides possess E-selectin ligand activity, CHO-E cells, untreated or treated with anti-cd62e mab, were perfused over mock treated or enzyme treated immobilized ganglioside spots at 80 sec -1 using the parallel plate flow chamber. Gangliosides were mixed in 1:1 methanol/water containing 25 µm phosphatidylcholine and 25 µm cholesterol [67]. Approximately 5 mm diameter (19.6 mm 2 ) ganglioside spots were prepared on a Petri dish using material equivalent to a cellular surface area of 19.6 mm 2 (BT-20 and MDA-MB-468 cell diameter are 18 ± 2 and 20 ± 2 µm, respectively). Dried ganglioside spots were then treated with sialidase (0.1 U/ml), bromelain (1% w/v) or buffer (mock treatment, no enzyme) for 30 min at 37 C. Dishes were blocked with 1% BSA at 37 C prior to use in adhesion assays. Observations of cell adhesion were made as described above.

45 Statistics Data are expressed as mean ± SEM of at least 3 independent experiments except where indicated. Statistical significance was determined by the Student s t-test with p 0.05 considered statistically significant. 2.4 Results BT-20 and MDA-MB-468 breast cancer cells express E-selectin ligands that mediate tethering on activated endothelium BT-20 and MDA-MB-468 cells were tested for E-selectin ligand activity under physiological flow conditions using the parallel plate flow chamber. Both breast cancer cell lines tethered (i.e., attached from the fluid stream) on IL-1 -activated HUVECs, and attachment levels were maintained when the cancer cells were perfused over activated HUVECs pretreated with isotype control antibody. However, the cells failed to tether on anti-e-selectin monoclonal antibody (anti-cd62e mab) treated (Fig. 2.1A) HUVECs or unactivated HUVECs, thus establishing cancer cell binding was mediated specifically by endothelial E-selectin. Additionally, a terminal sialylation requirement (characteristic of glycans supporting E-selectin adhesion) for attachment was verified by treating cancer cells with sialidase, which led to large and statistically significant reductions in tethering of both cell lines (Fig. 2.1A). Treatment efficacy was confirmed by the complete loss of expression of sialic acid-dependent HECA-452 antigens by flow cytometry. Therefore, sialylated glycans are required for optimal E-selectin-mediated tethering of breast cancer cells under flow conditions.

46 46 A * * * * B BT-20 MDA-MB-468 Cell count sle X sle A Figure 2.1. Breast cancer cell adhesion to IL-1β activated HUVECs is E-selectinmediated. (A) Untreated or sialidase treated (0.1 U/ml) BT-20 and MDA-MB-468 cells ( /ml) were perfused over activated, migg1 isotype treated, or anti-cd62e mab-treated HUVECs for 2 min at wall shear rate of 80 sec -1. Data are mean number of breast cancer cell tethering on HUVECs ± SEM, n = 3-6. *p < 0.05 with respect to untreated cell tethering. (B) Flow cytometric analysis of breast cancer cells labeled with anti-sle X or sle A mabs. Open curves show isotype and filled curves show specific mab reactivities.

47 BT-20 but not MDA-MB-468 cells express sle X and sle A Breast cancer cell expression of glycans correlative with E-selectin ligand activity was characterized by immunostaining and flow cytometry. Labeling with mabs specific to sialyl Lewis X (sle X ; CSLEX-1) and sialyl Lewis A (sle A ; KM-231) revealed that BT-20 cells strongly expressed these antigens, but MDA-MB-468 cells weakly expressed them (Fig. 2.1B). Similar trends were observed when cells were stained with HECA-452 mab, which collectively recognizes sialofucosylated oligosaccharides including sle X and sle A (Fig. 2.2). These results imply that BT-20 cells express classical carbohydrate antigens correlating with E-selectin ligand activity, but MDA-MB-468 cells may express other glycans that bind E-selectin Protease-insensitive molecules possess E-selectin ligand activity Breast cancer cells were treated with a broadly active protease, bromelain, to assess the role of glycoproteins versus glycosphingolipids as E-selectin ligands. As shown in Fig. 2.2, protease treated BT-20 cells expressed similar levels of HECA-452 antigen compared to untreated cells. This signal persistence after enzymatic protein cleavage suggested that sialofucosylated antigens were present on glycosphingolipids. Protease treated MDA-MB-468 cells expressed very low levels of HECA-452 antigen similar to untreated cells (Fig. 2.2). Consistent with earlier adhesion results (Fig. 2.1A), BT-20 as well as MDA-MB-468 cells exhibited reactivity with E-Ig chimera. Notably, E-Ig binding of both breast cancer cell lines did not decrease after protease treatment (Fig. 2.2), providing evidence of glycosphingolipids as E-selectin ligands. The efficacy of protein cleavage on both BT-20 and MDA-MB-468 cells was confirmed by flow

48 48 cytometry through the complete loss of E-cadherin expression. Additionally, the levels of GM1 on breast cancer cells (tested using cholera toxin B) were not changed, implying that protease treatment did not cleave glycosphingolipids. Collectively, these findings strongly indicate that E-selectin ligands expressed by breast cancer cells are glycosphingolipids that are HECA-452 positive (BT-20) as well as negative (MDA-MB- 468) Gangliosides are E-selectin ligands In order to confirm sialylated glycosphingolipids as E-selectin ligands, gangliosides from breast cancer cells were extracted and used in adhesion and immunooverlay experiments. When E-selectin-transfected Chinese hamster ovary (CHO-E) cells were perfused over immobilized gangliosides, CHO-E cells tethered to mock treated BT- 20 and MDA-MB-468 ganglioside spots (Fig. 2.3A). The binding was E-selectin specific, as anti-cd62e mab pretreated CHO-E cells failed to attach. When ganglioside spots were treated with protease, the number of tethering cells was the same as for mock treated spots, demonstrating that tethering was not due to contaminating proteins. Treatment with sialidase completely abolished CHO-E cell attachment, consistent with earlier results (Fig. 2.1). These findings definitively show that gangliosides expressed by breast cancer cells are ligands for E-selectin. Additional confirmation that gangliosides from breast cancer cells support E-selectin binding was achieved with immuno-overlay assays using E-Ig chimera (Figs 2.3B and C). Although BT-20 and MDA-MB-468 cells expressed distinct ganglioside profiles (Fig. 2.3B), reactive bands with the same TLC migration (rf) were detected by E-Ig chimera in both cells, suggesting that they may share

49 49 the same E-selectin ligands (Fig. 2.3C). In addition, the higher staining intensity observed in BT-20 gangliosides corroborates flow cytometry analysis using HECA-452 mab and E-Ig chimera (Fig. 2.2). BT-20 MDA-MB-468 Untreated Protease treated Untreated Protease treated HECA-452 Cell count E-Ig chimera CTB Figure 2.2. Protease treatment does not alter the E-selectin activity of breast cancer cells. Untreated or protease (bromelain, 1% w/v) treated BT-20 and MDA-MB-468 cells were labeled with HECA-452 mab, E-Ig chimera, or CTB and analyzed by flow cytometry. Open curves show isotype or negative controls, and filled curves show specific mab reactivities.

50 50 A Tethering cells/mm 2 CHO-E Immobilized Ganglioside BT-20 MDA-MB-468 * * * * - αcd62e Protease Sialidase B Standard BT-20 MDA-MB-468 C Standard BT-20 MDA-MB-468 GM3 GM1 GD1a GD1b GT1b HCl-resorcinol staining E-Ig chimera overlay Figure 2.3. Breast cancer cell gangliosides are E-selectin ligands. (A) Gangliosides extracted from breast cancer cells were immobilized on Petri dishes. Untreated or anti-cd62e mab-treated CHO-E cells were perfused over ganglioside spots at a wall shear rate of 80 sec -1. Spots were treated with buffer (mock treatment), protease (bromelain, 1% w/v), or sialidase (0.1 U/ml). Data are mean number of CHO-E cells interacting ± SEM, n = 5. *P < 0.05 with respect to untreated cell tethering. (b and c) A ganglioside standard mixture and breast cancer cell ganglioside extracts were resolved by TLC in replicate. Plates were separated in two, (B) stained with HCl-resorcinol 1 or (C) overlaid with E-Ig chimera 1. 1 This work was performed in collaboration with Nimrichter L. (Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazi.)

51 Gangliosides mediate tethering and rolling on endothelial E-selectin To assess the E-selectin ligand activity of glycosphingolipids in the native cell membrane, protease (bromelain) treated breast cancer cells were perfused over IL-1 activated HUVECs. As shown in Fig. 2.4A, tethering of protease treated breast cancer cells was not statistically different from that of untreated cells. Furthermore, protease treatment did not alter rolling velocities in a statistically significant manner (Fig. 2.4B). These findings demonstrate that breast cancer cell gangliosides are natural E-selectin ligands under physiological flow conditions. However, firm adhesion was significantly reduced after protease treatment as compared to firm adhesion of untreated cells (Fig. 2.4A). Altogether, these results indicate that breast cancer cells express gangliosides as E-selectin ligands required for cell tethering and rolling, but protein ligands facilitate firm adhesion. It is also notable that the percentage conversion of tethering into firm adhesion was significantly higher for untreated BT-20 cells (70 ± 11%) compared to that of untreated MDA-MB-468 cells (8 ± 2%), and that the rolling velocities of BT-20 cells were significantly lower than those of MDA-MB-468 cells (Fig. 2.4B). These data indicate that BT-20 cells express more and/or higher affinity E-selectin ligands than do MDA-MB-468 cells, which is consistent with earlier results (Figs. 2.2 and 2.3).

52 52 A Cells/mm Tethering BT-20 MDA-MB-468 Firm adhesion 100 * 0 - Protease - Protease * B Rolling velocity (µm/s) BT-20 MDA-MB-468 $ $ - Protease Figure 2.4. Effect of protease treatment on breast cancer cell adhesion to HUVECs. Untreated or protease treated (1% w/v) BT-20 or MDA-MB-468 cells ( /ml) were perfused over activated HUVECs for 2 min at a wall shear rate of 80 sec -1. Adhesive interactions were identified as initial tethering or firm adhesion as described in Materials and Methods. (A) Data are mean number of breast cancer cells tethering or firmly adhering on HUVECs ± SEM, n = 3-6. *p < 0.05 with respect to untreated cell adhesion. (B) Data represent mean rolling velocities ± SEM, n = $p < 0.05 with respect to BT-20 cell rolling velocity.

53 Discussion Several specific E-selectin ligands have been identified for a variety of cancer cell lines [26,50,51,75]. However, knowledge of breast cancer cell E-selectin ligands is lacking. Due to the involvement of glycosphingolipids in a diverse array of breast cancer metastatic processes and their role as adhesion molecules in several other types of cancers, the E-selectin ligand activity of breast cancer cell glycosphingolipids is of particular interest. In the present work, we have demonstrated that human BT-20 and MDA-MB-468 invasive breast cancer cell lines express gangliosides (sialylated glycosphingolipids) as E-selectin ligands that are operational under physiological flow conditions, irrespective of the presence of sialofucosylated epitopes recognized by classical carbohydrate antibodies. Our data clearly show that breast cancer cells specifically and avidly attach to activated HUVECs via E-selectin. It is widely accepted that E-selectin binds to ligands bearing sialofucosylated oligosaccharides, including epitopes that can be detected by the HECA-452 mab [38]. BT-20 cells expressed HECA-452 reactive antigens, particularly sle X and sle A oligosaccharides (Figs. 2.1B and 2.2). Hence, BT-20 cells likely express classical carbohydrate glycans as E-selectin binding domains. However, sle X, sle A, and HECA-452 antigens are correlative markers but not necessarily the exact epitopes that bind E-selectin [67,68]; non-classical E-selectin ligands also exist [76]. To wit, human myeloid cells express non-sle X E-selectin ligands [68]. Such novel ligands may be expressed by the MDA-MB-468 cell line, which possess E-selectin ligand activity while lacking significant HECA-452 reactivity (Figs. 2.1 and 2.2). In turn, the types of glycans

54 54 expressed on the breast cancer cells studied correlate with E-selectin ligand efficiency. BT-20, the HECA-452-positive cell line, showed higher conversion of cell tethering into firm adhesion and slower rolling velocity than that of the HECA-452-negative MDA- MB-468 cell line (Fig. 2.4A and B). Additionally, E-selectin binding affinity of BT-20 gangliosides, expressed on whole cells (Fig. 2.2), or extracted (Fig. 2.3A and C), was higher than that of MDA-MB-468 cells. Thus the breast cancer cell lines studied represent two broader classes of E-selectin binding glycans, one consisting of high efficiency sle X/A or sle X/A -like structures recognized by HECA-452 (BT-20), and the other expressing novel, low efficiency glycans (MBA-MB-468). However, both require sialic acid to maintain E-selectin ligand function (Fig. 2.1). Our findings do not discount the possibility of overlapping or redundant function of the various E-selectin ligands. Rather, adhesion of breast cancer cells is a complex interplay among different molecules, which is consistent with reports identifying distinct molecules as mediators of disparate adhesion events [48,74,77]. Whereas breast cancer gangliosides are clearly important for tethering and rolling on endothelium (similar to several other cancer cells and leukocytes [48,67-69]), proteins facilitate firm adhesion (Fig. 2.4). We tested BT-20 and MDA-MB-468 cells for known protein E-selectin ligands PSGL-1 [18], CD43 [24], CD66 [51], and PCLP [50], but none were detected (Fig. 2.5). E-selectin and its ligand CD44 variant 4 have recently been shown to participate in transendothelial migration of breast cancer cells by static (no flow) assays [78]; investigation of CD44 as an E-selectin ligand (i.e., HCELL [14]) under physiologic flow conditions is currently ongoing in our laboratory.

55 To the best of our knowledge, the present study is the first to report that gangliosides expressed by breast cancer cells are E-selectin ligands under physiological 55 flow conditions. This new role for breast cancer glycosphingolipids is particularly significant in light of a recent publication from the Hakomori group [70], in which it was clearly demonstrated that expression of gangliosides is altered during EMT [70], a process inducible by local factors at the metastatic site and may be essential for successful tumor spread [19,70]. Moreover, the gangliosides themselves may regulate EMT [70]. Altogether, breast cancer cell ganglioside expression appears to be a critically regulated process, in which certain gangliosides are present on circulating tumor cells for tissue-specific homing, while expression of others are altered post-homing. Further studies are warranted to investigate the regulation and synergy between different gangliosides in the metastatic cascade. In conclusion, we have shown that gangliosides on BT-20 and MDA-MB-468 breast cancer cell lines are E-selectin ligands that mediate adhesion to E-selectin expressing activated endothelium. These lipids are major participants in breast cancer cell tethering and rolling and cooperate with proteins to mediate firm adhesion. Moreover, there are breast cancer glycans not recognized by classic sialofucosylated epitope mabs (e.g., HECA-452 and CSLEX-1), which are responsible for E-selectin ligand activity. Further efforts are required to reveal the identities of novel glycans and lipid backbones that form functional E-selectin ligands. Ultimately, the structural identification of E-selectin ligands and their correlation with disease progression may lead to new biomarkers and/or treatments against breast cancer metastasis.

56 56 BT-20 MDA-MB-468 ZR-75-1 CD43 Cell count CD44 CD66 PCLP PSGL-1 Figure 2.5. Screening for molecules capable of carrying selectin reactive glycans are expressed by breast cancer cells. Expression of various E-selectin ligands was tested on the breast cancer cell lines by flow cytometry. Blue curves show isotype, and red curves show specific mab.

57 57 CHAPTER 3: MAC-2 BINDING PROTEIN IS A NOVEL E-SELECTIN LIGAND EXPRESSED BY BREAST CANCER CELLS Abstract Hematogenous metastasis involves the adhesion of circulating tumor cells to vascular endothelium of the secondary site. We hypothesized that breast cancer cell adhesion is mediated by interaction of endothelial E-selectin with its glycoprotein counter-receptor(s) expressed on breast cancer cells. At a hematogenous wall shear rate, ZR-75-1 breast cancer cells specifically adhered to E-selectin-expressing human umbilical vein endothelial cells when tested in parallel plate flow chamber adhesion assays. Consistent with their E-selectin ligand activity, ZR-75-1 cells expressed flow cytometrically detectable epitopes of HECA-452 mab, which recognizes high efficiency E-selectin ligands typified by sialofucosylated moieties. Multiple E-selectin reactive proteins expressed by ZR-75-1 cells were revealed by immunoprecipitation with E-selectin chimera (E-Ig chimera) followed by Western blotting. Mass spectrometry analysis of the 72 kda protein, which exhibited the most prominent E-selectin ligand activity, corresponded to Mac-2 binding protein (Mac-2BP), a heretofore unidentified E-selectin ligand. Immunoprecipitated Mac-2BP expressed sialofucosylated epitopes and possessed E-selectin ligand activity when tested by Western blot analysis using HECA-452 mab and E-Ig chimera, respectively, demonstrating that Mac-2BP is a novel 2 Previously published as: Mac-2 binding protein is a novel E-selectin ligand expressed by breast cancer cells. Shirure VS, Reynolds NM, and Burdick MM. PLoS One, 2012; e Text and figures used with permission.

58 58 high efficiency E-selectin ligand. Furthermore, silencing the expression of Mac-2BP from ZR-75-1 cells by shrna markedly reduced their adhesion to E-selectin expressing cells under physiological flow conditions, confirming the functional E-selectin ligand activity of Mac-2BP on intact cells. In addition to ZR-75-1 cells, several other E-selectin ligand positive breast cancer cell lines expressed Mac-2BP as detected by Western blot and flow cytometry, suggesting that Mac-2BP may be an E-selectin ligand in a variety of breast cancer types. Further, invasive breast carcinoma tissue showed co-localized expression of Mac-2BP and HECA-452 antigens by fluorescence microscopy, underscoring the possible role of Mac-2BP as an E-selectin ligand. In summary, breast cancer cells express Mac-2BP as a novel E-selectin ligand, potentially revealing a new prognostic and therapeutic target for breast cancer. 3.2 Introduction The five-year survival rate for breast cancer patients is almost 98% if the disease is detected in early stages. However, if the primary growth has metastasized to distant organs, the survival rate decreases drastically to 27% [79]. This bleak statistic emphasizes a need for greater understanding and better interventions for the prevention of metastasis. Metastatic invasion to distant organs is a systematic series of events, in which cancer cells dissociate from a primary tumor, enter the circulatory system, travel through the vasculature, attach to endothelium of a specific secondary site, and traverse the vascular wall to colonize the tissue. It is believed that the attachment of circulating tumor cells to endothelium occurs through a mechanism that is similar to the recruitment of leukocytes to inflamed tissue. According to this model, flowing leukocytes form initial

59 59 contacts (capture), which lead to continuous but transient interactions (rolling), and finally arrest of the cells on endothelium (firm adhesion). E-selectin expressed by endothelial cells is a well-recognized mediator of adhesion of cancer cells and cells of hematopoietic origin [24,26,32,48,60,66,80,81]. E-selectin engages its counter-receptors expressed on flowing cells, which not only captures and slows down the cells but also activates other mechanisms that promote tissue homing [15,24,82,83]. The significance of E-selectin mediated interactions in metastasis is apparent in several in vivo studies, wherein metastasis in mice was reduced when E-selectin and/or E-selectin ligand activity were blocked, compared to control conditions [23,84]. Therefore, understanding of E- selectin ligands expressed on cancer cells may be critical in devising new prognostic and therapeutic strategies against cancer metastasis. Several E-selectin ligands have been identified on human colon cancer, prostate cancer, leukemic, and hematopoietic cells, such as CD43, PSGL-1, CD44, PCLP, and CEA [3,24,26,47,49-51]. Although diverse proteins appear to function as E-selectin ligands, the core species of these ligands are primarily decorated with sialofucosylated carbohydrates such as the tetrasaccharide sialyl Lewis X (sle X ) and its stereoisomer sialyl Lewis A (sle A ), both of which are detectable by the HECA-452 monoclonal antibody (mab) [38,47,60,69]. In general, glycoproteins recognized by HECA-452 mab are believed to be high affinity E-selectin ligands [47,60,69]. Breast cancer cell lines possess E-selectin ligand activity and are known to express sle X, sle A, and HECA-452 mab reactive oligosaccharides [32,66,78,85-88]. Recently, we identified sialylated glycolipids (gangliosides) as breast cancer cell E-selectin ligands [85]. However, core

60 60 glycoprotein E-selectin ligands recognized by HECA-452 mab and expressed by breast cancer cells remain to be elucidated and are the focus of the work herein. Because reports identifying breast cancer glycoprotein E-selectin ligands are lacking, it is anticipated that breast cancer cells may express novel ligands. Of potential interest is Mac-2BP, a highly glycosylated protein that has been recognized as a cell adhesion molecule [89-91], but which has never been tested for E-selectin ligand activity. Notably, breast cancer patients with Mac-2BP over-expressing tumors are more likely to develop distant metastasis compared to patients with low Mac-2BP expressing tumors [92]. To explain the poor prognosis in the former type of patients, the study authors postulated that Mac-2BP acts as a cell adhesion molecule that promotes tumor cell interactions with vascular endothelium [92]. In addition, Mac-2BP is a well-documented counter-receptor of galectin-1 (Gal-1) [93], and some reports suggest cross-reactivity of Gal-1 ligands with E-selectin [49,94]. For instance, CD43, a Gal-1 ligand expressed by T-cells, is also an E-selectin ligand [49,94]. Despite these lines of evidence, neither E- selectin activity nor any other functional or mechanistic role of Mac-2BP in promoting tumor metastasis has been reported. In this study, we investigated the E-selectin ligands expressed by the bone metastatic ZR-75-1 breast carcinoma cell line [85]. We identified and characterized Mac- 2BP as a novel high affinity E-selectin ligand on ZR-75-1 cells, and also tested its presence on several other metastatic breast cancer cell lines and in pathological tissue samples. Ultimately, knowledge of E-selectin ligands expressed by breast cancer cells may lead to novel molecular tools to inhibit metastasis.

61 Materials and methods Cell culture Except where indicated, cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA). ZR-75-1 and T-47D breast cancer cell lines were cultured in RPMI (Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) and 1x penicillin-streptomycin (Life Technologies). The BT-20 breast cancer cell line was grown in minimum essential medium (MEM; Life Technologies) with 10% FBS and 1x penicillin-streptomycin. Hs-578T, MDA-MB-231, MDA-MB-468, and MCF-7 breast cancer cells were maintained in Dulbecco s modified eagle medium (Life Technologies) with 15% FBS and 1x penicillin-streptomycin. E- selectin transfected Chinese hamster ovary cells (CHO-E) were a generous gift from Dr. Robert Sackstein (Harvard Medical School, Boston, MA). CHO-E cells were maintained in MEM supplemented with 10% FBS and 0.1 mm non-essential amino acids (Life Technologies). Human umbilical vein endothelial cells (HUVECs) were obtained from Lonza, Inc. (Allendale, NJ) and cultured in Medium 199 (Lonza, Inc.) supplemented with 10% FBS, 50 µg/ml endothelial mitogen (Biomedical Technologies, Stoughton, MA), 50 µg/ml heparin (Sigma-Aldrich, St. Louis, MO), 2 mm L-glutamine, and 1x penicillinstreptomycin. One day prior to experiments, HUVECs were plated in sterilized 6.5 mm diameter flexiperm gaskets (Greiner Bio-one, Monroe, NC) placed at the center of 35 mm tissue culture dishes [95].

62 Antibodies All primary antibodies were monoclonal antibodies (mabs), unless otherwise noted. Anti-human CD43 (1G10), CD44 (515), CD62E (68-5H11), CD66 (COL-1), PSGL-1 (KPL-1), HECA-452 (anti-cutaneous lymphocyte antigen recognizing sle X, sle A, and related sialofucosylated moieties), sle X (CSLEX-1), and all isotype controls were obtained from BD Biosciences (San Jose, CA). Anti-human Mac-2BP mab (Sp-2) [96] was from ebioscience (San Diego, CA), and anti-human Mac-2BP polyclonal antibody (pab) [97] and anti-pclp (3D3) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Recombinant mouse E-selectin/human Fc chimera (E- Ig chimera) was from R & D Systems (Minneapolis, MN). Anti-sLe A antibody (KM- 231) was from Calbiochem (San Diego, CA). Fluorescein isothiocyanate (FITC)- conjugated and alkaline phosphatase (AP)-conjugated polyclonal secondary antibodies were from Southern Biotech (Birmingham, AL). AlexaFluor 488- and AlexaFluor 568- conjugated secondary antibodies were obtained from Life Technologies Enzyme treatments and cell surface biotinylation Breast cancer cells were treated with 0.1 U/ml Vibrio Cholerae neuraminidase (Roche Biochemicals, Indianapolis, IN) for 60 min at 37 C to cleave terminal sialic acid residues. Cell surface proteins were removed by treating cells with a general protease, bromelain (Sigma-Aldrich), at 1% for 60 min at 37 C [48,85]. After enzyme treatment cells were washed and incubated with 0.1% BSA to block non-specific interactions. Cell surface proteins were biotinylated by EZ-link sulfo-nhs-lc-biotin kit (Pierce Biotechnology, Rockford, IL) according to the manufacturer s protocol.

63 Flow cytometry All antibody solutions were prepared at 10 µg/ml concentration in blocking buffer (0.1% BSA in Dulbecco s phosphate buffered saline; DPBS). Cells were washed with blocking buffer and incubated with primary antibody or isotype control, to determine the background fluorescence level, for 30 min at 4 C. Cells were washed and incubated with secondary antibody for 30 min at 4 C [26,85]. Finally, cells were washed and analyzed using a FACSAria Special Order Research Product flow cytometer/sorter (BD Biosciences) RNA interference Mac-2BP knockdown of ZR-75-1 cells was performed using MISSION shrna lentiviral transduction particles (Sigma-Aldrich, St. Louis, MO) prepared from plko.1 vector (TRC 1 version). The cell transductions with viral particles containing empty vector or specific sequence were carried out at an optimized concentration of viral particles (20 multiplicity of infection; MOI). From a library of five constructs, a construct (5 -CCGGGTACTTCTACTCCCGAAGGATCTCGAGATCCTTCGGGAGTAGAAGTA CTTTTT-3 ; underlined portion is sense and italic portion is antisense sequence; TRCN ) that produced significantly high levels of knockdown of Mac-2BP as determined by flow cytometry was chosen Parallel plate flow chamber adhesion assay The flow adhesion assays were performed using a parallel plate flow chamber (Glycotech, Rockville, MD) placed on a Nikon TE300 inverted microscope equipped with a CCD video camera. Prior to experiments HUVECs were activated to express E-

64 64 selectin by treatment with 50 U/ml interleukin-1β (IL-1β; Calbiochem) at 37 C for 6 hr [48,85]. In certain experiments, activated HUVECs were treated with anti-cd62e mab to block E-selectin function. Breast cancer cells with or without enzyme treatment were perfused over activated or activated and anti-cd62e mab-treated HUVECs. All experiments were performed at a bone marrow microvasculature shear rate of 80 s -1 [74], equivalent to 0.8 dynes/cm 2 in our system, and recorded for 2 min for later analysis. The number of adhering cells included all cells attaching from the free fluid stream [48,85]. In the other flow adhesion experiments designed for cell detachment analysis, wild type, empty vector transduced or Mac-2BP shrna transduced ZR-75-1 cells were perfused over a monolayer of CHO-E cells for 4 min at a shear rate of 80 s -1. Subsequently, the shear rate was doubled in time steps of 30 s up to the final shear rate of 2560 s -1. The numbers of adhering cells corresponding to each shear stress were counted at the end of the 30 s time intervals. The percentage of attached cells was found with respect to attached cells at 80 s -1. Cell velocity was calculated by capturing still images from a video over 5 s intervals and measuring cell displacement using Image J software [48] Cell lysis and immunoprecipitation Cell lysates were prepared in lysis buffer containing 1% Triton X-100, 0.02% NaN 3, 150 mm NaCl, 0.5 mm Tris (ph 10.4), 1 mm EDTA and protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN). To immunoprecipitate Mac-2BP, cell lysates were incubated with anti-mac-2bp mab [98] and protein G agarose beads (Life Technologies) overnight at 4 C under constant rotation. Antigen-antibody-bound protein G beads were subsequently washed with lysis buffer, and the beads were then

65 65 incubated with Laemmli reducing sample buffer and heated to 100 C for 5 min to release the Mac-2BP. The immunoprecipitates were subsequently subjected to SDS-PAGE and Western blotting [47]. To immunoprecipitate E-selectin reactive proteins, cell lysates were prepared in a modified lysis buffer containing 1 mm CaCl 2 but no EDTA. To reduce non-specific binding to the human Fc portion of E-Ig chimera, lysate was pre-cleared by incubating with human IgG (h-igg) isotype control (5 µg/10 million cell lysate) and protein G beads. The pre-cleared cell lysate was then incubated overnight with E-Ig chimera (5 µg/10 million cell lysate) and protein G beads at 4 C with constant rotation. After sufficiently washing antigen-antibody-bound protein G beads, E-Ig chimera reactive antigens were eluted using elution buffer (5 mm EDTA, 50 mm Tris (ph 7.4), and 0.1% Triton-X-100). Eluted samples were subjected to SDS-PAGE and Western blotting SDS-PAGE and Western blotting Cell lysates or immunoprecipitates were resolved on 4-15% Tris-HCl precast gels (Bio-Rad Laboratories, Hercules, CA) by reducing SDS-PAGE and subsequently transferred to polyvinylidene difluoride (PVDF; Bio-Rad) membrane [26,47]. Western blots were probed with appropriate antibodies or isotype controls and AP-conjugated secondary antibodies Mass spectrometry analysis of E-selectin reactive protein E-Ig chimera immunoprecipitate isolated from ZR-75-1 cell lysate was resolved by SDS-PAGE. In parallel, E-Ig chimera immunoprecipitate from surface biotinylated cell lysate was Western blotted with streptavidin-ap to serve as a reference for protein

66 66 migration. The band corresponding to a molecular weight of 72 kda was excised and sent to Protea Biosciences (Morgantown, WV) for mass spectrometry analysis. The gel fragments were digested with trypsin and were analyzed by liquid chromatography interfaced to matrix assisted laser desorption/ionization-time of flight-mass spectrometry (LC-MALDI-TOF-TOF MS/MS). The mass spectrometry data was subjected to the ProGroup algorithm in Applied Biosystems ProteinPilot 3.0 software, Paragon search engine, and Swissprot database. The protein confidence expressed as Prot Score was more than 99% significant with p < 0.01 [99] Immunofluorescence microscopy and image deconvolution Cells were labeled with primary antibody or isotype control as described for flow cytometry. After primary incubation, the cells were washed and incubated with appropriate AlexaFluor 488 (green)-and/or AlexaFluor 568 (red)- conjugated secondary antibody at 4 µg/ml for 30 min at 4 C. After washing, cells were fixed in 4% methanolfree paraformaldehyde in DPBS [100]. The cells were mounted on glass slides in ProLong Gold antifade reagent with DAPI (Life Technologies). Breast cancer tissue slides were prepared for fluorescence immunohistochemistry as follows. Formalin fixed paraffin embedded (FFPE) breast invasive ductal carcinoma tissue slides (US Biomax, Rockville, MD) [88] were deparaffinized by heating at 60 C for 30 min, and serially incubating with xylene (three times, 10 min each), 95% ethanol, 70% ethanol, and deionized water (once, 5 min each) [101]. For antigen retrieval, the deparaffinized tissue slides were heated at 95 C for 30 min with 10 mm sodium citrate at ph 6.0. The tissue slides were blocked with 1% BSA and 1% FBS in DPBS for 30

67 67 min, and incubated for 1 hour with HECA-452 mab [69] and anti-mac-2bp pab [97] at 5 µg/ml and 10 µg/ml, respectively. The slides were washed and incubated with appropriate AlexaFluor conjugated secondary antibodies at 4 µg/ml for 30 min at room temperature. After washing, the tissue slides were mounted in ProLong Gold antifade reagent for microscopy [101]. Tissue and cell slides were imaged using 10x or 40x objectives, respectively, under wide field fluorescence using a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany) equipped with a motorized high precision specimen stage and an automated optical filter cube wheel with appropriate excitation and emission filters. Images were acquired using Simple PCI software (Hamamatsu Corporation, Sewickley, PA), and were subjected to 2D or 3D blind deconvolution algorithms in AutoQuant X software (Media Cybernetics, Bethesda, MD) to reduce out of focus light. Image projections of processed 3D images were also generated using AutoQuant X [100]. Manders overlap coefficient, which indicates quantitative co-localization on the scale of 0 (no overlap) to 1 (100% overlap), was found by using the JACoP plugin in ImageJ 1.40g software. The thresholds for the calculations were set to automatically retrieved background values [13] Statistics Data are expressed as mean ± SE for at least 3 independent experiments except where indicated. Statistical significance of differences between means was determined by paired Student s t-test, and probability values of P 0.05 were considered statistically significant.

68 Results Adhesion of ZR-75-1 breast cancer cells to HUVECs is mediated by E-selectin The E-selectin ligand activity of the bone metastatic ZR-75-1 breast carcinoma cell line was tested by perfusing the breast cancer cells over IL-1 activated HUVECs under physiological flow conditions in the parallel plate flow chamber adhesion assay. The breast cancer cells attached to activated HUVECs, but the binding was completely diminished when the HUVECs were treated with E-selectin function blocking mab (Fig. 3.1A), demonstrating that the adhesion of ZR-75-1 cells to HUVECs is specifically mediated by E-selectin. When sialidase treated ZR-75-1 cells were perfused over activated HUVECs, the adhesion of treated ZR-75-1 cells was significantly reduced as compared to the adhesion of untreated cells (Fig. 3.1A), showing the need for sialylated glycans for optimal adhesion of cancer cells. The data are consistent with previously reported data that terminal sialylation is necessary for E-selectin ligand function [48,60,69,85]. Collectively, these results demonstrate that the adhesion of ZR-75-1 cells to activated endothelium is specifically mediated by the binding of sialylated ligands to endothelial E-selectin. Expression of sialylated oligosaccharides on breast cancer cells that are indicative of E-selectin ligand activity [48,69,85] was tested by flow cytometric analysis. ZR-75-1 cells stained with anti-sle X (CSLEX-1) and anti-sle A (KM-231) mabs showed positive expression for both types of antigens (Fig. 3.1B). Furthermore, ZR-75-1 cells were reactive with HECA-452 mab (Fig. 3.1C), which broadly recognizes sialofucosylated

69 69 molecules [38,69]. Thus, ZR-75-1 cells express classical E-selectin binding sialofucosylated epitopes, including sle X and sle A ZR-75-1 cells do not express known E-selectin ligands We have previously shown that BT-20 and MDA-MB-468 breast cancer cell lines express ganglioside E-selectin ligands [85]. However, ZR-75-1 gangliosides did not possess detectable E-selectin ligand activity by immuno-overlay and lipid perfusion assays, techniques described previously [85]. Consistent with these results, ZR-75-1 cells treated with bromelain, a general protease, showed reduced HECA-452 mab reactivity compared to untreated cells when tested by flow cytometry (Fig. 3.1C), indicating that proteins, rather than lipids, preferentially express appropriate carbohydrate modifications required for E-selectin ligand activity. The efficacy of bromelain treatment in these experiments was confirmed by loss of expression of E-cadherin, which was nearly reduced to isotype control levels after treatment. The absence of candidate glycolipid ligands implied that glycoproteins are the major E-selectin ligands expressed by ZR-75-1 cells. However, known glycoprotein E- selectin ligands PSGL-1, CD43, CD44, CD66, and PCLP [3,24,26,47,49-51] were not detected on ZR-75-1 cells by flow cytometric analysis (Chapter 2). These results strongly suggested that ZR-75-1 cells may express previously unidentified glycoprotein ligands.

70 70 A B sle X Attached cells/mm 2 * * Cell count sle A Fluorescence intensity C Untreated Protease treated Cell count HECA-452 HECA-452 Fluorescence intensity Figure 3.1. ZR-75-1 cell adhesion to IL-1β activated HUVECs is E-selectin mediated. (A) ZR-75-1 cells (10 6 /ml) were perfused over activated (left bar) or anti-cd62e mabtreated activated HUVECs (middle bar), or sialidase (neuraminidase; 1% at 37 C for 30 min) treated ZR-75-1 cells were perfused over activated HUVECs (right bar) for 2 min at a wall shear rate of 80 s -1. Data are mean ± SE for n = 3 6 independent experiments. *P < 0.05 with respect to untreated cells control. (B) ZR-75-1 cells were surface labeled with anti-sle X (CSLEX-1) or anti-sle A (KM-231) mabs and analyzed by flow cytometry. Open curves show isotype and filled curves show specific mab. (C) Untreated or protease treated (1% bromelain at 37 C for 1 hr) ZR-75-1 cells were surface labeled with HECA-452 mab and analyzed by flow cytometry. Open curves show isotype, and filled curves show specific mab.

71 Identification of a novel E-selectin ligand To screen for glycoprotein E-selectin ligands, E-Ig chimera immunoprecipitates obtained from lysates of surface biotinylated cells were Western blotted with streptavidin-ap. Several bands were revealed (Fig. 3.2A), confirming that ZR-75-1 cells express protein E-selectin ligands. The band corresponding to a molecular weight of 72 kda showed the highest E-selectin activity as found by image intensity analysis of the Western blot (Fig. 3.2A). To identify this protein, the portion of the gel corresponding to the 72 kda immunoprecipitated protein from ZR-75-1 cell lysate was excised and submitted for mass spectrometry analysis. Multiple peptides matched to Mac-2BP, generating a ProtScore equivalent to 99.99% confidence for the protein identification. Previously, a similar form of Mac-2BP (~75 kda) was found to be expressed by other cancer cell lines and tissues [98,102,103] and was shown to be a cell surface form of the protein that is distinct from the secreted form of Mac-2BP (~90 kda) [98] Mac-2BP expressed by ZR-75-1 cells is an E-selectin ligand To confirm breast cancer cell expression of Mac-2BP and its E-selectin ligand activity, a series of Western blots were performed. ZR-75-1 cell lysate probed with Mac- 2BP antibody revealed a band corresponding to 72 kda (Fig. 3.2B), verifying the expression of Mac-2BP at the same molecular weight of the major E-Ig immunoprecipitated protein. Furthermore, staining of immunoprecipitated Mac-2BP by E-Ig chimera or HECA-452 mab (Fig. 3.2B) showed that Mac-2BP expressed by ZR-75-1 cells is a sialofucosylated E-selectin ligand, confirming the results described earlier (Figs 3.1 and 3.2A).

72 72 A ZR-75-1 cell lysate + protein G + E-Ig chimera kda 170 Staining Intensity Gel band Extraction and digestion Bioinformatics analysis Elution with EDTA Protein fragments Mac-2BP E-selectin reactive molecules 72 - (LC-MALDI-TOF- TOF MS/MS) Immunoprecipitation SDS-PAGE Mass spectrometry Data analysis B kda Lysate Mac-2BP ippt Mac-2BP ippt Mac-2BP E-Ig HECA- Figure 3.2. Identification of a novel protein, Mac-2BP, as an E-selectin ligand. (A) The steps followed for identification of the E-selectin reactive protein. Biotinylated ZR-75-1 cells ( ) were lysed and immunoprecipitated using E-Ig chimera. Immunoprecipitates were resolved by SDS-PAGE and subsequently blotted with streptavidin-ap. The intensity histogram was obtained by analysis of digitalized image using Image Lab software. (B) Lysate from ZR-75-1 cells ( ) was subjected to Western blotting with anti-mac-2bp pab. Immunoprecipitates from ZR-75-1 cell lysate ( cells) by anti-mac-2bp mab were Western blotted with E-Ig chimera or HECA- 452 mab.

73 73 Since functional E-selectin ligands must be natively exposed on the cell surface, the expression of Mac-2BP on ZR-75-1 cells was tested by flow cytometry. As shown in Figure 3.3A, ZR-75-1 cells were positive for Mac-2BP compared to the isotype control. To demonstrate that the cell surface Mac-2BP possessed E-selectin binding epitopes, fluorescence microscopy of ZR-75-1 cells with anti-mac-2bp pab and E-Ig chimera was performed. The results revealed positive cell surface expression of Mac-2BP as well as E- Ig chimera reactivity compared to the respective isotype controls (Fig. 3.3). Moreover, the overlapped image (Fig. 3.3B) shows distinct co-localization of Mac-2BP and E- selectin ligand activity. The quantitative analysis by Manders coefficient showed that more than 85% of Mac-2BP (green) co-localizes with E-selectin ligands (red), implying the majority percentage of Mac-2BP may possess E-selectin ligand activity. The analysis also indicated that more than 55% of E-selectin ligands overlap with Mac-2BP, implying that the majority of Mac-2BP possesses E-selectin ligand activity, as well as suggesting the presence of other E-selectin ligands, in support of the earlier results (Fig. 3.2A). Together, these data provide further evidence that Mac-2BP expressed by ZR-75-1 cells is a cell surface glycoprotein that possesses E-selectin ligand activity.

74 74 A Mac-2BP Cell count B Fluorescence intensity Mac-2BP E-Ig chimera Mac-2BP + E-Ig chimera rabbit IgG human IgG rabbit IgG + human IgG Figure 3.3. Immunostaining of ZR-75-1 cells shows co-localization of signals for Mac- 2BP and E-selectin ligand activity. (A) ZR-75-1 cells were surface labeled with anti-mac-2bp pab and analyzed by flow cytometry. Open curve shows isotype, and filled curve shows specific antibody. (B) ZR cells were dually surface labeled with anti-mac-2bp pab (green) and E-Ig chimera (red) or with corresponding isotype controls. Images of slices, 0.5 µm apart, were obtained in epifluorescence microscopy, and projected to obtain a composite image. The composite image was deconvoluted using AutoQuant X software. Co-localization of two molecules is shown in the overlapped image (orange). Scale bar indicates 10 µm.

75 Mac-2BP regulates E-selectin mediated rolling and adhesion of ZR-75-1 To investigate the native E-selectin ligand activity of Mac-2BP, ZR-75-1 cells were transduced with shrna for Mac-2BP, or empty vector as a negative control, and tested in parallel plate flow chamber adhesion assays. Mac-2BP silenced cells, compared to vector cells, expressed significantly reduced levels of Mac-2BP (Fig. 3.4A), showing an efficient knockdown. In experiments performed to test adhesion specificity, transduced ZR-75-1 cells perfused at shear rate of 80 s -1 adhered to CHO-E cells, but the adhesion was completely abrogated when CHO-E cells were treated with E-selectin function blocking mab, indicating that the adhesion was specifically mediated by E- selectin. Interestingly, the adhesion of Mac-2BP silenced cells (85 ± 9 cells/mm 2 ) to CHO-E cells at a shear rate of 80 s -1 was significantly less than that of vector cells (121 ± 13 cells/mm 2 ; Fig. 3.4B), demonstrating that Mac-2BP is a crucial E-selectin ligand under physiological flow conditions. Furthermore, the rolling velocity of silenced cells was significantly higher than that of vector cells (Fig. 3.4C), demonstrating the role of Mac-2BP-E-selectin ligation in controlling cell rolling velocity, a classical function of E- selectin ligands [48,64]. Similar levels of adhesion and rolling velocities were found for wild type ZR-75-1 and vector cells, indicating that the cell transduction did not nonspecifically alter cell adhesion function. For detachment analysis, which was performed by sequentially increasing the shear rate from 80 s -1, the adhesion data of vector and silenced cells were normalized with number of attached cells of respective types at 80 s -1 (Fig. 3.4D). A consistently lower percentage attachment of silenced cells, relative to that of vector cells, at all shear rates studied was observed (Fig. 3.4D). These data collectively

76 76 demonstrate Mac-2BP as an E-selectin ligand necessary for regulating cell adhesion under hemodynamic flow conditions. The E-selectin ligand activity of Mac-2BP was also found in separate experiments performed using IL-1β activated HUVECs as E-selectin expressing cells. A significantly lower number of Mac-2BP silenced cells (34 ± 4 cells/mm 2 ) compared to vector cells (58 ± 9 cells/mm 2 ), attached to the HUVECs, consistent with CHO-E data (Fig. 3.4B). Yet no significant difference was found between the rolling velocities of vector and silenced cells over HUVECs (9.0 ± 1.0 versus 9.1 ± 1.3 µm/s, respectively). These results may indicate alternative molecular pathways for regulating rolling of ZR-75-1 cells to activated endothelium, and these data are consistent with other reports suggesting alternative cell adhesion pathways for cancer cell adhesion to activated endothelium [48,65,101] Mac-2BP is expressed by several other breast cancer cell lines that also possess E- selectin ligand activity Although identification of a ubiquitous E-selectin ligand expressed by multiple breast cancer cell lines is lacking, many breast cancer cell lines have been previously shown to possess E-selectin ligand activity [32,66,78,85-87]. Yet the types of assays and assay conditions used in these studies vary greatly [32,66,78,85-87]. Therefore, flow cytometry using E-Ig chimera and carbohydrate antibodies was performed to test for E- selectin ligand activity and to compare carbohydrate expression profiles of several widely studied breast cancer cell lines: ZR-75-1, BT-20, MDA-MB-468, T-47D, MDA-MB-231, MCF-7, and Hs-578T. The results showed that all of these cell lines possess detectable E-

77 77 selectin ligand activities and express carbohydrates indicated in E-selectin ligand activity (Table 1). Further, Western blotting of cell lysates or flow cytometry analysis of cells revealed that all of these cell lines express Mac-2BP (Figs 3.5A and B). Since BT-20 cells possess high E-selectin ligand activity (Table 1 and [85]) and are derived directly from a primary invasive breast cancer tumor [71], they may be used as a model cell line to validate the E-selectin ligand activity of Mac-2BP expressed in the primary site. Mac- 2BP immunoprecipitated from BT-20 cells stained with E-Ig chimera revealed Mac-2BP as an E-selectin ligand, similar to ZR-75-1 cells (Fig. 3.5C) Expression of Mac-2bp correlates with E-selectin ligand activity in invasive breast cancer tissue To investigate the presence of Mac-2BP as an E-selectin ligand in breast cancer tissue, fluorescence immunostaining of breast invasive ductal carcinoma tissue was performed. Parallel sections from breast cancer tissue were dual stained to reveal expression of Mac-2BP and HECA-452 antigens. Tissue staining with specific antibodies was positive compared to respective isotype controls, illustrating that Mac-2BP and HECA-452 antigens are expressed in breast cancer tissue (Fig. 3.6). A composite image revealed a visibly distinct co-localization of signals (arrow heads in Fig. 3.6), quantification of which by Manders coefficient showed that more than 70% of Mac-2BP co-localizes with HECA-452 antigens and more than 40% of sialofucosylated antigens relate with Mac-2BP. Collectively, these results suggest that breast cancer tissue expression of Mac-2BP is related to high affinity E-selectin ligand activity.

78 78 Table 3.1 Diverse Breast Cancer Cell Lines Possess E-selectin Ligand Activity. Cell line Tumor type and Specimen site sle X sle A E-selectin stage [71] [71] ligand activity ZR-75-1 IDC (IV) Ascites BT-20 IDC (NR) Primary tumor MDA-MB-468 AC (IV) Pleural effusion T-47D IDC (IV) Pleural effusion MDA-MB-231 IDC (IV) Pleural effusion MCF-7 IDC (IV) Pleural effusion Hs-578T IDC (NR) Primary tumor Breast cancer cells were surface labeled with anti-sle X mab (CSLEX-1), anti-sle A mab (KM-231), or E-Ig chimera and analyzed by flow cytometry. The positive mean fluorescence intensities, compared to respective isotype control, are categorized as high (++) or low (+). The negative/very low intensities with respect to respective isotype control are indicated by (-). IDC, invasive ductal carcinoma; AC, adenocarcinoma; NR, not reported.

79 79 A B Mean fluorescence intensity Isotype Vector cells Silenced cells * Mac-2BP Attached cells/mm Vector cells * Silenced cells C D Rolling velocity (µm/s) Vector cells * Silenced cells % attachment Vector cells * Silenced cells * * Shear rate (s -1 ) Figure 3.4. Mac-2BP silencing of ZR-75-1 cells reduces their adhesion to E-selectin. (A) Vector or Mac-2BP silenced cells were surface labeled with anti-mac-2bp pab and analyzed by flow cytometry. The data are represented as mean fluorescence intensity ± SE for n = 5 independent experiments. *P < 0.05 with respect to vector. (B) Vector or Mac-2BP silenced cells (10 6 /ml) were perfused over CHO-E cells for 4 min at a wall shear rate of 80 s -1 and the number of adhering cancer cells were counted. Data are mean ± SE for n = 5 independent experiments. *P < 0.05 with respect to vector. (C) The rolling velocity of vector and silenced cells over CHO-E cells was determined at a wall shear rate of 80 s -1. Data are mean ± SE for n = 10 cells. *P < 0.05 with respect to vector. (D) Vector or silenced cells (10 6 /ml) were perfused over CHO-E cells for 4 min at a wall shear rate of 80 s -1 and then the shear rate was doubled in 30 s time intervals. Data are mean ± SE for n = 5 independent experiments. *P < 0.05 with respect to vector.

80 80 A 72 kda (Mac-2BP) BT-20 ZR-75-1 T-47D MCF-7 MDA- MDA-MB- 231 C 72 kda (Mac-2BP ippt) ZR-75-1 BT kda (β-actin) B BT-20 T-47D MDA-MB-468 Cell count MCF-7 MDA-MB-231 HS-578T Fluorescence intensity Figure 3.5. Several breast cancer cell lines express Mac-2BP and Mac-2BP of BT-20 cells possess E-selectin ligand activity. (A) Lysates of 2x10 6 cells obtained from a variety of breast cancer cell lines were subjected to Western blot analysis with anti-mac-2bp pab. β-actin staining was used as the loading control. (B) Breast cancer cells were surface labeled with anti-mac-2bp pab and analyzed by flow cytometry. Open curve shows isotype, and filled curve shows specific antibody. (C) Immunoprecipitates from ZR-75-1 or BT-20 cell lysate ( cells) by anti-mac-2bp mab were subjected to Western blotting with E-Ig chimera.

81 81 Mac-2BP HECA-452 Mac-2BP+HECA-452 Mac-2BP HECA-452 rabbit IgG rat IgM rabbit IgG+rat IgM Figure 3.6. Immunofluorescence analysis shows co-localization of Mac-2BP and E- selectin ligand activity on breast cancer tissue. Deparaffinized breast invasive ductal carcinoma tissue was labeled with anti-mac-2bp pab (red) and HECA-452 mab (green) or isotype controls. Co-localization of two signals is shown in the overlapped image (orange). Scale bar indicates 100 µm.

82 Discussion In distant metastasis, circulating tumor cells attach to the vasculature of a remote tissue or organ by engaging with endothelial cell adhesion molecules. Although breast cancer cell adhesion to endothelial cells has been shown to be mediated by E-selectin [32,66,78,85-87], a limited number of reports have characterized breast cancer cell E- selectin ligands. In stark contrast, several E-selectin ligands have been identified on leukemic, colon, and prostate cancer cells [24,26,47,48,50,51]. In the present report we bridged this knowledge gap by investigating the E-selectin mediated adhesion and E- selectin ligands of ZR-75-1 breast cancer cells. We have identified and characterized a new glycoprotein ligand, Mac-2BP, which to the best of our knowledge has not been shown previously to possess E-selectin ligand activity. Mac-2BP is upregulated in advanced stage cancer tumors, and its expression positively correlates to development of distant metastasis [92,104]. Our study thus ascribes a functional role for Mac-2BP in cell adhesion by identifying it as an E-selectin ligand, and potentially explains one of its roles in cancer metastasis. ZR-75-1 breast cancer cells specifically adhered to endothelial E-selectin under physiological flow conditions (Fig. 3.1A), consistent with other breast cancer cell lines [66,85]. The E-selectin ligand activity of these cells was associated with sialofucosylated carbohydrates including sle X and sle A (Figs 3.1B and C), which have been reported to provide E-selectin ligand activity to core protein or lipid molecules [25,47,69]. Notably, mainly glycoproteins rather than glycolipids were sialofucosylated in ZR-75-1 cells (Fig. 3.1C). Hence, glycoproteins possessed appropriate carbohydrate modifications for E-

83 83 selectin ligand activity. It is believed that cell surface proteins are preferred E-selectin ligands for the initiation of endothelial adhesion under hematogenous flow conditions. They can extend farther than lipids from the cell surface and therefore can easily make initial contact with endothelial E-selectin. In fact, many human cell lines and native cells have been shown to employ glycoproteins ligands for E-selectin mediated adhesion in vitro and in vivo [3,23,24,26,47,49-51]. ZR-75-1 cells express Mac-2BP as a heretofore unrecognized E-selectin ligand, but known E-selectin ligands were undetected. Furthermore, several other potentially novel E-selectin ligands were found (Fig. 3.2A). Identification and characterization of these proteins are in progress in our laboratory. The putative ability of Mac-2BP expressed by ZR-75-1 cells to mediate E-selectin binding was investigated by using HECA-452 mab, which detects sialofucosylated epitopes that are purported to confer high efficiency E-selectin ligand activity. Several investigators have used HECA-452 mab for analyzing E-selectin ligands. For example, a glycoform of CD44 known as HCELL (hematopoietic cell E-/L-selectin ligand), a major E-selectin ligand expressed by human colon cancer and hematopoietic stem cells, was found to be reactive to HECA-452 mab by Western blot [47,60]. A similar approach in our study revealed that Mac-2BP is a high efficiency E-selectin ligand, whose ligand activity is primarily associated with sialofucosylated epitopes detectable by HECA-452 mab (Fig. 3.2B). However, we do not discount the possible existence of Mac-2BP glycoforms lacking HECA-452 mab reactivity, especially because some fraction of Mac-2BP expressed in tumor tissue was not reactive to HECA-452 mab (Fig. 3.6). In agreement with this notion, breast cancer

84 84 cell lines express diverse carbohydrate profiles yet are positive for E-selectin ligand activity (Table 1 and [85,86]). Since Mac-2BP possessed putative structural features necessary for E-selectin ligand activity, the intricate functional details of Mac-2BP expressed on intact cells were unraveled in parallel plate flow chamber adhesion experiments. Selectins and their ligands mediate various steps of adhesion of circulating cells to endothelium [24-26,48,80]. At bone marrow vascular flow conditions, Mac-2BP expressed by ZR-75-1 cells exhibited E-selectin ligand activity necessary for cell capture and controlling cell rolling velocity (Fig. 3.4C). Furthermore, the sequential increase in cell detachment as shear rate was increased suggested that Mac-2BP regulates firm attachment of cells (Fig. 3.4D). In totality, these data clearly demonstrate that Mac-2BP is a potent E-selectin ligand that regulates various stages of the adhesion cascade. In addition to ZR-75-1 cells, several other breast cancer cell lines possessing E- selectin ligand activities (Table 1) expressed Mac-2BP (Fig. 3.5). The majority of these cell lines metastasize to mouse bone marrow [85], which is known to express E-selectin [74,87]. In humans, breast cancer frequently spreads to bone marrow [93], and human bone marrow endothelium constitutively expresses E-selectin [31]. These findings lead to the notion that the Mac-2BP-E-selectin cell adhesion pathway is crucial in breast cancer metastasis. In support of this notion, staining of breast carcinoma tissue indicated expression of E-selectin reactive Mac-2BP in invasive breast cancer tumors (Fig. 3.6). Furthermore, previous studies have reported that cancer patients with Mac-2BP overexpressing tumors are more likely to develop distant metastasis, have shorter disease free

85 85 survival, and have adverse prognosis compared to patients with low Mac-2BP expressing tumors [92,104]. In association with literature reports, our data imply that Mac-2BP expressing cells in breast tumors may invade distant tissues via Mac-2BP-E-selectin mediated adhesion and suggest E-selectin reactive Mac-2BP as a potential prognostic marker or therapeutic target for prevention of breast cancer metastasis. However, to make such determinations, it would be important to evaluate Mac-2BP-E-selectin mediated adhesion in comparison with other adhesion molecules that may play a role in mediating cancer cell adhesion to activated endothelium [48,65,101]. Because Mac-2BP has also been identified previously as a Gal-1 ligand [93], we have hypothesized a unique model for the regulation of breast cancer cell adhesion to endothelium (Fig. 3.7). First, we must revisit leukocyte adhesion to endothelial cells. Leukocytes express specific E-selectin ligands that initiate adhesion to endothelial E- selectin [24,82], and certain leukocyte subsets express Mac-2BP [105]. It has been reported that soluble Gal-1 inhibits adhesion of leukocytes to activated endothelium under hematogenous flow conditions, although the exact interference mechanisms are unclear [94]. These findings, together with our data that Mac-2BP is an E-selectin ligand, imply that E-selectin-Mac-2BP interactions potentially compete with Gal-1-Mac-2BP binding (Fig. 3.7). In contrast to the inhibitory effect in cell recruitment by blocking E- selectin ligand activity (i.e., prevention of metastasis), Gal-1 is known to help tumor cells to escape from immune action [106]. Therefore, it appears that expression of Mac-2BP by cancer cells is critically regulated, and an optimal expression level of Mac-2BP is required for evading the immune response and to colonize distant tissue. Thus, our results

86 86 may serve to explain the poor prognosis of breast cancer patients with Mac-2BP overexpressing tumors. Although the present study and the postulated model propose mechanistic links to understand breast cancer metastasis mediated by Mac-2BP, further studies are warranted to confirm and add comprehensive details, including roles for other E-selectin ligands. In conclusion, our study provides new insights into the molecular mechanisms underlying cell adhesion in breast cancer metastasis. The data show that the adhesion of ZR-75-1 breast cancer cells to endothelial cells is mediated by endothelial E-selectin and Mac-2BP, a novel high efficiency E-selectin ligand. We believe that these interactions lead to other pathways necessary for organotropism of metastasis. Altogether our study, combined with previously reported data, demonstrates that Mac-2BP is an important molecule in breast cancer metastasis, and it is anticipated that further investigation will reveal its prognostic and therapeutic potential.

87 87 (3) Deletion (2) No adhesion (1) Metastatic tumor Endothelial cell Cancer cell E-selectin Mac-2BP Galectin-1 Figure 3.7. Hypothesized model for the role of Mac-2BP in metastasis. Three possible scenarios for Mac-2BP expressing breast cancer cells. (1) Breast cancer cells expressing high levels of Mac-2BP bind to Gal-1 yet possess enough free epitopes for E-selectin binding. Thus, these cells are more likely to form metastatic lesions than (2) cells expressing low levels of Mac-2BP, which may not bind to endothelium due to blockade of E-selectin ligand function by Gal-1. (3) Absence of Gal-1 may lead to detection and deletion of cancer cells by immune cells.

88 88 CHAPTER 4: CD44V EXPRESSED BY BREAST CANCER CELLS ARE FUNCTIONAL E-SELECTIN LIGANDS UNDER FLOW CONDITIONS Abstract Adhesion of circulating tumor cells to vascular endothelium leading to metastasis is mediated by specialized molecules that function under shear forces exerted by hematogenous flow. Endothelial E-selectin and its prevalent counter receptor HCELL (Hematopoietic Cell E- and L-Selectin Ligand), which is a specialized glycoform of CD44, mediate cell adhesion in numerous physiologic and pathologic processes, including cancers. However, this pathway is poorly understood for breast cancer cell adhesion and is the focus of the present investigation. Multiple breast cancer cell lines strongly expressed CD44 by flow cytometry and quantitative RT-PCR. Specifically, MDA-MB-231 breast cancer cells predominantly expressed CD44s but limitedly expressed CD44v isoforms, and BT-20 breast cancer cells expressed CD44s as well as CD44v. CD44 expressed by BT-20 cells, but not MDA-MB-231 cells, possessed E-selectin ligand activity by Western blotting and antigen capture assays. Hence BT-20 cell CD44 possess HCELL activity. The HCELL expressed on intact BT-20 cells were functional E-selectin ligands regulating cell rolling and adhesion under physiological flow conditions, as found by experiments involving targeted silencing of the CD44 gene using shrna. Importantly, the BT-20 cell CD44v isoforms (HCELLv) possessed 1 To be submitted as: CD44v expressed by breast cancer cells are functional E-selectin ligands under flow conditions. Shirure VS, Liu T, Delgadillo LF, Cuckler CM, Benencia F, Goetz DJ, and Burdick MM. Part of the work was previously submitted as an abstract to American Association of Cancer Research Conference, 2011 and 2012.

89 89 shear-resistant E-selectin ligand activity, perhaps greater than that of CD44s, in antigen capture assays. Interestingly, CD44 on breast cancer cells was not recognized by the HECA-452 mab, which detects sialofucosylated epitopes, suggesting that BT-20 cells express a novel glycoform of HCELL. Furthermore, the HCELL activity of BT-20 cells was predominantly due to N-linked glycans, and the glycans are likely terminally fucosylated by α-(1, 3) fucosyl transferases, FT-3 and/or FT-6. Remarkably, HCELL was predominantly expressed by epithelial-like cells (BT-20), but its expression was weak in mesenchymal-like cells (MDA-MB-231) quantitative RT-PCR and microscopy. In summary, breast cancer cells express HCELLv as shear-resistant E-selectin ligands, expression of which may be associated with epithelial/mesenchymal state of the cells, offering a novel perspective on HCELL biology. 4.2 Introduction During hematogenous metastasis, tumor cells dissociate from a primary tumor, migrate through the tissue space, and enter the circulatory system. The circulating tumor cells (CTCs), once bloodborne, travel to distant sites, where they adhere to the endothelial cells lining the vessel wall, potentially extravasate, and form secondary tumors if directed by niche factors [4,24]. Determining the mechanisms of each of these steps can provide insights into novel diagnostics and therapeutics for cancer. Of particular interest is elucidating the molecular mechanisms by which metastatic cells adhere to the endothelium while resisting the disruptive shear exerted by the blood flow. The shear-resistant adhesion of cancer cells is hypothesized to be similar, in part, to leukocyte recruitment at sites of inflammation, which is mediated by specialized

90 90 molecules known as selectins [24,26-28,35,48,60]. Several in vivo and in vitro studies have implicated E-selectin in breast, colon, and prostate cancer cell adhesion to the endothelium [24-29,48,69,80]. Interestingly, E-selectin is constitutively expressed in human bone marrow [31], and bone marrow is one of the most frequent sites of cancer metastasis [93]. Given that E-selectin appears to be involved in metastasis, there has been an increasing focus on determining the moieties on cancer cells that can serve as E- selectin ligands. In general, a number of E-selectin binding molecules (ligands) have been identified [24-29,48,60,80]. Among them, CD44 is a major functional ligand on a number of cancer cell lines as well as certain normal blood cells [26,47,51,60,101,107], perhaps due to its enormous structural diversity. Human CD44 is a family of proteins that are encoded by a single gene comprising of at least 20 exons. Exons 1-5 and generate standard form (CD44s) that lacks any variable region, and exons 6-15 are alternatively spliced to produce CD44 variant (CD44v) isoforms, designated as CD44v1-10 [108]. CD44s as well as CD44v isoforms can function as E-selectin ligands, yet CD44v isoforms are apparently the prevalent E-selectin ligands expressed by carcinoma cells [26,108,109]. Importantly, E-selectin binding epitopes are actually the carbohydrate moieties presented on core proteins or lipids. Along these lines, CD44 can have extensive posttranslational modifications including N- and O-glycans forming high avidity E- selectin binding ligands [26,108,109]. Such E- and L-selectin reactive CD44 glycoforms were first identified on hematopoietic cells, hence termed as hematopoietic cell E- and L- selectin ligands (HCELL; [60]).

91 91 Several lines of evidence have implicated CD44 in breast cancer. For example, the expression of CD44 has been detected on a number of breast cancer cell lines and primary breast cancer tumors [78,110,111]. Further, the circulating tumor cells have been found to express CD44 [112], and CD44 has been recognized as a breast cancer stem cell marker [11,12]. Also, it is believed that a subpopulation CD44 expressing tumor cells possess high metastatic potential [18]. While these observations link CD44 to breast cancer metastasis, others give insight into the functional role of breast cancer cell CD44 as an E-selectin ligand. We have previously shown that breast cancer cells adhere to endothelium under flow via E-selectin and found evidence that glycoproteins play a prominent role in this adhesion process [27,28]. The E-selectin ligand activity of CD44 under static (no-flow) conditions has been reported for post adhesion process of transendothelial migration of breast cancer cells [78]. The above observations have lead us to hypothesize that CD44 present on breast cancer cells serves as a functional ligand for E-selectin, allowing breast cancer cells to adhere to the endothelium under physiological flow conditions. Thus, in the present study, we characterized CD44 expressed by breast carcinoma cell lines and assessed its E-selectin ligand activity under flow conditions. Eventually, the findings of the present study may lead to a potential therapeutic target against breast cancer metastasis. 4.3 Materials and methods Cell culture The BT-20 breast cancer cell line from the American Type Culture Collection (ATCC, Manassas, VA) was grown in minimum essential medium (Invitrogen, Carlsbad,

92 92 CA) with 10% FBS and 1x penicillin-streptomycin. The MDA-MB-231, Hs-578T, and MDA-MB-468 breast cancer cell lines (all ATCC) were cultured in Dulbecco s modified eagle medium (Invitrogen) with 15% FBS and 1x penicillin-streptomycin. E-selectin transfected Chinese Hamster Ovary cells (CHO-E), which were a generous gift from Dr. Robert Sackstein (Harvard Medical School, Boston, MA), were maintained in MEM supplemented with 10% FBS, 0.1 mm non-essential amino acids (Invitrogen). The LS- 174T colon carcinoma cell line (ATCC) was cultured in MEM supplemented with 10% FBS, 0.1 mm non-essential amino acids, and 1 mm sodium pyruvate (Invitrogen) Antibodies and chimera constructs Anti-human CD44 (515), CD62E (68-5H11), and FITC-conjugated CD44 (G44-26) monoclonal antibodies (mabs) and all isotype controls were obtained from BD Biosciences (San Jose, CA). Anti-human CD44 (2C5), CD44v3 (3G5), and CD44v4/5 (3D2) mabs and recombinant mouse E-selectin human Fc chimera constructs (E-Ig chimera) were obtained from R & D systems (Minneapolis, MN). Anti-human mabs against variant isoforms v3 (VFF 27), v4 (VFF 11), v5 (VFF 8), v6 (VFF 7), v7 (VFF 9), v7/8 (VFF 17), v10 (VFF 14) of CD44 were obtained from AbD Serotec (Raleigh, NC). FITC-conjugated and AP-conjugated secondary antibodies were from Southern Biotech (Birmingham, AL) Quantitative reverse transcriptase polymerase chain reaction (qrt-pcr) Extraction and purification of RNA was performed using RNeasy Plus kit (Qiagen, Valencia, CA) following the manufacturer s protocol. The RNA purity and concentration were found by NanoVue Plus (GE Healthcare Biosciences, Piscataway,

93 93 NJ). The purified RNA was treated with RNAse inhibitor (Applied Biosystems, Foster City, CA) and DNAse I (New England Biolabs Inc., Ipswich, MA) to remove genomic DNA. The RNA was reverse transcribed by a high capacity reverse transcription kit (Applied Biosystems) following the manufacturer s instructions. PCR was performed on cdna, synthesized from 75 ng of RNA, by SYBR Green FastMix (Quantas Biosciences) chemistry and monitored by an icycler iq5 real-time PCR instrument (Bio-Rad Laboratories, Hercules, CA). The forward and reverse primers were purchased from Integrated DNA Technologies (Coralville, IA). The gene expression data was normalized to glyceraldehyde-3-phosphatedehydrogenase (GAPDH; a housekeeping gene) [113] Cell lysis and immunoprecipitation Cells were lysed in buffer containing 1% Triton X-100, 0.02% NaN 3, 150 mm NaCl, 0.5 mm Tris, ph 10.4, 1 mm EDTA, and protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN). CD44 was immunoprecipitated from cell lysates by incubation with anti-cd44 (515) mab and protein G agarose beads (Invitrogen) overnight at 4 C under constant rotation. Protein G beads were subsequently washed with lysis buffer, and the beads were then incubated with Laemmli reducing sample buffer and heated to 100 C for 5 min to release the CD44 [28]. To purify E-selectin reactive proteins, cell lysates were first pre-cleared of nonspecific antigens by incubating with human IgG isotype control and protein G beads. The pre-cleared lysate was then incubated with E-Ig chimera and protein G beads at 4 C with constant rotation, overnight. After sufficient washings, E-Ig chimera reactive antigens were eluted from protein G beads using elution buffer (5 mm EDTA, 50 mm

94 94 Tris (ph 7.4), and 0.1% Triton-X-100). The immunoprecipitates were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), and then subjected to Western blotting [28,47] RNA interference CD44 knockdown of BT-20 cells was performed using MISSION shrna lentiviral transduction particles (Sigma-Aldrich, St. Louis, MO) prepared from plko.1 vector (TRC 1 version). Five different clones were screened, and a clone with significantly reduced cell surface expression of CD44 by flow cytometry was chosen for experiments, the oligonucleotide sequence of which is 5 - CCGGCGCTATGTCCAGAAAGGAGAACTCGAGTTCTCCTTTCTGGACATAGCGTT TTTG-3 (TRCN ), where underlined portion is sense and italic portion is antisense sequence. The concentration of viral particles was optimal at 10 multiplicity of infection (MOI), and the cells transfected with specific sequence or with empty vector were puromycin selected [28] SDS-PAGE and Western blotting SDS-PAGE of cell lysate or immunoprecipitate was performed on 4-15% Tris- HCl precast gels (Bio-Rad Laboratories) under reducing conditions, and proteins were subsequently transferred to polyvinylidene difluoride (PVDF) membrane (Bio-Rad Laboratories). Western blot membranes were blocked with FBS overnight to reduce the background noise and subsequently stained with appropriate antibodies or isotype controls and AP-conjugated secondary antibodies [28]. The signal was detected by Western Blue substrate (Promega Madison, WI).

95 Flow cytometry Cells were washed with blocking buffer and incubated with primary (10 µg/ml) or FITC-conjugated antibody or isotype control for 30 min at 4 C. Cells were washed and, as necessary, stained with secondary antibody for 30 min at 4 C [26,28,85]. Cells were washed and analyzed by a FACSort or FACSAria Special Order Research Product flow cytometer/sorter (BD Biosciences) Flow adhesion assay and antigen capture The experimental set up for the flow adhesion assays consisted of a parallel plate flow chamber (Glycotech, Rockville, MD) placed on a Nikon TE300 inverted microscope equipped with a CCD video camera and a video recorder. Untreated, vector, or CD44 silenced BT-20 cells (1 million cells/ml) were perfused over a monolayer of CHO-E cells for 4 min at a bone marrow micro-vascular shear rate 100 s -1 [74]. Subsequently, for detachment analysis, the shear rate was sequentially doubled in time steps of 30 s until the final shear rate reached to 3200 s -1. The number of adherent cells, which included cells interacting with the surface at a given time, was counted at the end of the time intervals. The number of adherent cells at the end of a time interval was normalized with respect to the number of adherent cells at the end of 100 s -1 to find percentage attachment. Cell rolling velocity was calculated by capturing stills from video over 5 sec and converting pixel distance into microns using Image J software [28,48]. The antigen capture assay is described elsewhere [114]. In short, 20 µg/ml mab was incubated on a Petri dish at 37 C for 2 hrs. The mab substrate was blocked by 1% BSA overnight at 4 C. Lysate equivalent to 2 million cells was applied to the spot using

96 96 cloning chambers and incubated overnight at 4 C. CHO-E cells (1 million/ml) were perfused over captured antigens for 2 min at 100 s -1 using the parallel plate flow chamber set up described earlier. The number of adhering cells included all cells bound to surface at the end of perfusion [48,85] Fluorescence microscopy Cells grown on tissue culture grade EZ slides (Millipore, Billerica, MA) were washed and fixed in 4% methanol-free paraformaldehyde in DPBS. The slides were incubated with primary antibody or isotype control for 1 hr at 4 C. The slides were washed and incubated with AlexaFluor 488 (green)-conjugated secondary antibody for 1 hr at 4 C [28,115]. A drop of ProLong Gold anti-fade reagent with DAPI (Invitrogen) was added before mounting cover slips. The slides were imaged using a 40x objective under wide field fluorescence using a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany) and Simple PCI software (Hamamatsu Corporation, Sewickley, PA) Statistics Data are represented as mean ± SE for at least 3 independent experiments. Statistical significance (P 0.05) between control and sample was tested by paired Student s t-test except where indicated. Multiple samples were compared with one way ANOVA coupled with post-hoc Tukey s multiple comparison test (P 0.05).

97 Results Breast cancer cell lines express CD44 isoforms Previously, we have shown that shear-resistant adhesion of breast cancer cell lines is mediated by E-selectin and breast cancer cell glycoprotein ligands [28]. Therefore, BT- 20, MDA-MB-468, MDA-MB-231, and Hs-578T breast cancer cell lines were initially screened for the expression of CD44 using CD44 (515) mab that recognizes CD44s and majority of the CD44v isoforms [47,53,60]. Consistent with previous reports [16,110,111], flow cytometric analysis showed that each of these breast cancer cell lines robustly express CD44 (Fig. 4.1A). To quantitatively compare the expression of CD44 isoforms at the mrna level, qrt-pcr was performed using primers designed for the specific isoforms (Table 1; [116]). BT-20 and MDA-MB-468 cells, compared to MDA- MB-231 and Hs-578T cells, generally expressed higher levels of CD44v isoforms. On the other hand, MDA-MB-231 and Hs-578T cells expressed similar or slightly higher levels of CD44s (Fig. 4.1B), compared to the other two breast cancer cell lines. The breast cancer cell lines were probed by flow cytometry to find expression of CD44 variants at protein level. In line with the qrt-pcr data (Fig. 4.1B), BT-20 (Fig. 4.2A) cells showed positive expression of multiple CD44v isoforms. Particularly, isoforms CD44v3, v4, v5, v6 were strongly expressed on the surfaces of these cells while other isoforms were also present but at a lower level. Furthermore, MDA-MB-231 (Fig. 4.2B) expressed limited levels of the variant isoforms, indicating that the CD44 proteins (Fig. 4.1A) on these cells are mainly CD44s. These flow cytometric data (Fig. 4.2) and qrt-pcr results (Fig. 4.1B) collectively demonstrated that BT-20 and MDA-MB-468,

98 98 compared to MDA-MB-231 and Hs-578T cells, express higher levels of CD44v isoforms. We chose to conduct further investigations with BT-20 and MDA-MB-231, which represent cell lines expressing two different levels of CD44v isoforms. To find the molecular weight of CD44 isoforms, anti-cd44 (515) mab immunoprecipitates were Western blotted by another multiple isoform recognizing mab, anti-cd44 (2C5). BT-20 immunoprecipitates resolved into three distinct bands at ~95, ~110 and ~150 kda, and MDA-MB-231 immunoprecipitates displayed a single prominent band at ~95 kda (Fig. 4.3A). Additionally, Western blotting of lysates from either of the cell lines showed similar molecular weight bands of CD44 as that observed with immunoprecipitates (Fig. 4.3A). The intensity of ~110 kda isoform of BT-20 lysate was weak, which is likely due to its low expression. The Western blot results thus confirm that BT-20 cells express multiple CD44 isoforms including CD44v, and MDA- MB-231 cells mainly express CD44s CD44 expressed by BT-20 cells, but not by MDA-MB-231 cells, possess HCELL activity To test the E-selectin ligand activity of CD44, E-Ig chimera immunoprecipitates were Western blotted with anti-cd44 (2C5) mab or an isotype control. As shown in Fig. 4.3B, bands for BT-20 at ~95, ~110, and ~150 kda were detected by the anti-cd44 (2C5) mab but not by the isotype control. Note that these bands (Fig. 4.3B) corresponded to similar molecular weights for CD44 as shown in Fig. 4.3A. In contrast to the results observed with BT-20, probing the E-Ig chimera immunoprecipitates from MDA-MB-231 did not reveal any detectable CD44 bands (Fig. 4.3B). These results

99 suggest that the CD44 isoforms expressed by BT-20 cells are E-selectin ligands, but the CD44s expressed by MDA-MB-231 cells has limited, if any, E-selectin ligand activity. 99 A Cell count BT-20 MDA-MB-468 MDA-MB-231 Hs-578T Fluorescence intensity Figure 4.1. CD44s and CD44v are expressed by breast cancer cells. (A) Breast cancer cells were labeled with anti-cd44 (515) mab (filled curves) or isotype control (open curves) and analyzed by flow cytometry. (B) qrt-pcr was performed on breast cancer cell mrna using primers designed for detection of specific CD44 isoform and the data is represented by relative expression with housekeeping gene (GAPDH). Data are mean ± SE for n = 4. * P < 0.05 by one way ANOVA coupled with Tukey s multiple comparison test.

100 100 Table 4.1 Primer Sequences of CD44 [116] and Fucosyltransferases[117] Used in qrt-pcr. Primer Sequence CD44s F 5' -CCT CCA GTG AAA GGA GCA GCA C-3' CD44s R 5' -GTG TCT TGG TCT CTG GTA GCA GGG AT-3' CD44 v3 F 5' -GTA CGT CTT CAA ATA CCA TCT CAG C-3' CD44v3 R 5' -GGT GCT GGA GAT AAA ATC TTC ATC-3' CD44v4 F 5' -TTT CAA CCA CAC CAC GGG C- 3' CD44v4 R 5' -CAG TCA TCC TTG TGG TTG TCT G-3' CD44v5 F 5' -GTA GAC AGA AAT GGC ACC ACT GC-3' CD44v5 R 5' -TTG TGC TTG TAG AAT GTG GGG TCT C 3' CD44v6 F 5' -TCC AGG CAA CTC CTA GTA GTA C-3' CD44v6 R 5' -CAG CTG TCC CTG TTG TCG AAT GGG-3' CD44v7 F 5' -CAG CCT CAG CTC ATA CCA G-3' CD44v7 R 5' -CCA TCC TTC TTC CTG CTT G-3' CD44v8 F 5' -TGG ACT CCA GTC ATA GTA TAA CGC-3' CD44v8 R 5' -GCG TTG TCA TTG AAA GAG GTC CTG-3' CD44v9 F 5' -AGC AGA GTA ATT CTC AGA GC-3' CD44v9 R 5' -TGC TTG ATG TCA GAG TAG AAG TTG-3' CD44v10 F 5' -ATA GGA ATG ATG TCA CAG GTG G-3' CD44v10 R 5' -CGA TTG ACA TTA GAG TTG GAA TCT CC-3' GAPDH F 5' -AGC CAC ATC GCT CAG ACA C-'3 GAPDH R 5' -GCC CAA TAC GAC CAA ATC C-'3 FT-3 F 5 -GCCGACCGCAAGGTGTAC-3 FT-3 R 5 -TGACTTAGGGTTGGACATGATATCC-3 FT-4 F 5 -AAGCCGTTGAGGCGGTTT-3 FT-4 R 5 -ACAGTTGTGTATGAGATTTGGAAGCT-3 FT-5 F 5 -TATGGCAGTGGAACCTGTCA-3 FT-5 R 5 -CGTCCACAGCAGGATCAGTA-3 FT-6 F 5 -CAAAGCCACATCGCATTGAA-3 FT-6 R 5 -ATCCCCGTTGCAGAACCA-3 FT-7 F 5 -TCCGCGTGCGACTGTTC-3 FT-7 R 5 -GTGTGGGTAGCGGTCACAGA-3

101 101 A CD44v3 CD44v4 CD44v5 CD44v6 CD44v7 CD44v7/8 CD44v10 Cell count B CD44v3 CD44v4 CD44v5 CD44v6 CD44v7 CD44v7/8 CD44v10 Fluorescence intensity Figure 4.2. Differential levels of CD44v isoforms are expressed on the surface of breast cancer cells. (A) BT-20 and (B) MDA-MB-231 cells were labeled with mab against specific isoform (filled curves) or isotype control (open curves) and analyzed by flow cytometry. Data are representative of n = 3 independent experiments.

102 102 We next sought to determine if the CD44 expressed by breast cancer cells are functional E-selectin ligands under physiological flow conditions. To investigate this issue, antigens captured from BT-20 cell lysates using anti-cd44 (515) mab were adsorbed on a Petri dish. CHO-E cells, a cell line stably expressing E-selectin (Fig. 4.3C), were subsequently perfused over the adsorbed antigens at a bone marrow vasculature shear rate of 100 s -1. CHO-E cells robustly adhered to CD44 immunoprecipitated from BT-20 and LS-174T (positive control [26]; Fig. 4.3D). In contrast, CHO-E cells exhibited negligible adhesion to antigens isolated with an isotype control (Fig. 4.3D). In addition, treatment of CHO-E cells with an E-selectin function blocking mab (anti-cd62e mab) prior to perfusion through the flow chamber significantly diminished (more than 7-fold reduction) CHO-E cell adhesion to BT-20 CD44. Combined, these results show that adhesion of CHO-E cells to BT-20 cell CD44 is specifically mediated by E-selectin. In contrast, CD44 isolated from MDA-MB-231 supported relatively low levels of CHO-E cell adhesion. Specifically, the level of adhesion was only 12% of that observed for CD44 isolated from BT-20 cells. These data collectively demonstrate that BT-20 cell CD44 is a highly avid, but MDA-MB-231 cell CD44 is a weak, form of HCELL under physiological flow conditions.

103 103 A kda Molecular weight marker BT-20 MDA-MB-231 BT-20 (ippt) MDA-MB-231 (ippt) CD44 (2C5) --NA-- --NA-- β-actin loading control B kda Molecular weight marker BT-20 (ippt) MDA-MB-231 (ippt) Molecular weight marker BT-20 (ippt) MDA-MB-231 (ippt) CD44 Isotype

104 104 C Cell count CHO-E Fluorescence intensity D Cells adhering/mm * * Isotype control CD44 (515) *$ LS-174T BT-20 MDA-MB-231 Figure 4.3. CD44 expressed by BT-20 cell possesses functional E-selectin ligand activity. (A) Cell lysate from cells or immunoprecipitates (ippt) from lysate of 1x10 7 cells by anti-cd44 (515) mab were subjected to Western blotting with anti-cd44 (2C5) mab. -actin loading controls for cell lysates but not for immunoprecipitates (indicated by NA) are shown in the bottom panel. Data are representative of n = 3 independent experiments. (B) E-Ig chimera immunoprecipitates from 1x10 7 cells were Western blotted with isotype or anti-cd44 (2C5) mab. Data are representative of n = 3 independent experiments. (C) CHO-E cells were labeled with anti-cd62e mab (filled curve) or isotype control (open curve) and tested by flow cytometry. (D) CHO-E cells were perfused over antigens captured by CD44 (515) or isotype controls at bone marrow micro-vascular shear rate of 100 s -1. Data are mean ± SE for n = 5. * P < 0.05 with respect to adhesion to isotype control antigens, and $ P < 0.05 with respect to adhesion to BT-20 cell CD44.

105 CD44 on intact BT-20 cells possesses HCELL activity under physiological flow conditions To investigate the E-selectin ligand activity of CD44 on intact BT-20 breast cancer cells, BT-20 cells transduced with CD44 shrna or empty vector were perfused over a monolayer of CHO-E cells. Initially, the efficacy of CD44 silencing was verified by flow cytometry, which showed that CD44 silenced BT-20 cells, compared to vector cells, expressed a significantly lower (~50% reduction) level of CD44 (Fig. 4.4A). Furthermore, the adhesion of wild type, vector, and CD44 silenced BT-20 cells to CHO-E cells was specifically mediated by E-selectin, as the cells adhered to CHO-E cells but not to CHO-E cells treated with anti-cd62e mab. Notably, the rolling velocity, a hallmark of E-selectin mediated interactions, of silenced cells at 400, 800, and 1600 s -1 were significantly higher than vector cells (Fig. 4.4B) showing involvement of E-selectin- CD44 ligation in controlling the cell rolling velocity. Furthermore, CD44 silenced BT-20 cells were less stably adherent than vector cells, which was especially apparent at higher shear rates. For example, ~60% of vector cells were adherent whereas only ~30% of the silenced cells were remained attach at 800 s -1 (Fig. 4.4C). These data demonstrate that CD44 expressed by intact BT-20 cells possessing functional E-selectin ligand activity under physiological flow conditions is HCELL.

106 106 A vector * Mean fluorescence intensity CD44 silenced *$ 0 Isotype CD44 B Rolling velocity (µm/s) * * Vector * CD44 silenced Shear rate (s -1 )

107 107 C 100 Vector CD44 silenced 75 * % Attachment 50 * 25 * Shear rate (s -1 ) Figure 4.4. CD44 knock down in BT-20 cells reduces their adhesion to E-selectin expressing cells. (A) BT-20 cells transfected with empty vector or shrna for CD44 were labeled with anti-cd44 mab (G44-26; filled curves) or isotype control (open curves) and analyzed by flow cytometry. * P < 0.05 with respect to isotype labeled cells, and $ P < 0.05 with respect to vector cells. (B) Rolling velocities of vector or CD44 silenced cells over CHO- E cells were found as mentioned in methods. Data are mean ± SE for n = 15. * P < 0.05 with respect to vector cells. (C) Vector or CD44 silenced cells were perfused at a shear rate of 100 s -1, and the shear rate was stepwise increased and the percentage of attached cells in each step is presented. Data are mean ± SE for n = 5 independent experiments. * P < 0.05 with respect to vector cells BT-20 cell HCELLv isoforms are sufficient for shear-resistant adhesion of CHO-E cells To investigate whether specific HCELLv isoforms are sufficient for functional E- selectin ligand activity, antigens immunopurified using mabs against specific CD44v

108 108 isoform were adsorbed over a tissue culture dish, and CHO-E cells were perfused over the captured antigens at 100 s -1. Since BT-20 cells mainly expressed CD44v3-6 isoforms on the cell surface (Fig. 4.2), only these isoforms were tested for E-selectin ligand activity. Significantly, CHO-E cells strongly adhered to the CD44v3 and CD44v4/5 but minimally/did not adhere to antigens isolated with CD44v6 or the isotype negative control (Fig. 4.5A). Additionally, the possibility that the anti-cd44v6 mab does not block E-selectin binding sites was negated by treatment of anti-cd44 (515) mab captured antigens with anti-cd44v6 mab, which did not reduce the CHO-E cell adhesion. In contrast to C44v of BT-20 cells, the adhesion of CHO-E cells to CD44 isoforms captured from MDA-MB-231 cell lysate was limited (Fig. 4.5A), as expected. These results collectively demonstrate that CD44v isoforms, particularly CD44v3 and v4/5, expressed by BT-20 cells possess sufficient E-selectin ligand activity to mediate cell adhesion under physiological flow conditions. To estimate the contribution of CD44v versus CD44s isoforms for E-selectin ligand activity, the adhesion data of each variant (Fig. 4.5A) were normalized to the adhesion data for CD44 (515) (Fig. 4.5B). Assuming that the anti-cd44 (515) mab captures most number of the CD44 molecules [53], the normalized numbers represent percent contributions of each isoform towards E-selectin ligand activity. As shown in Fig. 4.5B, the adhesion to CD44v3 or v4/5 was more than 50% and 35%, respectively, indicating the total contribution of CD44v was more than 50%. In support of these data, the evaluation staining intensity of BT-20 Western blots (Fig. 4.3B) indicate that E- selectin ligand activity of CD44 at ~150 kda (most likely CD44v) is the highest and that

109 109 at ~95 kda (most likely CD44s) is the lowest. These estimates suggest greater contribution of HCELLv for E-selectin ligand activity compared to HCELLs BT-20 cell HCELL is HECA-452 negative To examine whether the E-selectin ligand activity of CD44 expressed by breast cancer cell lines is associated with terminal sialofucosylated groups, anti-cd44 (515) immunoprecipitated from breast cancer cell lysate was Western blotted with HECA-452 mab. Interestingly, neither CD44 isolated from BT-20 nor MDA-MB-231 cell lines was reactive to HECA-452 mab (Fig. 4.6A). In contrast, CD44 isolated from LS-174T cells [positive control [47]], was reactive to HECA-452 mab (see band at ~150 kda in Fig. 4.6A). These observations strongly suggest that the E-selectin ligand activity of BT-20 cell CD44 is due to non-heca-452 reactive antigens and perhaps due to novel terminal glycosylations FT-3 and FT-6 regulate E-selectin ligand activity in BT-20 cells To unravel molecular mechanisms involved in the synthesis of E-selectin ligands, we chose to analyze expression of α-(1, 3)- and α-(1, 4)-fucosyltransferases (FTs), which catalyze terminal fucosylation necessary for E-selectin ligand function [80]. The qrt- PCR analysis [117] showed that the expression levels of FT-3 and FT-6 mrna were significantly higher in BT-20 cells compared to that in MDA-MB-231 cells, and the expression of FT-5, FT-7 was not significantly different in the two breast cancer cell lines (Fig. 4.6B). Yet, FT-4 was expressed by MDA-MB-231 cells and was not detected in BT-20 cells. Taken together, these data imply FT-3 and FT-6 primarily regulate the E- selectin ligand activity of BT-20 breast cancer cells.

110 110 A 25 * BT-20 Cells adhering/mm *$ * $ MDA-MB migg1 CD44v3 CD44v4/5 CD44v6 B 100 % Normalized adhesion all variants CD44v3 CD44v4/5 CD44v6 Figure 4.5. CD44v isoforms on BT-20 cells possess sufficient E-selectin ligand activity to support cell adhesion under hematogenous flow conditions. (A) CHO-E cells were perfused over antigens captured by specific CD44v mab or isotype controls at a bone marrow micro-vascular shear rate of 100 s -1. Data are mean ± SE for n = 5. * P < 0.05 with respect to adhesion to isotype control antigens, and $ P < 0.05 with respect to adhesion to BT-20 cell CD44. (B) The number of CHO-E cells adhering to variant antigens was normalized with respect to adhesion to CD44 (515). The all variant bar indicates total normalized adhesion of CD44v3, v4/5, and v6 to adhesion to CD44 (515).

111 111 A Molecular weight marker LS-174T (ippt) BT-20 (ippt) MDA-MB-231 (ippt) kda HECA-452 B Relative expression (ΔCt) FT-3 * BT-20 MDA-MB-231 * ND * FT-4 FT-5 FT-6 FT Relative expression (ΔCt) Figure 4.6. HECA-452 negative glycans confer E-selectin ligand activity to BT-20 cell CD44. (A) CD44 immunoprecipitate from 10 7 cells was tested by Western blotting with HECA- 452 mab. Data are representative of n = 3 independent experiments. (B) Expression of fucosyltransferases was found by qrt-pcr. The data are represented by relative expression with housekeeping gene (GAPDH). ND indicates expression not detected. Data are mean ± SE for n = 3. * P < 0.05 with respect to BT-20 cells.

112 Breast cancer cell expression of epithelial and mesenchymal cell markers Recently, it has been shown that expression of E-selectin ligands is regulated by epithelial to mesenchymal transition (EMT), which is a process believed to be critical for metastasis [118]. Also, it has been shown that expression of CD44 isoforms switching is necessary for EMT [119]. In light of these reports, we sought to uncover whether the differential expression and E-selectin ligand function of CD44 isoforms may be correlated with epithelial or mesenchymal phenotype of the breast cancer cell lines. BT- 20, compared to MDA-MB-231, expressed dramatically higher levels of mrna for the epithelial markers E-cadherin, yet markedly low levels of mrna for the mesenchymal markers, N-Cadherin (Fig. 4.7A) and SLUG, the latter of which is a transcriptional repressor of E-cadherin. Similar to mrna expression, BT-20 expressed higher protein levels of E-cadherin but almost no N-cadherin, compared to MDA-MB-231 (Fig. 4.7B). These data indicate that HCELL may be expressed primarily by breast cancer cells primarily in an epithelial-like state.

113 113 A * * *

114 114 B Isotype E-cadherin N-cadherin MDA-MB-231 BT-20 Figure 4.7. BT-20 are epithelial-like and MDA-MB-231 are mesenchymal-like cells. (A) qrt-pcr was performed on mrna extracted from breast cancer cells, and the data are represented by C T method by normalization with the housekeeping gene GAPDH. Data are mean ± SE for n = 4. * P < 0.05 with respect to BT-20 cells. (B) Breast cancer cells grown on tissue culture slides were labeled with anti-e-cadherin or N-cadherin mabs and were imaged in epifluorescence microscopy. Scale bar = 50 µm. 4.5 Discussion The HCELL, glycoforms of CD44, are prevalent E-selectin ligands promoting shear-resistant adhesion of circulating cells to endothelium in variety of physiologic and pathologic processes. For instance, the HCELL- E-selectin interactions mediate trafficking of hematopoietic stem cells into bone marrow [60] and are suggested to be involved in osteotropism of prostate cancer cells [69]. Interestingly, breast cancer frequently metastasizes to bone marrow [14], the endothelium of which constitutively expresses E-selectin [31], and certain breast cancer cell lines adhere to endothelium via

115 115 E-selectin [16,27,35,110,111]. These reasons compelled us to perform a detailed investigation of E-selectin ligand function of breast cancer cell CD44 under bone microvascular flow conditions. Our results project a novel perspective on HCELL-E-selectin pathway in breast cancer cell adhesion by demonstrating that CD44v isoforms (i.e. HCELLv) are more relevant E-selectin ligands than CD44s under physiological flow conditions. Consistent with the literature [16,110,111], each of the four breast cancer cell lines used in the present study strongly expressed CD44 (Fig. 4.1A) detected by an anti- CD44 (515) mab that recognizes the majority of CD44 isoforms [53]. The distinguishing feature, however, was the difference in expression levels of specific CD44v isoforms. While BT-20 cells expressed high levels of mrna for the majority of CD44v, MDA- MB-231 cells expressed conspicuously low levels of CD44v mrna (Fig. 4.1B). These expression patterns were even more convincing for cell surface proteins (Fig. 4.2). Furthermore, Western blot analysis confirmed these patterns of expressions (Fig. 4.3A), in which BT-20 and MDA-MB-231 cells expressed ~95 kda form, most likely CD44s [60], but only BT-20 expressed ~110 and ~150 kda forms, which are most likely CD44v isoforms [47]. These cell lines provided a system to compare HCELL activity of highly and limitedly CD44v positive breast cancer cells. It is believed that since CD44v isoforms possess additional glycosylation sites and may extend farther from cell surface than CD44s, CD44v may have relatively higher functional selectin ligand activity [26,47,53]. In accordance with this notion, the CD44 on BT-20 (CD44v expressing) cells possessed higher E-selectin ligand activity compared to

116 116 MDA-MB-231 (predominantly CD44s expressing) cells (Figs 4.3B and D). Specifically, the BT-20 cell CD44 were sufficient to engage the flowing CHO-E cells, (Fig. 4.3D), were necessary for E-selectin mediated cell rolling (Fig. 4.5B), and appeared critical for high avidity binding (Fig. 4.5C). Further, the antigen capture assays clearly suggest that the major HCELL activity of breast cancer cells is associated with CD44v, particularly CD44v3 and CD44v4/5 (Fig. 4.5) isoforms. Notably, solid cancer cells possessing strong E-selectin ligand activity, such as colon [26,47] and breast cancer cells (present data), is associated with CD44v, but the major E-selectin ligand activity of hematopoietic stem cells and leukemic cells is mostly due to CD44s [46] [60]. Thus, the expression of E- selectin reactive CD44v could be a potential predictive metastasis marker, at least in certain cancer types. Since E-selectin binds to carbohydrate epitopes on a core molecule, such as CD44, appropriate glycosylations are a pre-requisite of E-selectin ligand function [24,29]. In this regard, HECA-452 mab recognizing sialofucosylated groups has been classically used to detect E-selectin reactive carbohydrates [47,60]. However, a number of HECA-452 negative E-selectin ligands have been reported, and HECA-452-negative molecules have been predicted to be the principal E-selectin ligands on the MDA-MB- 468 breast cancer cell line [27,38]. Similarly, the E-selectin ligand activity of BT-20 cell CD44 was associated with HECA-452-negative glycans (Fig. 4.6A), suggesting a previously unobserved glycoform of HCELL on the BT-20 breast cancer cell line. A recent evidence shows that the expression of CD44v on breast cancer cells is down regulated during epithelial to mesenchymal transition (EMT; [119]). In association

117 117 with this data, the results of the present study imply that expression of HCELLv is controlled during EMT. In support of the notion that E-selectin ligand activity is perhaps regulated during EMT, it has been shown that EMT in mouse epithelial cell lines can be controlled by certain gangliosides [70], many of which have been shown to be E-selectin ligands on breast cancer cells [25,27,48,67]. Along these lines, the data of the present study indicate that regulation of E-selectin ligand activity in mesenchymal-like breast cancer cells is perhaps mediated by α-(1, 3)- and α-(1, 4)- fucosyltransferases, as expression level of these enzymes was generally lower in mesenchymal MDA-MB-231 cells compared to epithelial BT-20 cells (Fig. 4.6B). Insight into the negative relationship of E-selectin ligand activity with the mesenchymal state of breast cancer cells can be obtained from other studies. During metastatic invasion, tumor cells are thought to undergo EMT as well as the reverse, mesenchymal to epithelial transition (MET; [18]). Notably, cancer stem cells have been found in EMT, MET, and an intermediate of EMT-MET state [18]. While the mesenchymal state may be beneficial for cell migration through surrounding tissue, the epithelial state may be required for the adhesion of circulating tumor cells to vascular endothelium. The key for metastasis is regulation of these complementary goals. In line with this notion, it is thought that cells in the intermediate state could have high metastatic potential [18]. Thus, the mechanistic association of formation of E-selectin ligands with EMT and MET may reveal a new paradigm for breast cancer metastasis, investigation of which is currently ongoing in our laboratory.

118 118 In summary, the present investigation demonstrated that CD44 expressed as HCELL on BT-20 breast cancer cell line are E-selectin ligands under physiological flow conditions. The HCELL on these cells are mainly glycoforms of CD44v, particularly CD44v3 and v4/5 isoforms, rather than CD44s. Further, the E-selectin ligand activity of HCELLv is due to HECA-452 negative glycans. On the other hand, the predominantly CD44s expressing MDA-MB-231 cell line has very limited HCELL activity. Furthermore, the abundant E-selectin ligand production in BT-20 cells is mediated by α- (1, 3)- and α-(1, 4)- fucosyltransferases. Finally, the expression of HCELLv by breast cancer cells may be regulated by EMT.

119 119 CHAPTER 5: BREAST CANCER STEM-LIKE CELLS POSSESS LOWER E- SELECTIN LIGAND ACTIVITY THAN NON-STEM-LIKE CELLS Abstract Cancer stem cells are a subset of tumor cells thought to possess properties necessary for metastatic invasion. E-selectin and their ligands are believed to play a major role in the adhesion of circulating tumor cells to vascular endothelium. We hypothesized that the levels of expression of E-selectin ligands are related to breast cancer stem cell (BCSCs) or non-bcsc phenotypes. The five breast carcinoma cell lines analyzed by flow cytometry were BCSCs classified by the positive expression of CD44 and low/negative expression of CD24 (CD44 + /CD24 -/low ) or non-bcscs marked by alternative expression of these proteins. All these cell lines were positive for E-, P-, and L-selectin ligand activities, when tested by respective selectin chimeras in flow cytometry or flow adhesion assays. Consistent with their E-selectin ligand activity, all the cell lines expressed putative glycans for E-selectin binding, which included sle X, sle A, and/or VIM-2 glycans, found by flow cytometric analysis. Yet the level of selectin ligand activity and glycan expression in BCSCs was conspicuously lower than non-bcscs. The abundant production of selectin reactive glycans in non-bcscs was associated with α- (1,3)- and α-(1,4)-fucosyltransferases, as indicated by qrt-pcr analysis. Furthermore, 4 To be submitted as: Breast cancer stem and non-stem-like cells express distinct levels of E-selectin ligand activity. Shirure VS, Delgadillo LF, Henson KA, Xiong C, Benencia F, and Burdick MM Part of the work was previously submitted as abstracts to American Association of Cancer Research Conference, 2011 and 2012.

120 120 BCSCs were mesenchymal-like and non-bcscs were epithelial-like cells, as found by fluorescence microscopy of breast cancer cells for E-cadherin and N-cadherin. Significantly, treatment of a non-bcsc cell line with TGF-β induced epithelial to mesenchymal transition (EMT), marked by changes in cell morphology and increase in expression levels of N-cadherin and E-cadherin suppressor genes. These EMT cells showed somewhat reduced levels of E-selectin ligand activity under physiological flow conditions, indicating loss of selectin ligand activity upon acquisition of mesenchymal properties. In summary, the selectin ligand activity of BCSCs is lower than non-bcscs and the level of selectin ligand activity of breast cancer cells is perhaps controlled by the BCSC and EMT state of the cells. 5.2 Introduction One of the important evolving concepts in the pathogenesis of cancer is that only a subset of cancer tumor cells possesses potential to generate new tumors. These cancer cells are known as cancer stem cells, as they are thought to possess stem cell properties of self-renewal and asymmetric cell division [8,11,18]. In a seminal paper, Al Hajj et al. found that small number of breast epithelial cancer cells with positive expression of CD44 and negative or low expression of CD24 (CD44 + /CD24 -/low ) are sufficient to form new tumors in mouse models [12]. Furthermore, these breast cancer stem-like cells (BCSCs) were able to generate secondary tumor with heterogeneous cell populations resembling the original tumor mass [12]. Recently, it has been found that the expression of CD44 + /CD24 -/low markers is the result of cytokine induced process of epithelial to mesenchymal transition (EMT) [19],

121 121 which is marked by the loss of epithelial adhesion molecule E-cadherin and increase in the expression of mesenchymal marker N-cadherin. It is believed that cancer cells undergo EMT to achieve higher migratory properties necessary to pass through maze of cellular and matrix proteins to invade surrounding tissue [5,8,11,19]. However, which EMT state is required for circulating tumor cells to adhere to endothelium is not known. The selectin family of adhesion molecules consists of E-, P-, and L-selectin. E- and P-selectin are expressed by endothelial cells in response to inflammatory cytokines such as TNF-α or IL-1β, and L-selectin is natively expressed by most types of leukocytes [24,29,33,34]. Also, E-selectin is constitutively expressed by bone marrow endothelium and P-selectin is expressed by activated platelets [24,29,31,33,34]. While E- and P- selectins mediate adhesion of circulating tumor cells to vascular endothelium, L-selectin is believed to mediate adhesion of the tumor cells to leukocytes which could bind to endothelium. Compared to the other selectins, E-selectin is thought to be more relevant in bone metastatic cancers since human bone marrow endothelium constitutively expresses E-selectin [31]. However, each of the selectins has been independently implicated in metastasis [24,29]. In the present study, BT-20, MDA-MB-468, ZR-75-1, MDA-MB-231, and Hs- 578T cells were analyzed for selectin ligand activity, putative BCSC marker CD44 + /CD24 -/low expression, and epithelial or mesenchymal markers. The E-, P-, and L- selectin ligand activities of these cell lines were correlated with CSC and EMT traits. In addition, EMT was induced in a representative non-bcsc cell line to reveal E-selectin

122 122 ligand activity during EMT. Ultimately, the mechanistic association of selectin ligand activity with BCSC and EMT may reveal a new paradigm for cancer metastasis. 5.3 Materials and methods Cell culture The BT-20 breast cancer cell line obtained from the American Type Culture Collection (ATCC; Manassas, VA) was grown in minimum essential medium (Life Technologies, Carlsbad, CA) with 10% FBS and 1x penicillin-streptomycin. MDA-MB- 231, Hs-578T, and MDA-MB-468 breast cancer cell lines (all ATCC) were cultured in Dulbecco s modified eagle medium (Life Technologies) with 15% FBS and 1x penicillinstreptomycin. ZR-75-1 breast cancer cell line was cultured in RPMI (Life Technologies) supplemented with 10% fetal bovine serum (FBS) and 1x penicillin-streptomycin (Life Technologies). E-selectin transfected Chinese hamster ovary cells (CHO-E) were a generous gift from Dr. Robert Sackstein (Harvard Medical School, Boston, MA). CHO-E cells were maintained in MEM supplemented with 10% FBS and 0.1 mm non-essential amino acids (Life Technologies) and 1x penicillin-streptomycin Antibodies and chimera constructs Anti-CD43 (1G10), CD44 (515), CD66 (COL-1), and PSGL-1 (KPL-1), HECA- 452, anti-sialyl Lewis X (sle X ; CSLEX-1), FITC-conjugated CD44 (G44-26), PEconjugated CD24 (ML-5) monoclonal antibodies (mabs), and all isotype controls were obtained from BD Biosciences (San Jose, CA). Recombinant mouse selectin human Fc chimera constructs for E-selectin, P-selectin, and L-selectin (E-Ig, P-Ig, and L-Ig chimera, respectively) were obtained from R & D Systems (Minneapolis, MN). Anti-

123 123 sialyl Lewis A (sle A ; KM-231) was from Calbiochem (San Diego, CA), and anti-cd65s (VIM-2) was from AN DER GRUB Bio Research GmbH (Gerichtsberg, Austria). FITCconjugated secondary antibodies were from Southern Biotech (Birmingham, AL) Flow cytometry In single labeling experiments, cells were washed with blocking buffer and incubated with primary antibody or isotype control prepared at a concentration of 10 μg/ml for 30 min at 4 C. Cells were washed and incubated with secondary antibody for 30 min at 4 0 [27,28]. In triple labeling experiments, cells were washed with blocking buffer and incubated with selectin chimera (10 μg/ml for 30 min at 4 C). The cells were washed and incubated with CD44-FITC, CD24-PE mabs, and anti-human APC secondary antibody for 30 min at 4 C. Finally, cells were washed and analyzed using a FACSAria Special Order Research Product flow cytometer/sorter (BD Biosciences) Flow adhesion assays The selectin substrates were prepared by incubating selectin chimera or human IgG at a concentration of 10 µg/ml in DPBS on a Petri dish for 1 hr. Subsequently, the surface was blocked with 1% BSA solution prepared in DPBS for 1 hr. Breast cancer cells were perfused over the selectin substrate using a flow chamber, the experimental set up of which is detailed in the chapter 1. Each experimental run was performed at a bone marrow microvasculature shear rate of 100 s -1 [74] for 2 min. The number of cells attaching from the free stream were counted for the 2 min time period. Cell velocity was calculated by capturing stills from video over 5 sec and converting pixel distance into microns using Image J software [27,28,48].

124 Quantitative reverse transcriptase polymerase chain reaction (qrt-pcr) A detailed protocol is given in chapter 4. Briefly, extraction and purification of RNA was performed using RNeasy Plus kit (Quiagen, Valencia, CA) following the manufacturer s protocol. The RNA was reverse transcribed by a high capacity reverse transcription kit (Applied Biosystems) following the manufacturer s instructions. PCR was performed on cdna, synthesized from 75 ng of RNA, by SYBR Green FastMix (Quantas Biosciences) chemistry and monitored by an icycler iq5 real-time PCR instrument (Bio-Rad Laboratories, Hercules, CA). The forward and reverse primers were purchased from Integrated DNA Technologies (Coraville, IA) and are given in chapter Fluorescence microscopy Cells grown over tissue culture grade EZ slides (Millipore, Billerica, MA) were washed and fixed in 4% methanol-free paraformaldehyde in DPBS. The slides were incubated with primary antibody or isotype control for 1hr at 4 C. The slides were washed and incubated with AlexaFluor 488 (green)-conjugated secondary antibody for 1hr at 4 C. A drop of ProLong Gold antifade reagent with DAPI (Invitrogen) was added before placing cover slips. The slides were imaged using 40x objective under wide field fluorescence using a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany) and Simple PCI software (Hamamatsu Corporation, Sewickley, PA) [27] TGF-β treatment for EMT BT-20 cells seeded in duplicate tissue culture flasks were allowed to grow overnight. The cell culture medium of one flask was replaced with the media

125 125 supplemented with TGF-β1 (R & D systems) at a concentration of 10 ng/ml, and the other flask was maintained in the regular medium. The cells were grown for seven days in respective media with a replacement of old media with fresh media on fourth day Statistics Data are represented as mean ± SE for at least 3 independent experiments except as otherwise mentioned. Statistical significance was determined by paired Student s t- test or by one way ANOVA coupled with Tukey s multiple comparison test, and probability values of P 0.05 were considered statistically significant. 5.4 Results Selectin ligand activity of stem-like and non-stem like breast cancer cell lines To assess the expression of putative breast cancer stem-like cell (BCSC) markers and selectin ligands, five breast cancer cell lines were triple stained with CD44 and CD24 mabs and selectin chimeras and analyzed by flow cytometry. While BT-20 and MDA- MB-468 cells were non-bcscs of CD44 + /CD24 + phenotypes, ZR-75-1 cells were also non-bcscs expressing CD44 - /CD24 + (Fig. 5.1). On the other hand, MDA-MB-231 and Hs-578T cells expressed the BCSC phenotype CD44 + /CD24 -/low. Consistent with a previous report [16], these data show that the five breast cancer cell lines of the present study have differential profiles of CD44 and CD24 expression related to BCSCs and non- BCSCs. The E-, P-, and L-selectin ligand activity detected by the respective chimeras, relative to human IgG negative control, was positive for all of the cell lines (Fig. 5.1). Interestingly, the level of E-selectin ligand activity, found by normalizing mean

126 126 fluorescence intensity of E-Ig chimera staining to the negative control staining, of non- BCSCs was higher than that of BCSCs, but no such difference was found in the levels of P- and L-selectin ligand activities. These data indicate that the expression of E-selectin ligands, than that of P- or L-selectin ligands, may be more tightly regulated in a phenotype specific manner Selectin ligand activity of breast cancer cells under physiological flow conditions To compare the selectin ligand activities of representative BCSCs and non- BCSCs under bone marrow microvascular flow conditions, BT-20 and MDA-MB-231 cells were perfused over purified selectin substrate at a shear rate of 100 s -1. The tumor cells specifically bound to selectins, as they failed to bind to human IgG negative control (Fig. 5.2). The capture efficiency to each type of selectins for BT-20, compared to MDA- MB-231 cells, was dramatically higher, indicating that non-bcscs possess stronger selectin ligand activity than BCSCs. Since all selectin substrates were prepared by using similar concentration of selectin chimeras, this system allowed comparison binding avidities of a specific cell types towards different selectins, assuming that coating efficiency of each type of selectin was similar. The majority of cells of both cell lines remained stationary on E-selectin, mainly rolled on P-selectin, and was able to form only weak interactions of transient tethering on L-selectin. These data show that the binding avidity of BCSCs and non-bcscs is highest towards E-selectin followed by P-selectin and then L-selectin.

127 127 BT-20 MDA-MB-468 ZR-75-1 MDA-MB-231 HS-578T CD24 CD44 Cell count E-selectin P-selectin L-selectin Figure 5.1. BCSCs and non-bcscs express E-, P-, and L-selectin ligand activities. Various breast cancer cell lines were triple labeled with CD44, CD24 mabs, and selectin chimeras (E-, P-, or L-selectin chimera) and analyzed by flow cytometry. Blue curve shows isotype, and red curve shows specific probe.

128 * higg E-selectin * higg P-selectin Adhering/mm *$ Rolling /mm *$ 0 BT-20 MDA-MB BT-20 MDA-MB-231 # Tethering/mm * higg L-selectin *$ 0 BT-20 MDA-MB-231 Figure 5.2. Different levels of selectin ligand activity are expressed by BCSCs and non- BCSCs under physiological flow conditions. BT-20 and MDA-MB-231 breast cancer cells were perfused over E-, P-, or L-selectin chimera, or human IgG substrates at a wall shear rate of 100 s -1 in the parallel plate flow chamber assay. Data are mean ± SE for n = 3. * P < 0.05 with respect to adhesion to negative control, and $ P < 0.05 with respect to adhesion of BT-20 cells.

129 Breast cancer cell lines express carbohydrate epitopes for selectin ligand activities Since selectins primarily bind to carbohydrate epitopes displayed on a ligand [24], expression of selectin reactive glycans on the cell surface of breast cancer was tested by flow cytometry. For this purpose, mabs recognizing sle X (CSLEX-1 and KM-93 mabs), sle A (KM-231 mab), VIM-2 (anti-cd15s mab), general sialofucosylation (HECA-452 mab), sialyl dimeric Le X (FH-6 mab), or sulfated sle X (MECA-79 mab) were used. Two non-bcsc lines, BT-20 and ZR-75-1, robustly expressed sle X, sle A, and VIM-2 glycans, and the other non-bcsc lines, MDA-MB-468 cells, showed strong expression of VIM-2 (Fig. 5.3). The expression of sulfated sle X and sialyl dimeric Le X on all of the non-bcscs was minimal. These data show that non-bcscs abundantly express sialofucosylated glycans, mainly sle X, sle A, and/or VIM-2, correlating with selectin ligand activity. On the other hand, BCSCs (MDA-MB-231 and Hs-578T) minimally expressed sialofucosylated or sulfated carbohydrates detectable by any of the CSLEX-1, HECA- 452, VIM-2, FH-6, or MECA-79 mabs (Fig. 5.3). Yet both of the BCSCs in this study showed positive expression of sle X by KM-93 mab that generally detects a variety of sle X and related structures decorated on proteins and lipids. Also, positive expression of sle A by KM-231 mab was found on MDA-MB-231 but not on Hs-578T cells (Fig. 5.3). A comparison of the levels of KM-93 and KM-231 antigen expression in BCSCs and BT- 20 and ZR-75-1 cells indicated that BCSCs express lower levels of these antigens than the two non-bcscs. Together these results indicate that BCSCs as well as non-bcscs

130 express reported putative selectin binding epitopes, but BCSCs, compared to non-bcscs, generally possess lower levels of these carbohydrates. 130 BT-20 MDA-MB-468 ZR-75-1 MDA-MB-231 Hs-578T KM-93 Cell count CSLEX-1 KM-231 HECA-452

131 131 BT-20 MDA-MB-468 ZR-75-1 MDA-MB-231 Hs-578T VIM2 Cell count MECA-79 Figure 5.3. Putative E-selectin reactive glycans are expressed by BCSCs and non- BCSCs. Expression of glycans was tested on the five breast cancer cell lines by flow cytometry. Blue curves show isotype, and red curves shows specific mab recognizing the glycans described in section FH6

132 Levels of α-(1,3)- and α-(1,4)-fucosyltransferases are different in BCSCs and non- BCSCs The α-(1,3)-fucosyltransferases are involved in the transfer of terminal fucosyl groups to generate sle X and VIM-2 and α-(1,4)- fucosyltransferases direct transfer of fucose to produce sle A glycan [24]. To find the expression levels of these fucosyltransferases, mrna in breast BCSCs and non-bcscs were analyzed with qrt- PCR. Breast non-bcscs, than BCSCs, expressed FT-3, which is a α-(1,3)- and also α- (1,4) fucosyltransferase, at a significantly higher level and generally also expressed FT-6, which is a α-(1,3)- fucosyltransferase, at a high level (Fig. 5.4). However, no difference was found in the levels of FT-4, FT-5, and FT-7 in BCSCs and non-bcscs. Therefore, FT-3, and FT-6 are perhaps the major catalysts for abundant production of sle X, sle A, and VIM-2 (Fig. 5.3) glycans in non-bcscs Expression of epithelial and mesenchymal markers by various breast cancer cells To test whether cells were in the epithelial and mesenchymal state, breast cancer cells were immunostained for epithelial marker E-cadherin and mesenchymal marker N- cadherin for fluorescence microscopy analysis. BCSCs were negative for E-cadherin expression but were strongly positive for N-cadherin expression, compared to the respective isotype controls (Fig. 5.5). On the other hand, the non-bcscs showed positive expression of E-cadherin yet BT-20 cells were negative and MDA-MB-468 cells were somewhat positive for N-cadherin, compared to the respective isotype controls (Fig. 5.5). These data associate non-bcscs with epithelial-like cells and BCSCs with mesenchymal-like cells.

133 133 Relative expression (ΔCt) $ BT-20 MDA-MB-468 ZR-75-1 MDA-MB-231 Hs-578T FT-3 Relative expression (ΔCt) BT-20 MDA-MB-468 ZR-75-1 MDA-MB-231 Hs-578T ND FT-4 FT-5 FT-6 FT-7 Figure 5.4. α-(1,3)- and/or α-(1,4)-fucosyltransferases are expressed at higher levels in non-bcscs than BCSCs. qrt-pcr was performed on breast cancer cell mrna using primers designed for detection of specific FTs, and the data are represented by relative expression with housekeeping gene (GAPDH). Data are mean ± SE for n >3, $ P < 0.05 with respect to non-bcscs.

134 EMT induced by TGF-β increases may alter expression of E-selectin ligands As E-selectin is believed to be a relevant selectin for cancer metastasis, the effect of EMT on E-selectin ligand activity was studied. For this purpose, BT-20 cells were treated with TGF-β1, a well-documented EMT inducing cytokine, and observed under microscopy or analyzed by qrt-pcr. The treatment changed cell morphology from cobble stone-like to elongated, and increased expression of N-cadherin and E-cadherin repressors SLUG and TWIST (Fig. 5.6), indicating induction of EMT. However, the level of the other E-cadherin repressor SNAIL was not changed upon the treatment, suggesting a minor role for SNAIL in induction of EMT in BT-20 cells. When TGF-β treated or untreated BT-20 cells were perfused over E-selectin expressing CHO-E cells at a wall shear rate of 100 s -1 the rolling velocity of treated BT-20 cells was increased (Fig. 5.7), indicating loss of E-selectin ligand activity. A minor but not significantly higher increase in rolling velocity of the cells upon TGF-β treatment may be due to only partial EMT of the cells, as limited reduction in E-cadherin levels (Fig. 5.6B) was obtained upon TGF-β treatment. The complete EMT of BT-20 cells may significantly affect selectin ligand activity. Nevertheless, the present data imply E-selectin ligand activity is regulated during EMT. 5.5 Discussion Selectins are involved in the adhesion of circulating tumor cells during cancer metastasis. The literature on the selectin science is abundant, according to which selectins primarily bind to sialofucosylated glycans displayed on carrier molecules, such as CD44. Interestingly, CD44 is also a molecular signature for putative BCSCs, which have been

135 135 purported to possess stem cell properties relevant for formation of metastatic tumors. The stem cell properties of BCSCs are thought to be acquired during EMT. Significantly, some studies suggest expression of E-selectin ligands of tumor cells is regulated by EMT [70,118]. In this chapter, the selectin ligand activity of various breast cancer cells in the context of cancer stem cell and EMT models was evaluated. The five breast cancer cell lines of the present study were BCSCs expressing putative markers CD44 + /CD24 -/low, or non-bcscs expressing alternate markers CD44 + /CD24 + or CD44 - /CD24 + (Fig. 5.1). All of these cells possessed positive ligand activities for E-, P-, and L-selectin, underscoring the importance of selectins in the adhesion of BCSCs as well as non-bcscs. Notably, breast cancer frequently metastasize to bone [93,120], where E-selectin is constitutively expressed on the vascular endothelium [31]. Hence, E-selectin may be a prominent player in breast cancer metastasis. In line with this idea, the E-selectin ligand activity of the majority of the cell lines was higher than P- or L-selectin ligand activities. Yet the possibility of involvement of other two selectins in breast cancer metastasis cannot be discounted, as BCSCs and non-bcscs showed positive expression of ligands for the other two selectins. Thus, we believe that P- and L-selectin, next to E-selectin, could be the important cell adhesion molecules. In binding assays performed under physiological flow conditions, the non-bcsc cell line BT-20 binding to E-selectin was significantly higher than that of the BCSC cell line MDA-MB-231 (Fig. 5.2). Complementing this trend, differences were also noted in the level of selectin reactive glycans. While non-bcscs abundantly expressed purported

136 136 selectin reactive glycans, such as sle X, sle A, and VIM-2, and related fucosyltransferases, BCSCs expressed these molecules at a low level (Fig. 5.3 and 5.4). A systematic evaluation of CD44 and other ligands is necessary to answer whether their expression levels vary in BCSCs versus non-bcscs. Nevertheless, these data relating conspicuously low levels of selectin ligand activities with BCSCs indicate that perhaps the E-selectin ligand activity is cross regulated during acquisition of cancer stem cell properties by controlling level of glycosylations. This issue is addressed in the present study to a certain degree by relating EMT, a process by which non-bcscs can acquire BCSC traits, with E-selectin ligand activity. Consistent with the previous reports [18,19,119], BCSCs were in a mesenchymal-like state, and non-bcscs were in an epithelial-like state (Fig. 5.5). These data indicate that non-bcscs in the epithelial state possess higher levels of E-selectin ligand activities than BCSCs in the mesenchymal state. Further clarity in this idea was obtained by inducing EMT in non-bcsc line BT-20 by TGF-β1. Even with partial EMT achieved in the study, the cells showed somewhat decreased E-selectin ligand activity, indicating that E-selectin ligand activity may be regulated by EMT.

137 137 Isotype E-cadherin N-cadherin Hs-578T MDA-MB-231 MDA-MB-468 BT-20 Figure 5.5. BCSCs are mesenchymal-like cells and non-bcscs are epithelial like cells. Breast cancer cells grown on tissue culture slides were labeled with anti-e-cadherin or N- cadherin mabs and were imaged in epifluorescence microscopy. Scale bar for all is 15 µm. The BT-20 and MDA-MB-231 figures are the same as Fig. 4.7, which are included here to compare staining of different cell lines.

138 138 A B Relative expression (ΔCt) 6.E-02 5.E-02 4.E-02 3.E-02 2.E-02 1.E-02 E-cadherin N-cadherin * 6.E-05 5.E-05 4.E-05 3.E-05 2.E-05 1.E-05 0.E+00 BT-20 BT-20 (TGF-β) BT-20 BT-20 (TGF-β) 0.E+00 C Relative expression (ΔCt) 6.E-06 5.E-06 4.E-06 3.E-06 2.E-06 1.E-06 0.E+00 TWIST BT-20 * BT-20 (TGF-β)

139 139 D Relative expression (ΔCt 3.E-02 2.E-02 2.E-02 1.E-02 5.E-03 * SLUG SNAIL 3.E-03 2.E-03 2.E-03 1.E-03 5.E-04 0.E+00 BT-20 BT-20 (TGF-β) BT-20 BT-20 (TGF-β) 0.E+00 Figure 5.6. TGF-β induces EMT in BT-20 cells. (A) Untreated or TGF-β1 treated BT-20 cells were imaged by phase contrast microscopy. Scale bar indicates 10 µm. (B, C, and D) qrt-pcr was performed on mrna extracted from TGF-β1 treated or untreated BT-20 cells and the data is represented by C T method by normalization with GAPDH housekeeping gene. Data are mean ± SE for n = 4. * P < 0.05 with respect to untreated BT-20 cells. 12 Rolling velocity (µm/s) BT-20 BT-20 (TGF-β) Figure 5.7. EMT increases E-selectin mediated rolling velocity of BT-20 cells Untreated or Untreated or TGF-β1 treated BT-20 cells were perfused over CHO-E cells at a wall shear rate of 100 s -1 in parallel plate flow chamber assay. Data are mean ± SE for n = 5 cells.

140 140 In view of these data, we hypothesize that circulating breast cancer cells in epithelial state attach to endothelial cells via the selectin pathway (Fig. 5.8). Subsequently, EMT reduces E-selectin ligand activity and attached non-bcscs acquire BCSC properties, allowing them to easily extravasate and migrate into the surrounding tissue (Fig. 5.8). In support of the notion that tumor cells transit back and forth between epithelial and mesenchymal states, recent reports have found BCSCs can reverse EMT, by a process known as mesenchymal to epithelial transition (MET). The MET BCSCs are characterized by the expression of aldehyde dehydrogenase (ALDH), while epithelial cell adhesion molecule (EpCAM) and CD49f have also been reported as markers. Notably, a small subset of EMT BCSCs with the expression of CD44 + /CD24 -/low markers has been found to simultaneously express a MET BCSC marker (ALDH). Thus, now it is believed that cells that are in an intermediate state, from where they can undergo EMT or MET, may have the highest metastatic potential [18]. In conclusion, we found that BCSCs as well as non-bcscs express positive ligand activities for E-, P-, and L-selectins. Yet E-selectin, which is constitutively expressed by bone marrow endothelium, may have the prominent role in the adhesion of circulating breast tumor cells to endothelium. Also, non-bcscs express higher levels of E-selectin ligand activity than BCSCs. The selectin ligand activity may be regulated by EMT by which non-bcscs acquire BCSC properties. These data reveal a novel paradigm for the role of selectins and their ligands in breast cancer metastasis.

141 141 (1) Epithelial-like circulating non-bcscs express E- selectin ligands and TGFR (2) Tethering (3) Rolling (4) Firm attachment (5) TGF-β reduces E- selectin ligandsand induces EMT (6) BCSCs Extravasate Endothelial cell Cancer cell TGF-β E- selectin E-selectin TGF ligand receptor Non-selectin ligand molecules Metastatic tumor Figure 5.8. Hypothesized model to relate E-selectin mediated adhesion with BCSC and EMT models in metastasis. Epithelial like non-bcscs expressing E-selectin ligands in the circulatory system bind to vascular endothelium via E-selectin. Subsequently, local factors, such as TGF-β, induce EMT by which non-bcscs become BCSCs with loss of E-selectin ligand activity. These cells extravasate to form a metastatic tumor mass.

142 142 CHAPTER 6: VARIOUS LEVELS OF E-SELECTIN LIGAND ACTIVITIES OF CANCER TISSUES AND CELL LINES ARE DETECTABLE BY HECA-452 AND E- SELECTIN CHIMERA 6.1 Abstract E-selectin ligands mediate adhesion or cancer cells to vascular endothelium and thus are believed to play a crucial role in cancer metastasis. E-selectin ligand activity can be assessed by using HECA-452 and recombinant E-selectin (E-Ig) chimera probes. In this work, the potential of these probes in finding E-selectin ligand activities of cancer cell lines and tissues is evaluated. Various levels of activities of colon and breast cancer cell lines were detectable by HECA-452 and E-Ig chimera in flow cytometric analysis. Further, the cancer tissues were specifically detected by the probes, compared to the respective isotype controls, and the activity of cancer tissues was distinguishable from non-cancer tissues, when tested in fluorescence IHC analysis. Importantly, either of the probes detected a range of activities on cancer tissues, distinguishing different forms of cancers based on level of the activities. Interestingly, activity of a minority of cell lines, tissues, or portion of the tissues was detectable by one but not the other probes, indicating the differences in aspects of a ligand are detectable by these two probes. Together, the data indicates that E-selectin ligand activity detectable by E-Ig chimera and HECA-452 mab may serve as an indicator in novel cancer diagnostic and prognostic applications.

143 Introduction Endothelial E-selectin and its ligands expressed on tumor cells are believed to play a critical role in cancer metastasis [24,29]. Expression of E-selectin ligands on tumor cells may indicate high metastatic potential of the cells, and hence they could be target for novel diagnostic, prognostic and therapeutic strategies. This idea motivated several investigators to identify and characterize E-selectin ligands of a variety of cancer cells [24-29,48,50,51,69,80]. As a result a number of E-selectin ligands, such as HCELL, CD43, PCLP, CEA, etc., are now known [24-29,48,50,51,69,80], and several additions to this list are expected in the near future. However, the identification of E-selectin ligands is complicated by a number of factors, such as cancer cells typically express multiple E- selectin ligands, and an E-selectin ligand itself is a complex of molecules composed of glycans and a protein or lipid scaffold [24-29,48,50,51,69,80]. As an alternative strategy, finding ligands by a universal probe that recognizes E-selectin ligand activity, but not necessarily the identity of the ligands, could provide valuable information for better understanding metastasis. The available literature analyzing E-selectin-ligand interactions provides clues for such a universal probe. E-selectin specifically binds sialofucosylated oligosaccharides presented on core lipid or protein molecules [24]. Sialofucosylated structures are numerous and some examples are sialyl Lewis X (sle X ) and its stereo-isomer sialyl Lewis A (sle A ). Classically, carbohydrate recognizing monoclonal antibodies (mabs), such as HECA-452, CSLEX-1, and KM-231, have been utilized to detect E-selectin ligand activity [24-28,38,48]. Many of these mabs only detects specific sialofucosylated

144 144 structures, but HECA-452 mab recognizes diverse sialofucosylated moieties, including sle X and sle A [38]. Therefore, it has been a mab of choice for screening of E-selectin ligand activity of several cell lines, such as colon and breast cancer cell lines [26,28,60]. Alternative probes to explore E-selectin ligands are chimeric forms of recombinant E-selectin (E-Ig chimera). These constructs consist of lectin-like antigen binding region from murine, rat, or human E-selectin fused to the Fc domain of human IgG. To obtain optimal levels of detection signal, the choice of appropriate E-Ig chimera construct appears to be critical, which maybe because affinities of E-Ig chimeras obtained from different species vary dramatically [121]. For instance, even with very sensitive analysis techniques such as flow cytometry, staining of highly E-selectin ligand positive leukocytes with human E-Ig chimera revealed very weak signal, in contrast to staining of these cells with murine E-Ig chimera, which provided a highly positive signal [121]. The important questions to address in using these probes are whether they can distinguish between cancerous and non-cancerous tissues and whether they can identify distinct levels of E-selectin ligand activities. To answer these questions, tissue microarrays consisting of colon or breast cancer and cancer cell lines were analyzed by HECA-452 and murine E-Ig chimera in immunohistochemistry (IHC) assays. Such analysis of tissues may form a foundation for larger studies exploring E-selectin ligand activity for clinical applications.

145 Materials and methods Cell culture The BT-20 breast cancer cell line (American Type Culture Collection; ATCC; Manassas, VA) was maintained in minimum essential medium (Invitrogen; Carlsbad, CA) with 10% fetal bovine serum (FBS) and 1x penicillin-streptomycin (Invitrogen). ZR and T-47D (all ATCC) breast cancer cell lines were cultured in RPMI (Invitrogen) supplemented with 10% FBS and 1x penicillin-streptomycin. Hs-578T, MCF-7, MDA- MB-231, and MDA-MB-468 breast cancer cell lines (all ATCC) were cultured in Dulbecco s modified eagle medium (Invitrogen) with 15% FBS and 1x penicillinstreptomycin Antibodies and chimera constructs Human and murine recombinant E-selectin human Fc chimera constructs (E-Ig chimera) were purchased from R & D systems (Minneapolis, MN). HECA-452 (anticutaneous lymphocyte antigen) and all isotype controls were from BD Biosciences (San Jose, CA). Fluorescein isothiocyanate (FITC)-conjugated and phycoerythrin (PE)- conjugated polyclonal secondary antibodies were from Southern Biotech (Birmingham, AL). AlexaFluor 488- and AlexaFluor 568-conjugated secondary antibodies were obtained from Invitrogen Preparation of sle X microsphere The details of microsphere preparation are described previously [122]. In brief, 9.95 μm superavidin microspheres (Bangs Laboratories, Fishers, IN) were washed and incubated for 30 min with 1% BSA in Dulbecco s phosphate buffered saline (blocking

146 146 buffer). Microspheres were suspended at 1 million/ 100 µl with 10 µg/ml biotinylated multivalent sle X (Glycotech, Gaithersburg, MD) for 1 hour and then washed to use them for flow cytometry Flow cytometry Cells or microspheres were incubated with primary antibody or isotype control prepared at 10 µg/ml concentration for 30 min at 4 C. Cells or microspheres were washed and incubated with secondary antibody for 30 min at 4 0 C. Finally, cells or microspheres were washed and analyzed by a FACSAria Special Order Research Product flow cytometer/sorter (BD Biosciences). The data obtained was analyzed by FlowJo (version 7.6.4) software (Ashland, OR) [27,28] Tissue microarrays Formalin fixed paraffin embedded (FFPE) tissue microarray (TMAs) slides were obtained from US Biomax, Rockville, MD; ref [88]. Colon cancer TMAs consisted of non-cancerous colon (N) and tissues derived from mucinous adenocarcinoma with necrosis (MAC), papillary adenocarcinoma (PC), carcinoid tumors (CC), and signet-ring cell carcinoma (SC) tumors. Breast cancer TMAs consisted of non-cancerous breast tissues (N) and tissues derived from mixed lobular and duct carcinoma (LDC), mucinous carcinoma (MC), invasive ductal carcinoma (IDC), medullary carcinoma (MdC), and neuroendocrine carcinoma (NC) tumors Immunohistochemistry and image analysis TMAs were deparafinized by heating at 60 C for 30 min, and serially incubating with xylene, 95% ethanol, 70% ethanol, and deionized water [101]. The slides were

147 147 blocked with 1% BSA and 1% FBS in DPBS for 30 min, and incubated for 1 hour with HECA-452 mab [69] and murine E-Ig chimera construct at 5 µg/ml and 10 µg/ml, respectively. The slides were washed with blocking buffer and incubated with appropriate AlexaFluor conjugated secondary antibodies for 30 min at 4 µg/ml and at room temperature. After washing, the TMAs were mounted in ProLong Gold antifade reagent [101] and were imaged under wide field fluorescence using a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany) equipped with a motorized high precision specimen stage and an automated optical filter cube wheel with emission and excitation filters. The microscope was controlled by the Simple PCI software (Hamamatsu Corporation, Sewickley, PA). The raw images obtained were subjected to blind deconvolution algorithms in AutoQuant X software (Media Cybernetics, Bethesda, MD) to reduce out of focus light. Quantitative analysis of fluorescence intensity was performed by using Image-Pro Plus (Media Cybernetics, Inc.; Bethesda, MD) in manual area selection mode. A positively stained tissue area was the tissue area for which staining intensities of the probes were higher than the highest intensity corresponding to 90 % area stained with the negative controls. 6.4 Results Affinities of human and murine E-Ig chimeras towards sle X are distinct It has been reported that the affinities of E-Ig chimeras towards E-selectin ligands vary dramatically, with human E-Ig chimera showing weakest signal and murine E-Ig chimera revealing the strongest signal [121]. However, these experiments were performed on cells which have complex cell surface topography and molecular display

148 148 that could affect the observations. To circumvent such possibilities, sle X coated microspheres were prepared to study the nature of E-Ig chimera binding. These microspheres were positively stained for HECA-452 mab compared to isotype control (Fig. 6.1A), indicating the effective conjugation of sle X on the microsphere surface. Furthermore, the microspheres showed stronger signal with murine than human E-Ig chimera (Fig. 6.1B). Thus, further experiments were conducted using murine E-Ig chimera E-Ig chimera and HECA-452 mab can detect different levels of activities on cancer cells Accumulating evidence strongly implicates the E-selectin adhesion pathway in colon and breast cancer metastasis [24-29,35,48], and cells of these cancers have been reported to express a number of E-selectin ligands [24-29,35,48]. To test whether these probe could detect different levels of E-selectin ligand activities, a colon cancer cell line LS-174T and five breast cancer cell lines, BT-20, ZR-75-1, MDA-MB-468, MDA-MB- 231, and MCF-7 were tested for E-selectin ligand activity with HECA-452 mab and murine E-Ig chimera by flow cytometry. The majority of cell lines specifically reacted with the probes, compared to respective isotype controls. Furthermore, these cell lines showed distinct levels of E-selectin ligand activities detectable by either of the probes (Fig. 6.2). For instance, the E-selectin ligand activity detectable by E-Ig chimera decreased by many fold from BT-20, ZR-75-1, to MDA-MB-468. Quite interestingly, the levels of E-selectin ligand activity of a cell line detected by the two probes also varied. This was more apparent for some cell lines such as MDA-MB-468, which showed E-

149 149 selectin ligand activity detectable by E-Ig chimera, but not by HECA-452 mab. Thus, E- selectin ligand activity of cancer cells is detectable by HECA-452 mab and E-Ig chimera, yet these probes show different aspects of E-selectin ligands. A HECA-452 B * $ Mean fluorescence intensity * higg heigg meigg Figure 6.1. Human and murine E-Ig chimeras reveal different reactivities with sle X coated microspheres. (A) sle X coated microspheres were labeled with HECA-452 mab, and analyzed by flow cytometry. The red curve shows specific mab, and the blue curve show isotype. (B) sle X coated microspheres were labeled with human or murine and analyzed by flow cytometry. The error bars are SE for n =3. *P < 0.05 with respect to h-igg isotype, and $ < 0.05 with respect to human E-Ig chimera.

150 Activity of colon carcinoma tissues is detectable by HECA-452 mab and E-Ig chimera To find E-selectin ligand activity in colon carcinoma tissues, fluorescence IHC was performed on colon cancer TMAs, which consisted of multiple cancerous and noncancerous colon tissues. The TMAs were dually labeled with HECA-452 mab and murine E-Ig chimera revealed positive staining with either probes, compared to the staining with respective isotype controls (Fig. 6.3E), showing the specificity of detection. Notably, E-selectin ligand activity, detectable by either of the probe, of some cancer tissues, such as colon MAC tissue, was several folds higher than that of non-cancer tissues. Furthermore, more than 40% of area of three out of four cancer tissues was HECA-452 positive, yet less than 20% of area of the non-cancerous colon tissues were positive for HECA-452 (Fig. 6.3E). These data demonstrate that E-selectin ligand activity detectable by HECA-452 mab and E-Ig chimera probes could be used to distinguish at least some forms of cancer tissues from non-cancer tissues A range of activities of breast cancer tissues is detectable by HECA-452 mab and E-Ig chimera In addition to colon cancer tissues, the E-selectin ligand activity of breast carcinoma tumor tissues was explored to find the relevance of E-selectin ligand activity in multiple cancers. The breast cancer tissues tested positively with HECA-452 as well as E-Ig chimera, compared to staining with respective isotype control (Fig. 6.4E), showing the specificity of detection. The majority of breast cancer tissues showed higher E- selectin ligand activity (more than 15% positive area) compared to normal tissues (less

151 151 than 10% positive area) by HECA-452 as well as E-Ig chimera (Fig. 6.4E). A striking example is breast LDC tissue which showed robust staining with either of the probes (Figs 6.4A, B, C, and D) that was several fold higher than non-breast cancer tissue (Fig. 6.4E). Additionally, a range of E-selectin ligand activities detected by either of the probes are noticeable for breast cancer tissues (Fig. 6.4E). Interestingly, the positive staining of non-breast cancer tissue was restricted mainly to ductal lumina (Fig. 6.5), which generate and carry several mucinous proteins in the form of milk. This is in contrast to the breast cancer tissues which stained irregularly all over the area (Figs 6.4A and B). These data collectively showed that a range of E-selectin ligand activities of carcinoma tissues is detectable by HECA-452 mab and E-Ig chimera Some tissue antigens are detected by the one but not by the other probe Although a major portion of the cancer tissues detectable by HECA-452 overlapped with that detected by E-Ig chimera (Figs 6.3C and 6.4C), some tissues or portions of the tissues were stained with the one but not with the other probe. For example the breast Md tissue showed E-selectin ligand activity by HECA-452 mab but not by E-Ig chimera (Figs 6.6A, B, and C). On the other hand, colon CT tissue reveled high positive E-selectin ligand activity by E-Ig chimera, but exhibited weak activity by HECA-452 mab (Figs 6.6D, E, and F). These data show that HECA-452 and E-Ig chimera recognize some unshared antigens.

152 152 LS-174T BT-20 ZR-75-1 Cell count MDA-MB-468 MDA-MB-231 MCF-7 E-Ig chimera LS-174T BT-20 ZR-75-1 Cell count MDA-MB-468 MDA-MB-231 MCF-7 HECA-452 Figure Breast and colon cancer cell lines show various levels of reactivities to murine E-Ig chimera and HECA-452 mab. Colon cancer cell line (LS-174T) or breast cancer cell lines (all others) were surface labeled with murine E-Ig chimera or HECA-452 mab and analyzed by flow cytometry. Dashed curves show isotype, and solid curves shows specific probes.

153 A HECA-452 B E-Ig chimera 153 C Co-localization D Rat IgM and human IgG E % positive tissue area HECA-452 E-Ig Co-localization MAC PC SC N N CT Figure 6.3. Colon cancer tissues were positive for HECA-452 reactivity and E-Ig chimera. (A and B) A MAC tissue from colon carcinoma TMA was dually labeled with HECA- 452 mab (green) and murine E-Ig chimera (red). (C) Co-localization of the two molecules is shown in overlapped image (orange). (D) A MAC tissue from colon carcinoma TMA, which was a matched TMA with a probe stained TMA, was dually labeled with isotype controls. Scale bar indicates 100 µm. (C and D) the blue color is for DAPI. (E) The percentage of tissue area under observation that was stained positively with respect to isotype control signal.

154 154 A HECA-452 B E-Ig chimera C Co-localization D Rat IgM and human IgG E % positive tissue E HECA E-Ig Co-localization LDC MC IDC MdC IDC MdC IDC N IDC NC N MC Figure 6.4. Breast carcinoma tissues were positive for HECA-452 and E-Ig chimera. (A and B) A LDC tissue from breast carcinoma TMA was dually labeled with HECA-452 mab (green) and murine E-Ig chimera (red). (C) Co-localization of the two molecules is shown in the overlapped image (orange). (D) LDC tissue from breast carcinoma TMA, which was a matched TMA with a probe stained TMA, was dually labeled with isotype controls. Scale bar indicates 100 µm. (C and D) the blue color is for DAPI. (E) The percentage of tissue area under observation that was stained positively with respect to isotype control signal.

155 155 A HECA-452 B E-Ig chimera C Co-localization D Rat IgM and human IgG Figure 6.5. The HECA-452 reactivity and E-Ig activity of non-cancerous tissue is restricted to certain areas. Non-cancerous breast tissue was dually labeled with (A) HECA-452 mab (green) and (B) murine E-Ig chimera (red). (C) Co-localization of the two probe signals or (D) a matched non-cancerous tissue stained with respective isotype controls is shown in the overlapped images (orange). Scale bar indicates 100 µm. (C and D) the blue color is for DAPI.

156 156 A HECA-452 B E-Ig chimera C Co-localization D HECA-452 E E-Ig chimera F Co-localization Figure 6.6. Levels of activities of a same tissue detected by HECA-452 mab and murine E-Ig chimera differ. (A, B, and C) Md breast carcinoma or (D, E, and F) CT colon carcinoma tissues were dually labeled with HECA-452 mab (green) and murine E-Ig chimera (red). Colocalization of the two molecules is shown in the overlapped image. Scale bar indicates 100 µm. (C and F) The blue color is for DAPI. 6.5 Discussion In metastasis, cancer cell adhesion to endothelial cells is mediated by endothelial E-selectin and its ligands expressed on the cancer cells. Accumulating evidence suggests the relevance of the E-selectin mediated pathway in variety of cancer types, including colon, prostate, pancreatic, and breast cancers [24-29,35,48]. Importantly, numerous E- selectin ligands are expressed by cancer cells, but identification of each ligand is

157 157 complicated by experimental issues [26-28,47,50,51]. These issues pose major impediments in evaluating the potential of E-selectin ligands for clinical applications. As an alternative strategy, detection of cancer cells or tissues by universal probes that recognizes E-selectin ligand activity, but not necessarily the identity of the ligands, could provide valuable information needed to target selectin ligands. For this purpose, HECA- 452 mab and E-Ig chimera are the two important probes that can be used for finding E- selectin ligand activity [26-28,47,50,51]. Using these probes, we analyzed E-selectin ligand activity of colon and breast cancer cell lines and tissues, and evaluated whether such analysis can distinguish cancer tissues from non-cancer tissues. These data are expected to provide the basis for larger clinical studies for developing diagnostics and prognostics based on E-selectin ligand activity. Multiple colon and breast cancer tissues and cell lines displayed strong E-selectin ligand activity detectable by E-Ig chimera or HECA-452 mab. The level of E-selectin ligand activity of some cancer tissues was several folds higher than that of non-cancer tissues. Notably, the E-selectin ligand activity of non-cancer tissues was mainly restricted to portions of the tissues that are known to express highly glycosylated molecules for physiological function. On the contrary, the staining of cancerous tissues was irregular and dispersed over the tissue. Another important finding from the tissue data is that different levels of E-selectin ligand activities corresponding to various cancer tissues were detectable by HECA-452 mab as well as E-Ig chimera (Figs 6.3 and 6.4). Therefore, E-selectin ligand activity detectable by these probes could be used as an

158 158 indicator to differentiate forms of cancers and to distinguish cancer tissues from noncancer tissues. The cell line and tissue data also revealed other important aspects of detection of E-selectin ligand activity by the two probes. Although E-selectin ligand activity of majority of cells, tissues or portions of tissues is recognized by either of the probes, in some cases the activity was recognizable by E-Ig chimera but not by HECA-452 mab. For example, MDA-MB-468 cells which showed limited HECA-452 reactivity displayed strong E-Ig chimera activity. This is because E-selectin ligand activity of MDA-MB-468 cells is perhaps due to VIM glycans (chapter 2), whose fucose group is located at an internal site and not at terminal end of the glycan. Such structures are not recognized by HECA-452 mab [38]. In support to this notion, it is believed that HECA-452 and E- selectin do not recognize exactly the same epitopes on a ligand [122], and many HECA-452 negative E-selectin ligands have been reported [27,38]. Nevertheless, HECA- 452 positive ligands are thought to be prevalent E-selectin ligands that are functional under pathological and physiological conditions [28,47,60]. In some other cancer tissues or portions of the tissues, E-selectin ligand activity was detectable by HECA-452 but not by E-Ig chimera (Figs 6.6A and B). This is perhaps because E-selectin ligands need glycans to be displayed on appropriate scaffold molecules, absence of which may not render the E-selectin ligand activity. Hence, these probes could recognize a majority of E-selectin ligands on a tissue, yet a minority of ligands could be detected by one but not by the other probe.

159 159 At this point, it is important to note that E-selectin ligand activity found by static assays, such as IHC, provides information whether appropriate glycoconjugates are displayed by a tissue or cells under investigation. However, the mere presence of such moieties may not be sufficient to find ligands that can function under physiologic or pathologic conditions. This is because E-selectin binds to its ligands under the influence of fluid shear exerted by blood or lymph [24,35,69]. Therefore, functional E-selectin ligands should possess appropriate biophysical properties to counter the fluid and other forces introduced by the dynamics of the cell [62,123,124]. To address this issue, selectin ligand activity of cells is routinely tested under hematogenous flow conditions created in vitro parallel plate flow chamber [24-28,35,48,69]. However, no method is available to test E-selectin ligand activity of tissues under physiological flow conditions, which is the subject of the next chapter. In conclusion, the present study showed that E-selectin ligand activity of cancer tissues is detectable by HECA-452 and E-Ig chimera and could be used as an indicator to distinguish cancer and at least some types of non-cancer tissues. Moreover, different levels of E-selectin ligand activities are detectable by these probes and could provide indications to differentiate various forms of cancer. Ultimately, linking E-selectin ligand activity with cancer progression may provide novel diagnostic, prognostic, and therapeutic targets.

160 160 CHAPTER 7: DYNAMIC BIOCHEMICAL TISSUE ANALYSIS Abstract The unique kinetic and tensile properties of E-selectin bonds allow it to mediate cancer cell adhesion to the endothelium and leukocyte recruitment to sites of inflammation. Therefore, we developed a dynamic biochemical tissue analysis (DBTA) that allows exquisite control over the interaction between the probe and the antigen to enhance characterization of molecular recognition in situ. Specifically, we conjugated a recombinant E-selectin construct to polystyrene microspheres and used the resulting microspheres as the probe to investigate E-selectin ligand activity on colon cancer, breast cancer, and normal tissues using DBTA. When E-selectin microspheres were perfused over invasive colon and breast cancer tissue sections at a wall shear stress of 1 dynes/ cm 2, the microspheres exhibited robust adhesion with the tissue. The adhesion of E- selectin microspheres was significantly higher than that of negative control (h-igg conjugated) microspheres, demonstrating the specificity of interaction of E-selectin microspheres with the tissue sections. To demonstrate that modulating easily controlled assay parameters results in discernible changes in adhesion, E-selectin microspheres of 10 and 15 µm diameters were perfused over the colon adenocarcinoma tissue sections at different levels of shear stress, achieved by varying the volumetric flow rate. The results showed that with increasing shear stress or particle diameter, the adhesion of E-selectin 1 To be submitted as: Dynamic Biochemical Tissue analysis. Shirure VS, Malgor R, Resto VA, Goetz DJ, and Burdick MM. Part of the work was previously submitted as an abstract to American Association of Cancer Research Conference 2012.

161 161 microspheres to the tissue sections decreased as predicted. To demonstrate that DBTA can be used for high throughput analysis, slides with multiple colon and breast cancer tissues arranged in a microarray format were used. These microarray slides consisted of tissues from various histopathological classifications. E-selectin microspheres exhibited higher adhesion than the h-igg microspheres to the majority of the cancer tissues, showing the preferential expression of functional E-selectin ligands on cancer tissues. Combined the results (i) convincingly demonstrate the specificity of the DBTA, (ii) reveal the critical role physical factors play in tissue analysis and (iii) demonstrate that DBTA allows facile control over these key physical factors. Thus, we have established a new approach for characterizing tissue that may lead to novel diagnostic and prognostic assays for a large spectrum of pathologies including cancer Introduction Histology, the microscopic analysis of animal and plant tissues and cells, is a central tool for biotechnology, clinical pathology as well as for a variety of diagnostic and prognostic assays. Staining a tissue with hematoxylin and eosin (H & E) allows characterization of the microanatomy of the tissue. More powerful techniques, which broadly can be categorized as biochemical analyses (e. g., immunohistochemistry, IHC; in situ hybridization), identify molecular entities present in a tissue section with a high degree of specificity. Typically in a biochemical tissue analysis (BTA), a section is incubated with a solution containing a probe (e. g., antibody) for a particular entity. The presence or absence of the probe reveals the presence or absence of the entity. While BTA can clearly provide important information, it is insightful to recognize that these

162 162 methods are endpoint assays only. The antigens are detected by utilizing only one property of the probe-antigen bond specifically affinity, and other aspects of the bond remain unexploited (e. g., the kinetics, tensile strength, and reactive compliance of the bond [62, ]). Work in the field of cell adhesion has undeniably shown that the full spectrum of bond properties is germane to biological recognition and communication [29,62, ]. For instance the kinetic and tensile properties of selectin bonds allow this family of adhesion molecules to mediate leukocyte recruitment to sites of inflammation [62,129,130]. The importance of kinetics is not restricted to selectins. Indeed, antibody maturation and selection, at least in some cases, has been ascribed to enhancement of the kinetic on rate [131,132]. These observations led Williams in 1991 to state that, The concept of affinity dominates most thinking about complex biological reactions even though it is relevant only at equilibrium [132]. A similar statement could be made regarding the present state of BTA: Bond affinity underlies all BTA even though a broader spectrum of bond properties are relevant and could be exploited to characterize tissues. An important, pragmatic limitation of traditional BTA is that these assays do not readily allow control over the way in which the probe interacts with the antigen (e. g., control over the force exerted on the probe-antigen bond), and hence with BTA the probe-antigen bond kinetics and other bond properties cannot be easily accessed. A BTA that allows systematic variation of the mode of contact between the probe and antigen will more readily reveal differences between background and specific molecular interactions as well as differences between various probe-antigen bond pairs. The above

163 163 considerations strongly motivate the development of a BTA approach that allows exquisite control over the interaction between the probe and the antigen that fully explores the adhesion space that underlies all tissue analyses. Thus, we applied cell adhesion technology to BTA to arrive at an assay that achieves this goal. The particular assay described here is termed dynamic (as opposed to stationary) biochemical tissue analysis (DBTA). E-selectin and its ligands are ubiquitous in cancer and pathological inflammation (e. g., colon, prostate, pancreatic and breast cancers, and arthritis, atherosclerosis [24,29,39]). The extensive understanding of the biophysics/ biochemistry of E-selectin bonds [62, ], and the fact that the kinetic properties of E-selectin (and selectins in general) give rise to easily discernible relationships between assay conditions (e. g., exerted force) and observed adhesion led us to use E-selectin as a probing molecule in our prototype DBTA. Specifically, we conjugated a recombinant E-selectin construct to polystyrene microspheres and used the resulting E-selectin microspheres as the probe. The E-selectin microspheres were brought into contact with cancer tissue sections and arrays under defined shear stress conditions created by fluid perfusion through a parallel plate flow chamber. The force on the probe-antigen bond was systematically varied by simply altering the size of the E-selectin microspheres or the volumetric flow rate controlled by a high precision syringe pump.

164 Methods Microsphere preparation Polystyrene microspheres 10 or 15 µm diameter (Bangs Laboratories, Inc., Fisher, IN) were washed in Dulbecco s phosphate buffered saline (DPBS; Invitrogen) and incubated with recombinant E-selectin chimera or h-igg at 10 µg/ml to give a final microsphere concentration of 1 million/100 µl. After 1 hr incubation at 4 C, the microspheres were washed and incubated with blocking buffer (1% BSA in DPBS) for 1 hr. The microspheres were re-suspended at 1 million/ml concentration in blocking buffer for experiments Flow cytometry Microspheres were washed with blocking buffer and incubated with FITCconjugated antibodies at a concentration of 10 µg/ml for 30 min at 4 C. Microspheres were washed and analyzed by a FACSAria Special Order Research Product flow cytometer/sorter (BD Biosciences) [122] Tissue slide preparation Formalin fixed paraffin embedded (FFPE) breast or colon carcinoma tissue slides in microarray format (US Biomax, Rockville, MD) or in single tissue format were deparafinized by heating at 60 C for 30 min, and serially incubating with xylene, 95% ethanol, 70% ethanol, and deionized water [28]. The slides then were blocked to reduce non-specific sites with 1% BSA and 1% FBS in DPBS for 1 hour. The blocked tissue slides were used in DBTA.

165 Flow chamber set up The experimental set up for this dynamic adhesion assay consisted of a rectangular parallel plate flow chamber and a silicon rubber gasket (Glycotech, Rockville, MD), which when assembled with tissue slides create a leak-proof flow channel over the tissue area (Fig. 7.1). The laminar flow conditions in the flow channel were controlled by a high precision syringe pump. The flow chamber was placed on Nikon TE300 inverted microscope or a Leica DMI 6000 inverted microscope (Leica Microsystems, Wetzlar, Germany) equipped with video camera and recording system. The microspheres at 1 million cells/ml were perfused over tissue slides at a flow rate corresponding to desired shear stress Immunohistochemistry The tissue slides used in DBTA were washed with 20 mm EDTA-containing buffer to remove microspheres, if any, and the slides were incubated in 1% BSA and 1% FBS in DPBS for 30 min. The slides were stained with HECA-452 or anti CD44 mabs or respective isotype controls (all from BD Biosciences, San Jose, CA) at 5 µg/ml for 1 hr at room temperature. The slides were washed and incubated with AlexaFluor-488- conjugated secondary antibodies at 4 µg/ml for 1 hr and at room temperature. After washing, the tissue slides were mounted in ProLong Gold antifade reagent for microscopy [28]. The tissues were imaged using 10x or 40x objectives under wide field fluorescence using a Leica DMI 6000 inverted microscope. Images were acquired using Simple PCI software (Hamamatsu Corporation, Sewickley, PA), and were subjected to

166 2D blind deconvolution algorithms in AutoQuant X software (Media Cybernetics, Bethesda, MD) to reduce out of focus light [28]. 166 Vacuum IN Vacuum OUT Figure 7.1. Experimental setup used in DBTA. The setup consists of a vacuum sealed rectangular flow chamber (shown in the box) placed on platform of an inverted microscope, which is equipped with video camera connected to a video recording system. Probe microspheres are perfused through the channel at a defined flow rate controlled by a high precision syringe pump.

167 Results Initially, conjugation of E-selectin chimera construct to polystyrene microspheres (E-selectin microspheres) was tested by flow cytometry (Fig. 7.2). The E-selectin microspheres, but not negative control microspheres, were specifically detected by anti-cd62e mab, which recognizes functional sites on E-selectin molecules. These data showed that E-selectin chimera construct is effectively conjugated to polystyrene microspheres by the technique used in the study. When E-selectin microspheres of 15 µm diameter, in buffer containing 2 mm of calcium chloride, were perfused over invasive colon adenocarcinoma tissue section at a wall shear stress of 1 dynes/ cm 2, the microspheres exhibited adhesion with the tissue. The specificity of the interactions of E-selectin microspheres was examined by perfusing two types of negative control microspheres over the same tissue section. The negative controls were, (i) microspheres conjugated with human IgG, which is the Fc portion of the recombinant E-selectin construct (higg microspheres), and in buffer containing 2 mm of calcium chloride, and (ii) E-selectin microspheres in buffer containing 10 mm of EDTA, a divalent cation chelating agent, which is known to diminish E-selectin mediated interactions. The adhesion of E-selectin microspheres was significantly higher than that of negative control microspheres (Fig. 7.3A and B), demonstrating the specificity of interaction of E-selectin microspheres with the tissue sections. The majority of E-selectin microspheres exhibited a rolling type of adhesion (Fig. 7.4), a hallmark of selectin mediated adhesion [62, ]. The average rolling velocity was less than 15 µm/s, which is 100-fold slower than a non-interacting particle. In contrast, the few negative

168 168 control microspheres that did adhere did not exhibit this rolling behavior i. e. they were firmly adherent. Combined, these results clearly demonstrate that when the E-selectin microsphere probes are brought into contact with tissue section by DBTA, the probes interact with the tissue with a high degree of specificity. To demonstrate that controlled modulations in assay parameters results in discernible changes in adhesion, E-selectin microspheres of 10 and 15 µm diameter were perfused over the colon adenocarcinoma tissue sections at different levels of shear stress, which was varied by changing the volumetric flow rate [63]. The results showed that with increasing shear stress or particle diameter the adhesion of E-selectin microspheres to the tissue section decreases (Figs 7.5A and B). To quantitatively demonstrate that the sensitivity of DBTA can be easily modulated over a broad range, we calculated analytical relative sensitivity, which was defined as the ratio of number of microspheres adhering to a tissue under a given assay conditions to the least number of microspheres adhered to the same tissue under any set of the assay parameters studied (15 µm E-selectin microspheres at 1.5 dynes/cm 2 ). Such analysis of the data indicated that the DBTA can be adapted over a very broad range of relative sensitivity (1-50 times; Fig. 7.5C).

169 169 A B C Figure 7.2. E-selectin chimera construct conjugated polystyrene microspheres are specifically recognizable by E-selectin recognizing mab. Microspheres of 10 µm diameter were conjugated with (a) E-selectin chimera construct or (b) higg or (c) buffer were surface labeled with anti-cd62e mab (filled curve) or isotype control (open curve) and analyzed by flow cytometry. A B higg EDTA EIgG Adhering C A A A A Colon cancer Breast cancer

170 170 B higg microspheres E-selectin microspheres in EDTA solution E-selectin microspheres Figure 7.3. E-selectin microspheres specifically adhere to cancer tissues. (A) E-selectin or higg microspheres (15 µm polystyrene particles) or E-selectin microspheres suspended in 10 mm EDTA were perfused over invasive adenocarcinoma tissue of colon and mucinous carcinoma tissue of breast at 1 dynes/cm 2 using a flow chamber. Data are mean ± SE for n = 3, means that do not share a letter are significantly different (One way ANOVA coupled with Tukey s multiple comparison test, P =0.0001, α = 0.05). (B) higg microspheres (left image)/ E-selectin microspheres treated with 10 mm EDTA (middle image)/ E-selectin microspheres (right image) of 15 µm size were perfused over signet carcinoma of colon cancer tissue at 1 dynes/cm 2. Scale bar indicates 100 µm. Figure 7.4. E-selectin microspheres exhibit rolling interactions with carcinoma tissue. E-selectin microspheres were perfused over colon papillary carcinoma tissue in the DBTA. Images for a single particle were captured every 2 seconds and overlaid to create the composite image. The arrow heads show transit of a rolling microsphere at different time points, and the scale bar indicates 100 µm.

171 171 A Adhering particles/mm A B B BC C C 10 μm 15 μm Shear stress (dynes/cm 2 ) B

172 172 C Relative sensitivity Decreasing force on bond Figure 7.5. Controlled modulations in particle size and shear stress result in discernible changes in adhesion. (A) E-selectin microspheres of 10 and 15 µm size were perfused over invasive adenocarcinoma of colon tissue at various shear stresses. Data are mean ± SE for n = 3, means that do not share a letter are significantly different (one way ANOVA coupled with Tukey s multiple comparison test, P =0.0001, α = 0.05). (B) Images acquired for adhesion of 15 (top panels) and 10 µm (bottom panels) E-selectin microspheres for invasive adenocarcinoma of colon cancer tissue for shear stress of 0.5 (right panels) and 1.5 dynes/ cm 2 (left panels). Images for 10 µm particles were pseudo colored for better visibility. Scale bar indicates 100 µm. (C) Relative sensitivity is defined as the ratio of number of adhering particles to a tissue at a given conditions to the minimum number particles adhered to the same tissues (15 µm E-selectin microspheres at 1.5 dynes/cm 2 ). To demonstrate that DBTA can be used for high throughput analysis, slides with multiple colon and breast cancer tissues arranged in a microarray format were used. These microarray slides consisted of tissues from various histopathological classes. E- selectin microspheres, compared to h-igg microspheres, of 15 µm diameter perfused at 1 dynes/cm 2 exhibited higher adhesion to the majority of the cancer tissues (Fig. 7.6A and B). Furthermore, these cancer tissues supported rolling type of interactions of E-selectin

173 173 microspheres. In contrast, the tissues, which are derived from healthy colon or breast, neither showed a significant difference between the adhesion of E-selectin and negative control microspheres nor exhibited rolling type of interactions. Remarkably, E-selectin microspheres exhibited a dramatically higher adhesion to some types of cancer tissues (e. g., mucinous and signet ring carcinoma tissues) than the other types of cancer tissues. The E-selectin ligand activity of mucinous carcinoma tissue could be because of the mucinous proteins that have been shown to possess E-selectin ligand activity [133], but signet ring carcinoma has not been previously reported to possess such a high E-selectin ligand activity (Figs 6A and B). These results demonstrate potential for the immediate application of DBTA in clinical applications, at least for select cancer types. As tissue samples are a limited resource for clinical as well as research purposes, probing the tissues multiple times is advantageous to extract maximum information. However, there are practical limitations to the number of times a tissue can be probed by traditional BTA, such as removing mounting medium, avoiding cross reactivity among probes etc. Probing microspheres adhering to the tissue used in DBTA could be easily removed by washing the tissue with a stripping solution (e. g., EDTA containing buffer for selectin probing) and/ or the application of shear stress. The tissue, free from probing microspheres, can then be utilized for other assays such as IHC or for re-examining with other probes by DBTA. To demonstrate this aspect, IHC was performed with HECA-452 mab or CD44 following DBTA on a tissue. The IHC staining for the tissue (Figs 7.6C, D, and E) with mab was robust compared to the staining with respective isotype control, indicating that the tissues used in DBTA could be reused in other assays.

174 174 A B Adhering particles/mm 2 Adhering particles/mm A E-selectin h IgG B C C C C C C C C Sc Mc Pc Cc Nc A E-selectin h IgG B B B B B B B B B Mb Db Lb Meb Nb

175 175 C D E Figure 7.6. DBTA can be applied to tissue microarrays, and the microarrays used in DBTA can be reused for immunohistochemistry. E-selectin or h-igg microspheres (15 µm) were perfused over (A) colon cancer tissue microarray (signet ring carcinoma, Sc; mucinous adenocarcinoma with necrosis, Mc; papillary adenocarcinoma, Pc; carcinoid tumor, Cc; normal colon tissue, Nc), and (B) breast cancer tissue microarray (mucinous carcinoma, Mb; invasive ductal carcinoma, Db; mixed lobular and duct carcinoma, Lb; medullary carcinoma, Meb; normal breast tissue, Nb) at 1 dynes/cm 2 using a flow chamber. (A and B) data are mean ± SE for n = 3, means that do not share a letter are significantly different (One way ANOVA coupled with Tukey s multiple comparison test, P = for (A) and P = for (B), α = 0.05). After use in DBTA, (C) signet ring carcinoma tissue of colon and (D) normal colon tissue were labeled with HECA-452 or isotype, and (E) invasive adenocarcinoma tissue of colon were labeled with CD44 or isotype, and appropriate AlexaFluor conjugated secondary antibody. (C-E) Left and right panels show specific antibody and isotype labeling, respectively. Scale bar on each image indicates 100 µm.

176 Discussion DBTA is a highly specific method of in situ tissue survey in real time. The strength of the assay is the ease in controlling the assay parameters. The changes in DBTA parameters, such as microsphere size and flow rate (shear stress), in actuality modulates the exerted force on the probe-antigen bond [62,125,130,134]. Intuitively, with increasing force weak interactions disappear decreasing the overall adhesion. Hence these easily controllable assay parameters allow discerning between regions of varying affinities on a tissue/ cell samples under investigation. Another important advantage of DBTA over BTA is that it allows real time analysis which provides additional information such as nature of probe-antigen interactions, which may be critical while analyzing by certain probes. For example, interactions of E-selectin with its ligands are transient but slightly firmer than that of P- selectin [62]. DBTA also provides an easy way for high throughput screening of tissue samples, which is useful in developing clinical applications. An immediate application of DBTA is relating E-selectin activity with cancer progression and pathological grading, as we demonstrated that certain types of cancer tissues show dramatically higher levels of E- selectin ligand activity by DBTA. Also, in the current state of development, DBTA could be extended to analyze L- and P-selectin activity of tissues by conjugating appropriate probes to microspheres. Such studies could ultimately lead to novel diagnostic and prognostic assays for cancers.

177 177 In conclusion, the results of the present study demonstrated that the interactions between probe-antigens are specifically detected in DBTA. Further, the sensitivity of the assay could be modulated by easily alterable assay parameters. Finally, using DBTA, certain types of cancer tissues could be distinguished from non-cancer tissues.

178 178 CHAPTER 8: MATHEMATICAL MODELING OF MICROSPHERE ADHESION TO IMMOBILIZED SUBSTRATE IN PARALLEL PLATE FLOW CHAMBER 8.1 Introduction A parallel plate chamber is used to create physiological flow conditions in the dynamic biochemical tissue analysis (DBTA) as described in the previous chapter. In DBTA, microspheres coated with probe molecules are perfused over a tissue section. These probe microspheres bind to their ligands expressed by the tissue sample. At any given instant, the number of adhering microspheres in a flow chamber not only depends on the kinetics of receptor-ligand bond (association and dissociation rates) but also on the rate of particle transport to the surface of the tissue. In other words, since the tissue section is at the bottom of the flow chamber, only a small number of flowing microspheres contact the tissue, and majority of the particles leave flow chamber without seeing the tissue. If the transport of probe microspheres to the tissue surface is insufficient, some spots on a tissue may not come in contact with probe microspheres, which may result in false negative results. Thus, it is imperative to consider transport of particles in a flow chamber especially for DBTA, as it has been developed for quantitative tissue analysis. One way to approach this problem is by experimentally optimizing the assay parameters. However, such an approach is experimentally laborious and cost ineffective because of the complex nature of the experiments and highly expensive reagents. Alternatively, a mathematical model can enhance the understanding of the system and could greatly reduce the number of experiments needed to optimize the assay.

179 179 As a first step towards developing a comprehensive model for DBTA, a mathematical model for nanosphere adhesion proposed by Haun and Hammer (2008) was adapted [63] for microsphere adhesion to substrate coated with purified adhesion molecules. The model parameters were experimentally determined and system behavior in response to variation in transport properties was studied. More specifically, probe molecules (sle X ) conjugated to microspheres were perfused over the surface of a Petri dish coated with receptor molecules (E-selectin; Fig. 8.1). The E-selectin-sLe X system was chosen because it is previously characterized [64], and interactions between these two molecules are also biologically significant, particularly to cancers and inflammation [24,128]. Ultimately, the model is expected to provide understanding of transport parameters that can affect DBTA results. 0.1 cm E-selectin Free sle X Bound sle X microspheres microspheres Figure 8.1. Schematics of sle X coated microsphere interactions with E-selectin coated substrate in flow chamber.

180 Experimental methods and description of mathematical model Preparation of E-selectin substrate and sle X microspheres for experimentation The E-selectin substrate was prepared by incubating 10 µg/ml of murine E- selectin chimera construct (R & D systems; Minneapolis, MN) on a Petri dish for 1 hr. Subsequently, the dish was blocked with 1% BSA solution in DPBS buffer for 30 min [64]. These substrate plates were assembled in the flow chamber for experiments. The sle X microspheres were prepared by first blocking superavidin coated microspheres (Bangs Laboratories, Inc., Fisher, IN) with 1% BSA solution in DPBS buffer for 30 min. The microspheres were then incubated biotinylated sle X at µg/ml for 1 hr. The microspheres were again blocked with 1% BSA solution in DPBS buffer for 30 min. The resulting sle X coated microspheres were used for experiments [122] Model assumptions Settling velocity of a particle is negligible The other assumptions of the model were stated by Haun and Hammer [63] as follows, Fully developed laminar flow Viscosity of perfusion buffer is equal to water viscosity (buffers are used and not blood) Hydrodynamics of the system is governed by Navier-Stokes equations Particle concentration is uniform across channel width (two dimensional coordinates) Cartesian coordinate system with origin at inlet y-axis parallel to and x-axis perpendicular to the direction of flow Ligands and receptors are uniformly coated on the respective surfaces

181 181 Particle adhesion is governed by first order reversible attachment-detachment kinetics Description of symbols used in mathematical model B = bound particle density (#/m 2 ) C = free particle concentration (#/m 3 ) C w = unbound particle concentration at wall (#/m 3 ) D = diffusivity (m 2 /s) H = height of chnnel (m) K 0 D = kinetic reaction constant for detachment at reference time t ref (s -1 ) K A = kinetic reaction constant for attachment (nm/s) K D = kinetic reaction constant for detachment (s -1 ) L = length of channel (m) P = Peclet s number Q = flow rate (m 3 /s) R p = particle radius (nm) T = temperature (K) t = time (s) t ref = reference time (s) U = Average velocity (m/s) V x = velocity in x direction (m/s) α = constant to determine time dependence of detachment rate constant γ w = shear rate near wall (s -1 ) δ A = Damkohler number for attachment

182 182 δ D = Damkohler number for detachment μ = viscosity (cp) Model equations The model equations in Haun and Hammer model were as follows [63] The kinetics of multivalent particle adhesion is given by (1) It is important to note that K A and K D in the above equation are different from ligandreceptor kinetic rate constants k f and k r. K A and K D describe the particle adhesion, which is a multivalent binding. k D 0 kd () t (2) ( t ) t ref Where, α sets the functional dependence of detachment rate constant on time. For the purpose of calculation, t ref = 1s α = 0.33 was suggested by Haun and Hammer [63]. When above assumptions are applied to two dimensional convective diffusion equations, the concentration profile is given by following equation [63] 2 2 C + C C C x D 2 2 t x x y (3) The flow profile within two parallel plates under fully developed laminar flow is given by following equation [63] x 6 y y v U y y wh H H H H (4) Where shear rate at wall is given by

183 183. 6Q (5) w 2 Hw Particle diffusivity is given by Stokes-Einstein equation as follows kt B D (7) 6 R p Dimensionless parameters [63] ^ C ^ B x y H td C B C C H L H L H UH 6 H Rp K AH KDH A D D KBT D D P 2 (8) Incorporating equations 4, 5, 6, 7 into equation 3 and normalizing the obtained equation using equation 8 following equation can be obtained [63]. ^ ^ ^ ^ P 2 2 C C C C (9) Normalizing equation 1 gives ^ B ^ ^ (, ) AC(,, 0) D B(, ) (10) Above two equations were solved using following boundary conditions obtained [63]. At inlet starting concentration is starting concentration ^ C, 0, 1 (11)

184 184 Flux at out let is zero ^ C (, 1, ) 0 (12) Top surface is unreactive ^ C (,, 1) 0 (13) Surface adhesion reaction governs the bottom surface by kinetics in equation 1 ^ C ^ ^ (,, 0) AC(,, 0) D B(, ) (14) Bound particle density is zero everywhere, except at the bottom wall hence ^ (, ) 0 B (15) Solution technique for mathematical equations Unsteady state two dimensional partial differential equations 9 and 10 subjected to boundary conditions (Fig. 8.2) were solved using the Band (j) subroutine in FORTRAN. Implicit finite difference method (FDM) was used for time difference, three point forward or backward approximations were applied for spatial boundary conditions, and central difference approximation was used for all central nodes. The values of various constants used in the program are given Appendix 1. Jacobians were manually found and entered into the code (Appendix 2). Solutions for each time intervals converged in less than 20 iterations. Solutions obtained were stable for a wide range of initial guesses attempted (C = to 0.1).

185 Results Experimental results of adhesion of sle X microspheres The sle X microspheres suspended at a concentration of 0.5 million/ml in 1% BSA solution in DPBS buffer were perfused at a shear stress 1.5 dyne/cm 2 for 1, 2, 3, or 4 min and the number of adhering particles were counted at the end of the time interval. As expected, with increasing time of perfusion, the number of adhering particles increased (Fig. 8.3) Experimental determination of K D In all of the experiments performed, the sle X microspheres firmly attached to E- selectin substrate with almost no detachment. Hence, K D was set to a minimum arbitrary value of 4.5 x s -1, which is equivalent to detachment Damkohler number of Validation of solution technique using the published data Haun and Hammer solved the system of equations using Comsol Multiphysics finite element method (FEM) [63]. In this work, FDM was programmed using Band (j) subroutine in FORTRAN. Thus, the new solution technique needed to be validated using the previous data from [63]. For this purpose, various results from the original publication were reproduced, which included bound particle density and free particle concentration profiles near wall, along channel length. The FORTRAN results matched the previously published data (Refer Appendix 3 for comparison of results), showing that the technique used in this study gives expected results.

186 Top (No binding) 186 Y X Inlet (Concentration of microspheres is constant) Bottom (Binding reaction) Outlet (Flux of microspheres is zero) Figure 8.2. Schematic diagram representing boundary conditions and FDM grids. The flow chmber height (Y direction) and length (X direction) are divided by uniform numerical grids. The boundary conditions are mentioned in the red boxes. Adhering microspheres (#) Perfusion time (min) Figure 8.3. sle X microspheres were perfused over E-selectin substrate at a shear stress of 1.5 dyne/cm 2.

187 Determination of K A The K A value was found by minimizing the root mean square error between the number of adhering particles predicted by model and found by experiments. For K A = 3.53 x 10-6 cm/s (attachment Damkohler number = 310) the root mean square was and the normalized root mean square error was 4.1%. The experimental and predicted data matched well at this value of K A as shown in Fig B (normalized bound particles) < Experimental Predicted by model τ (dimensionless time) Figure 8.4. Comparison of normalized bound particles predicted by the model and found by experiments Effect of Peclet s number Since Peclet s number is related to particle transport, its effect on free particle concentration near wall and the number of bound particles was found by simulating the model (Fig. 8.5). As higher Peclet s number indicates better mass transport, with

188 increasing Peclet s the free particle concentration near wall increased which resulted into increase in number of bound particles Normalized free particle concentration near wall P = P = P = P = Dimension less flow chamber length B (normalized bound particles) < E E E E+08 Peclet' number (P) Figure 8.5. Effect of Peclet s number (P) on particle profile in the flow chamber. Free particle concentration near the surface coated with E-selectin is shown in the top panel, and bound particles to the surface is shown in the bottom panel. The data was generated by simulating the model with following values of parameters, time = 2 min, K A = 3.53 x 10-6 cm/s, and K D = 4.5 x s -1.

189 Discussion A previously published mathematical model for adhesion of particles in the flow chamber [63] is adapted, which could be extended for DBTA in future. Specifically, adhesion of sle X coated microspheres to E-selectin substrate was studied. The mathematical model is composed of two parts, one accounting for transport properties of the system and the other describing receptor-ligand interactions. The transport of the particle to the surface E-selectin substrate is accounted by the Peclet s number (P), which depends on applied wall shear stress and particle diffusivity. The variations in P resulted in variation in free particle concentration near wall and particle binding to E-selectin (Fig. 8.5), demonstrating the effect of particle transport on particle adhesion in flow chamber. The other part of the model accounts for particle-substrate interactions. It is important to note that at a given instant interaction between multiple receptors and ligands, not a single receptor-ligand interaction, lead to particle adhesion. Thus, K A and K D in this model are for adhesion and dissociation of particles and are not same as association and dissociation constants (k f and k r, respectively) for single receptor-ligand kinetics. Previously, a theoretical approach was used to determine the values of these constants [63], but in the present work a semi-empirical approach was used. This approach provided estimates of the values of the constants, which could be used to generate concentration/ bound particle profiles for various conditions in the flow chamber.

190 190 It is important to note that certain factors that are not considered in the present work affect values of K A and K D. Particularly, receptor-ligand density greatly affects the values of these constants, as suggested by Haun and Hammer [63]. Other aspects that affect the values of mainly K D are size of microspheres and wall shear stress. With increasing value of either of these parameters, K D decreases as the force experienced by the particle increases [62,63]. The recommended approach to account for these issues is to generate mathematical correlations for values K A and K D that include various factors in a pragmatic range of values. Additional factors that account for the tissue surface, which is different from purified substrates, need to be incorporated in to the model to extend it to DBTA. Since the tissue surface is presumed to be highly irregular, especially compared to that of a Petri dish, the bottom boundary conditions in the model have to be appropriately altered. Further, it is necessary to account for surface saturation as the rolling microspheres and firmly attached microspheres that occupy sites for a time period during which other particles cannot interact with that tissue spot. Moreover, the settling velocity needs to be considered, which could be accounted by adding a factor of velocity in y direction in equation (3). With these adaptations in the current form of the model, an advanced model that can quantify adhesion of microspheres in DBTA could be developed.

191 191 CHAPTER 9: CONCLUSION AND RECOMMENDATION FOR FUTURE WORK The present work is focused on molecular mediators of breast cancer cell adhesion to endothelium under physiological flow conditions, which is believed to lead to distant metastasis. The specific aim 1 of the present study was to identify the E-selectin ligands expressed by breast cancer cells. It was found that gangliosides, sialylated glycolipids, expressed by BT-20 and MDA-MB-468 breast cancer cell lines are E- selectin ligands (Chapter 2). These glycolipids were major participants in breast cancer cell tethering and cooperate with glycoproteins to mediate rolling and firm adhesion. Given the importance of glycoproteins in E-selectin mediated adhesion [24,26,47,50,51,101], the identification of glycoprotein ligands was next sought (Chapter 3). For this purpose, ZR-75-1 breast cancer cells which specifically adhere to E-selectin under physiological flow conditions were analyzed. The ZR-75-1 cells expressed Mac- 2BP as a novel high efficiency E-selectin ligand. The expression of Mac-2BP was also found on several other breast cancer cell lines including BT-20 and MDA-MB-468 cells, indicating it may serve as E-selectin ligand on variety of breast cancer cells. To explain the poor prognosis of breast cancer patients with Mac-2BP overexpressing tumors, a model was postulated (Chapter 4). According to this model, expression of Mac-2BP is critically regulated in cancer cells, and an optimal level of expression of Mac-2BP is required for evading the immune response and to colonize distant tissue. A dual role of Mac-2BP as E-selectin and Galectin-1 (Gal-1) ligand to perform these two functions was proposed in the model. Supporting this idea, preliminary data obtained indicated that Mac-2BP possess Gal-1 ligand activity, as ZR-75-1 cells

192 192 possess Gal-1 ligand activity detectable by chimeric Gal-1 construct, but Mac-2BP silenced cells were negative for Gal-1 ligand activity (Fig. 9.1). Further studies are needed to confirm the role of Mac-2BP as a Gal-1 ligand and how Mac-2BP-Gal-1 ligation mediates tumor cell evasion from the immune response. It is recommended to investigate whether E-selectin ligand activity is cross regulated by Gal-1. These studies will strongly implicate Mac-2BP adhesion pathways in breast cancer metastasis, and provide a novel perspective on cooperation of lectin family of molecules in breast cancer metastasis. Wild type Mac-2BP silenced Figure 9.1. Mac-2BP silencing of ZR-75-1 cells reduces Gal-1 ligand activity. Wild type or Mac-2BP silenced ZR-75-1 cells were surface labeled with chimeric form of Gal-1 construct or a human Ig negative control and analyzed by flow cytometry. Blue curve shows isotype, and red curve shows specific probe. The data is from n=1 experiment. In addition to gangliosides and Mac-2BP, breast cancer cells also expressed HCELL (Chapter 4), which is the E-selectin reactive glycoform of CD44. The HCELL on

193 193 BT-20 cells are mainly glycoforms of CD44v, particularly CD44v3 and v4/5 isoforms, rather than CD44s. Further, it was found that the HCELL expressed by BT-20 breast cancer cells is a HECA-452-negative ligand. The identity of breast cancer cell HCELL glycans needs to be established in the future, as this information could potentially attribute glycan specific form of HCELL to breast cancer. Altogether our study demonstrates that gangliosides, Mac-2BP, and CD44v expressed by breast cancer cells are E-selectin ligands functional under physiological flow conditions. In the future, it is recommended to study the concurrent roles of the three ligands on E-selectin mediated adhesion. Such a study could be performed by simultaneously silencing expression of these molecules in BT-20 cells, as this cell line expressed all of these E-selectin ligands. Investigation of the relevance of these molecules in animal models will reveal prognostic and therapeutic potential of E-selectin ligands expressed by breast cancer cells. In the studies performed for specific aim 1 it was also found that HCELLv is abundantly expressed by breast cancer cells in the epithelial-like state but is minimally expressed by cells in the mesenchymal state. This result motivated studies to find whether E-selectin ligand expression is related to epithelial to mesenchymal transition (EMT) and breast cancer stem-like cells (BCSCs) in specific aim 2 (Chapter 5). The breast cancer cells lines were categorized as BCSCs expressing the putative stem cell markers CD44 + /CD24 -/low or non-bcscs possessing alternate expression of these markers. Further, the BCSCs were in mesenchymal-like state and non-bcscs were in epitheliallike state. Interestingly, BCSCs as well as non-bcscs expressed positive E-selectin

194 194 ligand activities, yet non-bcscs expressed higher levels of the E-selectin ligand activity than that expressed by BCSCs. Also, the results indicate that the E-selectin ligand activity may be regulated by EMT, by which non-bcscs acquire BCSC properties. A hypothetical model connecting E-selectin ligand expression, BCSC, and EMT was proposed (Chapter 5). Also, it was hypothesized that mesenchymal to epithelial transition (MET) may have connection to selectin ligand creation (Chapter 5). Further efforts are required to validate these models. Particularly, complete EMT of non-bcscs, rather than the partial EMT achieved in the present work, is perhaps necessary to convincingly reveal the association of E-selectin mediated adhesion and EMT. For this purpose, a combinatorial treatment of cytokines could be used to induce EMT instead of single cytokine treatment used in the present study. Such combinatorial treatment is likely to be closer to the metastatic niche, where a variety of cytokines interacts with cells [5]. For the studying effect of MET on selectin ligand activity the mesenchymal-like cell line MDA-MB-231 could be ideal cell line, as this cell line has been characterized for cell adhesion properties (Chapter 4). Mechanistic association of EMT and MET with selectin mediated adhesion could be a seminal finding proving the criticality of selectin mediated adhesion in cancer metastasis. The cell line data presented in specific aim 1 and 2 indicated a prominent role of E-selectin ligands in breast cancer metastasis. The specific aim 3 was to find the relevance of E-selectin ligand activity in cancer tissue samples (Chapter 6). For this purpose, first the E-selectin ligand activity of breast and colon cancer tissues was analyzed by HECA-452 mab and E-Ig chimera. E-selectin ligand activity of cancer

195 195 tissues was detectable by HECA-452 and E-Ig chimera and could be used as an indicator to distinguish at least some cancer tissues from non-cancer tissues. E-selectin ligand activity found by static assays in Chapter 6 provides information as to whether appropriate glycoconjugates are displayed by a tissue sections or cells under investigation. However, the mere presence of such moieties may not be sufficient to find ligands that can function under physiologic or pathologic conditions. This is because E-selectin binds to its ligands under the influence of fluid shear exerted by blood [24,35,69]. A method, named dynamic biochemical tissue analysis (DBTA), that surveys tissues for selectin activity under well-defined flow conditions is presented in Chapter 7. It was demonstrated that E-selectin microspheres specifically adhere to tissues in a shear stress and particle size dependent manner. Also, it was shown that the method can be applied to tissue microarrays. DBTA provides an easy and novel way for high throughput screening of tissue samples, which is useful in developing clinical applications. An immediate scope for application of DBTA is relating E-selectin activity with cancer progression and pathological grading, as we demonstrated that certain types of cancer tissues show dramatically higher levels of E-selectin ligand activity by DBTA. Additional data from the study indicate that L- and P-selectin may also be relevant in breast cancer (Chapter 5). DBTA could be extended to analyze L- and P-selectin activity of tissues by conjugating appropriate probes to microspheres. Such studies could ultimately lead to novel diagnostic and prognostic assays for cancers.

196 196 REFERENCES 1. (2010) Cancer Facts and Figures Atlanta, GA: American Cancer Society. 2. Fidler IJ (2001) Seed and soil revisited: contribution of the organ microenvironment to cancer metastasis. Surgical Oncology Clinics of North America 10: , viiviiii. 3. Fidler IJ (2002) The organ microenvironment and cancer metastasis. Differentiation 70: Fidler IJ (2003) The pathogenesis of cancer metastasis: the 'seed and soil' hypothesis revisited. Nat Rev Cancer 3: Weinberg RA (2007) The biology of cancer. Garland Science, Taylor & Francis Group, LLC. pp Poste G, Fidler IJ (1980) The pathogenesis of cancer metastasis. Nature 283: Stetler-Stevenson WG, Liotta LA, Kleiner DE, Jr. (1993) Extracellular matrix 6: role of matrix metalloproteinases in tumor invasion and metastasis. FASEB Journal 7: Baccelli I, Trumpp A (2012) The evolving concept of cancer and metastasis stem cells. Journal of Cell Biology 198: Kai K, Arima Y, Kamiya T, Saya H (2010) Breast cancer stem cells. Breast Cancer 17: Odoux C, Fohrer H, Hoppo T, Guzik L, Stolz DB, et al. (2008) A stochastic model for cancer stem cell origin in metastatic colon cancer. Cancer Research 68: Al-Hajj M, Clarke MF (2004) Self-renewal and solid tumor stem cells. Oncogene 23: Al-Hajj M, Wicha M, Benito-Hernandez A, Morrison S, Clarke M (2003) Prospective identification of tumorigenic breast cancer cells. Proceedings of the National Academy of Sciences of the United States of America 100: Bonnet D, Dick JE (1997) Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nature Medicine 3: Lapidot T, Sirard C, Vormoor J, Murdoch B, Hoang T, et al. (1994) A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature 367: O'Brien CA, Pollett A, Gallinger S, Dick JE (2007) A human colon cancer cell capable of initiating tumour growth in immunodeficient mice. Nature 445: Sheridan C, Kishimoto H, Fuchs RK, Mehrotra S, Bhat-Nakshatri P, et al. (2006) CD44+/CD24- breast cancer cells exhibit enhanced invasive properties: an early step necessary for metastasis. Breast Cancer Res 8: R Abraham BK, Fritz P, McClellan M, Hauptvogel P, Athelogou M, et al. (2005) Prevalence of CD44+/CD24-/low cells in breast cancer may not be associated with clinical outcome but may favor distant metastasis. Clinical Cancer Research 11:

197 Liu S, Clouthier SG, Wicha MS (2012) Role of micrornas in the regulation of breast cancer stem cells. Journal of Mammary Gland Biology and Neoplasia 17: Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, et al. (2008) The epithelialmesenchymal transition generates cells with properties of stem cells. Cell 133: Paget S (1989) The distribution of secondary growths in cancer of the breast Cancer and Metastasis Reviews 8: Weiss L (1980) Cancer cell traffic from the lungs to the liver: an example of metastatic inefficiency. International Journal of Cancer 25: Hart IR, Fidler IJ (1980) Role of organ selectivity in the determination of metastatic patterns of B16 melanoma. Cancer Research 40: Auerbach R, Lu WC, Pardon E, Gumkowski F, Kaminska G, et al. (1987) Specificity of adhesion between murine tumor cells and capillary endothelium: an in vitro correlate of preferential metastasis in vivo. Cancer Research 47: Barthel SR, Gavino JD, Descheny L, Dimitroff CJ (2007) Targeting selectins and selectin ligands in inflammation and cancer. Expert Opinion on Therapeutic Targets 11: Burdick MM, Bochner BS, Collins BE, Schnaar RL, Konstantopoulos K (2001) Glycolipids support E-selectin-specific strong cell tethering under flow. Biochemical and Biophysical Research Communications 284: Burdick MM, Chu JT, Godar S, Sackstein R (2006) HCELL is the major E-and L- selectin ligand expressed on LS174T colon carcinoma cells. Journal of Biological Chemistry 281: Shirure VS, Henson KA, Schnaar RL, Nimrichter L, Burdick MM (2011) Gangliosides expressed on breast cancer cells are E-selectin ligands. Biochemical and Biophysical Research Communications 406: Shirure VS, Reynolds NM, Burdick MM (2012) Mac-2 binding protein is a novel E- selectin ligand expressed by breast cancer cells. PLoS One 7: e Krause T, Turner GA (1999) Are selectins involved in metastasis? Clin Exp Metastasis 17: Mannori G, Santoro D, Carter L, Corless C, Nelson RM, et al. (1997) Inhibition of colon carcinoma cell lung colony formation by a soluble form of E-selectin. American Journal of Pathology 151: Schweitzer KM, Drager AM, Van der Valk P, Thijsen SF, Zevenbergen A, et al. (1996) Constitutive expression of E-selectin and vascular cell adhesion molecule- 1 on endothelial cells of hematopoietic tissues. American Journal of Pathology 148: Sackstein R (2004) The bone marrow is akin to skin: HCELL and the biology of hematopoietic stem cell homing. Journal of Investigative Dermatology Symposium Proceedings 9: Ley K, Laudanna C, Cybulsky MI, Nourshargh S (2007) Getting to the site of inflammation: the leukocyte adhesion cascade updated. Nat Rev Immunol 7:

198 Sperandio M (2006) Selectins and glycosyltransferases in leukocyte rolling in vivo. Febs Journal 273: Giavazzi R, Foppolo M, Dossi R, Remuzzi A (1993) Rolling and adhesion of human tumor cells on vascular endothelium under physiological flow conditions. Journal of Clinical Investigation 92: Phillips M, Nudelman E, Gaeta F, Perez M, Singhal A, et al. (1990) ELAM-1 mediates cell adhesion by recognition of a carbohydrate ligand, sialyl-lex. Science 250: Takada A, Ohmori K, Takahashi N, Tsuyuoka K, Yago A, et al. (1991) Adhesion of human cancer cells to vascular endothelium mediated by a carbohydrate antigen, sialyl Lewis A. Biochemical and biophysical research communications(print) 179: Kannagi R, Hakomori S (2001) A guide to monoclonal antibodies directed to glycotopes. Advances in Experimental Medicine and Biology 491: Dube DH, Bertozzi CR (2005) Glycans in cancer and inflammation potential for therapeutics and diagnostics. Nature Reviews Drug Discovery 4: Renkonen J, Paavonen T, Renkonen R (1997) Endothelial and epithelial expression of sialyl Lewis x and sialyl Lewis a in lesions of breast carcinoma. International Journal of Cancer 74: Ellies LG, Sperandio M, Underhill GH, Yousif J, Smith M, et al. (2002) Sialyltransferase specificity in selectin ligand formation. Blood 100: St Hill CA, Bullard KM, Walcheck B (2005) Expression of the high-affinity selectin glycan ligand C2-O-sLeX by colon carcinoma cells. Cancer Lett 217: St Hill CA, Farooqui M, Mitcheltree G, Gulbahce HE, Jessurun J, et al. (2009) The high affinity selectin glycan ligand C2-O-sLex and mrna transcripts of the core 2 beta-1,6-n-acetylglucosaminyltransferase (C2GnT1) gene are highly expressed in human colorectal adenocarcinomas. BMC cancer 9: Brockhausen I, Schachter H, Stanley P (2009) O-GalNAc Glycans. In: Varki A, Cummings RD, Esko JD, Freeze HH, Stanley P et al., editors. Essentials of Glycobiology. 2nd ed. Cold Spring Harbor (NY). 45. Seeberger PH, Finney N, Rabuka D, Bertozzi CR (2009) Chemical and Enzymatic Synthesis of Glycans and Glycoconjugates. In: Varki A, Cummings RD, Esko JD, Freeze HH, Stanley P et al., editors. Essentials of Glycobiology. 2nd ed. Cold Spring Harbor (NY). 46. Sackstein R, Dimitroff CJ (2000) A hematopoietic cell L-selectin ligand that is distinct from PSGL-1 and displays N-glycan-dependent binding activity. Blood 96: Hanley WD, Burdick MM, Konstantopoulos K, Sackstein R (2005) CD44 on LS174T colon carcinoma cells possesses E-selectin ligand activity. Cancer research 65: Burdick MM, McCaffery JM, Kim YS, Bochner BS, Konstantopoulos K (2003) Colon carcinoma cell glycolipids, integrins, and other glycoproteins mediate adhesion to HUVECs under flow. American Journal of Physiology: Cell Physiology 284:

199 Matsumoto M, Atarashi K, Umemoto E, Furukawa Y, Shigeta A, et al. (2005) CD43 Functions as a Ligand for E-Selectin on Activated T Cells 1. The Journal of Immunology 175: Thomas SN, Schnaar RL, Konstantopoulos K (2009) Podocalyxin-like protein is an E-/L-selectin ligand on colon carcinoma cells: comparative biochemical properties of selectin ligands in host and tumor cells. American Journal of Physiology: Cell Physiology 296: C505-C Thomas SN, Zhu F, Schnaar RL, Alves CS, Konstantopoulos K (2008) Carcinoembryonic antigen and CD44 variant isoforms cooperate to mediate colon carcinoma cell adhesion to E-and L-selectin in shear flow. Journal of Biological Chemistry 283: Wilkins P, McEver R, Cummings R (1996) Structures of the O-glycans on P-selectin glycoprotein ligand-1 from HL-60 cells. Journal of Biological Chemistry 271: Hanley WD, Napier SL, Burdick MM, Schnaar RL, Sackstein R, et al. (2006) Variant isoforms of CD44 are P- and L-selectin ligands on colon carcinoma cells. FASEB Journal 20: Rosen SD (2004) Ligands for L-selectin: homing, inflammation, and beyond. Annual Review of Immunology 22: Dohi T, Nemoto T, Ohta S, Shitara K, Hanai N, et al. (1993) Different binding properties of three monoclonal antibodies to sialyl Le(x) glycolipids in a gastric cancer cell line and normal stomach tissue. Anticancer Research 13: Fukushi Y, Nudelman E, Levery SB, Hakomori S, Rauvala H (1984) Novel fucolipids accumulating in human adenocarcinoma. III. A hybridoma antibody (FH6) defining a human cancer-associated difucoganglioside (VI3NeuAcV3III3Fuc2nLc6). Journal of Biological Chemistry 259: Kniep B, Peter-Katalinic J, Muthing J, Majdic O, Pickl WF, et al. (1996) The CDw65 monoclonal antibodies VIM-8 and VIM-11 bind to the neutral glycolipid V3FucnLc8Cer. J Biochem 119: Macher BA, Buehler J, Scudder P, Knapp W, Feizi T (1988) A novel carbohydrate, differentiation antigen on fucogangliosides of human myeloid cells recognized by monoclonal antibody VIM-2. Journal of Biological Chemistry 263: Hanai N, Shitara K, Furuya A, Yoshida H, Dohi T, et al. (1990) Detailed characterization of reactivities of anti-gastric cancer monoclonal antibodies to carbohydrate antigen. Anticancer Research 10: Dimitroff C, Lee J, Rafii S, Fuhlbrigge R, Sackstein R (2001) CD44 is a major E- selectin ligand on human hematopoietic progenitor cells. Journal of Cell Biology 153: Ni Z, Walcheck B (2007) Varied levels of reactivity by different E-selectin/Fc constructs with cutaneous lymphocyte-associated antigen (CLA)(+) CD4(+) T cells. Immunol Lett 108: Chang K, Tees D, Hammer D (2000) The state diagram for cell adhesion under flow: leukocyte rolling and firm adhesion. Proc Natl Acad Sci U S A 97:

200 63. Haun J, Hammer D (2008) Quantifying nanoparticle adhesion mediated by specific molecular interactions. Langmuir: the ACS journal of surfaces and colloids 24: Brunk D, Goetz D, Hammer D (1996) Sialyl Lewis (x)/e-selectin-mediated rolling in a cell-free system. Biophysical Journal 71: Moss MA, Zimmer S, Anderson KW (2000) Role of metastatic potential in the adhesion of human breast cancer cells to endothelial monolayers. Anticancer Research 20: Tozeren A, Kleinman HK, Grant DS, Morales D, Mercurio AM, et al. (1995) E- Selectin mediated dynamic interactions of breast cancer and colon cancer cells with endothelial cell monolayers. International Journal of Cancer 60: Nimrichter L, Burdick MM, Aoki K, Laroy W, Fierro MA, et al. (2008) E-selectin receptors on human leukocytes. Blood 112: Handa K, Stroud MR, Hakomori S (1997) Sialosyl-fucosyl poly-lacnac without the sialosyl-le x epitope as the physiological myeloid cell ligand in E-Selectin dependent adhesion: studies under static and dynamic flow conditions. Biochemistry 36: Dimitroff CJ, Lechpammer M, Long-Woodward D, Kutok JL (2004) Rolling of human bone-metastatic prostate tumor cells on human bone marrow endothelium under shear flow is mediated by E-selectin. Cancer research 64: Guan F, Handa K, Hakomori S (2009) Specific glycosphingolipids mediate epithelialto-mesenchymal transition of human and mouse epithelial cell lines. Proceedings of the National Academy of Sciences, USA 106: Sutherland RL, Watts CKW, Lee CSL, Musgrove EA (1999) Breast cancer. In: Masters J, and Palsson B, editor. Human cell cultures volume II: Cancer cell lines part 2: Kluwer academic publishers. 72. Schnaar RL (1994) Isolation of glycosphingolipids. In: Lennarz W, Hart G, editors. Methods in enzymology: guide to techniques in glycobiology Vol 230 Academic Press, San Diego. pp Schnaar RL, Needham LK (1994) Thin layer chromatography of glycosphingolipids. In: Lennarz W, Hart G, editors. Methods in enzymology: guide to techniques in glycobiology Vol 230: Academic Press, San Diego. pp Mazo IB, Gutierrez-Ramos JC, Frenette PS, Hynes RO, Wagner DD, et al. (1998) Hematopoietic progenitor cell rolling in bone marrow microvessels: parallel contributions by endothelial selectins and vascular cell adhesion molecule 1. Journal of Experimental Medical Sciences 188: Dimitroff CJ, Descheny L, Trujillo N, Kim R, Nguyen V, et al. (2005) Identification of leukocyte E-selectin ligands, P-selectin glycoprotein ligand-1 and E-selectin ligand-1, on human metastatic prostate tumor cells. Cancer Research 65: Wagers AJ, Stoolman LM, Kannagi R, Craig R, Kansas GS (1997) Expression of leukocyte fucosyltransferases regulates binding to E-selectin: relationship to previously implicated carbohydrate epitopes. The Journal of Immunology 159:

201 Hidalgo A, Peired AJ, Wild MK, Vestweber D, Frenette PS (2007) Complete identification of E-selectin ligands on neutrophils reveals distinct functions of PSGL-1, ESL-1, and CD44. Immunity 26: Zen K, Liu DQ, Guo YL, Wang C, Shan J, et al. (2008) CD44v4 is a major E-selectin ligand that mediates breast cancer cell transendothelial migration. PLoS ONE 3: e Costa S, Terzano P, Bovicelli A, Martoni A, Angelelli B, et al. (2001) CD44 isoform 6 (CD44v6) is a prognostic indicator of the response to neoadjuvant chemotherapy in cervical carcinoma. Gynecologic Oncology 80: Barthel SR, Wiese GK, Cho J, Opperman MJ, Hays DL, et al. (2009) Alpha 1,3 fucosyltransferases are master regulators of prostate cancer cell trafficking. Proceedings of the National Academy of Sciences of the United States of America 106: Konstantopoulos K, Thomas SN (2011) Hematogenous metastasis: roles of CD44v and alternative sialofucosylated selectin ligands. Advances in Experimental Medicine and Biology 705: Yago T, Shao B, Miner JJ, Yao L, Klopocki AG, et al. (2010) E-selectin engages PSGL-1 and CD44 through a common signaling pathway to induce integrin alphalbeta2-mediated slow leukocyte rolling. Blood 116: Viswaroop B, B A, Gopalakrishnan G (2005) Evaluating erectile dysfunction: oral sildenafil versus intracavernosal injection of papaverine. National Medical Journal of India 18: Capobianco ML, De Champdore M, Arcamone F, Garbesi A, Guianvarc'h D, et al. (2005) Improved synthesis of daunomycin conjugates with triplex-forming oligonucleotides. The polypurine tract of HIV-1 as a target. Bioorganic and Medicinal Chemistry 13: Yin JJ, Mohammad KS, Kakonen SM, Harris S, Wu-Wong JR, et al. (2003) A causal role for endothelin-1 in the pathogenesis of osteoblastic bone metastases. Proceedings of the National Academy of Sciences of the United States of America 100: Narita T, Kawakami-Kimura N, Matsuura N, Funahashi H, Kannagi R (1996) Adhesion of human breast cancer cells to vascular endothelium mediated by Sialyl Lewis x/e-selectin. Breast Cancer 3: Frenette PS, Mayadas TN, Rayburn H, Hynes RO, Wagner DD (1996) Susceptibility to infection and altered hematopoiesis in mice deficient in both P- and E-selectins. Cell 84: Ratliff TL (2005) CD44 potentiates the adherence of metastatic prostate and breast cancer cells to bone marrow endothelial cells. Journal of Urology 173: Sasaki T, Brakebusch C, Engel J, Timpl R (1998) Mac-2 binding protein is a celladhesive protein of the extracellular matrix which self-assembles into ring-like structures and binds beta 1 integrins, collagens and fibronectin. The EMBO Journal 17:

202 Grassadonia A, Tinari N, Iurisci I, Piccolo E, Cumashi A, et al. (2002) 90K (Mac-2 BP) and galectins in tumor progression and metastasis. Glycoconjugate journal 19: Inohara H, Akahani S, Koths K, Raz A (1996) Interactions between galectin-3 and Mac-2-binding protein mediate cell-cell adhesion. Cancer research 56: Tinari N, Lattanzio R, Querzoli P, Natoli C, Grassadonia A, et al. (2009) High expression of 90K (Mac-2 BP) is associated with poor survival in node-negative breast cancer patients not receiving adjuvant systemic therapies. International Journal of Cancer 124: Weiss L, Haydock K, Pickren JW, Lane WW (1980) Organ vascularity and metastatic frequency. American Journal of Pathology 101: Phillips TE, Stanley CM, Wilson J (1993) The effect of 16,16-dimethyl prostaglandin E2 on proliferation of an intestinal goblet cell line and its synthesis and secretion of mucin glycoproteins. Prostaglandins Leukotrienes and Essential Fatty Acids 48: Zou X, Patil S, Vivek R, Dagia N, Smith L, et al. (2005) PSGL-1 derived from human neutrophils is a high-efficiency ligand for endothelium-expressed E-selectin under flow. American Journal of Physiology: Cell Physiology 289: C Sackstein R (2004) The bone marrow is akin to skin: HCELL and the biology of hematopoietic stem cell homing. Journal of Investigative Dermatology 122: Kim E, Schueller O, Sweetnam PM (2012) Targeting the leukocyte activation cascade: getting to the site of inflammation using microfabricated assays. Lab Chip 12: Stokes GN, Shelton JB, Jr., Zahn CM, Kendall BS (2002) Association of CD44 isoform immunohistochemical expression with myometrial and vascular invasion in endometrioid endometrial carcinoma. Gynecologic Oncology 84: Peyronneau MA, Renaud JP, Jaouen M, Urban P, Cullin C, et al. (1993) Expression in yeast of three allelic cdnas coding for human liver P-450 3A4. Different stabilities, binding properties and catalytic activities of the yeast-produced enzymes. European Journal of Biochemistry 218: Wang L, Shirure VS, Burdick MM, Wu S (2011) UVB-irradiation regulates VLA-4- mediated melanoma cell adhesion to endothelial VCAM-1 under flow conditions. Molecular Carcinogenesis 50: Draffin JE, McFarlane S, Hill A, Johnston PG, Waugh DJ (2004) CD44 potentiates the adherence of metastatic prostate and breast cancer cells to bone marrow endothelial cells. Cancer Research 64: Feyertag J, Haberhauer G, Skoumal M, Kittl EM, Bauer K, et al. (2000) [Serum soluble CD44 isoform variant 5 level in patients with seropositive rheumatoid arthritis treated with cyclosporin A]. Acta Medica Austriaca 27: Franzmann EJ, Weed DT, Civantos FJ, Goodwin WJ, Bourguignon LY (2001) A novel CD44 v3 isoform is involved in head and neck squamous cell carcinoma progression. Otolaryngology - Head and Neck Surgery 124:

203 Marchetti A, Tinari N, Buttitta F, Chella A, Angeletti CA, et al. (2002) Expression of 90K (Mac-2 BP) correlates with distant metastasis and predicts survival in stage I non-small cell lung cancer patients. Cancer research 62: Skrzypski M, T TL, Kaczmarek P, Pruszynska-Oszmalek E, Pietrzak P, et al. (2011) Orexin A stimulates glucose uptake, lipid accumulation and adiponectin secretion from 3T3-L1 adipocytes and isolated primary rat adipocytes. Diab tologia 54: Ho HW, Lim Z, KH TL, Choo CT (2011) Bilateral Kuttner tumours of the lacrimal glands. Orbit 30: Lesley J, Hyman R, English N, Catterall J, Turner G (1997) CD44 in inflammation and metastasis. Glycoconjugate journal 14: Jacobs PP, Sackstein R (2011) CD44 and HCELL: preventing hematogenous metastasis at step 1. FEBS Letters 585: Sackstein R (2012) Glycoengineering of HCELL, the human bone marrow homing receptor: sweetly programming cell migration. Annals of Biomedical Engineering 40: Agelopoulos K, Greve B, Schmidt H, Pospisil H, Kurtz S, et al. (2010) Selective regain of egfr gene copies in CD44+/CD24-/low breast cancer cellular model MDA-MB-468. BMC Cancer 10: Olsson E, Honeth G, Bendahl PO, Saal LH, Gruvberger-Saal S, et al. (2011) CD44 isoforms are heterogeneously expressed in breast cancer and correlate with tumor subtypes and cancer stem cell markers. BMC Cancer 11: Theodoropoulos PA, Polioudaki H, Agelaki S, Kallergi G, Saridaki Z, et al. (2010) Circulating tumor cells with a putative stem cell phenotype in peripheral blood of patients with breast cancer. Cancer Letters 288: Osterbur J, Sprague L, Muccioli M, Pate M, Mansfield K, et al. (2012) Adhesion to substrates induces dendritic cell endothelization and decreases immunological response. Immunobiology Fuhlbrigge R, King S, Sackstein R, Kupper T (2006) CD43 is a ligand for E-selectin on CLA+ human T cells. Blood 107: Takada A, Ohmori K, Takahashi N, Tsuyuoka K, Yago A, et al. (1991) Adhesion of human cancer cells to vascular endothelium mediated by a carbohydrate antigen, sialyl Lewis A. Biochemical and Biophysical Research Communications 179: Takeo K, Kawai T, Nishida K, Masuda K, Teshima-Kondo S, et al. (2009) Oxidative stress-induced alternative splicing of transformer 2beta (SFRS10) and CD44 pre-mrnas in gastric epithelial cells. Am J Physiol Cell Physiol 297: C Barthel SR, Gavino JD, Wiese GK, Jaynes JM, Siddiqui J, et al. (2008) Analysis of glycosyltransferase expression in metastatic prostate cancer cells capable of rolling activity on microvascular endothelial (E)-selectin. Glycobiology 18: Sakuma K, Aoki M, Kannagi R (2012) Transcription factors c-myc and CDX2 mediate E-selectin ligand expression in colon cancer cells undergoing

204 204 EGF/bFGF-induced epithelial-mesenchymal transition. Proceedings of the National Academy of Sciences of the United States of America 109: Brown RL, Reinke LM, Damerow MS, Perez D, Chodosh LA, et al. (2011) CD44 splice isoform switching in human and mouse epithelium is essential for epithelial-mesenchymal transition and breast cancer progression. Journal of Clinical Investigation 121: Weigelt B, Peterse JL, van 't Veer LJ (2005) Breast cancer metastasis: markers and models. Nat Rev Cancer 5: Ni Z, Walcheck B (2007) Varied levels of reactivity by different E-selectin/Fc constructs with cutaneous lymphocyte-associated antigen (CLA)+ CD4+ T cells. Immunology letters 108: Kummitha CM, Shirure VS, Delgadillo LF, Deosarkar SP, Tees DF, et al. (2012) HECA-452 is a non-function blocking antibody for isolated sialyl Lewis x adhesion to endothelial expressed E-selectin under flow conditions. Journal of Immunological Methods 384: Chang KC, Hammer DA (1999) The forward rate of binding of surface-tethered reactants: Effect of relative motion between two surfaces. Biophysical Journal 76: Chen S, Springer TA (2001) Selectin receptor-ligand bonds: Formation limited by shear rate and dissociation governed by the Bell model. Proc Natl Acad Sci U S A 98: Hammer DA, Apte SM (1992) Simulation of cell rolling and adhesion on surfaces in shear flow: general results and analysis of selectin-mediated neutrophil adhesion. Biophysical Journal 63: Bell GI (1978) Models for the specific adhesion of cells to cells. Science 200: Lawrence MB, Kansas GS, Kunkel EJ, Ley K (1997) Threshold levels of fluid shear promote leukocyte adhesion through selectins (CD62L,P,E). J Cell Biol 136: Cheung LS, Raman PS, Balzer EM, Wirtz D, Konstantopoulos K (2011) Biophysics of selectin-ligand interactions in inflammation and cancer. Phys Biol 8: Wayman AM, Chen W, McEver RP, Zhu C (2010) Triphasic force dependence of E- selectin/ligand dissociation governs cell rolling under flow. Biophys J 99: Tozeren A, Ley K (1992) How do selectins mediate leukocyte rolling in venules? Biophys J 63: Foote J, Milstein C (1991) Kinetic maturation of an immune response. Nature 352: Williams AF (1991) Cellular interactions. Out of equilibrium. Nature 352: Chen SH, Dallas MR, Balzer EM, Konstantopoulos K (2011) Mucin 16 is a functional selectin ligand on pancreatic cancer cells. FASEB Journal Shinde Patil VR, Campbell CJ, Yun YH, Slack SM, Goetz DJ (2001) Particle diameter influences adhesion under flow. Biophys J 80:

205 205 APPENDIX 1 Table A1. Parameter Values Used for Calculations Parameter Description Value Unit D Diffusivity 2.90E-10 cm 2 /s DeltaA Damkohler number 310 DeltaD Damkohler number Dτ Time interval E-05 eps Epselon Ht Chamber height cm k B T Thermal energy (298 K) 4.1E-14 gm cm 2 /s 2 LLO No. of time nodes 1200 Lt Chamber Length 0.1 cm R Particle radius 7.5E-04 cm μ Viscosity (water) 0.01 gm/cm s P Peclet s number

206 206 Table A2. Dimensionless Time Calculated from Perfusion Time Time (min) Τ (dimensionless time) E E E E-5

207 207 APPENDIX 2 c!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! c Band(J) for three independent, variable, two dependent variable, linear PDEs. c c Venktesh S. Shirure (11/28/12) adapted the original comments by Dr. G. G. Botte (04/15/2001). c c The following variables are used in the program. c i and j are nodes in positive y and positive x direction of cartecian coordinates c C(1,i,j): for i=1 to Ni dependent variables 1 c for i=ni+1 to 2Ni dependent variable 2 c zetax(i): independent variable in positive y direction of cartecian coordinates c zetay(j): independent variable in positive x direction of cartecian coordinates c TIME: Independent variable time c A(1,i,i): Jacobian c B(1,i,i): Jacobian c D(1,i,2*i+1): Jacobian. c X(1,i,i): Jacobian c Y(1,i,i): Jacobian c F(i): Equation c CG(1,i,j): Guess values. Used in nonlinear equations c c If the dimensions of any of the common variables are changed c (A, B, C, D, G, X, and Y) in the main program. c They also need to be changed in the Band(j) and Ludec subroutines. c!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! IMPLICIT REAL*8(A-H,O-Z) REAL*8 P, Deltaa, deltad, eps, TIME, DT, L c c The dimensions of the variables shown below allow solving until c three differential equations simultaneously with a maximum of c 1001 nodes (i=3, j=1001) c DIMENSION C(1,501,501),CG(1,501,501),A(1,501,501) 1,B(1,501,501), X(1,501,501),Y(1,501,501),F(501), 1G(501),ZETAZ(501),CN(1,501,501), 2D(1,501,501),ZETAX(501),cd(1,501,501),vsstemp(501,501) CHARACTER*10 NAME COMMON/A1/ A,B,C,D COMMON/A2/ G,X,Y,N,NJ,NI

208 FORMAT(A10) 114 FORMAT('0',/,16X,'ANALYTICAL SOLUTION ',//, 1 13X,' ZETA ',10X,'dzeta(1)') 108 FORMAT (' ',/,5X,'THIS RUN DID NOT CONVERGE') 111 FORMAT(' ',6X,6f14.5,6f14.5) 109 FORMAT (' ',/,5X,'THIS RUN DID CONVERGE;IE. RESIDUAL L.T. 1E-12') 113 FORMAT(5X,'NUMBER OF ITERATIONS =',I4) 210 FORMAT('0',//,16X,'NUMBER OF EQUATIONS =',I3,//,16X,'NUMBER OF' 1,' NODE POINTS =',I4,//) 211 FORMAT('0',16X,'DELTA ZETA=',F20.8,//) 212 FORMAT('0',//,12X,' ZETA ', 1 6X,' Numerical ',4X,'Analytical',6X,'Error') 200 FORMAT('0',/,6X,'INITIAL PROFILE',//,12X,' ZETA ', 1 8X,' dzeta(1) ',10X,'Y2',//) 312 FORMAT('0',/,6X,'LAST COMPUTED PROFILE',//,12X,'POSITION', 1 8X,' dzeta(1) ',10X,//) 943 FORMAT(' ',5X,'COMPONENT ',I2,' RESIDUAL= ',E14.5,'FOR ', 1I2,' ITERATION' ) 717 FORMAT('1') PRINT *, 'ENTER OUTPUT FILE NAME (.txt):' READ (*,102) NAME OPEN (UNIT=16, FILE=NAME, STATUS='UNKNOWN') c=============================================================== ===== c Program Starts: c N: # equations c NJ: # of nodes in x direction of cartecian coordinates. Be sure that your dimensions are consistent with the number of nodes c Ni: # of nodes in y direction of cartecian coordinates. Be sure that your dimensions are consistent c with the number of nodes c=============================================================== NI=11 NJ=101 N=2*NI LLO=120 TIME= *0.02 DT=TIME/LLO

209 209 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c Constants c c Any variable that is used in the pogram needs to be defined above ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc eps=0.254 P=1.5*23325.D4 deltaa=310 K0D=0.001 Cf=0.1 Ht= Lt=0.1 DZETAZ=1.D0/(NJ-1) DZETAX=0.1/(NI-1) Dzetay=DzetaZ cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c Calculation of the zeta value ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc DO 777 J=1,NJ 777 ZETAz(J)=(J-1)*DZETAZ DO 7 I=1,NI ZETAX(I)=(I-1)*DZETAX 7 CONTINUE DO 28 L=1,LLO TIME=L*DT RTIME=L deltad=(k0d/(rtime**0.33))*31128 c c This loop stores latest calculated C(I,J)as the old concentration c DO 2 I=1,N DO 2 J=1,NJ 2 CN(1,I,J)=C(1,I,J) c c This loops assignns initial guess c id=0 Do 3 i=1,2*ni DO 3 J=1,NJ 3 C(1,i,J)= c JCOUNT=0 8 JCOUNT=JCOUNT+1 DO 100 I=1,N

210 210 DO 100 J=1,NJ 100 CG(1,I,J)=C(1,I,J) J=0 10 J=J+1 DO 9 I=1,N DO 9 K=1,N X(1,I,K)=0.0 D0 Y(1,I,K)=0.0 D0 9 CONTINUE id=0 DO 11 I=1,2*Ni G(I)=0.0 D0 DO 11 K=1,Nj A(1,I,K)=0.0 D0 B(1,I,K)=0.0 D0 11 D(1,I,K)=0.0 D id=id+1 IF(J-1) 12,12,14 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc C THIS IS ZETA=0.0 c Input your equations and jacobians for the left boundary cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c ccccccccccccccc c Bottom surface c ccccccccccccccc 12 if(id==1) then F(id)=c(1,id,j)-1 F(id+ni)=C(1,id+ni,j) B(1,id+ni,id+ni)=1 B(1,id+ni,id)=-deltaA B(1,id,id)=1 c cccccccccccccccccc c c c c c Intermediate nodes cccccccccccccccccc else if ((1<id).and.(id<NI)) then F(id)=c(1,id,j)-1 F(id+Ni)=c(1,id+ni,j) B(1,id+Ni,id+Ni)=1 B(1,id,id)=1 ccccccccccccccc Top surface ccccccccccccccc else if (id==ni) then F(id)=c(1,id,j)-1

211 211 F(id+Ni)=c(1,id+ni,j) B(1,id+Ni,id+Ni)=1 B(1,id,id)=1 endif if (id==n) then goto 1111 endif goto DO 30 I=1,N G(I)=-F(I) DO 30 K=1,N 30 G(I)=G(I)+B(1,I,K)*C(1,K,J)+D(1,I,K)*C(1,K,J+1)+X(1,I,K)* 1C(1,K,J+2) CALL BAND(J) GO TO IF (J-NJ) 15,20,20 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c Input your equations and jacobians for the central nodes cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c ccccccccccccccc c Bottom surface c ccccccccccccccc 15 if(id==1) then F(id)=(-3*C(1,id,j)+4*C(1,id+1,j)-C(1,id+2,j))/(2*dzetax)- 1deltaA*C(1,id,j)+deltaD*C(1,id+Ni,j) F(id+ni)=(C(1,id+ni,j)-CN(1,id+ni,j))/(DT)-deltaA* 1C(1,id,j)+deltaD*C(1,id+ni,j) B(1,id,id)=-3/(2*dzetax)-deltaA B(1,id,id+1)=2/dzetax B(1,id,id+2)=-1/(2*dzetax) B(1,id,id+ni)=deltaD B(1,id+ni,id+ni)=1/DT+deltaD B(1,id+ni,id)=-deltaA c c c cccccccccccccccccc Intermediate nodes cccccccccccccccccc else if ((1<id).and.(id<NI)) then F(id)=(C(1,id,j)-CN(1,id,j))/DT+P*eps*(zetax(id)-zetax(id)**2)* 1(C(1,id,j+1)-C(1,id,j-1))/(dzetay*2)+eps**2*(C(1,id,j+1)-2* 1C(1,id,j)+C(1,id,j-1))/dzetay**2-(C(1,id+1,j)-2* 1C(1,id,j)+C(1,id-1,j))/dzetax**2 F(id+Ni)=c(1,id+ni,j)

212 212 c c c B(1,id+Ni,id+Ni)=1 B(1,id,id)=1/DT-2*eps**2/dzetay**2+2/dzetax**2 D(1,id,id)=P*eps*(zetax(id)-zetax(id)**2)/(dzetay*2)-eps**2 1/dzetay**2 A(1,id,id)=-P*eps*(zetax(id)-zetax(id)**2)/(dzetay*2)-eps**2 1/dzetay**2 B(1,id,id+1)=-1/dzetax**2 B(1,id,id-1)=-1/dzetax**2 cccccccccccccccccc Top surface cccccccccccccccccc else if (id==ni) then F(id)=(3*C(1,id,j)-4*C(1,id-1,j)+C(1,id-2,j)) F(id+Ni)=c(1,id+ni,j) B(1,id+Ni,id+Ni)=1 B(1,id,id)=3 B(1,id,id-1)=-4 B(1,id,id-2)=1 endif if (id==n) then goto 1112 endif goto DO 31 I=1,N G(I)=-F(I) DO 31 K=1,N 31 G(I)=G(I)+A(1,I,K)*C(1,K,J-1)+B(1,I,K)*C(1,K,J)+D(1,I,K)* 1C(1,K,J+1) CALL BAND (J) GO TO 10 cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc C THIS IS ZETA=1.0 c Input your equations and jacobians for the right boundary cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c ccccccccccccccc c Bottom surface c ccccccccccccccc 20 if(id==1) then F(id)=(3*C(1,id,j)-4*C(1,id,j-1)+C(1,id,j-2)) F(id+ni)=(C(1,id+ni,j)-CN(1,id+ni,j))/(DT)-deltaA* 1C(1,id,j)+deltaD*C(1,id+ni,j) B(1,id+ni,id+ni)=1/DT+deltaD B(1,id+ni,id)=-deltaA B(1,id,id)=3

213 213 A(1,id,id)=-4 X(1,id,id)=1 c cccccccccccccccccc c Intermediate nodes c cccccccccccccccccc else if ((1<id).and.(id<NI)) then F(id)=(3*C(1,id,j)-4*C(1,id,j-1)+C(1,id,j-2)) F(id+Ni)=c(1,id+ni,j) B(1,id+Ni,id+Ni)=1 B(1,id,id)=3 A(1,id,id)=-4 X(1,id,id)=1 c ccccccccccccccc c Top surface c ccccccccccccccc else if (id==ni) then F(id)=(3*C(1,id,j)-4*C(1,id,j-1)+C(1,id,j-2)) F(id+Ni)=c(1,id+ni,j) B(1,id+Ni,id+Ni)=1 B(1,id,id)=3 A(1,id,id)=-4 X(1,id,id)=1 endif if (id==n) then goto 1113 endif goto DO 32 I=1,N G(I)=-F(I) DO 32 K=1,N 32 G(I)=G(I)+Y(1,I,K)*C(1,K,J-2)+A(1,I,K)*C(1,K,J-1) 1+B(1,I,K)*C(1,K,J) CALL BAND(J) DO I=1,N DO J=1,NJ if (C(1,I,J).gt.1.) then C(1,I,J)=1 elseif (C(1,I,J).LT.0) then C(1,I,J)=0 endif continue DO 201 I=1,N RES=0.D0

214 214 DO 202 Jd=1,NJ 202 RES=RES+(C(1,I,Jd)- CG(1,I,Jd))**2 IF(RES.GT.1E-12)GO TO CONTINUE GO TO IF (JCOUNT-30) 8,8,40 22 continue 222 PRINT 109 PRINT 113,JCOUNT-1 c c The solution is printed in this loop c The output file is stored in the same folder as the program file c c IMPORTANT: THE OUTPUT COLUMNS GIVE X DIRECTION c AND ROWS GIVE Y DIRECTION CONCENTRATION c if (L==1.or.L==60.or.L==120.or.L==180.or.L==240.or.L==300.or. 1L==360.or.L==420.or.L==480.or.L==540.or.L==600.or.L==660.or.L==720 1.or.L==780.or.L==840.or.L==900.or.L==960.or.L==1020.or.L==1080.or. 1L==1140.or.L==1200) then c WRITE(16,111) (C(1,1,J), j=1,nj), L endif if (L==LLO) then WRITE(16,111) (zetax(i), i=1,ni) DO 27 j=1,nj WRITE(16,111) ZETAz(j),((C(1,i,J)), i=1,ni) c c Disable above statement and enable the following statement c for bound particles at wall c c WRITE(16,111) ZETAz(j),((C(1,1+Ni,j))) 27 CONTINUE endif c GO TO WRITE(16,943) I,RES,JCOUNT-1 PRINT 108 PRINT 312 WRITE(16,111) (C(1,1,J),J=1,NJ) 28 CONTINUE PRINT 717 STOP END

215 215 ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c Band(J) subroutine c c The dimensions of A, B, C, D, X, Y and G c need to be consistent with the main program c E is not a common variable but its dimensions need c to be consistent with the maximum number of nodes c and equations used in the program. c c C(1,i,j): dependent variables c A(1,i,i): Jacobian c B(1,i,i): Jacobian c D(1,i,2*i+1): Jacobian. c X(1,i,i): Jacobian c Y(1,i,i): Jacobian c F(i): Equation c G(i): Equation used in band(j) c E(i,2*i+1,j) c c Important: Do not change anything here unless c you have modified the dimensions of the variables c described above. ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc SUBROUTINE BAND(J) IMPLICIT REAL*8(A-H,O-Z) DIMENSION A(1,501,501),B(1,501,501),C(1,501,501), 1D(1,501,501), 1 X(1,501,501),Y(1,501,501),G(501),E(501,501,501) COMMON/A1/ A,B,C,D COMMON/A2/ G,X,Y,N,NJ,NI 101 FORMAT (15H0DETERM=0 AT J=,I4) IF (J-2) 1,6,8 1 NP1= N+1 DO 2 I=1,N D(1,I,2*N+1)=G(I) DO 2 L=1,N LPN= L + N 2 D(1,I,LPN)= X(1,I,L) CALL LUDEC (N,2*N+1,DETERM) IF (DETERM) 4,3,4 3 PRINT 101, J

216 4 DO 5 K=1,N E(K,NP1,1)= D(1,K,2*N+1) DO 5 L=1,N E(K,L,1)= - D(1,K,L) LPN = L+N 5 X(1,K,L) = -D(1,K,LPN) RETURN 6 DO 7 I=1,N DO 7 K=1,N DO 7 L=1,N 7 D(1,I,K)= D(1,I,K) + A(1,I,L)*X(1,L,K) 8 IF (J-NJ) 11,9,9 9 DO 10 I=1,N DO 10 L=1,N G(I)= G(I) - Y(1,I,L)*E(L,NP1,J-2) DO 10 M=1,N 10 A(1,I,L)= A(1,I,L) + Y(1,I,M)*E(M,L,J-2) 11 DO 12 I=1,N D(1,I,NP1)= -G(I) DO 12 L=1,N D(1,I,NP1)= D(1,I,NP1) + A(1,I,L)*E(L,NP1,J-1) DO 12 K=1,N 12 B(1,I,K)= B(1,I,K) + A(1,I,L)*E(L,K,J-1) CALL LUDEC (N,NP1,DETERM) IF (DETERM) 14,13,14 13 PRINT 101, J 14 DO 15 K=1,N DO 15 M=1,NP1 15 E(K,M,J)= - D(1,K,M) IF (J-NJ) 20,16,16 16 DO 17 K=1,N 17 C(1,K,J)=E(K,NP1,J) DO 18 JJ=2,NJ M=NJ - JJ + 1 DO 18 K=1,N C(1,K,M)=E(K,NP1,M) DO 18 L=1,N 18 C(1,K,M)=C(1,K,M)+E(K,L,M)*C(1,L,M+1) DO 19 L=1,N DO 19 K=1,N 19 C(1,K,1)=C(1,K,1)+X(1,K,L)*C(1,L,3) 20 RETURN END 216

217 217 ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc c Ludec(N,M,Determ) subroutine c c The dimensions of A, B, C, and D c need to be consistent with the main program c C(i,j): dependent variables c A(1,i,i): Jacobian c B(1,i,i): Jacobian c D(1,i,2*i+1): Jacobian. c X(1,i,2*i+1): Internal variable used in this subroutine c jcol(i): Internal variable used in this subroutine c c Important: Do not change anything here unless c you have modified the dimensions of the variables c described above. ccccccccccccccccccccccccccccccccccccccccccccccccccccccccc SUBROUTINE LUDEC(N,M,DETERM) IMPLICIT REAL*8(A-H,O-Z) DIMENSION A(1,501,501),B(1,501,501),C(1,501,501), 1D(1,501,501), 1 X(1,501,501),JCOL(501) COMMON/A1/ A,B,C,D NM1=N-1 DETERM=1.0 DO 1 I=1,N JCOL(I)=I DO 1 K=1,M 1 X(1,I,K)=D(1,I,K) DO 6 II=1,NM1 IP1=II+1 BMAX=DABS(B(1,II,II)) JC=II DO 2 J=IP1,N IF(DABS(B(1,II,J)).LE.BMAX)GOTO2 JC=J BMAX=DABS(B(1,II,J)) 2 CONTINUE DETERM=DETERM*B(1,II,JC) IF(DETERM.EQ.0.0) RETURN IF(JC.EQ.II) GO TO 4 JS=JCOL(JC) JCOL(JC)=JCOL(II) JCOL(II)=JS

218 218 DO 3 I=1,N SAVE=B(1,I,JC) B(1,I,JC)=B(1,I,II) 3 B(1,I,II)=SAVE DETERM=-DETERM 4 DO 6 I=IP1,N F=B(1,I,II)/B(1,II,II) DO 5 J=IP1,N 5 B(1,I,J)=B(1,I,J)-F*B(1,II,J) DO 6 K=1,M 6 X(1,I,K)=X(1,I,K)-F*X(1,II,K) DETERM=DETERM*B(1,N,N) IF(DETERM.EQ.0.0) RETURN DO 7 II=2,N IR=N-II+2 IM1=IR-1 JC=JCOL(IR) DO 7 K=1,M F=X(1,IR,K)/B(1,IR,IR) D(1,JC,K)=F DO 7 I=1,IM1 7 X(1,I,K)=X(1,I,K)-B(1,I,IR)*F JC=JCOL(1) DO 8 K=1,M 8 D(1,JC,K)=X(1,1,K)/B(1,1,1) RETURN C CLOSE(16) END

219 219 APPENDIX 3 A) B) Figure A3.1. Free particle concentration profile along channel length. Free particle concentration near the wall decreases with channel length but does not vary with time (time variation from 2 min to 20 min). K A, = nm/s, K D 0 =0.005 s -1, P=3.11Х10 6. The comparison of (A) the current study s results with (B) previously published results. (B) Haun and Hammer Originally published in Langmuir, doi: /la Reproduced with permission.

220 220 A) B) Figure A3.2. Bound particle density profile along the channel length. Bound particle density at wall deceases with channel length but increases with time. K A, = nm/s, K D 0 =0.005 s -1, P=3.11Х10 6. The comparison (A) the current study s results with (B) previously published results at similar conditions shows good match. (B) Haun and Hammer Originally published in Langmuir, doi: /la Reproduced with permission.

221 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Thesis and Dissertation Services!