CHARACTERIZING ROLES OF ANNEXIN 1 IN BREAST CANCER DEVELOPMENT BY MASS SPECTROMETRY BASED QUANTITATIVE PROTEOMICS

Through quantitative mass spectrometry-based proteomics, SILAC stable isotope labelling by amino acids in cell culture and Super-SILAC were applied to normal mammary gland cells and mamm

CHARACTERIZING ROLES OF ANNEXIN-1 IN

BREAST CANCER DEVELOPMENT BY MASS

DEPARTMENT OF PHYSIOLOGY NATIONAL UNIVERSITY OF SINGAPORE

2013

Acknowledgements

This graduate journey is akin to a jigsaw puzzle Besides the science involved, completion of the full picture would not be possible without any of the following ‘jigsaw pieces’ I would like to acknowledge and express my gratitude to:

A/P Lina Lim, my supervisor, for her patience and guidance I am also

thankful for the freedom, help and support she has graciously given me in this journey

Asst Prof Jayantha Gunaratne, my co-supervisor and mentor, for his grace,

guidance and support His hard-working and passion in science greatly motivates me

Prof Walter Blackstock, my ex-boss of the lab He was the one who cleared

the obstacles-filled path for me, making the commencement of this graduate studies possible

Past and present members of the Quantitative Proteomics Group (QPG),

especially Suat Peng and Sheena for their friendship, support and hearing ears whenever I whine Siok Ghee, for her expertise in mass spectrometry and

Rachel for her help in all the bioinformatics and visualization plots

Past and present members of the Inflammation and Cancer Lab, especially

Durkesh, Sunitha and Suruchi for their friendship, for always being there for

me This graduate fellowship we have shall be a beautiful memory

Institute of Molecular and Cell Biology, A*STAR, for funding part of this graduate programme

My family members for their support, and especially my grandmother for her constant love and concern for me

Above all, I present my heartfelt thanksgiving and gratitude to God, the One who loves me, never gives up on me and the One who holds the pieces of the jigsaw puzzle together

Table of Contents

DECLARATION i 

Acknowledgements ii 

Summary………… vii 

List of Tables ix 

List of Figures x 

List of Abbreviations xiii 

List of Symbols xvi 

Chapter 1:  Introduction 1 

1.1:  Breast cancer 1 

1.2:  Annexin 1 2 

1.2.1: Structure of ANXA1 2 

1.2.2: Functions of ANXA1 3 

1.2.2.1:  Anti-inflammatory role 3 

1.2.2.2:  Regulator of Cellular Processes 4 

1.2.3: ANXA1 and Cancer 6 

1.3:  Models for breast cancer 8 

1.3.1: Cell line models for breast cancer 8 

1.3.2: Mouse models for breast cancer 9 

1.3.2.1:  Xenograft models 9 

1.3.2.2:  Genetically-modified models 10 

1.3.3: Mammary gland development 10 

1.3.4: Hallmarks of cancer 12 

1.4:  Proteomics 15 

1.5:  MS-based proteomics 17 

1.6:  Quantitative MS-based proteomics 22 

1.6.1: Label-free Quantification 25 

1.6.2: Chemical labeling 26 

1.6.3: Metabolic Labelling 32 

1.6.3.1:  Stable isotope labelling by amino acids in cell culture

(SILAC) 32 

1.6.3.2:  Super-SILAC 35 

1.7:  Aims of Study 36 

Chapter 2:  Methods and Materials 37 

2.1:  Isolation and culture of primary mammary gland epithelial cells 37 

2.2:  Adaptation of cell lines to SILAC media 37 

2.3:  Filter-aided sample preparation (FASP) for SILAC incorporation check ……… 37 

2.4:  Cell lysates preparation 38 

2.5:  Murine mammary tumors lysates preparation 39 

2.6:  In-Gel Dehydration, Reduction, Alkyation and Tryptic Digestion 39 

2.7:  Super-SILAC mix preparation 40 

2.8:  Protein quantification 40 

2.9:  Western blot analysis 40 

2.10:  RT-PCR analysis 42 

2.11:  Heat treatment 43 

2.12:  Comet assay 44 

2.13:  Transfection experiments 44 

2.14:  Cell adhesion assay 45 

2.15:  Hydrogen peroxide (H2O2)treatment and Cell proliferation (WST-1) assay … 45 

2.16:  Migration assay 46 

2.17:  Wound healing assay 46 

2.18:  Reactive oxygen species (ROS) assay 46 

2.19:  Liquid Chromatography-Mass Spectrometry (LC-MS) 46 

2.20:  Identification and Quantification of Peptides and Proteins 47 

2.21:  Bioinformatics Analysis 47 

Chapter 3:  Results 48 

3.1:  Proteome profiling of ANXA1+/- and ANXA1-/- normal mammary gland cells 48 

3.1.1: SILAC workflow for mass spectrometer analysis 48 

3.1.2: MS data analysis 52 

3.1.2.1:  Clustering proteins into up- and down-regulated with ANXA1-/-………… 55 

3.2:  Analysis of down-regulated proteins with ANXA1-/- 57 

3.2.1: Pathway analysis 57 

3.2.1.1:  Identification of proteins in DNA-damage response

pathway……… 57 

3.2.2: Biochemical validation of selected down-regulated proteins 59 

3.2.3: Functional validation of pathway analysis 60 

3.2.3.1:  Comet assay analyses of thermal-stressed mammary gland cells……… 61 

3.2.3.2:  Reactive oxygen species (ROS) assay of mammary gland cells……… 63 

3.2.3.3:  Cell proliferation assays of oxidative-stressed mammary gland cells……… 65 

3.2.3.4:  Over-expression of down-regulated protein reverses DNA-damage response in ANXA1-/- mammary gland cells 66 

3.3:  Analysis of up-regulated proteins with ANXA1-/- 69 

3.3.1: Pathway analysis 69 

3.3.1.1:  Identification of proteins in cell adhesion/motility pathway 69  3.3.2: Biochemical validation of selected up-regulated proteins 72 

3.3.3: Functional validation of pathway analysis 72 

3.3.3.1:  Migration assay of the mammary gland cells 73 

3.3.3.2:  Wound healing assay of mammary gland cells 74 

3.3.3.3:  Silencing of up-regulated protein reverses adhesive phenotype in ANXA1-/- mammary gland cells 76 

3.4:  Proteome profiling of ANXA1+/- and ANXA1-/- mammary tumors 82 

3.4.1: Western blot analysis of the mammary tumors 82 

3.4.2: Workflow for Super-SILAC mass spectrometry analysis 83 

3.4.3: Relative quantification of proteins 88 

3.4.4: Analysis of the Super-SILAC mix 89 

3.5:  Categorization of up- and down-regulated proteins with ANXA1-/- 91 

3.6:  Bioinformatics analyses of down- and up-regulated proteins with ANXA1-/- 92 

3.6.1: Down-regulated proteins with ANXA1-/- 92 

3.6.2: Up-regulated proteins with ANXA1-/- 93 

3.6.3: Cancer-Associated Proteins (CAPs) and Cancer Driver Mutations (CDMs) 94 

3.6.3.1:  Mapping up- and down-regulated CAPs to CDMs 99 

3.6.3.2:  Network with the up-regulated proteins 102 

3.6.3.3:  Network with the down-regulated proteins 102 

3.7:  Comparison of normal mammary gland cells (SILAC-based) and mammary tumors (Super-SILAC based) 102 

3.7.1: Up-regulated in both mammary gland cells and mammary tumors103 

3.7.2: Down-regulated in both mammary gland cells and mammary

tumors 104 

3.7.3: Inverse correlation between up- and down-regulated proteins in mammary gland cells and mammary tumors 105 

3.7.4: Up- or down-regulated proteins in mammary gland cells but no change in mammary tumors 107 

3.7.5: Up- or down-regulated protein in mammary tumors but not detected in mammary gland cells 108 

Chapter 4:  Discussion 110 

4.1:  Characterization of ANXA1 from SILAC experiments 110 

4.2:  Possible roles of ANXA1 in tumorigenesis revealed by Super-SILAC experiments 118 

4.3:  Distinct differential regulation by ANXA1 between normal mammary gland cells and mammary tumors 125 

4.4:  Use of SILAC, Super-SILAC and the mouse models 128 

4.5:  Conclusion 129 

4.6:  Future directions 130 

Bibliography 132 

Appendix 153 

Tables………… 153 

Publications 190 

Supplementary CD 191 

Summary …………

Annexin-1 (ANXA1) has been reported to be involved in important pathological implications including cell proliferation, apoptosis, cancer and metastasis However, with controversies in ANXA1 expression in breast carcinomas, its role in breast cancer initiation and progression remains unclear The study presented here seeks to characterize ANXA1 in initiation and tumorigenesis of breast cancer by understanding the dysregulated pathways involved Through quantitative mass spectrometry-based proteomics, SILAC (stable isotope labelling by amino acids in cell culture) and Super-SILAC were applied to normal mammary gland cells and mammary tumors respectively from ANXA1-heterozygous (ANXA1+/-) and null (ANXA1-/-) mice to study their changes at the proteome level

physio-Quantitative comparison of the proteomes of normal mammary gland cells using SILAC quantified over 4000 proteins with 214 up-regulated and 169 down-regulated in ANXA1-/- Bioinformatics analysis of the up- and down-regulated proteins revealed that ANXA1 is potentially implicated in DNA-damage response and cell adhesion/motility pathways Relevant functional assays showed accumulation of more DNA damage with slower recovery on heat stress and an impaired oxidative damage response in ANXA1-/- cells in comparison to ANXA1+/- cells Over-expressing Yes-associated protein 1 (Yap1), the most down-regulated protein in DNA-damage response pathway cluster, reversed the proliferative response in ANXA1-/- cells when insulted, indicating the protective role ANXA1 plays via regulating a group of proteins involved in the DNA-damage response pathway Both migration and wound healing assays showed that ANXA1+/- cells possess higher motility with better wound closure capability than ANXA1-/- Silencing of β-parvin in ANXA1-/-, the protein with the highest fold change in the cell adhesion protein cluster reversed its motility phenotype This indicated that the pro-migratory role ANXA1 plays via regulating a group of proteins that are involved in cell adhesion/motility pathway

After establishing roles of ANXA1 in DNA-damage response and cell adhesion/motility that could contribute to breast cancer initiation, its role in breast cancer progression was investigated Super-SILAC analysis on mammary tumors derived from PyMT+ANXA1+/- and PyMT+ANXA1-/- mice quantified over 5000 proteins with 369 up- and 365 down-regulated proteins observed in PyMT+ANXA1-/- Bioinformatics analysis of up-regulated proteins showed that ANXA1 may be involved in cell cycle regulation pathways, whereas the analysis of down-regulated proteins showed its possible roles in cystic fibrosis and cytoskeletal remodelling during tumorigenesis As these outcomes are different from pathway analyses performed for the regulated protein clusters from the normal mammary gland cell experiments, overlapping up- and down-regulated protein clusters between mammary cells and tumors were closely compared Interestingly, the non-overlapped up- and down-regulated clusters were enriched in cell adhesion/migration and DNA-damage response pathway respectively, further indicating that ANXA1 plays a different role in tumorigenesis Out of the differentially-regulated proteins, 32 from the up-regulated and 25 from the down-regulated were identified to be cancer-associated Together with functional interaction network mapping, ANXA1 is proposed to play a modulatory role in several pathways in tumorigenesis Altogether the study here suggests that ANXA1 plays different, yet important roles in breast cancer initiation and progression

List of Tables

Table 1: The status of ANXA1 in clinical cancer tissues

Table 2: Antibodies used for western blotting

Table 3: Primers used for RT-PCR analysis

Table 4: Down-regulated proteins involved in DNA-damage response

pathway Table 5: Up-regulated proteins involved in cell adhesion and motility Table 6: Cell lines used in the Super-SILAC mix

Table 7: Up-regulated proteins (from Super-SILAC experiments) known

List of Figures

Figure 1: A ribbon presentation of ANXA1 molecular structure

Figure 2: The diverse biological functions of ANXA1

Figure 3: Morphology of the murine mammary gland

Figure 4: Overview of proteomics and its applications

Figure 5: Typical workflow in top-down proteomics approach

Figure 6: Typical workflow of shotgun proteomics

Figure 7: Summary of quantitative MS-based proteomics approaches Figure 8: Chemical structure of the ICAT Reagent

Figure 9: Workflow of ICAT methodology

Figure 10: Components of the iTRAQ label

Figure 11: Workflow of a typical isobaric labelling experiment

Figure 12: Chemistry involved in reductive dimethyl labeling

Figure 13: Typical workflow of a SILAC experiment

Figure 14: Super-SILAC methodology

Figure 15: Incorporation plot of heavy-labelled ANXA1+/- normal

mammary gland cells

Figure 16: Workflow of the SILAC experiment involving Fwd and Rev

experiments Figure 17: Correlation plots of all Fwd and Rev experiments against

protein fold changes Figure 18: Scatter and density plots of both Fwd and Rev experiments Figure 19: Biologically-mapped pathways of the down-regulated proteins Figure 20: Protein and mRNA levels of selected down-regulated proteins Figure 21: Graph of comets’ mean tail moments after heat treatment

Figure 22: Comet snapshots of control and heat-stressed mammary gland

cells Figure 23: FACS analysis of ROS assay

Figure 24: Proliferation rate of oxidative-stressed mammary gland cells Figure 25: Western blot of over-expression of Yap-1 in ANXA1-/-

mammary gland cells Figure 26: FACS analysis of the Yap1-transfected ROS assay

Figure 27: Comet snapshots of mock-treated and Yap1–transfected

Figure 28: Proliferation of Yap1-transfected in control and

oxidative-stressed ANXA1-/- Figure 29: Biologically-mapped pathways of the up-regulated proteins Figure 30: Sub-folders of tissue remodeling/wound repair pathway

Figure 31: Protein and mRNA levels of selected up-regulated proteins

Figure 32: Migratory capabilities of ANXA1+/- and ANXA1-/- mammary

gland cells

Figure 33: Wound healing assay snapshots and analysis

Figure 34: Silencing of -parvin in ANXA1-/- and its migratory capability

Figure 35: Wound healing snapshots and analysis of -parvin-silenced in

ANXA1-/- Figure 36: Illustration of adhesion/migration assay setup

Figure 37: Adhesion/migration assay of -parvin-silenced in ANXA1-/- Figure 38: Status of ANXA1 level in the murine mammary tumors

Figure 39: Incorporation plots of murine mammary epithelial and tumor

cell lines

Figure 40: 1:1 ratio mix of the murine mammary tumors with

Super-SILAC mix Figure 41: Unimodal distributions for Super-SILAC experiments

Figure 42: Density plot on the Super-SILAC experiments

Figure 43: Biologically-mapped pathways by the down-regulated proteins

Figure 44: Biologically-mapped pathways by the up-regulated proteins Figure 45: Network mapping of up-regulated proteins with CDMs

Figure 46: Network mapping of down-regulated proteins with CDMs

Figure 47: GeneGo analysis of overlapped down-regulated proteins

between SILAC and Super-SILAC

Figure 48: GeneGO analysis of oppositely-regulated proteins between

SILAC and Super-SILAC Figure 49: GeneGO analysis of up-/down-regulated proteins in SILAC but

unregulated in Super-SILAC Figure 50: GeneGo analysis of up-/down-regulated proteins in Super-

SILAC but not detected in SILAC Figure 51: Overview of the differentially-regulated proteins between

SILAC and Super-SILAC

Figure 52: Summary workflow of this study

List of Abbreviations

ANXA1 Annexin-A1

CAMs Cancer-associated mutations

CAPs Cancer-associated proteins

CM-H2-DCFDA 5-(and

6)-chloromethyl-2’,7’-dichlorodihydrofluorescein diacetate COL5A2 Collagen, type V, alpha 2

Da Daltons

DLBCL B-cell lymphoma-diffuse large B-cell lymphoma

EGFR Epidermal growth factor receptor

Erbb2ip Erbb2 interacting protein

FACS Fluorescence-associated cell sorting

FT-MS Fourier transform mass spectrometer

Fwd Forward

HER2 Human epithelial growth factor receptor 2

iTRAQ isobaric tagging for relative and absolute

quantification IAA Iodoacetamide

ICAT Isotope-coded affinity tagging

K0R0 12C614N2-lysine and 12C614N4-arginine

K8R10 13C615N2-lysine and 13C615N4-arginine

PAIs Protein abundance indices

PMN Polymorphonuclear

PS Phosphotidylserine

PyMT Polyomavirus middle T antigen

Rev Reverse

SILAC Stable isotope labeling by amino acids in cell

culture STAT Signal transducer and activator of transcription TGF Transforming growth factor beta

TIMP2 Tissue inhibitor of metalloproteinase 2

TJP1 Tight junction protein 1

WST-1 Water soluble tetrazolium

List of Symbols

α Greek alphabet alpha

μ Prefix meaning 10-6

Part of this work was presented as a poster in 1) IMPAKT Breast Cancer Conference held in Brussels, Belgium from

This work was presented as a poster in Proteomics Forum held in Berlin,

Germany from 17 – 21 March 2013

Chapter 1: Introduction

1.1: Breast cancer

Breast cancer is the second most common cancer worldwide Besides being the leading cause of cancer death in women, it has been reported that the incidence of breast cancer is increasing and that the burden it poses globally far exceeds other cancers (Jemal et al., 2010)

As breast cancer is known to be a clinically diversified and molecularly heterogeneous disease, gene expression studies from a variety of human breast tumors identifying four major molecular subtypes of breast cancer has been a major breakthrough for targeted therapies and treatments The molecular subtypes are the luminal, human epithelial growth factor receptor 2 (HER2)-enriched, the basal-like and the normal-like tumors (Perou et al., 2000) The luminal subtype typically express the estrogen and/or progesterone hormone receptors and are usually denoted as estrogen receptor (ER)-positive (Habashy

et al., 2012) Tamoxifen has been the gold standard in treating this type of breast cancer whereby it acts as a competitive inhibitor to the estrogen which

is often responsible for the proliferation of the breast cells (Pearson et al., 1982)

The HER2-enriched, or commonly known as the HER2-positive breast cancer accounts for about 25% of all breast cancers It is characterized by the over-expression of HER2 which results in the activation of the Ras/mitogen-activated protein kinase (MAPK) signalling pathway, increasing the proliferation of the breast cells (Janes et al., 1994) Despite the recombinant monoclonal antibody trastuzumab shown to obtain great success in curbing HER2-positive breast cancer, there have been non-responsive cases as well (Dean-Colomb and Esteva, 2008) This has thus also driven researches in search of alternative therapies or treatments

The basal-like subtype, more commonly presented as the triple-negative phenotype clinically with the absence of ER, progesterone receptor (PR) and HER2 This type of breast cancer is known to have the worst prognosis as

there is yet any successful targeted therapy, though chemotherapy is currently the primary therapy (Irvin and Carey, 2008) This has thus driven many breast cancer researches in an attempt to understand its biology for prevention of this disease and improving the current therapies

of ANXA1 forms an amphipathic -helix which replaces a helix (helix D in repeat III) in its core domain and causes unwinding of its core domain (Rosengarth et al., 2001) However, this helix D is forced back to its original position and structure upon calcium-dependent membrane binding, freeing the

NH2-terminal domain for interaction with S100A11 (Weng et al., 1993) This thus highlights some interesting mechanistic involvement of the NH2-terminal domain of ANXA1 (Gerke and Moss, 2002)

Figure 1: A ribbon presentation of the COOH-terminal -helical molecular structure of ANXA1 in the presence (left) and absence (right) of calcium

(Reproduced/Adapted with permission from Rescher and Gerke, 2004 Journal

of Cell Science from Company of Biologists Ltd)

The COOH-terminal domain of ANXA1 is the conserved core domain that contains four annexin repeats of 70 amino acid residues These repeats are packed into a -helical disk or the membrane binding module This disk has two sides, with the convex side having binding sites for calcium and facing the membrane when ANXA1 binds to the phospholipids, and the concave side facing away the membrane for accessibility of the NH2-terminal domain to bind to cytoplasmic proteins (Weng et al., 1993)

1.2.2: Functions of ANXA1

1.2.2.1: Anti-inflammatory role

ANXA1 has been known to play an anti-inflammatory role in the innate immune system via regulating the migratory activities of the polymorphonuclear (PMN) leucocytes (Perretti and D'Acquisto, 2009) These cells, such as the neutrophils, monocytes and macrophages, are known to have high levels of cytoplasmic ANXA1 during inactivated state (Morand et al., 1995) Upon activation such as during inflammation where the immune cells would migrate to the sites of inflammation, ANXA1 has been shown to mobilize from the cytoplasm to the cell surface membrane and secreted following PMN adhesion to the endothelium for extravasation (Perretti et al.,

The NH2‐terminal ‐helix

1996) This results in the inhibition of the cells into the subendothelial matrix tissue (Lim et al., 1998) and thus reduces the responsiveness of the immune cells to inflammation by negatively regulating their migratory capability Though the exposure of PMN to glucocorticoids increases ANXA1 level and its secretion, it has a different effect on the adaptive immune system

In the adaptive immune system, the effect of glucocorticoids to curb inflammation is also reported to be mediated by ANXA1 However, in contrast to that of innate immunity, the immunosuppressive effect of glucocorticoids on T-cells was brought about with a decrease in ANXA1 level (D'Acquisto et al., 2008; D'Acquisto et al., 2007), resulting in the inhibition of T-cell activation and thus, a reduction in adaptive immune response

Phospholipase A2 (PLA2) is an important enzyme that is involved in the production of arachidonic acid and eicosanoid, which are pro-inflammatory and released during inflammation and fever ANXA1 has also been shown to interact with and inhibit cytosolic phospholipase A2 (cPLA2) activity, preventing the phosphorylation and activation of PLA2 (Croxtall et al., 1995; Kim et al., 1994a) and thus shown to be antipyretic (Davidson et al., 1991)

1.2.2.2: Regulator of Cellular Processes

It has been reported that ANXA1 exerts both proliferative and

anti-proliferative effects in vitro ANXA1 was shown to inhibit proliferation in

macrophages via the constitutive activation of the mitogen-activated protein kinase extracellular signal-regulated kinase (MAPK/ERK) (Alldridge et al., 1999) Whereas ANXA1 has been reported to play an inhibitory role in cell growth and proliferation in lung cancer cells A549 (Croxtall et al., 1993a), it has a proliferative effect on liver cancer cells with its increased expression (de Coupade et al., 2000) Though the exact mechanisms of how ANXA1 exerts its regulatory role in cell growth may not be fully elucidated, ANXA1 has been reported to be a substrate for the epidermal growth factor receptor tyrosine kinase (EGFR) (De et al., 1986), one of the important signalling pathways involved in cellular proliferation and differentiation Such would result in an inhibition of the EGF-mediated proliferation (Croxtall et al., 1993b)

ANXA1 is also a phosphorylation target for many signal transducing kinases such as the platelet-derived growth receptor, hepatocyte growth factor receptor (Skouteris and Schroder, 1996) and protein kinase C (Varticovski et al., 1988), reinforcing its role in cellular proliferation

ANXA1 has also been implicated to regulate apoptosis, being pro-apoptotic in some cases but anti-apoptosis in other studies Not much is known about such contradiction in observations but there are some elucidations as to how ANXA1 plays a role in apoptosis For example, exogenously- administered ANXA1 promoted apoptosis by inducing the dephosphorylation of Bcl-2-associated death (BAD) protein which then translocate to the mitochondria in neutrophils (Solito et al., 2003) This apoptotic effect is further reinforced with the translocation of ANXA1 to the nucleus (Ishido, 2005) despite the functional role of nuclear ANXA1 not known

ANXA1 is also involved in apoptotic cells which relates to phagocytosis Exposure of phosphotidylserine (PS) on the outer leaflet of the plasma membrane of apoptotic cells is one of the well-known signals for phagocytes

to phagocytosize It has been shown that ANXA1 is able to be recruited to rich domains of apoptotic cell surfaces as a ‘eat-me’ signal for the phagocytes (Arur et al., 2003) The use of siRNA-mediated silencing of ANXA1 or antibodies against ANXA1 inhibited PS-mediated phagocytosis of apoptotic cells (Fan et al., 2004) This further illustrates the involvement of ANXA1 in apoptosis and PS-mediated phagocytosis

PS-All in all, ANXA1 plays regulatory role in the above-mentioned cellular processes (Figure 2) and any dysregulation in cellular processes due to ANXA1 may be informative in how ANXA1 may be involved in cancer development

Figure 2: A summary diagram showing the functions of ANXA1 (Reproduced

Adapted with permission from Lim and Pervaiz, 2007 The FASEB Journal from FASEB Office of Publications)

1.2.3: ANXA1 and Cancer

There is more and more evidence relating ANXA1 to cancer development in recent years Expressions of ANXA1 correlating with certain cancers have been reported even though ANXA1 could be found over-expressed in some cancers and loss in other cancers (Lim and Pervaiz, 2007) ANXA1 has been shown to be lost in esophageal cancer (Hu et al., 2004), prostate cancer (Xin et al., 2003), head and neck cancer (Garcia Pedrero et al., 2004) and over-expressed in hepatocarcinoma (de Coupade et al., 2000) as well as pancreatic cancer (Bai et al., 2004) The differential expression of ANXA1 in different types of cancer is also reflected by the variable effect of ANXA1 on the proliferation of tumor-derived cell lines The status of ANXA1 in certain clinical cancer tissues has been tabulated in Table 1

Table 1: The status of ANXA1 in clinical cancer tissues (Reproduced /

Adapted with permission from Lim and Pervaiz, 2007 The FASEB Journal from FASEB Office of Publications)

ANXA1 has also been implicated in tumor growth It has been reported that ANXA1, being the agonist for formyl peptide receptor (FPR1), was released

by necrotic tumor cells Its release allows its activation on the FPR1 expressed

on the live glioblastoma cells and promotes tumor growth and progression (Yang et al., 2011) It has also been demonstrated that skin and lung tumors in ANXA1-knockout mice had tumor growth retardation and were less aggressive and less metastatic than the wild-type mice (Yi and Schnitzer, 2009) It has also been reported that ANXA1 was able to promote the invasiveness of colorectal adenocarcinoma cells through its activation on the FPR2 (Babbin et al., 2006)

The role of ANXA1 in breast cancer has been intriguing as there seems to have conflicting reports on the status of ANXA1 in breast cancer Loss of ANXA1 was observed in ductal carcinomas while over-expression of ANXA1

was observed in basal cell carcinomas (Ahn et al., 1997; de Graauw et al., 2010) Interestingly, proteomics studies have shown a correlation between ANXA1 protein level and the metastatic capability of isogenic breast cancer cell lines-derived tumors (Lund et al., 2012) ANXA1 has also been reported

to be an important modulator for epithelial-to-mesenchymal (EMT)-like phenotypic switch via the transforming growth factor beta signalling pathway, promoting the migration and invasion of metastatic breast cancer cells (de Graauw et al., 2010) A recent study has also demonstrated how ANXA1 is required as part of the complex for the constitutive NFκB activity in basal cell carcinoma cell line, and that increases its metastastic potential (Bist et al., 2011) Together with the genomics studies involved in studying the molecular signatures associated with transformation and progression to breast cancer that highlights an up-regulation of ANXA1 in cellular transformation (Rhee et al., 2008), it points out the potential of ANXA1 playing a non-trivial role in breast cancer However, its specific role in breast cancer initiation and progression remains unclear

1.3: Models for breast cancer

1.3.1: Cell line models for breast cancer

Breast cancer cell lines are the most widely used models for the understanding

of molecular processes in breast cancer How dysregulation of important physio-pathological processes such as proliferation, apoptosis, migration and metastasis in breast cancer could be investigated in cell line models as these models are easy to manipulate and use Moreover, under well-defined experimental setups, cell line models generally are able to yield reproducible results (Vargo-Gogola and Rosen, 2007) Moreover, there are quite a number

of breast cancer cell lines that exhibit hormone dependence, making them suitable for steroid hormonal-dependent breast cancer studies under proper culturing conditions (Furuya et al., 1989)

1.3.2: Mouse models for breast cancer

The exact mechanisms underlying the initiation and progression of breast cancer remains largely unknown In many studies, the use of genetically-modified or xenograft mouse models has been informative in elucidating molecular processes involved in breast cancer metastasis The use of mouse models over other animal models present a number of advantages including short generation time with prolific breeding, small sizes for easy handling, bearing of young in utero which mimics the human gestation and the high similarity in structure and functions of genes between mouse and human These models are useful not only for the investigation of factors involved in malignant transformation, invasion and metastasis, they are also good models for testing of experimental drugs designed to succumb tumor malignancy (Ottewell et al., 2006)

1.3.2.1: Xenograft models

The xenograft mouse models are one of the widely used for studying human cancer metastasis Basically, human cancer cells can be injected either subcutaneously, intravenously (ectopic implantation) or directly into the target organ (orthotopic) of immunocompromised mice that would not reject the human cells

In the context of breast cancer, it has been shown that MDA-MB-231 cells were able to develop bone metastasis upon intravenous injection into the nude mice (Yoneda et al., 2000) This model has also been employed to further delineate the genes involved in organ-specific metastasis from breast cancer (Kang et al., 2003; Minn et al., 2005) However, the main limitation of xenograft models is that it is unable to fully recapitulate the development and progression of the tumors There are several reasons for this, including the lack of immune responses from immunocompromised mice which thus affect the stromal components or the tumor microenvironment, as well as the observation that cancer cells from different species of origin exhibit species-specific metastasis, indicating that the human cancer cells may not be fully adaptable to grow in a murine environment (Kuperwasser et al., 2005)

Despite the limitations, these experimental models would still be able to provide certain degree of impact towards our understanding in the biology of breast cancer

1.3.2.2: Genetically-modified models

These models often exploit the expression of known oncogenes under certain strong promoters to initiate breast carcinogenesis in mice For example, the MMTV-Neu (expression of Neu is driven by Mouse Mammary Tumor Virus promoter) mouse model is a widely used HER2-overexpressed breast cancer model whereby Neu (rat homolog of ErbB2 gene) is amplified These transgenic mice are able to develop multifocal adenocarcinomas with lung metastases (Muller et al., 1988) The MMTV-PyMT mouse model has the polyomavirus middle T antigen under the control of MMTV promoter This model spontaneously induces mammary tumors via the widespread transformation of the mammary epithelium Not only do the mice develop mammary adenocarcinomas, metastatic lesions were also observed in the lungs and lymph nodes (Guy et al., 1992)

1.3.3: Mammary gland development

The female mouse has a total of 5 pairs of mammary glands These mouse mammary glands develop throughout the whole lifespan of the female mouse with distinct morphologies during the different life-stages (Figure 3) The whole development has been proposed to be regulated by the hormonal status

of the mouse at its different life-stages

At the embryonic stage, the mammary bud that originates will start its proliferation after birth This is characterized by some ductal elongation from some small terminal end buds, effected probably by residual maternal and fetal hormones The extension of branched ducts into the mammary fat pad happens during puberty when the mouse is about 6-8 weeks old (in general) Ovarian steroid hormones released would lead to the pronounced formation of large club-shaped terminal end buds and epithelial ducts that would invade into the

mammary fat pad During pregnancy under the prolactin signaling (Topper and Freeman, 1980), mammary secretory epithelial cells would proliferate extensively to form the milk-secreting lobular alveoli, a compartment that would produce milk during lactation After weaning, these secretory epithelial cells would involute by apoptosis Thus, the whole development of the mouse mammary gland cells revolves around proliferation, invasion (extension of the ducts to mammary fat pads), angiogenesis (blood supply control during pregnancy to prevent premature involution) and apoptosis (after weaning) These are interesting notes as they are features to some of the hallmarks of cancer when the system is dysregulated (Fantozzi and Christofori, 2006; Hennighausen and Robinson, 1998)

Figure 3: A schematic representation of the morphology of the murine mammary gland from the embryonic stage to lactation (Reproduced / Adapted

with permission from Fantozzi and Christofori, 2006 Breast Cancer Research from Biomed Central)

1.3.4: Hallmarks of cancer

It has been proposed by Hanahan and Weinberg the six hallmarks of cancer (Hanahan and Weinberg, 2000) which are some biological capabilities that cells acquired as they transit into cancerous cells As the knowledge expands, this has recently been reviewed and added another four, totaling to ten hallmarks of cancer (Hanahan and Weinberg, 2011) (Figure of Hallmark) Sustaining proliferative signaling is one distinctive feature that is possessed by cancer cells This feature is a total dysregulation from a normal cell which has tight control over the production and release of signals for cell cycle progression and proliferation In contrast, cancer cells are able to override the tight control via a few ways: the cancer cells could have over-expression of receptors on cell surface which would render the cells hyperresponsive to the growth factors A classic example is the estrogen receptor positive breast cancer cells whereby a significant number of estrogen receptors is present and

is responsive to the hormone estrogen (Auricchio et al., 1987), important in females for both menstrual and estrous reproductive cycles Alternatively, it has been known that certain mutations present in cells could trigger the constitutive activation of downstream proliferative pathway Human melanomas, a type of very aggressive skin cancer was found to contain B-Raf protein mutations which could cause constitutive signaling in the mitogen-activated protein kinase (MAPK) pathway (Davies and Samuels, 2010) The cancer cells could also send signals to the normal cells in its vicinity to produce the various growth factors for themselves (Cheng et al., 2008)

Cancer cells, besides enabling themselves to sustain proliferative signaling are also able to dismantle the cell cycle checkpoints (evading growth suppressors) The tumor suppressor genes are the guardians of the cell which are gate-keepers at checkpoints – allowing the cell cycle to proceed or come to a halt for repair or apoptosis (Sun and Yang, 2010) The two classical tumor suppressor genes for cell cycle proliferation are retinoblastoma protein (RB) and tumor protein 53 (TP53) and they have been found to have loss-of-functions (for wild-type) or gain-of-functions (for mutated) in many human cancers (Nevins, 2001; Oren and Rotter, 2010)

Besides its ability to sustain proliferation signals and evades growth suppressors, cancer cells growth can also be attributed to its unlimited replicative ability This feature is mainly governed by the significant over-expression of telomerase and its activity in cancer cells (Kim et al., 1994b; Shay and Bacchetti, 1997) Telomerase is a specialized reverse transcriptase that adds repeated nucleotides of TTAGGG to the ends of the eukaryotic chromosomes to prevent DNA telomeric shortening which would lead to senescence or apoptosis of the cells In most normal human cells, there is no detectable level of telomerase and it has been known that with each successive round of DNA replication, telomeres are lost This shortening of telomeric DNA contributes to the limited replicative potential of normal cells Thus, with the over-expression of telomerase and its activity, it results in the stabilization of telomeric length and the immortalization of the cancer cells (Masutomi et al., 2003)

The sustenance of the normal cells is very much supported by the vascular system which delivers nutrients and oxygen as well as excretes the waste products and carbon dioxide Similarly, in order to cater to the needs of the highly proliferative cancer cells, the cancer cells are able to induce angiogenesis that remains constitutive, causing the normal quiescent vasculature to constantly grow new vessels to meet its demands (Hanahan and Folkman, 1996) This is of stark contrast to the normal angiogenic regulation whereby the only times it is activated in human adults would be during wound healing and female reproductive cycling Moreover, these are transiently controlled as opposed to the sustained activation of angiogenesis in cancer, indicating another dysregulation that is specific to cancer cells One important component that stabilizes and supports the vascular structure is the pericyte These pericytes, akin to contractile cells, have finger-like projections that wrap around the endothelial tubes in the normal tissue vasculature In addition, synthesis of the vascular basement membrane through the association of pericytes and endothelial cells helps to build a strong mechanical support to withstand the blood flow pressure Interestingly, in the tumor vasculature, the pericytes are found to be sparsely and loosely assembled (Barlow et al., 2013; Morikawa et al., 2002) Such abnormality observed in the tumor vasculature

has been proposed to be a factor that favors the intravasation of the cancer cells into the circulatory system for hematogeneous dissemination, one of the crucial and mandatory steps in metastasis (Gerhardt and Semb, 2008)

Epithelial-mesenchymal transition (EMT) is a series of events whereby the epithelial cell loses its epithelial phenotype and gains the mesenchymal phenotype This EMT process is one of the early steps of the invasion-metastasis cascade which is necessary for local invasion to take place before extravasation and colonization (Fidler, 2003; Talmadge and Fidler, 2010; Thiery, 2002) It is characterized by an alteration in cell-cell and cell-extracellular matrix (ECM) interactions, reorganization of the cytoskeleton to facilitate motility through the ECM and adopting a new transcriptional program (Radisky, 2005) which includes the ability to invade, resist apoptosis and disseminate (Polyak and Weinberg, 2009; Thiery et al., 2009; Yilmaz and Christofori, 2009) One of the best characterizations is the loss of the cell-cell adhesion molecule epithelial-cadherin (E-cadherin) and the up-regulation of another adhesion molecule neural-cadherin (N-cadherin) in various invasive cancers (Hazan et al., 2000; Islam et al., 1996; Richmond et al., 1997)

Cancer cells may also have this capability to circumvent the natural programmed cell death by apoptosis, which occurs as a homeostatic maintenance mechanism to prevent cancer and defense mechanism to any damaged or insulted cells (Elmore, 2007; Norbury and Hickson, 2001) A few

of the common strategies that the cancer or tumor cells would employ to resist cell death include loss of tumor suppressor and critical DNA damage sensor TP53, up-regulation of anti-apoptotic regulators Bcl-2 and Bcl-XL and down-regulation of pro-apoptotic factors such as Bax (Hanahan and Weinberg, 2011) It is typically known that cancer cells are able to resist cell death with the above-mentioned mechanisms and thus the still existing difficulty in annihilating the disease totally despite advanced medical technology

All the above-mentioned hallmarks of cancer were acquired abilities of the cancer cells made possible by the proposed two enabling characteristics, namely genomic instability and mutation in the cancer cells, and tumor-promoting inflammation which provides all the necessary ‘nutrients’ to the

tumor microenvironment needed for the tumor cells to sustain the cancer hallmarks necessary for invasion and metastasis (Hanahan and Weinberg, 2011) Interestingly, Hanahan and Weinburg also proposed two emerging hallmarks to be included: the reprogramming of energy metabolism and the ability to evade immune destruction

Cancer cells have been observed to be able to reprogram their glucose metabolism pathway in an attempt to support rapid cell division This was achieved by the switch from adenosine triphosphate (ATP) generation via oxidative phosphorylation (in normal cells) to ATP generation via glycolysis under aerobic conditions (Warburg, 1956) Thus, most incoming glucose in the cancer cells was converted to lactate in the aerobic glycolysis instead of being metabolized in the mitochondria This resulted in lower efficiency of ATP production despite its rapid production This was compensated by an up-regulation of glucose transporters (GLUT1) which allows for an abnormally high rate of glucose uptake to meet their needs (Cairns et al., 2011) Whereas much more is known in reprogramming of energy metabolism, the latter emerging hallmark still presents a big question mark as to how the immune surveillance system is involved and how the cancer cells are able to evade the immune surveillance

1.4: Proteomics

The advancements in the field of genomics and proteomics, together with the integration of informatics technologies, have brought tremendous insights to many biological studies Genomics is the systemic and comprehensive study

of genes and genomes High-throughput platforms has facilitated functional genomics approaches such as microarray-based expression profiles and automated sequencing in providing genome-wide expression analysis (Shoemaker and Linsley, 2002) These applications in cancer biology has prospective in identification of gene patterns which might be unique to distinguish between normal and different stages of cancer cells This has been exemplified in a few studies such as in the B-cell lymphoma-diffuse large B-cell lymphoma (DLBCL) and in breast cancer where the gene expression

profiles were able to stratify into the stage of disease and prognostic categories (Alizadeh et al., 2000; Hedenfalk et al., 2001) Despite the tremendous usefulness in aiding the understanding of molecular alterations especially in cancer biology, genomics study has the main limitation of the inability to predict or postulate cellular events taking place as the effectors to all biological functions are proteins and not genes (Cox and Mann, 2007; Graves and Haystead, 2002) Moreover, proteins are the ones to account for the phenotypes as the transcript levels correlate poorly to protein abundance (Greenbaum et al., 2003) and thus, best representatives for characterizing the biological systems of the cells Apart from these, many post-translational modifications, protein-protein interactions, spatial and temporal localizations

of proteins, key to many biological functions and processes, are not coded in the genome (Martin and Nelson, 2001) This has thus placed proteomics an edge over genomics

Proteomics is defined by large scale comprehensive and systemic study of the protein complement of a cell or tissue, including the assessment of all related fields of the proteins for characterizing the biological systems Unlike the genome which is constant, the proteome is a dynamic orchestration of proteins reflective of the cell state at any moment (Graves and Haystead, 2002) This has thus placed the field of proteomics a closer and more relevant step to understanding many biological studies and applications (Figure 4)

Figure 4: An overview of proteomics and its applications (Reproduced /

Adapted with permission from Graves, P.R., and Haystead, T.A, 2002

Microbiology and Molecular Biology Reviews from AM SOC FOR

MICROBIOLOGY)

Some of the proteome-wide platforms used in proteomics for the detection and identification of proteins include two-dimensional gel electrophoresis which separates proteins based on their molecular weight and isoelectric point, two-hybrid methods (Fields and Song, 1989), protein or antibodies arrays (MacBeath, 2002) and fluorescence methods such as the use of fluorescence resonance energy transfer for detection of protein-protein interactions, the use

of green fluorescence protein for localization of proteins in cell compartments

et cetera (Phizicky et al., 2003) These platforms, however, present the limitations of low resolutions and background bias With the advancement of technology and development of informatics, high resolution mass spectrometry (MS)-based proteomics is an uprising high throughput platform used as it is able to not only comprehensively map a proteome, but also expound into the dynamics of the proteins (Walther and Mann, 2010)

1.5: MS-based proteomics

MS-based technology has advanced to allow the rapid and accurate analysis of proteins with high sensitivity (Nilsson et al., 2010) The principal methodology of MS-based proteomics involves the use of a wide variety of

separation methods with mass spectrometry (MS), which consists of an ion source for volatizing proteins or peptides, a mass analyser for measuring the mass-to-charge ratio (m/z) of the ionized analytes and a detector for recording the ions (Aebersold and Mann, 2003) Such proteomic methodology can be categorized mainly into two approaches, the top-down and bottom-up approaches

The strategy of approaching proteomic studies via intact protein analysis by the mass and tandem MS analyses of proteins in the gas phase has been termed

‘‘top-down’’ proteomics This approach, as opposed to the bottom-up proteomics, analyses intact proteins directly without any prior digestion or use

of any proteases Thus, it directly derives sequences and protein information from the ion fragment spectra of intact proteins (Huber and Huber, 2010) Though there is another lexicon that defines such approach (top-down) by the first entity to be subjected to primary separation, that is, top-down for the separation of proteins by gel electrophoresis before further steps for preparation into the mass spectrometer, the former definition is more commonly adopted (Reid and McLuckey, 2002)

The general workflow of a top-down approach is illustrated in Figure 5 The protein mixture extracted from sample-of-interest may undergo fractionation

by a variety of techniques to reduce sample complexity and clean-up step amenable to electrospray prior to introduction into the mass spectrometer In the mass spectrometer, the protein ions derived from fragmentation are then

identified via databases search or through de novo sequencing of the amino

acids

Figure 5: A schematic illustration of a typical workflow in top-down

proteomics approach (Reproduced / Adapted with permission from Reid and

McLuckey, 2002 Journal of Mass Spectrometry from John Wiley and Sons)

As intact protein ions are produced in the top-down strategy, this in principle produces the full sequence of the protein for examination and allows ‘clearer’ identification This is especially useful for locating and characterizing post-translational modifications In addition, the ion fragments produced by the top-down strategy usually have much higher masses and thus, are more unique Thus, the information obtained would be able to give a higher accuracy of identification (Kelleher, 2004) However, with such higher masses of ions obtained would require a mass spectrometer of higher MS resolving power such as the Fourier transform mass spectrometer (FT-MS) (McLafferty et al., 2007) Furthermore, another major limitation to this top-down approach is the fractionation of the proteins As there is no available good fractionation method suitable for downstream top-down approach, analysis of the intact proteins becomes a major challenge The spectra of the proteins could be very complicated due to the fractionation limitation and thus, resulting in difficulty

in identifying the correct fragment ions corresponding to its respective parent ions (Yates and Kelleher, 2013)

In contrast to top-down, bottom-up proteomics refers the analysis of peptides (from protein digestion) for the characterization of proteins (Zhang et al., 2013) There are basically two main approaches for protein identification used

in bottom-up proteomics, namely the peptide mass fingerprinting (PMF) and tandem mass spectrometry (tandem MS)

In peptide mass fingerprinting, it lies on the basis that if the mass of the peptides could be measured with certain degree of accuracy, it would be sufficient to help in identifying the protein represented by these peptides The masses of the peptides analysed by the mass spectrometry instrument are

searched against a database generated by in silico digestion of the proteins

translated from genomes of specific organisms using the same enzyme employed in the experiment (Zhang and Linhardt, 2009) The overall results are then statistically analysed to find the best match to indicate the proteins highly represented by the peptides Since no fragmentation is needed for such analysis as only masses of the peptides are needed, any mass spectrometer is suitable for such work However, the simplicity of this technique also poses limitations to the scoring of PMF Firstly, it is limited only to the identification

of proteins with known sequences In order for high scoring of the PMF of a particular protein, the presence of several peptides to uniquely represent the protein should be present Moreover, as most of the algorithms written for PMF come with the assumption that the MS spectrum is derived from a single protein, mixtures or complicated protein samples would greatly compromise the quality or accuracy of the results Thus, this method is mostly suitable for isolated or pure protein samples (Thiede et al., 2005)

Tandem MS is another application for protein sequence identification whereby the peptide ions pass through the first mass analyser (to generate the MS1) before the mass-selected ions are transmitted to the collision cell for analysis for further fragmentation before analysis by the second mass analyser (to generate the MS2), generating two m/z analyses (Dongre et al., 1997) However, when this bottom-up proteomics is applied to a mixture of complex proteins, together with the use of high-performance liquid chromatography (HPLC) for fractionation purposes and tandem MS, the term shotgun

proteomics has been coined for such analysis (McCormack et al., 1997) A typical workflow for shotgun proteomics is depicted in Figure 6

Figure 6: Schematic diagram showing the typical workflow of shotgun proteomics (Reproduced / Adapted with permission from Prof Jürgen Cox)

For bottom-up proteomics, the main disadvantage is the difficulty in protein inference from the peptides analysed by the mass spectrometer due to sequence redundancy, which is much more apparent in complex protein mixtures For shotgun or discovery proteomics, the low abundance peptides may be less likely to be selected for sequencing On the other hand, as this technique is peptide-centric, the shorter peptides generated from proteolytic digestion are easier to be fragmented in the mass spectrometry instrument as compared to intact proteins Furthermore, when shotgun proteomics is used with quantitative methodologies, it is able to generate relatively accurate high throughput quantitative data This has thus placed shotgun proteomics as the method of choice among the rest for proteomics studies (Sabido et al., 2012) and is one of the main focuses in the proteomics field in recent years

(Nesvizhskii and Aebersold, 2005) Besides shotgun proteomics, targeted proteomics is another arena in bottom-up proteomics Whilst shotgun discovery-based strategize to identify as many proteins as possible, targeted proteomics simply selects only particular protein targets and enables the accurate and reproducible quantification of a particular protein target in any biological sample (Doerr, 2013) This method requires the development of an assay for quantifying the targeted protein-of-interest followed by using the same assay for subsequent quantification of the same protein in any biological sample Selected reaction monitoring or multiple reaction monitoring is the commonly used technique in targeted proteomics where mass filtering is applied for selection of ions of indicated mass range for further fragmentation This not only eliminates ‘high background’ contributed by other ions, but also eliminates stochastic sampling, allowing this technique to be very valuable for clinical validations such as quantification of candidate biomarkers in small amount of clinical samples (Picotti et al., 2013)

MS-based proteomics has been applied in helping to answer many biological questions However, mass spectrometry is not inherently quantitative and it is the advances in the whole proteomic workflow including the sample preparation, mass spectrometry instrumentation, computational and informatics analysis that allowed the quantitative information to be made available

1.6: Quantitative MS-based proteomics

Besides the identification of proteins in a complex sample, the ability to quantify proteins is a value-added component in MS-based proteomics This is especially valuable in understanding the functional aspects of proteins by its temporal changes in the proteome In quantitative MS-based proteomics, there are two main means to quantification, namely absolute and relative quantification

Absolute quantification in proteomics has emerged as an important arena in biology whereby it helps to answer questions where absolute quantity is concerned Common questions include, how many copies of the protein-of-