STUDY OF OSTEOSARCOMA DEVELOPMENT, PROGRESSION AND TREATMENT

1. Gene Profiling and Tumorigenesis

2. IGF-1R/MEK/ERK Signalling and Malignant Potential

3. IGF-1R Targeted Therapy Enhances Chemotherapy Using Doxorubicin

CHEUK FAI FREDERICK LUK

A thesis presented in fulfilment of the requirements for the degree of Doctor of Philosophy

Prince of Wales Clinical School Faculty of Medicine University of New South Wales Sydney, Australia

August, 2010

i

Copyright Statement

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only).

I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed: ______

14th December 2010 Dated: ______

Authenticity Statement

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed: ______

14th December 2010 Dated: ______ii

Original Statement

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed: ______

14th December 2010 Dated: ______iii

Abstract

Understanding of osteosarcoma progression and molecular pathogenesis remains limited due to the complex genetic changes. Recently, osteosarcoma is redefined as a differentiation disease due to the disruption of osteoblast differentiation, opening up a new direction for investigations. The type I insulin-like growth factor receptor (IGF-1R) participates in promoting malignant potential in osteosarcoma, but the effectiveness of its inhibition has not been fully elucidated.

Gene expression profile analysis of the two osteosarcoma cell lines showed that genes of cell-adhesion and nervous system were expressed differentially. These genes are known to participate in controlling osteoblast differentiation. Other genes known to be involved in osteoblast differentiation were also found in the analysis. This study suggests that the cell-adhesion and nervous system signalling pathways may involve in the disruption of osteoblast differentiation leading to the development of osteosarcoma.

The detection of protein expression of IGF-1R downstream signal mediators in the primary and secondary xenograft tumour tissues showed for the first time that IGF-

1R/MEK/ERK signalling is involved in the malignant potential and lung metastasis of osteosarcoma. In addition, the in vitro invasive ability of the osteosarcoma cell line was reduced after inhibition of MEK/ERK signalling. These results signify that MEK/ERK could be potential therapeutic targets for the treatment of metastatic osteosarcoma.

Furthermore, it is also the first time that enhancement of growth inhibition and increased Doxorubicin sensitivity were shown after IGF-1R inhibitor and Doxorubicin combination treatment in a panel of 6 human osteosarcoma cell lines and a self- established resistant sub-clone. Mechanism studies in osteosarcoma cell lines showed that the combination therapy has advantages over mono drug treatment in terms of enhances apoptosis, maintains clonogenic inhibition and improves cytotoxicity in iv

therapeutic efficacy without addition of individual drug doses. This study indicates that inhibition of IGF-1R signalling and together with Doxorubicin chemotherapy is a potential effective strategy to improve the treatment of both Doxorubicin sensitive and resistant osteosarcoma. v

Acknowledgments

First of all acknowledgements go to Prof. William R. Walsh, Director of the Surgical &

Orthopaedic Research Laboratories (S&ORL) facility, for providing scholarship, support and acceptance to work in his laboratory. I would also like to thank my supervisors,

A/Prof. Jia-Lin Yang and Dr. Yan Yu for their guidance, patience, feedback and motivation. Their enormous and generous contribution throughout the PhD is greatly appreciated.

In addition, I would like to thank all the staff and students from S&ORL, who have helped me. Especially thanks to Dr. Rema Oliver, Dr. Abe Lau and Ms Joy Francisco for their understanding, motivation and listening to my countless problems, I would not have been able to complete this PhD without their assistance. Special thanks for A/Prof.

Hai-tao Dong (Institute of Biotechnology, Zhejiang University, China) for his help and teachings, which showed me the future again.

Thanks to my “coffee buddies” from the university, Moonsun Jung, Dr. Brooke Farrugia and others for their friendship and support, daily sharing of the caffeine addiction brightened up my days. Also, I would like to say thank you to all of my friends from within and outside of UNSW for their understanding and support, especially to Dr.

David Chin and Gilbert Poon for their mental support throughout the PhD when I am in the stage of madness.

Special thanks to Dr. Christine Chuang for her unconditional love, support and putting up with my insanity.

Finally and most importantly, acknowledgments go to my parents, sister and her family.

I would not have been able to finish this study and staying alive without their endless love, support and reassurance. vi

Table of contents

Copyright Statement ...... i

Authenticity Statement ...... i

Original Statement ...... ii

Abstract ...... iii

Acknowledgments ...... v

Table of contents ...... vi

List of figures ...... xii

List of tables ...... xv

Abbreviation ...... xvi

List of publications ...... xix

Chapter 1. Introduction ...... 1

1.1. Introduction ...... 1

1.2. Hypotheses and aims ...... 2

Chapter 2. Literature Review ...... 4

2.1. Clinical features of osteosarcoma ...... 4

2.1.1. Incidence and mortality ...... 4 2.1.2. Classification ...... 5 2.1.3. Current diagnosis, treatment and management ...... 8

2.2. Molecular pathogenesis of osteosarcoma ...... 14

2.2.1. Tumour suppressor genes ...... 15 2.2.2. Oncogenes ...... 18 2.2.3. Other factors and pathways...... 22

2.3. Osteogenic and tumourigenic properties of osteosarcoma ...... 25

2.4. Differentiation disruption and osteosarcoma ...... 26 vii

2.5. Advance technology for osteosarcoma research ...... 29

2.5.1. DNA microarray technology ...... 29 2.5.2. Importance of microarray in cancer research ...... 31 2.5.3. Gene expression analysis in osteosarcoma ...... 33 2.5.4. Limitations of DNA microarray technology ...... 35

2.6. In vitro and in vivo models of osteosarcoma research ...... 36

2.6.1. Commonly used human osteosarcoma cell line ...... 36 2.6.2. Development of osteosarcoma animal model ...... 41

2.7. Type 1 insulin-like growth factor receptor (IGF-1R) ...... 43

2.7.1. Molecular structure of the IGF-1R ...... 43 2.7.2. Physiological function of the IGF-1R ...... 45 2.7.3. IGF-1R downstream signalling pathways ...... 46 2.7.4. Association of IGF-1R in tumorigenesis and osteosarcoma ...... 49 2.7.5. Current strategy of IGF-1R inhibition or down-regulation ...... 51 2.7.6. IGF-1R tyrosine kinase inhibitor ...... 53 2.7.7. MEK/ERK inhibitor ...... 55

2.8. IGF-1R targeted therapy and chemotherapy ...... 57

2.8.1. Chemotherapy ...... 57 2.8.2. IGF-1R chemoresistance ...... 58 2.8.3. IGF-1R targeted therapy ...... 60 2.8.4. IGF-1R combination therapy ...... 64

Chapter 3. Materials and Methods ...... 66

3.1. Materials ...... 66

3.1.1. Reagents and consumables for tissue culture ...... 66 3.1.2. Reagents for histochemistry, immunohistochemistry and immunocytochemistry ...... 66 3.1.3. Reagents for molecular detections ...... 67 3.1.4. Equipment ...... 67 3.1.5. Mammalian cell lines ...... 69 3.1.6. Animals ...... 69

3.2. Methods ...... 70

3.2.1. Mammalian cell culture ...... 70 3.2.1.1. Media and supplements preparation ...... 70 viii

3.2.1.2. Cell resuscitation ...... 70 3.2.1.3. Cell passage ...... 71 3.2.1.4. Cryostorage ...... 71 3.2.1.5. Establishment of a Doxorubicin resistant osteosarcoma cell line . 72 3.2.2. Molecular Techniques ...... 72 3.2.2.1. Total ribonucleic acid (RNA) extraction ...... 72 3.2.2.2. Assessment of RNA samples ...... 74 3.2.2.3. Microarray experiment ...... 74 3.2.2.4. Microarray expression data processing, quality control and normalization ...... 74 3.2.2.5. Gene ontology and clustering analyses for data mining ...... 75 3.2.2.6. Generation of complimentary deoxyribonucleic acid ...... 75 3.2.2.7. Dissociation curves analysis for qRT-PCR ...... 76 3.2.2.8. Efficiency of primers for qRT-PCR ...... 77 3.2.2.9. Quantitative real-time polymerase chain reaction ...... 78 3.2.2.10. Protein/DNA array analysis and data mining ...... 80 3.2.2.11. Protein extraction and sample assessment ...... 80 3.2.2.12. SDS PAGE electrophoresis and Western blotting ...... 81 3.2.3. In vitro assays and drug treatment ...... 81 3.2.3.1. Detection of alkaline phosphatase (ALP) activity ...... 81 3.2.3.2. Detection of calcium deposition and mineralization ...... 82 3.2.3.3. Crystal violet colorimetric assay ...... 83 3.2.3.4. Characterization of cell growth ...... 83 3.2.3.5. Combination drug effectiveness analysis ...... 84 3.2.3.6. Trypan blue exclusion assay ...... 85 3.2.3.7. Clonogenic assay ...... 85 3.2.3.8. Flow cytometry and data analysis ...... 86 3.2.3.9. Immunocytochemistry for detection of protein and apoptosis ...... 86 3.2.3.10. In vitro invasion assay ...... 87 3.2.4. In vivo animal model and assays ...... 87 3.2.4.1. Orthotopic mouse model ...... 87 3.2.4.2. X-ray radiography and microtomography imaging ...... 89 3.2.4.3. Histochemical and immunohistochemical analysis ...... 90 3.2.4.4. Detection of osteoclastic activity with TRAP staining ...... 91 3.2.4.5. Detection of human Alu DNA with in situ hybridization ...... 92 3.2.5. Statistical analysis ...... 92 ix

Chapter 4. New Gene Groups Associated with Dissimilar Osteoblastic Differentiation Linking to Osteosarcomagenesis ...... 93

4.1. Background ...... 93

4.2. Hypothesis and aim ...... 95

4.3. Methods ...... 95

4.4. Results ...... 96

4.4.1. Osteogenic induction properties of Saos-2 and U-2 OS ...... 96 4.4.2. Detection of the quality of extracted RNA ...... 98 4.4.3. Confirmation of the quality of microarray data ...... 102 4.4.4. Identification of 75 differentially expressed genes ...... 105 4.4.5. Verification of microarray data by qRT-PCR and immunocytochemistry ...... 112 4.4.6. Differential transcriptional protein expression in the dissimilar human osteosarcoma cell lines ...... 114

4.5. Discussion ...... 118

4.6. Conclusion ...... 124

Chapter 5. Association of Ras/Raf/MEK/ERK Pathway with Lung Metastasis in an Orthotopic Mouse Model of Osteosarcoma ...... 126

5.1. Background ...... 126

5.2. Hypothesis and aim ...... 128

5.3. Methods ...... 128

5.4. Results ...... 129

5.4.1. Tumour growth and metastasis: gross examination ...... 129 5.4.2. Radiographic characterizations of the mouse model of osteosarcoma 131 5.4.3.  Microscopic examination of the primary and the secondary xenograft tumours ...... 133 5.4.4. Detection of the IGF-1R signalling pathway in the mouse model of osteosarcoma ...... 137 5.4.5. Regulation of invasion by inhibition of MEK/ERK under in vitro condition ...... 141

5.5. Discussion ...... 142

5.6. Conclusion ...... 148

Chapter 6. IGF-1R Targeted Therapy and Its Enhancement of Chemosensitivity in Human Osteosarcoma Cell Lines ...... 149 x

6.1. Background ...... 149

6.2. Hypothesis and aim ...... 150

6.3. Methods ...... 151

6.4. Results ...... 151

6.4.1. IGF-1R expression in osteosarcoma cell lines ...... 151 6.4.2. Determination of growth condition in osteosarcoma cell lines for drug treatment analysis ...... 152 6.4.3. Anti-proliferation effects of Tyrphostin AG1204 on osteosarcoma cell lines ...... 153 6.4.4. Anti-proliferation effects of Doxorubicin on osteosarcoma cell lines .... 156 6.4.5. Anti-proliferation effects of the combination treatment with Tyrphostin AG1024 and Doxorubicin on osteosarcoma cell lines ...... 158 6.4.6. Synergy analysis of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on osteosarcoma cell lines ...... 163 6.4.7. Mechanism studies on selected osteosarcoma cell lines from the combination treatment with Tyrphostin AG1024 and Doxorubicin ...... 168 6.4.7.1. Cytotoxic effect of the combination treatment ...... 168 6.4.7.2. Clonogenic inhibition by the combination treatment ...... 170 6.4.7.3. Cell cycle arrest induction by the combination treatment ...... 171 6.4.7.4. Apoptosis induction by the combination treatment ...... 173 6.4.8. Effectiveness of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on a Doxorubicin-resistance osteosarcoma cell line . 175 6.4.8.1. Anti-proliferation effects of Tyrphostin AG1024 and Doxorubicin mono drug treatment on a Doxorubicin-resistant osteosarcoma cell line ...... 176 6.4.8.2. Synergy analysis of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on a Doxorubicin-resistant osteosarcoma cell line ...... 177

6.5. Discussion ...... 180

6.6. Conclusion ...... 186

Chapter 7. Conclusion and Future Directions ...... 188

7.1. Conclusion ...... 188

7.2. Future studies ...... 192

Chapter 8. Appendixes ...... 194

Appendix 1: List of materials used and solution recipes...... 194 xi

Appendix 2: List of 629 genes identified after functional annotation analysis ...... 200

Appendix 3: The unsupervised hierarchical clustering analysis of gene expression in Saos-2 and U-2 OS compared with five normal human cells*...... 219

Chapter 9. References ...... 222 xii

List of figures

Figure 2.1 Osteosarcoma ...... 5

Figure 2.2 Schematic model of somatic events that is important to the molecular pathogenesis of osteosarcoma ...... 14

Figure 2.3 Pathway of TP53 and RB...... 18

Figure 2.4 Schematic summary of the activities and regulations of transcription factors during differentiation and maturation within osteoblastic lineages starting from a mesenchymal precursor cell ...... 27

Figure 2.5 Model showing the interaction between cell cycle protein and RUNX2 during osteoblast differentiation ...... 28

Figure 2.6 Affymetrix™ microarray ...... 31

Figure 2.7 Microarray technology are involved in the three stages of cancer research ...... 33

Figure 2.8 The molecular structure of IGF-1R ...... 44

Figure 2.9 Endocrine regulation and function of IGF-1 ...... 46

Figure 2.10 Signal transduction pathways of the ligand-activated IGF-IR ...... 47

Figure 2.11 Various strategies for inhibition or down-regulation of the IGF-1R signalling pathways and the mode of action ...... 52

Figure 2.12 Molecular structure and formula of Tyrphostin AG1024 ...... 54

Figure 2.13 Mechanism and mode of action of chemotherapeutic drugs commonly used in cancer treatments ...... 58

Figure 2.14 Depicted are some of the novel agents (indicated in boxes) targeting the cellular signalling pathways ...... 61

Figure 3.1 Dissociation curve analysis of GAPDH qRT-PCR primer ...... 77

Figure 3.2 Standard curve analysis of GAPDH qRT-PCR primer ...... 78

Figure 3.3 Surgical procedures for the establishment of an orthotopic mouse model of osteosarcoma ...... 89

Figure 4.1 Osteogenic induction properties of Saos-2 and U-2 OS ...... 98

Figure 4.2 Quality assessment of the RNA samples by Bioanalyzer™ ...... 101

Figure 4.3 Graphical illustration showing the consistency of the gene expression data between the six samples ...... 104 xiii

Figure 4.4 Differentially expressed gene in Saos-2 compared to U-2 OS ...... 106

Figure 4.5 Gene ontology and functional annotation clustering were used to classify 1,968 differentially expressed genes ...... 107

Figure 4.6 The hierarchical clustering analysis showing the average signal intensity of the identified 75 differentially expressed genes in Saos-2 compared to U-2OS after several analyses ...... 111

Figure 4.7 Verification of microarray results by qRT-PCR ...... 113

Figure 4.8 Verification of microarray results by immunocytochemistry ...... 114

Figure 4.9 Differential transcriptional regulation in Saos-2 and U-2 OS ...... 116

Figure 5.1 Local tumour growth at the left tibia at week 1 (A), week 3 (B) and week 6 (C) post osteosarcoma cells inoculation...... 130

Figure 5.2 Tumour growth curve of six weeks post inoculation ...... 130

Figure 5.3 Gross examination of the tumour at the primary sites and lung ...... 131

Figure 5.4 X-ray images of the hind limbs of the mice at 2, 4 and 6 weeks post inoculation ...... 132

Figure 5.5 Micro-CT images of the hind limbs of the mice at 2 weeks post inoculation ...... 133

Figure 5.6 Histochemical analysis of the hind limbs of the mice with human osteosarcoma cell inoculation ...... 135

Figure 5.7 TRAP staining of the left tibias of the mice at 2, 4 and 6 weeks post inoculation ...... 136

Figure 5.8 Histochemical analysis of the lungs of the mice at 4 and 6 weeks post inoculation...... 136

Figure 5.9 Detection of human Alu gene expression in the mice at 4 and 6 weeks post inoculation...... 137

Figure 5.10 Detection of the IGF-1R signalling proteins and their phosphorylated forms in the mouse model of osteosarcoma ...... 139

Figure 5.11 Detection of the IGF-1R signalling proteins and their phosphorylated form in the chamber slide cultured 143B osteosarcoma cell line...... 140

Figure 5.12 Representative images of sample at 1:40 Matrigel™ dilution from the invasion assay ...... 142

Figure 5.13 The three major Ras signalling pathways in IGF-1R...... 144

Figure 6.1 Western blot analysis of IGF-1R expression in osteosarcoma cell lines ...... 152

Figure 6.2 Growth rate of the osteosarcoma cell lines ...... 153 xiv

Figure 6.3 Dose effect curves of the osteosarcoma cell lines from Tyrphostin AG1024 treatment ...... 155

Figure 6.4 Dose effect curves of the osteosarcoma cell lines from Doxorubicin treatment ...... 157

Figure 6.5 Growth inhibition comparison in osteosarcoma cell lines after mono drug and combination treatment at a specific dosage ...... 163

Figure 6.6 Synergy analysis of the osteosarcoma cell lines after combination treatment ...... 164

Figure 6.7 Dose effect curves and isobologram analysis in the osteosarcoma cell lines after combination treatment ...... 166

Figure 6.8 Cytotoxic effect of the combination treatment in the selected osteosarcoma cell lines ...... 169

Figure 6.9 Clonogenic survival status in the selected osteosarcoma cell lines after combination treatment ...... 171

Figure 6.10 Cell cycle distribution analysis in the selected osteosarcoma cell lines after combination treatment ...... 173

Figure 6.11 Apoptotic effect of the combination treatment in the selected osteosarcoma cell lines ...... 175

Figure 6.12 Dose effect curves on the 143B-Dox-400 Doxorubicin-resistant osteosarcoma cell line ...... 176

Figure 6.13 Dose effect curves and isobologram analysis in 143B-Dox-400 osteosarcoma cell line after combination treatment ...... 178 xv

List of tables

Table 2.1. Classification of osteosarcoma based on the WHO histological classification of bone tumour ...... 7

Table 2.2. The two commonly use osteosarcoma staging system ...... 8

Table 2.3. The multidisciplinary teams with important experts [16]...... 9

Table 2.4. Most effective chemotherapeutic agents currently used in the osteosarcoma treatments ...... 11

Table 2.5. Novel targets and therapies of osteosarcoma ...... 13

Table 2.6. Compounds in clinical and preclinical developments that target the IGF- 1R and their application in combination therapy ...... 63

Table 3.1. Machine setting and primers information of qRT-PCR...... 79

Table 4.1. RNA sample quality assessment by the ratio of absorbance ...... 99

Table 4.2. Five different criteria were compared amongst the six microarrays and used to confirm the integrity of the gene expression data ...... 102

Table 4.3. The list of 75 differentially regulated genes in Saos-2 compared to U-2 OS ...... 109

Table 4.4. Expression of genes related to transcription factors ...... 117

Table 5.1. Semi quantification of the total protein expression of the three IGF-1R related signalling pathway factors in the primary tumour, lung metastasis and 143B osteosarcoma cells ...... 141

Table 5.2. Effect of MEK/ERK inhibitor U0126 on invasion by 143B osteosarcoma cells ...... 142

Table 6.1. The “constant ratio two drug combination” drug combination treatment design model...... 159

Table 6.2. Drug combination treatment design model for Doxorubicin resistant osteosarcoma cell line ...... 178

Table 6.3. Drug sensitivity (IC50 values) of the 6 osteosarcoma cell lines and Doxorubicin resistant cell line...... 179 xvi

Abbreviation

AJCC American Joint Committee on Cancer

Akt Serine/threonine protein kinase

ALP Alkaline phosphatase

ATP Adenosine triphosphate

ATCC American tissue culture collection

BAD Bcl-2-associated death promoter

BMP Bone morphogenic protein

CDK Cyclin-dependent kinase cDNA Complementary deoxyribonucleic acid

CGH Comparative genomic hybridization

CT Computed tomography

CTGF Connective tissue growth factor

DNA Deoxyribonucleic acid

EGFR Epidermal growth factor receptor

ERBB2 v-erb-b2 erythroblastic leukemia viral oncogene homolog 2, neuro/glioblastoma derived oncogene homolog

ERK Extracellular signal-regulated kinase

FGF Fibroblast growth factor

FISH Fluorescent in situ hybridization

FOS FBJ murine osteosarcoma viral oncogene homolog

GDP Guanosine diphosphate

GTP Guanosine triphosphate

IGF-1 Type I insulin-like growth factor

IGF-1R Type I insulin-like growth factor receptor

IGF-2R Type II insulin-like growth factor receptor xvii

IGFBPs Insulin-like growth factor binding proteins

IR Insulin receptor

IRS Insulin receptor substrates

JAK Janus kinase

MAPK Mitogen-activated protein kinase

MDM2 Mouse double-minute 2 protein

MDR Multi-drug resistance gene

MEK Mitogen-activated protein kinase or extracellular signal-regulated kinase kinase

MET (HGF) Met proto-oncogene (hepatocyte growth factor receptor)

MMPs Matrix metalloproteinases

MRI Magnetic resonance imaging mRNA Messenger ribonucleic acid mTOR Mammalian target of rapamycin

MYC v-myc myelocytomatosis viral oncogene homolog

NICD Notch intracellular domain

OPN Osteopontin

OS Osteosarcoma

PCR Polymerase chain reaction

PDGF Platelet-derived growth factor

PDGFR Platelet-derived growth factor receptor

PET Positron emission tomography

PI3K Phosphatidylinositol 3-kinase

PPAR" Peroxisome proliferator-activated receptor gamma

PTEN Phosphatase and tensin homolog

PTH Parathyroid hormone

PTHrP Parathyroid hormone related peptide qRT-PCR Quantitative reverse transcription polymerase chain reaction xviii

RB Retinoblastoma

RNA Ribonucleic acid

RTK Tyrosine kinase receptor

RUNX2 Runt-related transcription factor 2 siRNA Small interference ribonucleic acid

SMAD Protein modulate the activity of transforming growth factor beta ligands

SP7 Transcription factor SP7 (Osterix)

SSS The Surgical Staging System

STAT Signal transducers and activators of transcription proteins

TAZ Transcriptional modulator TAZ

TP53 (p53) Tumour protein 53

TGF-! Transforming growth factor beta

VEGF Vascular endothelial growth factor

WHO World Health Organization

Wnt Wingless-type MMTV integration site family xix

List of publications

Journal articles

Luk, F., Yu, Y., Kuang, J., Walsh, W.R., Yang, J. “IGF-1R targeted treatment and its enhancement of chemosensitivity in human osteosarcoma cells” (1st version submitted to Cancer Letters)

Luk, F., Yu, Y., Dong, H., Walsh, W.R., Yang, J., “Gene expression profiling of the two human osteosarcoma cell lines with dissimilar tumour differentiation and osteogenic property” (1st version submitted to Journal of Cancer Research and Clinical Oncology)

Yu, Y., Luk, F., Yang, J., Walsh, W.R., “Ras/Raf/MEK/ERK pathway is associated with lung metastasis of osteosarcoma in an orthotopic mouse model” (2nd revised version submitted to Journal of Orthopaedic Research)

Conferences and meetings

Yu, Y., Luk, F., Chen, J.B., Yang, J., Walsh, W.R., “Gene expression profiles associated with osteogenesis and/or tumorigenesis in representative osteosarcoma cell lines”, 17th Scientific Meeting of the International Bone and Mineral Society, Montreal,

Canada, June 2007.

Walsh, W.R., Luk, F., Chen, J.B., Yang, J., Yu, Y., “Osteogenesis in representative human osteosarcoma cell lines”, 6th Combined Meetings of the Orthopaedic Research

Society, Honolulu, Hawaii, October 2007. xx

Luk, F., Yu, Y., Walsh, W.R., Crowe, P.J., Yang, J., “IGF-1R inhibitor Tyrphostin

AG1024 monotherapy and combine with doxorubicin in treatment of osteosarcoma (OS) cell lines”, The Australian Health and Medical Research (AHMR) Congress 2008,

Brisbane, Queensland, Australia, November 2008.

Yu, Y., Luk, F., Yang, J., Walsh, W.R., “Phosphorylated MEK associated with lung metastasis in an orthotopic mouse model of human osteosarcoma”, 56th Annual

Meeting of the Orthopaedic Research Society, New Orleans, Louisiana, USA, March

2010.

Luk, F., Yu, Y., Walsh, W.R., Crowe, P.J., Yang, J., “Anti-proliferation effect of the combination therapy using IGF-1R inhibitor Tyrphostin AG1024 and doxorubicin in human osteosarcoma cell lines”, The 6th Congress of Asian Society for Pediatric

Research, Taipei, Taiwan, April 2010.

Luk, F., Yu, Y., Walsh, W.R., Yang, J., “IGF-1R targeted therapy and its enhancement of chemosensitivity in human osteosarcoma cell lines”, 57th Annual Meeting of the

Orthopaedic Research Society, Long Beach, California, USA, January 2011. 1

CHAPTER 1. INTRODUCTION 1

1.1. Introduction

Osteosarcoma is the most common primary malignant tumour of the bone with poor survival due to early and frequent metastasis, and is recently regarded as a differentiation disease with limited understanding. Many genetic factors are known to contribute to tumorigenesis of osteosarcoma and act at different stages of osteoblast differentiation. Current progress in research technologies such as gene expression

DNA microarray, which enable comparing gene profiles from the “osteoblast-like cells” with dissimilar differentiation status, will help to improve the understanding on osteosarcomagenesis.

Similar to many other cancers, the development of tumour in osteosarcoma is a result of the imbalance between cell growth and cell death. Cell growth signals such as those from the insulin-like growth factor receptor (IGF-1R) play an important role in promoting malignant potential, which can be demonstrated by biological marker detection and protein function regulation using animal model. Furthermore, the IGF-1R signalling is associated with chemoresistance in cancers. Thus, once the independent relationship between activation of the IGF-1R signalling and malignant potential as well as chemoresistance of osteosarcoma is verified, the current clinical situation in the treatment of osteosarcoma can be improved by using specific molecular IGF-1R targeted mono-therapy or combination therapy with chemotherapeutic agent, such as

Doxorubicin. 2

1.2. Hypotheses and aims

The principal aim of this study was to improve the knowledge of tumour development and progression as well as the current treatment outcome of osteosarcoma. Firstly, it was hypothesised that studying the global gene expression of the two human osteosarcoma cell lines with the dissimilar differentiation status would enhance the understanding in the association between osteoblast differentiation and osteosarcoma tumorigenesis. Secondly, IGF-1R/MEK/ERK signalling would be important for malignant potential of osteosarcoma, particularly in in vivo lung metastasis of the disease. Lastly, targeting IGF-1R alone or in combination with Doxorubicin chemotherapy would promote favourable anti-proliferative effect on osteosarcoma.

Mechanism studies helped to improve the understanding and verified the potential implication of this IGF-1R targeted combination therapy in osteosarcoma treatment.

The specific objectives of the study are divided into three parts:

To perform gene expression profiling with Affymetrix™ microarray on the two human osteosarcoma cell lines (Saos-2 and U-2 OS) that are at different stages of osteoblast differentiation, followed by identification of the essential group of genes that distinguishes the osteogenic characteristics between these two human osteosarcoma cell lines.

To set up an orthotopic mouse model of osteosarcoma and investigate IGF-

1R/MEK/ERK signalling involvement in malignant potential, followed by investigation of the effect of down-regulation of MEK/ERK activity by the MEK/ERK inhibitor U0126 on in vitro invasive ability of the osteosarcoma cell line used in in vivo model.

To investigate the effect of mono-drug therapy with the IGF-1R inhibitor Tyrphostin

AG1024 or Doxorubicin, and in combination of both drugs on a panel of 6 human 3

osteosarcoma cell lines and a Doxorubicin resistant osteosarcoma sub-line, followed by investigating the mechanisms of the combination therapy. 4

CHAPTER 2. LITERATURE REVIEW 2

2.1. Clinical features of osteosarcoma

2.1.1. Incidence and mortality

Osteosarcoma (OS) is the most common primary malignant tumour of bone with higher rates of occurrence in adolescence and old age, and is more frequently observed in males [1,2]. This cancer commonly arises in the metaphysis of long bones, such as the distal femur, proximal tibia and proximal humerus regions where rapid bone growth occurs in younger patients, as well as in axial locations which have undergone irradiation or with bone abnormalities in elderly patients [3,4].

The occurrence of osteosarcoma is low, contributing to 3-5% of all childhood cancers and less than 1 % of adult cancers diagnosed in the United States between 1973-2004

[5]. Similarly in Australia, the statistical data from the Australia Institute of Health and

Welfare [AIHW] shows malignancies arising in bone and cartilage account for about

0.2% of new cancers between 1982-2005 (Cancer Data Online, AIHW). The epidemiologic features and anatomic site distributions of osteosarcoma are different among paediatric, middle-age and elderly age groups[6]. Such differences are the factors contributing to the variations in survival rates. The rarity and complexity of osteosarcoma has limited the study of this disease. Therefore, more effort is required to expand our knowledge and understanding about the etiology of osteosarcoma. 5

The use of combination chemotherapy and surgery has significantly improved the clinical outcomes over the past decades. However, patients with metastatic and recurrent disease still have poor prognosis, with about 30% long-term (disease-free for

5-years or more) survival compared with approximately 70% for the patients with non- metastatic osteosarcoma [2,7]. Radiotherapy and dose intensification of chemotherapy show no further improvement of clinical benefits to osteosarcoma patients but induce development of drug resistance and toxic side effects [2,8-10]. It is important to identify and develop new therapeutic strategies to increase the effectiveness of the current therapeutic modalities and enhance the survival of osteosarcoma patients.

Figure 2.1 Osteosarcoma. (A) Schematic diagram of an osteosarcoma at the femur [11], (B) Resection of an osteosarcoma from distal femur of a patient [12], (C) Radiographic illustration of an osteosarcoma at the proximal tibia. Regions with osteogenic characteristic are shown in the red circled area and osteolytic parts are indicated with green arrows [13].

2.1.2. Classification

Osteosarcoma is the malignant tumour of connective tissue of mesodermal origin within which is characterised by the formation of bone or osteoid (often referred to as

“tumour bone” or “tumour osteoid”) directly from the tumour cells [14]. According to Klein and Siegal (2006), osteosarcoma has the tendency to produce variable amount of cartilage matrix and fibrous tissue and may differentiate along all or any of these matrix 6

synthesis pathways[15]. Therefore, the histological characteristic of osteosarcoma is more similar to fracture callus or fibrous dysplasia than to differentiated bone matrix- producing tumours, such as osteoid osteomas or osteoblastomas. As a result, osteosarcomas are traditionally subdivided into osteoblastic, chondroblastic and fibroblastic osteosarcoma. However, most osteosarcoma display all three histological types in various amounts, therefore osteosarcoma is classified according to the type that is predominant in each case [15].

Osteosarcoma frequently arise from inside of the bones (in the intramedullary or intracortical compartment), but may also arise on the surfaces of bones, and in extraosseous sites. Osteosarcomas from medullary origin are usually of high grade, while the majorities from surface of the bone are of low grade. The World Health

Organization (WHO) has developed a classification system for bone cancer to distinguish different types of osteosarcoma [14]. Due to the variation of extracellular matrix, degree of differentiation and histological pattern between cases, the classification of osteosarcomas is determined by various factors, such as the predominant histological pattern, anatomic location, associated bone, and sometimes histological grade or related disease entity (Table 2.1). However, this classification system is not suitable for general surgical pathologist to make the diagnosis in such a variable disease.

On the other hand, two osteosarcoma staging systems are frequently utilised by clinical practitioners to diagnose the condition of patients before biopsy and making treatments decision. The American Joint Committee on Cancer (AJCC) has developed the TNM

(Tumour, Node, Metastasis) staging system for bone sarcoma is based on assessment of histologic grade, tumour size, presence of regional and/or distant metastases of the tumour (Table 2.2 A) [16]. The Surgical Staging System (SSS) developed by Enneking et.al (1980) that stratifies bone and soft tissue lesion is based on assessment of the 7

surgical grade, the local extent and the presence or absence of regional or distant metastases (Table 2.2 B) [17]. The two systems contain some distinctive criteria. Using both systems together to diagnose the condition of patients will provide better categorization of the tumour for clinical practitioners.

Table 2.1. Classification of osteosarcoma based on the WHO histological classification of bone tumour. The table is adapted from Klein and Siegal (2006) [14,15].

Type Grade

Central High grade: (Medullary) - Conventional (osteoblastic, chondroblastic, fibroblastic) - Telangiectasic - Small cell - Giant cell-rich - Epithelioid Low grade: - Intraosseous well-differentiated (fibrous dysplasia-like, desmoplastic fibroma-like)

Surface High grade: - High-grade surface - Dedifferentiated parosteal Low grade: - Parosteal (juxtacortical) - Periostea: low-grade to intermediate-grade osteosarcoma

Other - Intracortical - Gnathic - Extraskeletal - High grade - Low grade - Disease associated osteosarcoma - Osteosarcoma in Paget’s disease - Osteosarcoma in fibrous dysplasia - Osteosarcoma in Mazabraud’s disease - Postirradiation osteosarcoma 8

Table 2.2. The two commonly use osteosarcoma staging system

(A) AJCC - TNM staging system for bone sarcoma [16]

Stage Description

IA The low grade tumour that is confined to bone and is less than 8 cm

IB The low grade tumour that is confined to bone and is larger than 8 cm

IIA The high grade tumour that is confined to bone and is less than 8 cm

IIB The high grade tumour that is confined to bone and is larger than 8 cm

III The tumour in any grade that is confined to the bone but has "skipped" to other sites on the bone

IVA The tumour in any grade with any size that has spread to the lung

IVB The tumour in any grade with any size that has spread to lymph nodes and distant sites, or has spread to distant sites other than the lung

(B) Surgical staging system [17]

Stage Description

IA Low grade, intracompartmental, with no metastases

IB Low grade, extracompartmental, with no metastases

IIA High grade, intracompartmental, with no metastases

IIB High grade, extracompartmental, with no metastases

III Any grade, any site, with metastases

2.1.3. Current diagnosis, treatment and management

Osteosarcoma is a rare and complex disease. Affected patients should be evaluated and treated appropriately in specialised centres with proper diagnostic tools by a multidisciplinary team with expertise to provide a full spectrum of care (Table 2.3). The team should work closely to ensure patients benefiting from modern, efficacious, 9

interdisciplinary therapeutic regimens and from optimisation of these regimens.

Inappropriate diagnosis and treatment will greatly reduce the chance of patient’s recovery and survival [16].

Table 2.3. The multidisciplinary teams with important experts [16].

Core team member Members required in certain case

Orthopaedic oncologist Thoracic surgeon

Bone pathologist Plastic surgeon

Medical/Paediatric oncologist Interventional radiologist

Radiation oncologist Physiatrist

Musculoskeletal radiologist Vascular surgeon

Additional surgical subspecialties

During the osteosarcoma development, patients may experience several common clinical presentations. For example, an increase of dull or aching pain that last for several months is caused by intramedullary tumour penetration and periosteum irritation, or pathological fracture. Pains that disturb sleeping at night, frequent history of minor trauma and injury, sprain or muscle strain during sports or exercise are some other symptoms of osteosarcoma. Local tenderness, mass formation, swelling, deformity, limp or muscle atrophy and restricted joint motion uncover by physical examination are also common symptoms considered in the diagnosis of osteosarcoma patients [18].

Radiological examination is essential for the assessment of primary and metastatic osteosarcoma. Modern imaging technologies have advanced from plain-film radiography to computed axial tomography (CT) and magnetic resonance imaging

(MRI). These latest radiographic technique enhances the process of defining the extent 10

of the neoplasm especially when it has extended into bone marrow and soft tissue [19].

Both CT and MRI have been proven and applied as a standard procedure to clarify the tumour extent at the time of surgical resection [20]. MRI is currently one of the accurate methods for determining (i) the intraosseous extent of the tumour, (ii) the presence of skip metastases, and (iii) the involvement and invasion of adjacent muscle groups, subcutaneous fat, joints and major neurovascular structures [21].

Biopsy is the final and the key step to assess patients with osteosarcoma. It should be performed properly to avoid misdiagnosis, amputation and local recurrence that will reduce the rate of recovery and survival [19]. Caution should be taken to avoid impending pathological fracture during sample collection with either core needle or surgical biopsy techniques [16]. All procedures should be performed by an orthopaedic oncologist or a physician accompanied by an oncologist [22].

Multiple modalities and carefully planned regimens are required in osteosarcoma treatment to cure patients successfully. Before the introduction of chemotherapy to treat osteosarcoma, amputation or radiotherapy was used with low survival rates at about 12% and 75% of diagnosed patients died within 2 years [23]. Nowadays, the standard multi-modal approach, including neoadjuvant (preoperative) systemic polychemotherapy followed by a proper local surgical resection and then adjuvant

(postoperative) chemotherapy have enhanced the long-term disease-free survival of patients [24,25]. However, in some special cases of unresectable lesions or undesirable functional outcomes after surgery, radiotherapy may be used [26].

Chemotherapy have achieved 5 years disease free survival for up to 70% of diagnosed osteosarcoma patients [27]. Doxorubicin, Cisplatin and high dose of Methotrexate are commonly used in the standard osteosarcoma treatment protocol and occasionally

Ifosfamide is supplemented to improve the clinical results (Table 2.4). Unfortunately, 11

optimised multi-agent regimens and dose intensification have no further enhancement on the survival of patients, and metastatic patients still have poor prognosis. Therefore, new therapeutic approach is required for further improving the current situation.

Table 2.4. Most effective chemotherapeutic agents currently used in the osteosarcoma treatments [22,28].

Agent Type of Mode of action Side effects drug

Doxorubicin Cytotoxic Intercalates at point of local Cardiomyopathy, transient (Adriamycin) anthracycline uncoiling of the DNA double electrocardiographic antibiotic helix and inhibits the abnormalities, emesis, synthesis of DNA and RNA alopecia, mucositiis, myelosuppression

Cisplatin Platinum- Inhibits the synthesis of DNA Acute renal failure, chronic (Platinol) containing through the formation of DNA renal failure, peripheral complex cross-links and binds directly neuropathy, ototoxicity, to tumour DNA and denatures emesis, myelosuppression, the DNA double helix alopecia, hypomagnesemia

High-dose Anti- Inhibits synthesis of purine Renal failure, mucositis, mild Methotrexate metabolite and thymidylic acid by binding myelosuppression, central (Rheumatrex) dihydrofolate reductase nervous system effects (rarely)

Ifosfamide Alkalating Causes cross-link of DNA Hemorrhagic cystitis, renal (Ifex) agent strands and inhibiting the failure, myelosuppression, synthesis of DNA and protein alopecia, emesis, encephalopathy

Surveillance for recurrence is essential in all osteosarcoma patients after treatments and periodic screening may last for up to 10 year or longer if required [29]. In some cases with patients of high risk of relapse, functional reassessment and general diagnosis procedure is required for every visit. Bone tumour necrosis is used to assess the effectiveness of chemotherapy and as a prognostic factor to evaluate disease-free survival and overall survival of osteosarcoma patients. A classification system is 12

normally used to determine tumour necrosis status: Grade I – No necrosis; Grade II –

Necrosis between 50-95%; Grade III – Necrosis over 95%; and Grade IV – Total necrosis, 100% [30]. Moreover, it is important to monitor the cumulative radiation doses to the patient if radiation associated imaging methods such as positron emission tomography (PET) and computed tomography (CT) scanning are used. This is particularly critical to paediatric patients when an extensive amount of radiation is presented to the whole body [31]. Repetitive treatment is required for patients with relapse condition and alternative treatment options may be required for patients with progressive disease condition.

Advances in molecular and medicine development, ongoing research in tumour biology and increased knowledge in pathogenesis have helped to identify novel target and develop novel treatment strategies in osteosarcoma. Various forms of radiotherapy is currently available for osteosarcoma patients, such as conformal radiotherapy, proton- beam radiotherapy and specific organ-seeking radioisotopes. The high effectiveness and minimal side-effects of these techniques are the result of (i) careful management of applied dosage, (ii) high accuracy and precision of radiation dose to the tumour mass, and (iii) with the assistance of MRI and CT [28]. Other new directions for the treatment of osteosarcoma include the use of monoclonal antibody, small molecule inhibitors or gene therapy (ribozymes, siRNA or deoxyribonucleotides) to target tumour growth, bone metabolic markers, metastasis process or metastatic lesion [23,32] (Table 2.5). 13

Table 2.5. Novel targets and therapies of osteosarcoma [28,32-34].

Target Therapeutic Agent Development Therapeutic mechanism agent/class example status IGF-1 receptor Antibody, kinase CP-751,851 Preclinical / Inhibition of cell proliferation inhibitor Insm-18 clinical and survival H7C10

RANK Bisphosphonate Pamidronate, Preclinical / Inhibition of osteoclast Zoledronate clinical activity

Osteoblast / Bone-seeking Samarium-153 Preclinical / Reside at bone and inhibit bone radionucleotide clinical cancer invasion

VEGF receptor Antibody, kinase Many Preclinical / Inhibition of angiogenesis inhibitor clinical during metastatic progression and lesion

Src kinase Kinase inhibitor Dasatinib, Preclinical / Inhibition of osteoclast AZD0530 clinical activity and cancer invasion

Immune Macrophage MTP-PE Clinical Inhibition of monocytes and activation activator macrophages c-Met Kinase inhibitor XL880 Preclinical Inhibition of signalling pathway for cell motility, invasion, proliferation and survival

CXCR4 Competitive CTCE-9908 Preclinical Inhibition of chemokine peptide inhibitor pathway for cell adhesion and survival

Mammalian Competitive or Rapamycin, Preclinical / Inhibition of signalling target of kinase inhibitor rapalogues clinical pathways maybe involved in rapamycin cell growth, proliferation, survival, protein synthesis and transcription.

Heat shock Competitive 17-AAG Preclinical / Inhibition of cellular stress protein 90 inhibitor clinical response protection uPA/uPAR Deoxyribozymes Dz483, Dz720 Preclinical Inhibition of extracellular matrix degradation in cancer invasion 14

2.2. Molecular pathogenesis of osteosarcoma

The study of osteosarcoma molecular pathogenesis is often limited by the rarity of this disease and diagnosis at a late stage, which molecular cytogenetic alterations are accumulated and complicated. The development in medical sciences in the last decades revealed that understanding the inter-relationship between clinical and molecular information is important to improve osteosarcoma management and in other cancer [35,36]. The ongoing research in osteosarcoma with the help of latest technologies will improve the understanding of this disease, which is also contribute to the knowledge of cancer biology, and advance the current cancer treatments. For example, the function of RB in bone, the mechanism of RB in promoting tumorigenesis, and the overexpression of oncogene in related to metastases and tumour progression, are all the questions remain to be answered (Figure 2.2).

Figure 2.2 Schematic model of somatic events that is important to the molecular pathogenesis of osteosarcoma. Potential pathways that may be altered during this process are listed in the boxes. The exact timing of these events is unknown and proposed by the broken lines. Specific interactions between the pathways at different stage have not been identified. Osteosarcomas are mostly diagnosed at a late stage. Their molecular cytogenetic alterations were accumulated and become complicated. Diagram modified from Wang(2005) [37]. 15

Osteosarcoma is correlated with several genetic predispositions without familial pattern and displays a wide range of genetic and epigenetic alterations with no consensus changes identified [4,37-39]. Cytogenetic and molecular studies have demonstrated its complex karyotypes of numerical and structural chromosomal aberrations and abnormalities with substantial cell-to-cell variation and heterogeneity. Aneuploidy and lack of specific translocation are some of the characteristics commonly found in osteosarcoma [40-42]. Advances in molecular technology, such as fluorescent in situ hybridization (FISH) and comparative genomic hybridization (CGH), have helped to discover several recurrent breakpoint clusters, non-recurrent reciprocal translocations, most frequently amplified chromosomal regions (6p12-p21: 28%, 17p11.2: 32% and

12q13-q14: 8%), and also the chromosomal regions that commonly found losses (2q,

3p, 9, 10p, 12q, 13q, 14q, 15q, 16, 17p and 18q) and gains (Xp, Xq, 5q, 6p, 8q, 17p and 20q) in osteosarcoma [41]. Identification of these regions has helped to investigate the relative gene or genes that may play a role in osteosarcomagenesis and have emphasized the intricacy and variability in the genetic makeup of osteosarcoma.

2.2.1. Tumour suppressor genes

Formation of tumour occurs when the expression of growth-promoting genes or proteins is deregulated. Likewise, development of tumour arises with abnormal regulation or uncontrollable expression of genes or proteins that control cell growth, differentiation and cell death. The genes that are responsible for the inhibition of such disorderly cell proliferations are known as tumour suppressor genes.

Retinoblastoma

Patients are at higher risk of developing osteosarcoma with germline inactivation of retinoblastoma (RB) gene. Gene alteration is common in sporadic osteosarcoma 16

patients. About 50% to 70% of cases showed heterozygous loss of the chromosomal region 13q14 that comprised the retinoblastoma 1 (RB1) locus, while structural rearrangements and point mutations are seen in about 30% and 10% of cases, respectively [43-47]. RB1 alterations are more commonly found in the high-grade than the low-grade osteosarcoma and indicated as a poor prognostic factor [39,41,48] .

RB is a major regulator of G1 to S phase progression in the cell cycle, and is one of the earliest tumour suppressors to be cloned [49,50]. During G1/S phase cell cycle transition, cyclin-D1 forms a complex with cyclin-dependent kinase – 4 or 6 (cyclin D-CDK4/6) and phosphorylated RB, which leads to the activation and release of E2F transcription factor to promote DNA synthesis [37,38]. Any disruption or alteration of cyclin-D1 or

CDK4/6 will result in functional inactivation of the RB pathway. RB is also involved in controlling the balance of cell cycle, mediating the differentiation signals and modulating the chromosome segregation during mitosis [51]. Disrupting the function and balance of the RB pathway could possibly interfere with proliferation and disturb genomic stability, thus leads to osteosarcoma development [52].

TP53

Another frequently mutated tumour suppressor gene in osteosarcoma is tumour protein

53 (TP53, also known as p53), which is encoded from 17p13 chromosomal region.

TP53 is a 53-kDa nuclear phosphoprotein in response to cellular stresses (DNA damage, aberrant proliferative signals, heat shock or hypoxia), controls apoptosis and regulates cell cycle [41,50,53]. Mutation of TP53 was found in about 40%-60% of high- grade osteosarcoma compared to about 1% in low-grade osteosarcoma [54,55].

Mechanisms of TP53 inactivation in osteosarcoma involve allelic loss (70%-80%), point mutation (20%-30%) and gene rearrangement (10%-20%), whereas point mutations are predominantly missense mutations whose products formed heterodimers with and thereby inactivate normal p53 molecules [56,57]. Germline mutations of TP53 were found 17

in about 3% of sporadic osteosarcoma cases, and showed higher risk for developing osteosarcoma in patients with Li-Fraumeni syndrome [58,59]. This was one of the important evidence revealed the association between TP53 and osteosarcoma.

The TP53 and RB pathways are associated with cell cycle regulation [60]. Activation of

TP53 from cellular stress will up-regulates p21WAF/CIP, which binds and inhibits the activities of the cyclin D-CDK 4/6 or cyclin E-CDK2 complexes, thereby decreases RB phosphorylation and causes G1 arrest in cells [38]. Hence, inactivation of TP53 by the direct binding of mouse double-minute 2 protein (MDM2, also known as HDM2 for the human ortholog) or degradation of TP53 by the MDM2 mediated E3 ubiqutin ligase, will lead to the dysregulation of cell cycle [37,61].

INK4A/ARF

The INK4A/ARF locus (also known as CDKN2A) is located on chromosome 9p21, which encodes the cyclin-dependent kinase inhibitors p16INK4A and p14ARF. However, these two kinase inhibitors are structurally and functionally unrelated (Figure 2.3) [62]. p16INK4A is a tumour suppressor that acts on the cyclin D1-CDK4 complex to inhibit phosphorylation of RB, thus blocking the G1/S transition and cell growth. Inactivation of p16INK4A occurred in 10% of osteosarcoma mainly by deletion rather than point mutation, and in some cases RB alteration was not found [63,64]. Loss of p16INK4A expression is a factor linked to the low survival in paediatric osteosarcoma patients [39]. Unlike p16INK4A that regulates RB function indirectly, p14ARF binds to MDM2 directly and consequently suppresses TP53 degradation and enhances TP53-dependent activation, which G1 phase cycle arrest or apoptosis are increased. Approximately 10% of osteosarcoma have showed homozygous deletion of INK4A/ARF locus and expression of p14ARF is expected to be demolished [65]. As a result, genetic alteration in INK4A/ARF locus would regulate two separate critical pathways of the cell cycle (Figure 2.3). 18

Figure 2.3 Pathway of TP53 and RB. Interaction and regulation of TP53 and RB pathway. Dashed lines represent target inhibition or inactivated forms, and solid lines represent target activation or activated forms.

2.2.2. Oncogenes

Oncogenes are genes that are altered and/or expressed at abnormal levels, leading to the development of cancer. Many oncogenes are responsible for malignant cell growth in different neoplasias.

MYC

Many proto-oncogenes are up-regulated in osteosarcoma, however, the roles of these genes in the pathogenesis of osteosarcoma is still not fully understood. The MYC proto-oncogene encodes for a transcription factor that controls multiple cellular processes involved in cell proliferation, cell growth, apoptosis and differentiation [66]. It is also one of the most frequently activated oncogenes in cancer. MYC overexpression was found in 7-12% of osteosarcoma cases, and is more common in Pagetic 19

osteosarcoma [41]. In another study, the MYC gene expression was found elevated in

42% of the relapsed patients whilst 23% of the patients remained disease-free [67].

FOS

The FOS gene encodes a family of proteins that binds with specific JUN proteins to form the AP1 transcription factor, which up-regulates transcription of a diverse range of genes from cell proliferation, differentiation, transformation and bone metabolism to defend against invasion and cell damage [68]. In vivo studies showed that injection of the viral homolog v-FOS or overexpression of FOS in bone induced osteosarcoma formation in transgenic mice [69,70]. FOS is more frequently expressed in high-grade than the low-grade lesion of osteosarcoma. Highest levels of FOS expression were found in 47% of conventional osteosarcoma but negative or low-level expression were found in telangiectatic, low-grade central, low-grade periosteal and low-grade parosteal osteosarcoma [71]. FOS was also found expressed in 40% of patients who subsequently developed metastases [67].

MDM2

Overexpression of the chromosome 12q13 region, which contains both MDM2 and

CDK4, was found in 5-10% of osteosarcoma [52,72]. However, other amplicons were also identified in the chromosome region 12q13-q14. MDM2 could negatively modulate

TP53 function and signalling pathway by binding or degrading the TP53 protein. Also,

MDM2 overexpression could functionally suppress TP53 even in the presence of wild- type TP53 protein [52,73,74]. MDM2 were detected in various grades and types of osteosarcoma, and its expression has been associated with progression and metastases but not correlated with response to chemotherapy or survival [52,75-80]. 20

CDK4

CDK4 forms a complex with cyclin D1 and regulates the RB pathways. Increased level of CDK4 expression may stoichiometrically enhance RB phosphorylation, thereby impairing cell cycle control [52,81-84]. Identification of the discontinuity of 12q13 amplicons provides evidence that high levels of CDK4 may also drive 12q13-15 amplification independently of MDM2. This suggested that amplicons with selective advantages will be retained and intervention sequences will be lost in the amplification process [41,64,84-

89]. Although low percentage of osteosarcoma cases has shown CDK4 gene overexpression, but CDK4 protein is highly expressed in 65% of low grade osteosarcoma [52,85,90,91]. Cyclin-D1 gene overexpression has been reported in 4% of osteosarcoma and high level of proteins was detected in 22% of cases. It is also linked with metastatic phenotype and the absence of its gene expression is a significant prognostic factor in osteosarcoma [64,83,92]. Again, these findings indicate that alteration in the function and balance of the RB pathway plays an important role in osteosarcoma development.

ERBB2

The protein encodes from ERBB2 (also known as HER2/neu and c-erb-2) is structurally homologous to the epidermal growth factor receptor (EGFR), but its ligand has not yet been identified [52]. Studies showed that ERBB2 protein expression was found in about

42% of osteosarcoma patients and relatively higher to adjacent normal tissue. However, gene amplification and rearrangement were not responsible for the elevated protein expression [93-95]. Many studies have reported the significance of ERBB2 expression in osteosarcoma. For example, the elevated ERBB2 expression in high-grade non- metastatic osteosarcoma showed a higher rate of event-free and survival in patients and increased cytoplasmic ERBB2 expression in pretreated osteosarcoma had higher risk of pulmonary metastases and chemoresistance [95-98]. Absence of ERBB2 21

expression in osteosarcomas have also been reported [99]. The actual role of ERBB2 expression in osteosarcoma development remains inconclusive.

MET/HGF

The receptor of hepatocyte growth factor (HGF, also known as scatter factor) is a cytokine that promotes cell proliferation and motility, and is encoded by the MET protooncogene [100-103]. The MET/HGF receptor is believed to play a role in stromal- epithelial interaction, since it is mainly found in epithelial cells and its ligand is from cells of mesenchymal origins [41]. Expression of both MET and HGF have been found in some osteosarcoma and high level of MET expression was identified in 60% of osteosarcoma [101,102]. Activation of MET/HGF through an autocrine or paracrine mechanism found in some osteosarcoma may contribute to the metastases and survival of patients [100,101,104,105].

Chromosome 12q13-q15

Amplification of the q13-q15 region on chromosome 12 occurs frequently in osteosarcoma. Apart from protooncogenes CDK4 and MDM2, others such as GLI,

TSPAN31 and PRIM that lie in this region are also proposed to contribute to the development of osteosarcomas [41]. GLI is identified in human glioblastoma and is a zinc finger transcription factor that plays a role in sonic hedgehog (Shh) signal [106]. GLI was expressed in one third of osteosarcoma samples along with CDK4 and was detected in seven out of eight osteosarcoma that were mainly undifferentiated tumours

[107,108]. TSPAN31 (also known as SAS) is a member of the membrane protein tetraspanin family that are involved in cell adhesion, migration, signal transduction, activation, proliferation and differentiation [109,110]. Amplification of TSPAN31/SAS occurs more frequently in low-grade than in high-grade osteosarcoma, and its amplification in low-grade osteosarcoma is tightly linked with an increased expression and co-amplification of CDK4 [91,109,111-113]. PRIM1 (also known as DNA Primase I) is a 22

DNA-dependent RNA polymerase that is involved in DNA synthesis and G1/S transition of cell cycle and was found to be amplified in nine out of twenty two osteosarcoma

[114,115]. However, no other data has been published in regards to the correlation of

PRIM1 and osteosarcomas.

2.2.3. Other factors and pathways

Wnt signalling pathway

The Wnt pathway involves a family of highly conserved signalling proteins that interacts with receptors on target cells and plays an important role in embryogenesis, tissue regeneration and carcinogenesis. Any alteration of Wnt signalling is associated with many common human diseases and cancers, such as Schizophrenia, colorectal cancers and pancreatic cancers [116,117]. Wnt proteins and their receptors are also found expressed in bone progenitors and related to osteoblast and osteoclast differentiation

[117-121]. !-catenin is an essential signal mediator of the canonical Wnt pathway, where increased cytoplasmic expression and/or nuclear localization of this protein has been detected in a high percentage of osteosarcomas and is possibly linked with metastasis in osteosarcomas [122,123]. Sporadic mutations of !-catenin and/or its downstream signalling may also be involved with osteosarcoma progression [124]. High expression level of a Wnt co-receptor, LRP5, is associated with less differentiated osteosarcomas and poorer survival of patients [125]. In addition, decreased invasion and motility of the osteosarcoma cell line Saos-2 were demonstrated after ectopic expression of the Wnt agonist, dickkopf homolog 3 (also known as DKK3) [126].

Telomeres and telomerase

Telomeres are regions of repetitive DNA at the end of linear chromosomes. Part of the sequence will be lost after each DNA replication cycle and cell division, and will be 23

terminated in normal cells when the loss is critical [127]. Cancer cells maintain cell division by an enzyme called “telomerase” that replaces the lost telomeric DNA [128].

Telomerase activity is detected in most human cancers but not in normal somatic cells, benign lesions and low-grade sarcomas [129] and in only a portion of osteosarcomas [130-

132]. For those osteosarcomas with telomerase activity, an inverse correlation with the occurrence of pulmonary metastases in patients treated with chemotherapy was shown

[133]. Alternative lengthening of telomeres, which is associated with telomere dysfunction and chromosomal instability, in osteosarcomas that lack of telomerase activity may also have implications on the progression of osteosarcomas [130].

It has been shown that patients who lacked telomerase activity and alternative lengthening of telomeres were associated with a favourable prognosis, but patients with both telomerase activity and alternative lengthening of telomeres were not significantly different to patients with alternative lengthening of telomeres alone [132].

Since the telomerase activity is detectable in benign bone tumours (e.g. osteoblastoma, osteochondroma), it is not a suitable maker for differentiating between non-malignant and malignant tumours [134].

S100A6

The S100A6 protein (also known as calcyclin) is a member of the S100 family of proteins. It contains 2 EF-hand calcium binding motifs that are involved in cell contractility, cell motility and adhesion-dependent signalling by interacting with actin cytoskeleton [135,136]. This protein is involved in the regulation of protein phosphorylation,

DNA replication, calcium homeostasis, cell proliferation, and differentiation by interacting with transcription factors or chaperone proteins such as Hsp70 in the nucleus [137,138]. In addition, this protein also plays a role in the stimulation of calcium- dependent insulin release, stimulation of prolactin secretion, and exocytosis [139].

S100A6 was expressed in 84% of osteosarcoma specimens and correlated with 24

decreased metastasis clinically, but its overexpression in osteosarcoma cells decreased cell motility and anchorage-independent growth [138,140]. Therefore, it is suggested that although S100A6 overexpression is commonly found in osteosarcoma, its loss of expression may correlate with a metastatic phenotype.

Matrix metalloproteinases

Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases normally involved in degrading extracellular matrix within the context of physiological tissue remodelling, angiogenesis, as well as non matrix proteins [141,142]. Up-regulated expression of MMPs and inhibition of MMP inhibitors (tissue inhibitors of matrix metalloproteinase, or TIMPs) have been recognised as an important factor in cancer invasion and metastasis [141]. Overexpression of MMP2 and MMP9 in osteosarcoma cells and animal model showed their association with increased metastatic potential in osteosarcomas [143,144]. Elevated expression of membrane type MMP1 has been correlated with unfavourable prognosis in osteosarcoma patients [145]. Up-regulation of

TIMP1, which is the MMP9 inhibitor, showed poor clinical outcome in osteosarcoma patients and also participated in osteosarcoma progression by activation of

Ras/Raf1/FAK signalling system with an unknown receptor [146,147].

CXCR4/CXCL12

Chemokine receptor CXCR4 and its sole ligand CXCL12 (also known as stromal cell- derived factor 1 or SDF1) are important factors in the process of hematopoiesis, organogenesis and vascularisation. This receptor ligand complex plays an important role in metastases and is a potential therapeutic target for cancer treatment [148,149].

Investigations showed that motility, migration and adhesion of osteosarcoma cells were promoted by CXCL12, and conversely, cell growth, migration and pulmonary metastasis stimulation were suppressed by CX CR4 inhibitors [150-152]. 25

2.3. Osteogenic and tumourigenic properties of osteosarcoma

The complexity of osteosarcoma is given by its broad and complicated genetic and cytogenetic alterations. Human osteosarcoma cell lines demonstrate differences in the phenotypic characteristics, such as osteogenic and tumourigenic properties. Diverse alkaline phosphatase activity and differentially expressed osteogenic or osteoblastic factors (e.g. BMPs, SMADs, RUNX2, Osterix, CTGF and OPN) were found in different human osteosarcoma cell lines [153-156].

Amongst few of the human osteosarcoma cell lines that could induce ectopic bone formation [157], Saos-2 has been extensively studied for its osteogenic properties under different conditions. Devitalised cell extracts, freeze-dried cells and freshly cultured cells of Saos-2 have demonstrated its potential to induce ectopic bone formation after implantation into nude mice [156,158-163]. Primary and pulmonary metastatic osteosarcoma models were also successfully established by using Saos-2 [164-166] that demonstrated tumorigenicity.

Some human osteosarcoma cell lines only showed tumourigenic potential after genetic modification and remain non-osteogenic [1,167]. The understanding of the differences amongst the osteogenic and tumourigenic properties of human osteosarcoma cells remains poor. Further studies are required to explore the relationships between these phenotypic characteristics and their impact to tumour development and metastasis. 26

2.4. Differentiation disruption and osteosarcoma

Bone formation is a well-coordinated process involving epithelial mesenchymal interaction, condensation and differentiation. Osteoblastogenesis is a relatively well- understood process compared to the limited knowledge in osteosarcoma.

Understanding the fundamental molecular mechanism of osteoblast differentiation would help to unravel the differentiation defects during tumorigenesis and the molecular pathogenesis of osteosarcoma.

Osteoblast is the most important cell types in bone formation and it participates in both endochondral and intramembranous ossification. Differentiation of osteogenic lineages starts from the condensation and interaction of mesenchymal stem cells, which differentiates into osteoblast progenitor cells, pre-osteoblasts, mature osteoblasts and finally into osteocytes [168-170]. Osteoblast differentiation is tightly regulated by transcription factors (Figure 2.4) (RUNX2, Osterix, TAZ, TWIST, ATF4, etc.) [171-176] and signalling pathways (Wnt, Hedgehog, FGF, TGF-β/BMP superfamily, NOTCH, etc.)

[119,177-181]. Proper coordination of these factors is essential to osteoblast differentiation and required during skeletal development. Any disruption or interruption in the osteoblast differentiation process leads to improper bone formation or more seriously the development of skeletal disease.

Osteosarcoma exhibited some osteoblast-like features, which comes from a broad range of differentiation status from highly differentiated to poorly differentiated or undifferentiated phenotypes [155]. Studies have implicated stem-like tumour cells in the pathogenesis of other heterogeneous and highly malignant tumours, such as leukemia, brain tumours and breast tumours [182-185]. The cancer stem cell theory states that a small subpopulation of stem-like tumour cells selectively possesses tumour initiation and divides asymmetrically with the ability to generate a bulk population of 27

tumourigenic cell progeny through differentiation [186]. Therefore, osteosarcoma has recently been regarded as a differentiation disease, in which any disruption along the differentiation of mesenchymal stem cells to osteoblasts will lead to the development of osteosarcoma [52,155,187-189]. The cancer stem cells are potentially responsible for osteosarcoma development have yet to be identified.

Figure 2.4 Schematic summary of the activities and regulations of transcription factors during differentiation and maturation within osteoblastic lineages starting from a mesenchymal precursor cell. Positive regulations are indicated by red solid arrows and positive modulators are labelled red. Negative regulations are indicated by blue broken arrows and negative modulators are labelled blue. Diagram modified from Hartmann (2009) [172] and Marie (2008) [173].

Several arguments are used to characterise osteosarcoma as a differentiation disease, which is caused by disruption of osteoblast differentiation. Firstly, a few of the major signalling pathways such as Wnt, Notch and Hedgehog signalling play an important role in regulating osteoblast differentiation as well as in cancer development [190-194].

Secondly, osteosarcoma displays many characteristics specific to undifferentiated osteoblast cells and osteoblast differentiation could be promoted with differentiation agents, such as PPAR-gamma and 9-cis-retinoic acid [155,188,195-202]. Thirdly, RB is an essential factor for many cellular processes including mesenchymal differentiation [203- 28

205]. Interaction of RB and RUNX2 is required to activate transcription in osteoblast differentiation. It has been show that loss of functional RB, which is commonly found in osteosarcoma, attenuates the interaction between RB and RUNX2 [206]. Fourthly,

RUNX2 directs cell cycle progression through induction of p27(KIP1), which is lost in dedifferentiated human osteosarcoma [188], whilst playing an important role during osteoblast differentiation and bone development (Figure 2.5) [207,208]. Lastly, the osteogenic factor BMPs showed no effects in promoting the terminal differentiation of most osteosarcoma cells but rather stimulated its proliferation, which further implies possible differentiation defects in osteosarcoma cells [155]. These emerging findings clearly support the notion that osteosarcoma development possibly results from disrupted osteoblast differentiation pathway.

Figure 2.5 Model showing the interaction between cell cycle protein and RUNX2 during osteoblast differentiation. The interaction of hypophosphorylated pRB with RUNX2 positively controls cell cycle progression and aids the expression of the osteoblastic phenotype [188]. Positive regulations are shown in solid-line arrows and negative regulations are shown in broken-line arrows. 29

2.5. Advance technology for osteosarcoma research

2.5.1. DNA microarray technology

The study of gene expression profiling of cells and tissues is important for understanding molecular mechanisms and biological fundamentals in medical research.

DNA microarray, macroarray or gene chips allow the assessment of the expression level of thousands of genes concurrently within a particular mRNA sample [209,210].

Microarray technology is a remarkable advance for molecular biology and for cancer research. Studying of genotypes, polymorphisms and mutations with this high- throughput and miniaturised method has been used for (i) investigation of gene discovery and regulation, (ii) identification of diagnostic or prognostic biomarkers, (iii) classification of disease, (iv) drug discovery and evaluation of the response to therapy, and (v) identification of mechanism involved in disease pathogenesis and biological events [211,212]. This method has also been used for the identification of structural alterations by comparative genomic hybridisation [213]. Microarray technologies are currently one of the important tools for studies and discoveries in different research areas, such as in clinical medicine and oncology, etc. [212].

DNA microarray technology encompasses many aspects and techniques in genetic study all into one single assay. For example, automated DNA sequencing, DNA amplification, oligonucleotide analysis, nucleic acid labelling and bioinformatics are all required. Gene expression profiling of the DNA microarray relies on nucleic acid hybridization, immobilization of nucleic acid sequences on a solid surface as probes for complementary gene sequences, and detection of the signal from the labelled RNA [211].

The fundamental quantification of gene expression level is that the amount of mRNA with complementary sequence in the tested sample is directly proportion to the signal measured at each sequence-specific location [214]. However, the quantification data does not represent the absolute level of expression but can be used to compare and 30

exploit the expression level among different conditions and genes (e.g. health vs disease) [214].

A general classification of DNA microarrays is based on the following three criteria: (i) length of the probes immobilised on the array; (ii) manufacturing method; and (iii) number of samples that can be profiled together on one array [215]. Oligonucleotide and complementary DNA (cDNA) microarrays are the two most effective microarray platform and commonly used by a majority of investigators.

Oligonucleotide microarrays utilise probes of usually 50bps or less, which can be manufactured by deposition of previously synthesised sequences or directly synthesised (or “in situ” synthesis) on the surface of a silicon wafer in a pattern manner

[216]. Different “in situ” technologies are available in the manufacturing of oligonucleotide microarrays, such as “photolithography” by Affymetrix™ (Santa Clara,CA) (Figure 2.6),

“ink-jet printing” by Agilent (Palo Alto, CA) and “electrochemical synthesis” by

CombiMatrix (Mukilteo, WA) [215]. The standardization of probe length and the ability to differentiate splice variants are the major advantages of this technique, but the construction of specific oligonucleotides is limited by availability and knowledge of the targeted sequence [211]. 31

Figure 2.6 Affymetrix™ microarray. Diagrammatic representation of the steps involved in manufacturing of oligonucleotides microarray by photolithography (Affymetrix) and the experimental processes for the gene expression profile studies and analysis using Affymetrix microarray. Diagram modified from illustration in Affymetrix webpage (2006) http://www.affymetrix.com.

In cDNA microarray, polymerase chain reaction (PCR) products, which could be hundreds or thousands of base pairs (bps) representing the gene of interest, are spotted systematically on nitrocellulose filters, glass slides or silicon wafer [217]. The advantages of cDNA spotted arrays for individual investigator is (i) easy customization with particular genes expressed in a specific context or cell type, (ii) requires no prior knowledge of the sequence, and (iii) clones can be used and then sequenced later if of interest. However, maintaining the quality of arrays and managing large clone libraries are a strenuous task for most laboratories [211].

2.5.2. Importance of microarray in cancer research

Cancer is one of the most challenging and complex diseases presented in a clinical situation. Taxonomy of cancer is traditionally based on histopathology from more than 32

200 distinct entities of diverse cell types. The polygenic nature of cancers, which are specific to individual neoplasm, elucidates the clinical diversity of those histologically similar tumours. Multiple genomic alterations in cancers include point mutations, translocations, gene amplifications, epigenetic modifications, deletions, and aberrant splicing that could be inherited or somatically acquired during progression from a normal to a cancerous cell. These genetic abnormalities affects the expression of genes that control growth, invasiveness, metastatic potential and responsiveness or resistance to chemotherapy of cancers [211].

Advances in molecular biology techniques have revolutionised the investigation of the pathogenesis in cancers, especially after the complete sequencing of the human genome [218]. Microarray technologies have outraced traditional methods, which only allow investigation of a small number of genes at a time. Microarray technologies allow comprehensive analysis of the expression profiles between cancerous and normal tissue, as well as identification and quantification of the complex multi-gene expression patterns in cancer [211].

The extensive gene expression analysis using microarrays improves the understanding of the initiation of cancer, its progression and its sensitivity to therapeutics by distinguishing the genomic perturbations and mechanisms that drive tumour cell survival, cell cycle control, DNA repair, differentiation, apoptosis, vascularisation and metabolism [219]. Microarray application ultimately improves the three main stages in cancer research strategy, which include discovery, validation and clinical utility (Figure

2.7) [220]. It also helps to provide better diagnosis and treatment of cancer through more precise disease classification and patient stratification. The expanded insights in cancer help to design therapies that are specific and targeted to different cancer subtypes and enhance the effectiveness of treatment regimens. 33

Figure 2.7 Microarray technology are involved in the three stages of cancer research [220].

Many studies have employed the microarray technologies in investigating tumorigenesis in different types of cancer. One recent study has successfully utilised the global gene expression profiling in the discovery of genes involved in the transformation of Barrett’s esophagus to adenocarcinoma [221]. Evaluation of the gene expression pattern between esophageal epithelium, Barrett’s metaplasia and esophageal adenocarcinoma showed that suppression of the genes corresponding to the epidermal differentiation complex is associated with tumour pathogenesis and represents potential genetic markers of disease progression, as well as the pharmacologic targets for treatment intervention.

2.5.3. Gene expression analysis in osteosarcoma

Global gene expression profiling is commonly used in the study of osteosarcoma and helped to develop more informative classification system for better understanding of the disease. This technique is used in many different ways for studying of osteosarcoma, such as identification of prognostic factor or biomarker [222,223] and investigation of mechanisms involved in drug resistance [224,225]. 34

A recent study has identified a target gene of an oncosuppressor in osteosarcoma by using microarray gene profiling. Caveolin-1 is up-regulated after induced expression of the CD99 oncosupressor gene in osteosarcoma cell line. Further investigations have shown that the up-regulation of caveolin-1 was resulted from the inhibition of c-Src from the overexpression of CD99 [222]. Another study have also employed microarray gene profiling to identify gene targets of the receptor activator of nuclear factor 'B ligand

(RANKL) in osteosarcoma cell line, due to the crucial role of the receptor activator of nuclear factor 'B (RANK) in osteosarcoma. These gene targets would help to identify the potential therapeutic targets, but requires further investigations [223].

Studying the gene expression profiles also helped to understand the genetic alteration involved in drug resistance of osteosarcoma. A study has compared the gene expression profiles between the drug-resistant variants induced by the two most important chemotherapeutic drugs for high-grade osteosarcoma treatment, Doxorubicin and Methotrexate, and their parental osteosarcoma cell lines. The investigation has identified a list of overexpressed genes, including MDR1, DHFR, MLL and MYC, that should be focused for the analyses on experimental models and clinical samples [224]. A different study has compared the gene profiles from human osteosarcoma xenograft, and showed different sensitivities to Ifosfamide, Doxorubicin or Cisplatin. The investigation has identified a list of genes that may help predict the responsiveness to the three drugs. However, the potential of selected candidates will have to be further validated on clinical specimens from osteosarcoma patients [225].

Aforementioned studies showed that gene expression profiling have aided to increase the understanding of osteosarcoma by showing the genetic changes in different conditions. Other techniques and technologies would be required to confirm the relationship between the genetic changes and physiological situations 35

2.5.4. Limitations of DNA microarray technology

Although DNA microarray technology is a powerful tool in medical and cancer studies, many factors affect the application of this technology. Preparation, isolation and quality of sample and collection, analysis and validation of data are some of the limitations that affect the outcome [218]. RNA amplification and quality assurance techniques have been incorporated into microarray experimental processes to ensure the consistency and quality of the data acquired [218]. Selection of appropriate data processing algorithm, statistical model and analytical method are important for the quality of results [226]. The use of other conventional method, such as quantitative real-time polymerase chain reaction (qRT-PCR), [227] to validate microarray data is essential to assure the accuracy of gene expression profile analysis. Furthermore, heterogeneity of patient samples, amount and complexity in sample collection, patient population and selection of patient are other factors that contribute to the outcome and quality of microarray analysis [211].

Careful and meticulous experimental design and sample collection are critical to minimise false interpretation of gene expression profiling analysis by microarray.

Standardized the collection of microarray experimental data between laboratories and microarray platforms, cross-reference and integration of information between databases with complete gene annotation, comprehensive descriptions of genes and functions, proteomic information, and disease related information have all been contributed to improve the speed and quality of the gene expression profiling studies nowadays [228-231].

Microarrays are capable of studying comprehensive gene expression profiles, but the rudimentary knowledge of the physiologic role of most gene have restricted the further development of microarray technology [211,218]. Hence, the gene expression profile analysis to molecular mechanisms is limited to currently available information on the genes discovered. 36

2.6. In vitro and in vivo models of osteosarcoma research

2.6.1. Commonly used human osteosarcoma cell line

Saos-2

Saos-2 is one of the non-transformed human osteosarcoma cell lines well characterised in vitro [159,232-234]. It was derived from a primary osteosarcoma in an 11- year-old Caucasian female by Fogh et al. in 1973 [235] and retained as a non- transformed cell line. In vivo studies demonstrated the induction of tumour by subcutaneous injection of Saos-2 into nude mice to study the association of retinoic acid and insulin-like growth factor [236]. However, tumour induction by orthotopic implantation of Saos-2 has not been reported.

A metastatic model of nude mouse with the development of lung metastases was established after post-lateral tail vein injection of the Saos-2 at 6 months. Briefly, a highly metastatic sub-clone called "Saos-LM6" was established by isolating cells from these lung metastases and re-injecting into nude mice 5 times. The Saos-LM6 sub- clone had induced microscopic pulmonary metastases at 5 - 6 weeks and macroscopic lesions by 8 weeks after intravenous injection into nude mouse [164]. A study utilised this

Saos-LM6 sub-clone mouse model to show that pulmonary metastases can be eradicated by intranasally delivered adenovirus-mediated gene transfer of interleukin-

12 (IL-12) [237]. This intranasal delivery system is also explored as an ideal route of therapeutic administration for lung metastatic osteosarcoma patients.

Another highly metastatic sub-clone, Saos-LM7, was used to prove that intranasal instillation of gemcitabine is more effective in eradicating lung metastases than intraperitoneal drug administration [238]. The osteosarcoma model generated by intravenous inoculation of the Saos-2 sub-clone cells plays a role in the therapeutic 37

efficacy investigation of a novel approach for eliminating pulmonary metastatic osteosarcoma. On the other hand, the establishment of an orthotopic tumour model using the original Saos-2 cell illustrates the potential of an alternative and broader usage, such as studying events associated with late osteoblastic differentiation stage or mineralization in human cells [239,240].

U-2 OS

U-2 OS (originally 2T) is another well characterised non-transformed human osteosarcoma cell line. It was derived in 1964 from a biopsy of a moderately differentiated osteosarcoma obtained from the tibia of a 15-year-old Caucasian girl

[159,234,241,242].

Only one study has established an in vivo model to demonstrate the tumourigenic ability of U-2 OS, in which tumour was induced by subcutaneous and intravenous injection of tumour cells [154]. Subcutaneous injection of U-2 OS cells induced tumour formation in 63% (5/8) of mice without lung metastases at mean latency of 99 days and mean rate of 5 cm3 volume formation in 55 days. Pulmonary metastatic tumour lesions were developed in 100% of mice (9/9) from the intravenous injection of tumour cells after 8 weeks. The tumourigenic and metastatic ability was decreased in both subcutaneously and intravenously tumour induction by the use of U-2 OS cells transfected with liver/bone/kidney alkaline phosphatase (ALP) [154]. Despite the convincing results shown in several in vitro studies with a variety of agents such as

Cisplatin and Doxorubicin [243], Etoposide [244] and Paclitaxel [245], in vivo anti- osteosarcoma drug testing model was never established with U-2 OS due to its ineffective formation of spontaneous metastatic and orthotopic model. 38

HOS

HOS (originally M.T., then TE-85) was derived in 1970 from the distal femur of a 13- year-old Caucasian girl. It was one of the first human osteosarcoma cell lines characterised comprehensively [246,247]. This cell line had been studied extensively in the detection of possible causative viral agents and pathology of viruses such as

Human T-cell leukemia virus (HTLV) [248], human cytomegalovirus (HCMV) and the human immunodeficiency virus type 1 (HIV-1) [249].

In vivo tumour induction was unsuccessful from subcutaneous [250,251] and orthotopic

[165,252,253] inoculation of HOS cells. However, induction of tumour by HOS was achievable only after genetic modification, which resulted in various derivatives of the cell line. Many in vivo models were established from HOS derivatives [254-256].

Tumourigenic induction potential of HOS cells was obtained by using viral agents

Kirsten murine sarcoma virus (Ki-MSV) [247], Moloney sarcoma virus (M-MSV RD-114)

[257] and the chemical agent N-methyl-N’-nitro-N-nitroguanidine (MNNG) [255] to genetically alter this cell line.

The first successful orthotopic xenotransplantation osteosarcoma model was established by KRIB, which is generated from the transformation of HOS with the oncogene v-Ki-ras. It showed 100% tumour formation with histological and radiological similarity to the clinical presentation and also spontaneous pulmonary metastases after intratibial inoculation at 4 weeks [258]. The KRIB model was later on used to investigate the effectiveness of STI571 alone or in combination with Taxol for osteosarcoma treatment, in which STI571 has a specific inhibiting effect on certain tyrosine kinase receptor including platelet-derived growth factor (PDGF) and c-Kit [252]. Although the expression of PDGFR is abundant in osteosarcoma, the aberrant PDGFR downstream signalling pathways in the v-Ki-ras oncogene-transformed KRIB cell line contributed to 39

the ineffective treatment outcome after 6 weeks. The KRIB model is valuable for cases where artefactual effect from genetic modification is not important.

143B

143B was another sub-clone generated via Ki-ras oncogene transformation of the HOS osteosarcoma cells similar to KRIB [167]. Subcutaneous inoculation of 143B showed no metastases in the in vivo mouse model [259,260], while spontaneous lung metastases were developed in the orthotopic intratibial models [165,253]. A study showed that 143B exhibited a higher tumourigenicity and spontaneous metastatic potential compared to

MNNG/HOS, which is another sub-clone of HOS and is more efficient in forming primary tumours only [253]. 143B is potentially a useful model in studying factors involved in the spreading of human osteosarcoma by showing a varied degree of tumour growth at both primary and metastatic sites. However, it has to be aware of the fact that 143B is a transformed cell line.

MG63

MG63 is a non-transformed cell line derived in 1977 from an osteosarcoma obtained from a 14-year-old Caucasian male. It has been shown that induction of human interferon in MG63 was closely related to fibroblast than leukocyte cell type [261].

A recent study showed that in vivo gene inhibition of the anti-angiogenesis effectively reduced the xenograft tumour growth, which was induced by subcutaneous injection of

MG63 into nude mice[262]. However, metastases were not examined in the study.

Another study reported that orthotopic intratibial injection of MG63 had induced primary tumour formation in 29% of mice with no metastasis [165].

Nonetheless, an MG63 model with primary and lung micrometastases tumour formation was established with tail vein injection of sub-clone M8, which was selected 40

by in vitro characteristic after limited dilution method [263,264]. Another orthotopic MG63 model with primary and spontaneous pulmonary metastases tumour formation was established with intratibial injection of sub-clone MG63.2, which was generated by in vivo re-injection selection process [265]. Although improvement is required on the establishment of a representative animal model, MG63 remains one of the popular cell lines to be used in a wide range of in vitro investigations.

SJSA

According to the ATCC, SJSA (originally OsA-CL, also known as SJSA-1) was derived in 1982 from a primary tumour of a 19-year-old African male patient diagnosed with primitive multi-potential femur sarcoma [266]. Amplification of the MDM2 oncogene and the gli proto-oncogene were identified in this cell line [108,267]. Recently, for the first time an orthotopic SJSA model with primary tumour and pulmonary metastases was set up by intratibial injection and demonstrated that gene therapy using DNAzyme Dz13 effectively reduced tumour growth at the primary and metastatic site [268].

SJSA xenograft model was used in many studies for the potential of cancer therapy by the inhibition of p53 signalling in cancer. Studies showed that inhibition of the p53-

MDM2 interaction by the use of a cis-imidazoline inhibitor Nutlin-3a or a small molecule inhibitor MI-219 effectively reduced growth of xenograft tumour, which was induced by subcutaneous injection of SJSA into nude mice [269,270]. However, metastases were not examined in these studies. The importance of NOTCH signalling to the pathogenesis of osteosarcoma was demonstrated by the use of the SJSA xenograft model induced by subcutaneous injection [271]. 41

2.6.2. Development of osteosarcoma animal model

The application and integration of appropriate mouse models is essential in basic and clinical cancer studies nowadays [272]. An ideal mouse model represents all aspects of the human disease, such as tumour biology, genetics, aetiology and response to therapeutic agents. Development of an osteosarcoma model is indispensable in providing a greater insight into the mechanism and pathogenesis of human osteosarcoma. An osteosarcoma model is desirable as it includes features such as expression of osteoblastic markers, spontaneously develop primary tumour and pulmonary metastases, and allows the analysis throughout the tumour progression within the relative short life span of the mouse [167].

Mouse models of osteosarcoma were initially set up by exposure to external beam radiation [273], inoculation of radioactive isotopes [274] and oncogenic virus [275]. Due to the inconsistency of these models, other more robust models were later established by subcutaneous [276] and orthotopic [258] implantation of murine osteosarcoma cells in mice.

Murine (mouse or rat) osteosarcoma cell lines such as UMR106-01[277], K7M2 [278] and

K12 [279] were successfully used to create the metastatic models of osteosarcoma.

Utilisation of murine cells allows the physiological study of the interaction between tumour and host during tumour progression under the syngeneic condition.

Recently, there has been more interest in the use of human osteosarcoma cell induced tumour model [1]. The substitution from murine cells to human cells allows the investigation of the interaction of biomolecules that are active against human disease.

Hence, an immunocompromised or immunodeficient host is required to study the in vivo behaviour of these allotransplanted or xenotransplanted tumour cells. The development of the “nude” mouse is a significant advancement in cancer research, as it allows the tumour cells from different species to be studied within another host species. Our previous study showed that intramuscular implantation of Saos-2 cells 42

into the thigh muscle of nude mice induced ectopic bone formation, while a xenograft tumour was formed with the implantation of U-2 OS cells. Further investigation suggested that a combination of differentially expressed osteogenic factors, such as

BMPs, TGF-!, BMPRs and I/R/Co-Smads, may play important roles for the different osteoinductivity of the two human osteosarcoma cell lines [159].

Nude mice are characterised by their lack of hair and born without a thymus, which is the result of homozygous recessive mutation of “nu” gene (nu -/-) [280]. The high degree of genetic similarity, similar organ system, lower cost in production, ease of handling and maintenance, and lower dosage required to evaluate the potential of novel therapeutic agents, makes nude mice the most prominent tool in oncology research nowadays [167]. Furthermore, the generation of knockout and transgenic mice by the advances in genetic engineering have provided more comprehensive and diverse choices in cancer research.

Numerous human osteosarcoma cell lines have been established and characterised in vitro and several osteosarcoma mouse models have been studied extensively [1,167].

However, the understanding in the aetiology of human osteosarcoma and the interactions between host and tumour cells that governs growth and progression of osteosarcoma in vivo remains limited. Despite the similarity and the efforts to mimic the biological environment as closely as possible, factors such as the influence of the micro- and macro-environment of the human body and the human condition at temporal, physiological and histopathological levels are not easily replicable. Ongoing research and development of the osteosarcoma mouse model is required to study the genetic aberrations, identify the disturbed biochemical pathways, understand the tumour development, and evaluate the effectiveness of therapeutic strategies. 43

2.7. Type 1 insulin-like growth factor receptor (IGF-1R)

IGF-1R is a membrane bounded receptor with intrinsic tyrosine kinase activity. It has been widely shown that the receptor is involved in the regulation of cell growth in different tumour types, both in vivo and in vitro [33,281]. IGF-1R is one type of the tyrosine kinase receptors (RTKs) that share a high homology with insulin receptor (IR) and it is different to the other RTKs in regards to some structural features, receptor-ligand interaction and downstream signalling. IGF-1R signalling is under endocrine control, while most other RTKs are involved in autocrine and paracrine cell-to-cell signalling.

Ligand binding is required for activation and signal stimulation of IGF-1R, while other types of growth factors such as epidermal growth factor receptor (EGFR), simply requires overexpression of the receptor itself to activate the downstream biochemical signalling pathway [33,282].

2.7.1. Molecular structure of the IGF-1R

The IGF-1R is a heterotetrameric plasma membrane consisting of two identical extracellular -subunits and two identical intracellular !-subunits held together by disulfide bridges [283,284]. The extracellular domain is the binding site for ligands such as

IGF-1 and IGF-2, while intracellular subunits consist of the juxtamembrane domain, the tyrosine kinase domain, and the carboxy-terminus (C-terminus) domain. The molecular structure of the IGF-1R is shown in Figure 2.8. 44

Figure 2.8 The molecular structure of IGF-1R. L1 and L2 is the homologous domain, CR is the cys-rich region (Cys 152-298), FnIII is the fibronectin type III, and ID is the insert domain. Diagram modified from Brodt (2000) [283] and Adams (2000)[284].

The tyrosine kinase domain of the IGF-1R contains three principle autophosphorylation sites, namely the tyrosine residues Y1131, Y1135 and Y1136 located at the intracellular portion of the !-subunit (also known as the activation loop, “A-loop”). This

A-loop acts as an auto-inhibitory control of the receptor. When the tyrosine domain is unphosphorylated, it restricts the access of ATP and substrates to the active binding catalytic site. Conformational change and autophosphorylation occurs when ligand binds to the extracellular -subunit of the IGF-1R and activates phosphorylation of

Y1135 and then Y1131, which subsequently destabilises the auto-inhibitory conformation of the a-loop. Moreover, further phosphorylation of Y1136 stabilises the a-loop at a catalytic favourable conformation and facilitates the phosphorylation of other receptor tyrosines with subsequent activation of exogenous substrate proteins.

The intrinsic tyrosine kinase activity is increased when all three sites are 45

phosphorylated. Phosphorylations of C-terminus and juxtamembrane domains are also involved in regulation of IGF-1R activity [284,285].

2.7.2. Physiological function of the IGF-1R

IGF-1R is a tyrosine kinase receptor belonging to the insulin receptor family, which also includes insulin receptor (IR), IGF-2R/mannose 6-phophate receptor and other hybrid receptors [281]. IGF-1R and IR shared 70% homology in molecular level. Originally, the main function of IR involved in the control of glucose uptake and metabolism, and IGF-

1R was considered as a redundant receptor used by cells only when signalling from the

IR was depleted or abolished [281,286]. It has been shown that the two molecularly similar receptors do share some functional capabilities, in which IR could perform IGF-1R like functions and IGF-1R could partially compensate for loss of metabolic functions of the

IR [287].

IGF-1R signalling is involved in the regulation of mitogenesis, transformation, cell proliferation, differentiation, anti-apoptosis, adhesion, metastasis, and cell motility

[281,286,288]. High expression of IGF-1R is frequently found in embryonic stem cells with lower expression in most adult differentiated tissue except in the liver [289,290]. IGF-1R was activated by the binding of its ligands IGF-1 and IGF-2 with higher binding affinity towards IGF-1 [33]. A group of six IGF binding proteins (IGFBPs) are responsible in modulating the IGF-1 and IGF-2 stability in circulation, transporting to target tissues and interaction with IGF-1R, thereby controlling the activation of IGF-1R. However,

IGFBPs also exhibit some IGF independent functions associated with growth inhibition and apoptosis [291].

A recent discovery showed that IGF-1R plays a role in rescuing neuronal, haematopoietic and fibroblast cells from apoptosis [284]. Some of the other physiological 46

roles of IGF-1R include postnatal mammary development and lactation, bone formation, and renal function [33]. Furthermore, IGF-1R is involved in integrating signals for somatic preservation and survival, while diverting away from anabolic cellular processes, such as growth and reproduction, under stress condition [292] (Figure 2.9).

Thus, IGF-1R system is also regarded as a “cell survival system”.

Figure 2.9 Endocrine regulation and function of IGF-1. IGF-1 system connects with metabolic and reproductive functions through interaction with insulin and steroid hormones (IGFBP, IGF binding protein; GHR, growth hormone receptor; SHBG, sex hormone binding globulin) [33].

2.7.3. IGF-1R downstream signalling pathways

The IGF-1R signalling pathways have been studied extensively (Figure 2.10). IGF-1R activation is initiated by the binding of its ligands, which leads to the phosphorylation of tyrosines on the intracellular !-subunit and other tyrosine residues in the juxtamembrane and C-terminal domain [293]. The phosphorylated domains serves as a docking site for several receptor substrates including insulin receptor substrates (IRS)

1-4 and Src homology 2 domain containing (Shc) protein, and subsequently 47

phosphorylate a cascade of factors. Thus, some signalling pathways have been identified in the IGF-1R system.

Figure 2.10 Signal transduction pathways of the ligand-activated IGF-IR. A schematic representation of the major signalling pathways that can be activated by the autophosphorylated IGF-IR. The cell context, ligand concentrations, and crosstalk with other signalling systems affect the type and strength of the signal and the biological outcome [293].

The PI3K/Akt pathway involves phosphorylation of IRS-1 and leads to the activation of the phosphatidylinositol 3-kinase (PI3K) through its substrate p85, followed by the activation of p70 S6 kinase and protein kinase B of the Akt pathway [294]. Activation of

Akt pathway leads to the enhancement of protein synthesis by stimulating the mTOR pathway and the suppression of apoptosis by inactivating the BAD pathway [295].

Moreover, the MEK/ERK pathway involves phosphorylation of IRS-1 and Shc protein that directs the formation of Grb2/SOS and activates the Ras/Raf pathway. The activated Ras/Raf pathway will then stimulating the MEK/ERK pathway and other nuclear factors for the induction of cellular proliferation [296,297]. 48

Activation of IGF-1R allows direct activation of the Janus kinase (JAK) pathway in certain cell types leading to the activation of the signal transducers and activators of transcription (STAT) proteins, which is essential for cellular transformations [298].

Moreover, phosphorylation of IGF-1R is linked to integrin-mediated signalling and cytoskeleton through the association with p130Cas and paxillin, thus plays a role in the regulation of cell shape and motility [299,300]. Moreover, the activation of MEK/p38-MAPK pathway through the IGF-1R system promotes cell survival under stress conditions by regulating apoptosis [281].

Ras is a membrane-associated guanine nucleotide-binding protein activated by the exchange of guanosine diphosphate (GDP) with guanosine triphosphate (GTP) in response to various extracellular stimuli, which include growth factors, receptor- tyrosine kinases, G-proteins, adhesion molecules and second messengers [301]. Ras signalling effectors are involved in many biological cellular functions, such as cell proliferation, apoptosis, migration, differentiation, transcription, cell-cycle progression, and translation.

The Raf/MAPK pathway is one of the well studied Ras effector pathways in terms of its biochemistry, signalling and involvement in disease, especially of the Raf/mitogen extracellular kinase (MEK)/extracellular signal-related kinase (ERK) pathway [302]. Raf proteins are a family of serine/threonine kinases, and it activates by binding with the activated Ras-GTP complex leading to the phosphorylation of MEK followed by ERK pathways[303]. Activated ERK is translocated to the nucleus for the regulation of transcriptions and cell cycle, which is particularly important to cell survival and proliferation [302].

In addition, the Raf/MEK/ERK pathway as well as the abnormal Ras signalling are involved in human pathologies and oncogenesis. Previous studies have demonstrated 49

the oncogenic transforming activity of the viral oncogenic form of cRAF-1 in the transfected cells [304] and identified the B-Raf gene mutation in many types of cancer, such as melanoma and colorectal cancer [305]. The importance of the

Ras/Raf/MEK/ERK signalling pathway in the regulation of cell survival and proliferation, as well as its involvement in cancer have attracted enormous attention in therapeutic potential in targeting factors along this pathway [302,303,306].

2.7.4. Association of IGF-1R in tumorigenesis and osteosarcoma

Type I insulin-like growth factor receptor (IGF-1R) signalling has been found to participate in the development and progression of many different types of human epithelial cancer, such as ovarian carcinoma, pancreatic cancer and sarcomas

[33,293,307,308]. Any changes to the balance of specific effectors in the IGF-1R system trigger a cascade of molecular events leading to malignancy. In addition, IGF-1R was involved in the development of resistance to different cancer treatment types, such as chemotherapy and radiotherapy [309]. Association of IGF-1R and malignant transformation was first found in fibroblasts derived from homozygous IGF-1R null mice embryos [33]. Mouse embryo fibroblasts lack of IGF-1R are resistant to malignant transformation by a number of oncogenes (e.g. Simian virus 40T antigen (SV40T) and

Ewing sarcoma fusion protein), but the susceptibility to malignant transformation is restored after re-expression of the IGF-1R [310].

While poor organ development leads to fatality at birth in mice with homozygous knockout of IGF-1R, IGF-1 deficient mice were used as an alternative mouse model to study the in vivo effect of decreased IGF-1R signalling on cancer development [311].

Low level of IGF-1 showed reduced growth and metastasis of tumours and xenografts, and also increased resistance to tumour induction by carcinogens [312], whilst transgenic 50

overexpression of IGF-1 in basal epithelial cells of mice induced hyperplasia and well- differentiated adenocarcinomas of skin and prostate [313,314].

Transgenic overexpression of IGF-1R in mice showed escalation of metastases and spontaneous development of invasive adenocarcinomas of salivary and mammary glands [315,316]. Growth of many cancers is influenced by the level of circulating IGF secreted in distant tissues and the capability of IGF self production gradually acquires in an autocrine and/or paracrine manner. Thus, when the level of IGF increased the cancer becomes more aggressive [290,317]. The correlation between IGF-1R and p53 plays a significant role in tumour growth and metastasis. The wild-type p53 has shown its capability in suppressing IGF-1R promoter activity and reducing the endogenous level of IGF-1R mRNA [318]. Loss of the p53 function up-regulates IGF-1R leading to the increased distribution of MDM2 followed by further deactivation of p53 and the malignancy induction of cells [281].

In addition, the IGF-1R downstream signalling pathways, PI3K and MAPK, was demonstrated to play a role in tumorigenesis and maintenance of a transformed phenotype [319,320]. Enhance proliferation by the interaction of the anti-apoptosis Bcl-2 family members with p53/MDM2 and the loss of tumour suppressor genes PTEN are associated with the up-regulation of IGF-1R signalling in cancer [281,290,317]. Expression of IGF-1R is also involved in anchorage independent growth, which is a unique property of malignant cells [281]. These animal studies and signalling correlations provides proof of principle that signalling activation of IGF-1R facilitates malignant transformation of tumour, as well as enhances growth, progression, invasion and metastasis of the established tumours.

Several in vitro studies have demonstrated the expression of IGF ligands and IGF-1R in various epithelial cancer samples, which includes breast, colon and prostate cancer 51

[33,293]. Although the function of IGF-1R and its system have not been fully evaluated and understood in cancers, all evidence have shown its participation and significance in cancer biology. An increasing amount of studies further demonstrated the importance of IGF-1R-directed cancer therapy and many targeting strategies are developed and being investigated in clinical trials [33,321,322].

It has recently been reported that the insulin-like growth factor (IGF) system (including ligands, receptors, IGF binding proteins and proteases) plays an important role in the formation and homeostasis of bone [323], and also in Ewing’s sarcoma [324]. Consistent expression of IGF-1R was identified in osteosarcoma tissue samples and cell lines, but the expression of IGF-I and IGF-II was differentially expressed [104,105]. However, the expression of IGF-I, IGF-II and IGF-1R showed no correlation with the metastatic lesion of the disease [104,325] and the correlation of IGF-1R with the grade and survival of osteosarcoma patients was not assessed. Although IGF-I dependent growth was observed in some osteosarcoma cell lines in in vitro survival and proliferation [326], no clear evidence showed that IGF-I and IGF-II functions as a dominant autocrine growth or motility factor in osteosarcoma [105]. A study showed that lowering the serum level of

IGF-1 had no sustained clinical response in osteosarcoma patients [327]. Thus, the development and progression of osteosarcoma may be more associated with IGF-1R.

2.7.5. Current strategy of IGF-1R inhibition or down-regulation

Due to its role and involvement in tumorigenesis, researchers have suggested that

IGF-1R and its system are potential and promising targets for cancer treatment

[282,328,329]. Inhibition or down-regulation of the IGF-1R signalling inhibits cellular transformation to a malignant phenotype and induces apoptosis in tumour cells [330].

Studies have also shown that such anti-proliferation is effective to various types of 52

cancers such as glioblastoma, melanoma, neuroblastoma, prostate cancer, colon cancer, rhabdomyosarcoma, lung cancer, medulloblastoma and osteosarcomas

[286,331,332].

Several strategies have been investigated in targeting the IGF-1R system and signalling, including (i) inhibition of ligand, ATP or substrate binding, (ii) induction of

IGF-1R ubiquitination and degradation, (iii) reduction of IGF-1R expression, (iv) inhibition of interaction with IGF-1R downstream signalling molecules, and (v) inhibition of tyrosine kinase activity by the use of monoclonal antibodies, tyrosine kinase inhibitors, recombinant proteins and siRNAs [33,328,330,333,334] (Figure 2.11).

Figure 2.11 Various strategies for inhibition or down-regulation of the IGF-1R signalling pathways and the mode of action [330]. 53

IGF-1R is highly similar to IR structurally, with 84% homology of the protein sequence within the kinase domains and 100% homology at the ATP-binding pocket [335]. Thus, the use of antagonistic monoclonal antibodies and small molecule tyrosine kinase inhibitors in targeting the IGF-1R system are probably the most clinically viable option

[336] and many of these are in the clinical and preclinical development stages [33,321,334].

Monoclonal antibodies, such as EM164 and IMC-A12, were developed to target the interaction of IGF-I and IGF-1R and to prevent the activation of downstream signalling

[336], which subsequently inhibited proliferation in breast cancer cells [337] and reduced tumour growth in the human xenograft tumour model [338]. In addition, advances in technologies have helped the development of small molecule inhibitors, such as BVT-

51004 (also known as PPP) [322], with specificity and selective blockage of targeting

IGF-1R but not IR.

Several studies utilised mouse model to demonstrate that tyrosine kinase inhibitors, such as NVP-AEW541 and BMS554417, could be orally administrated, while IMC-A12, h7C10 and EM164 could only be administrated intravenously or intraperitoneally [321,339-

341]. Hence, different routes of administration (intravenous for antibodies versus oral for most small-molecule tyrosine kinase inhibitors) and mode of action (blocking access to the activating ligand for most antibodies versus inhibiting the receptor kinase activity for small molecules) have provided a full spectrum detailing various aspects of inhibiting

IGF-1R as a therapeutic cancer treatment [336].

2.7.6. IGF-1R tyrosine kinase inhibitor

Tyrphostin AG1024 (also known as 3-Bromo-5-t-butyl-4-hydroxy-benzylidenemalonitrile, molecular formula: C14H13N2BrO) is a small molecule tyrosine kinase inhibitor that is commercially available (Figure 2.12). This drug is classified as “substrate competitive 54

inhibitors” and it is specifically targeting the IGF-1R tyrosine kinase domain. The active centre of IGF-1R tyrosine kinase domain was blocked by the binding of Tyrphostin

AG1024, which restricts the access of substrate or ATP to the receptor [342].

Figure 2.12 Molecular structure and formula of Tyrphostin AG1024. Details obtained from catalogue of Calbiochem (San Diego, CA).

Unlike the ATP competitive inhibitors that target the conserved ATP domain and require a higher dosage to overcome the high concentration of intracellular ATP, substrate competitive inhibitors provide higher selectivity over the drug target and require lower dosage in application [343]. Thus, the selectivity and tolerability of substrate competitive inhibitors is beneficial to clinical practice [322].

Tyrphostin family is a group of low molecular weight synthetic compounds including

Tyrphostin AG1024, that have been identified as potent inhibitors of protein tyrosine kinases [344]. Members of this family inhibit autophosphorylation of receptors by targeting tyrosine kinase domain of different cellular growth factor receptors, such as

IGF-1R. Tyrphostins were designed based on modelling of the tyrosine phenolic group in the microbial inhibitor erbstatin and the benzylidene malononitrile nucleus, with the subgroup modifications to direct biological activity [344]. Thus, these modifications are the essential component in discriminating amongst different tyrosine kinase domain on different receptors. 55

Recent studies have proposed the wider application of Tyrphostin AG1024 by using this inhibitor to enhance the radiosensitivity of breast and lung cancer cells [345,346]. This

IGF-1R specific inhibitor is metabolised intracellularly and creates substances with an increased activity towards the down-regulation of the IGF-1R downstream signalling.

For example, treatment with Tyrphostin AG1024 in breast cancer cells has down- regulated anti-apoptotic signalling proteins, such as phospho-Akt1 and Bcl-2, and up- regulated the pro-apoptotic proteins, such as Bax [345]. Thus, apoptosis was increased.

In addition, Tyrphostin AG1024 directly suppressed the activation of the IGF-1R downstream effectors IRS-1 and Shc by inhibiting their phosphorylation in lung cell lines [347].

The IGF-1R inhibitor Tyrphostin AG1024 has higher specificity towards IGF-1R by exhibiting a lower IC50 for the IGF-1R than IR. This higher specificity is an essential factor for the IGF-1R inhibitors to distinguish between IGF-1R and IR and prevent cross-reactivity [342]. The growth inhibition effects of Tyrphostin AG1024 in tumour cells have proved its potential in cancer treatment.

2.7.7. MEK/ERK inhibitor

Commercially available protein kinase inhibitors were developed in targeting signalling pathway, and is useful for identifying the physiological substrates and studying of cellular functions associated with these protein kinases [348]. Many of these compounds were known to inhibit a specific target and commonly used in cancer related research, but some of the compounds have shown to be effective in protein kinase other than the presumed targets [349].

U0126 (also known as 1,4-diamino-2,3-dicyano-1,4-bis[2-aminophenylthio] butadiene) is a MEK/ERK inhibitor, which prevents activation of MEK1/2 and blocks activated 56

MEK1/2 [350]. Study have shown that inhibition of MEK1 and MEK2 by U0126 inactivated the downstream transactivation of AP-1 [351]. In addition, blocking the MEK pathways by U0126 inhibited anchorage-independent growth and induced apoptosis in human breast cancer cell lines, in which the MEK signalling pathway is constitutively activated [352]. Moreover, inhibition of the MEK pathways by U0126 induced growth arrest of embryonal rhabdomyosarcoma cells with c-Myc down-regulation [350] and aided the verification of the -tomatine anti-metastatic effect in human lung adenocarcinoma cells [353]. 57

2.8. IGF-1R targeted therapy and chemotherapy

2.8.1. Chemotherapy

Improvements in chemotherapy in the last decade make it one of the standard protocols in cancer treatment, for local and metastatic control of cancer. Adjuvant and neoadjuvant chemotherapy in combination with surgical excision of the tumour have served the best prognosis for patients with some cancer types, such as osteosarcoma

[2]. Chemotherapy is of particular importance in the pelvis, head, neck, base of the skull axial skeleton and inoperable metastatic cancers, in which optimal resection with acceptable margins are difficult to achieve [354]. Nonetheless, prognosis of patients remains poor in some of the more progressive/metastatic cancers, e.g. pancreatic cancer, as chemotherapy could not improve the clinical situation and overcome the problem of chemoresistance development.

Multi-agent chemotherapeutic regimens and dose intensification showed no significant improvement, but were associated with substantial toxic effects particularly for older patients with co-morbidities [2,355]. The overall 5-year survival rate of currently undergoing chemotherapy protocol varied in different cancers with 50-60% for all stages of sarcoma [356], 28% for stage IV breast cancer [357], 15% for lung cancer, and

11% for hepatocellular carcinoma[358].

Chemotherapeutic agents commonly used in clinical cancer treatments have different modes of action against tumour cells along the cellular processes, generally causing cell-cycle arrest, induction of apoptosis and/or cytotoxicity (Figure 2.13). Although these chemotherapeutic agents are effective in cancer treatments, the frequency of chemoresistance development greatly affected the application of these agents. Innate and acquired resistances to the cytotoxicity of these agents are both identified in 58

clinical practice. Moreover, side effects are problematic when the cytotoxic drugs are used in high dosage.

Figure 2.13 Mechanism and mode of action of chemotherapeutic drugs commonly used in cancer treatments. The lines indicate where the drug is effective along the cellular process [354].

2.8.2. IGF-1R chemoresistance

Talking about chemoresistance, it lets us immediately think of the multi-drug resistance gene (MDR) and its encoded protein P-glycoprotein, which have been thoroughly investigated [359,360]. P-glycoprotein is an ATP-dependent drug efflux pump for molecules with broad substrate specificity. Expression of MDR and/or P-glycoprotein has been identified in drug resistant cells with reduction in drug accumulation and correlated with the poor clinical outcome of chemotherapy. Enhanced cytotoxicity and apoptosis was identified by the inhibition of P-glycoprotein and suppression of MDR

[359,360]. One recent study showed that the use of non-steroidal anti-inflammatory drugs 59

(NSAIDS), which inhibit the function of P-glycoprotein, increased the sensitivity of the multidrug resistant human sarcoma cells to chemotherapeutic agents, such as

Doxorubicin [361].

Recently, several studies have shown the involvement of IGF-1R in the chemoresistance of many cancers, such as sarcomas, melanoma, breast cancer, lung cancer and prostate cancer [362-368]. Unlike MDR and P-glycoprotein that only aim at reducing the intracellular drug concentration, IGF-1R plays a role in chemoresistance by mediating its downstream signalling pathways. As IGF-1R is overexpressed in the malignant cells, different pathways controlled by its downstream signalling are stimulated, such as tumour growth, inhibition of apoptosis, up-regulation of survival signals, enhancement of DNA repair and promotion of cell cycle progression [354].

IGF-1R system was identified to play a critical role in chemoresistance development and survival of cancer cells during treatments. The vital survival mechanism “double strand breaks repair”, which depends on Ku86 expression controlled by p38 but not

PI3K signalling, is down-regulated after the inhibition of IGF-1R. This enhanced the effectiveness of radiotherapy in non-small cell lung cancer cells [369]. Attenuation of IGF-

1R signalling via Akt, ERKs and p38 also showed enhancement of sensitivity to drugs that caused DNA damages in prostate cancer [333]. In addition, activation of EGFR downstream signalling after knockdown of IGF-1R showed no effect to the chemotherapeutically induced apoptosis in breast cancer cells, signifying the importance of IGF-1R and its signalling system in the survival and chemoresistance mechanism [370]. Furthermore, it has been shown that high expression of IGF-1R is correlated with the induction of chemoresistance in patients after long-term treatment

[371]. Thus, the IGF-1R and its signalling pathways mediated a wide range of mechanism that involves resistance to cytotoxic chemotherapy and radiotherapy as well as other targeted therapies, including Tamoxifen and Trastuzumab [309]. 60

2.8.3. IGF-1R targeted therapy

Targeted therapies aim to target specific key molecules of tumour cells and/or other associated cells but with minimal or no effects to normal cells or tissues [372]. Generally, either small molecule inhibitors or specific antibodies are used to target proteins involved in tumorigenesis, tumour progression and/or metastasis (Figure 2.14). The aim of targeted therapy is to reduce the dosage and side effects from cytotoxic drugs, and propose a higher efficacy than chemotherapeutic drugs. However, effectiveness of treatment relies on the ability to individualise treatment based on the molecular characteristic presented in individual tumour [373], in which the heterogenic properties of tumour are associated with the variation in sensitivity of targeted therapy. Targeting factors that are overexpressed in tumours but ubiquitously expressed in normal tissue throughout the body , for example the IGF-1R, pose a concern in the physiological impacts for such treatment [33]. Thus, drug safety and resistance to treatments are some concerns in the development of targeted therapy. 61

Figure 2.14 Depicted are some of the novel agents (indicated in boxes) targeting the cellular signalling pathways. The targets of these drugs involve proliferation, angiogenesis, and differentiation in neoplasms with some of these targets amenable to therapeutic interventions in cancer therapy. Membrane-bound human epidermal growth factor receptors (HER), c-MET, and insulin-like growth factor 1 receptor (IGF-1R) mediate mitogenic signals by the binding of their extracellular ligands. The Ras/Raf/MEK/Erk (mitogen-activated protein kinase, MAPK) and PI3k/Akt/mTOR pathways are major intracellular signalling pathways modulated by those extracellular membrane-bound receptor. DNA methytransferases (DNMT) and histone deacetylases (HDAC) are “epigenetic switches” that regulates the expression of oncogenes and tumour suppressor genes [374].

The crucial role of tyrosine kinases and their receptors in many cancers attract substantial attention in anti-cancer drug development. For example, Imatinib (also known as Gleevec) is a small molecule inhibitor that blocks the activity of tyrosine kinase of the fusion oncoprotein, BCR-ABL, in Philadelphia Chromosome Positive chronic myelogenous leukaemia and the “Kit” protein in Gastrointestinal Stromal

Tumours (GIST) [355]. Furthermore, the use of Herceptin against HER2 in breast cancer, 62

and Gefitinib against EGFR in non-small cell lung cancer, have all demonstrated its potential in clinical applications [European Organization for Research and Treatment of

Cancer protocol 62022 (Gefitinib), NCI clinical trial ID: NCT00104949 (Herceptin)] [372].

IGF-1R targeted therapies have been extensively explored and showed promising results with high effectiveness and acceptable side effects. The high percentage of homology between IGF-1R and IR is responsible for the induction of hypo- and hyperglycaemic condition from non-specific anti-IGF-1R treatment. However, the benefits from this treatment are more profound than the manageable mild side-effects.

As IGF-1R is expressed throughout the body in normal tissue, the interference with some important physiological functions of normal cells by application of IGF-1R targeted therapy posed a significant safety concern [292]. Thus, drugs specific to IGF-1R is an essential factor in the development of IGF-1R targeted therapy.

Multiple antibodies and small molecule inhibitors were identified with different strategies and mechanism in blocking the function of IGF-1R (Table 2.6). Antibodies provided higher selectivity of IGF-1R by inducing rapid internalization and down- regulation of the receptor. In contrast, small molecule inhibitors target the intracellular tyrosine kinase domains. Thus, multiple inhibitors are used to compensate the lack of specificity, which do not induce any internalization or down-regulation of IGF-1R.

Furthermore, current development in anti-IGF-1R therapy allows the control over the duration of drug exposure by the application of long-lasting effects from antibodies or shorter effects from small molecules inhibitors [375]. Ongoing development of highly specific anti-IGF-1R therapy is required with the potential for lowering drug dosage of cytotoxic chemotherapeutic agents, which will minimize their adverse side-effects and effectiveness. 63

Table 2.6. Compounds in clinical and preclinical developments that target the IGF- 1R and their application in combination therapy [321].

Type of Compound Company Development Type of Drug shown agent phase of testing tumours efficacy in targeted combination Antibody CP-751,871 Pfizer Phase I MM Adriamycin, IgG2 (IGF1binding) 5-FU

Antibody EM164 / ImmunoGen / Preclinical – 90 in vitro and 27 Gemcitabine AVE1642 Sanofi-Aventis in vitro, in vivo in vivo cell lines

Antibody IMC-A12 ImClone Phase I Clinical solid C225 (IGF1 binding) tumours

Antibody R1507 Roche Phase I Solid tumours in NR vivo Colo205, NCI-H322M

Antibody AMG479 Amgen Phase I In vivo Colo-205, Gemcitabine, BxPC-3, Irinotecan MiaPaCa

Antibody 19D12 Schering Preclinical – NSCLC, ovarian NR in vitro, in vivo A2780

Antibody h7C10 Pierre Fabre / Preclinical – MCF-7, A549 Navelbine, C225 (IGF-1 binding) Merck in vitro, in vivo etc.

TK PPP Karolinska Preclinical – MM, uveal NR inhibitor Institute in vitro, in vivo melanoma non-ATP

TK AG 538 / Hebrew Preclinical – Breast, prostate, NR inhibitor AG1024 University of in vitro, in vivo leukemia etc. Jerusalem

TK Compound-1 OSI Preclinical – NSCLC Erlotinib inhibitor Pharmaceuticals in vitro, in vivo colorectal

TK NVP-ADW742 Novartis Preclinical – MM42, SCLC Imatinib, inhibitor in vitro, in vivo cell lines Etoposide, Carboplatin

TK NVP-AEW541 Novartis Preclinical – Sarcoma, NET Vincristine, inhibitor in vitro, in vivo Ifosfamide

TK BMS-554417 Bristol-Myers Preclinical – Leukemia, breast NR inhibitor Squibb in vitro, in vivo and ovarian cancer

TK BMS-536924 Bristol-Myers Preclinical – Prostate, colon, NR inhibitor Squibb in vitro, in vivo pancreatic Abbreviations: FU, fluorouracil; IGF, insulin-like growth factor; MM, multiple myeloma; NET, neuroendocrine tumours; NA, data not available; NR; data not reported; NSCLC, non-small-cell lung cancer; PPP, picropodophyllin; SCLC, small- cell lung cancer; TK, tyrosine kinase. 64

2.8.4. IGF-1R combination therapy

Medicinal drug responses among individuals are different, while some patients exhibit life-threatening adverse reactions others fail to show an expected therapeutic effect.

Occasionally, intermediate responses between the above two extreme cases are also observed. Application of conventional cancer chemotherapy is limited by drug resistance commonly exhibited in tumour cells. Although cancer remains a devastating disease, the development of molecular medicine and the improved understanding of tumour biology and pathogenesis in the last decade have helped to identify some new therapeutic targets and treatment modalities. One of the strategies to improve the current clinical cancer treatment situation is the use of combination therapy. Currently, combination therapy includes the use of two or more different forms of treatment methods, aiming to enhance the effectiveness of treatments or achieve the same cancer inhibition effect with reduced dosage. However, the use of more than three treatment methods has not been investigated due to potential complications, increased cytotoxicity and unknown interactions that may occur. Combination of targeted therapy and chemotherapy has been extensively studied in cancer and has shown promising outcomes [376-378]. The use of IGF-1R targeted therapy and chemotherapy is a good example of the advantage in combination therapy [354].

The IGF-1R system is known to play an important role in chemoresistance [309].

Inhibition of the IGF-1R system and signalling cascade will increase the sensitivity to chemotherapy and reduce the development of chemoresistance in tumour cells. This combination strategy is remarkable as it increases chemotherapeutic effectiveness and leads to dosage reduction for achieving the same cancer inhibitory effect compared to chemotherapy alone.

IGF-1R kinase inhibitor NVP-AEW541 was tested in combination with several chemotherapeutic agent in Ewing’s sarcoma, osteosarcoma and rhabdomyosarcoma 65

cell lines[379]. The result showed that only the combination of NVP-AEW541 and

Vinivristine could synergistically inhibit Ewing’s sarcoma in the xenograft tumour model, while other combinations yielded positive interactions and sub-additive effects. More investigation is required to clarify the role of molecular pathology downstream of the

IGF-1R in reaction to the therapy, the physiological effects after long-term therapy in conjunction with chemotherapy, and the sequence of co-administration of drugs.

A study showed that loss of PTEN function would be a resistance marker but is not necessarily associated with resistance to cancer treatment and results in hypersensitivity to IGF-1R upstream stimulation but not constitutive IGF-1R pathway activation [380]. Down-regulation of tumour suppressor PTEN, which normally negatively regulates the downstream PI3K/Akt signalling of IGF-1R, is implicated in the onset and progression of glioblastoma, breast cancer, and in other cancers. Thus, the use of IGF-

1R targeted therapy under PTEN negative situation will not increase the efficacy of treatment.

Successful development of small molecule tyrosine kinase inhibitors and antibodies directed against the IGF-1R has opened up its applications in combination with chemotherapy. The abovementioned researches have shown the efficacy of the combination therapy and its advantages, including convenience of drug administration.

At this stage, the potential of combination applications involves many organ sites, creating challenges for translational researchers seeking to design better preclinical and clinical investigations. There are still many questions to be answered before IGF-

1R targeted therapy could be declare as the most efficacious strategy in combination with chemotherapy for cancer treatments. Preclinical and clinical studies are in progress to investigate the different strategy of the IGF-1R targeted therapies (Table

2.6). 66

CHAPTER 3. MATERIALS AND METHODS 3

3.1. Materials

3.1.1. Reagents and consumables for tissue culture

RPMI1640 media powder including L-glutamine (31800-014), penicillin/streptomycin solution (15140-122), Dulbeco Phosphate Buffered Saline powder (DPBS) (21600-051),

Fetal bovine serum (FBS) (10099-141), Trypsin-EDTA solution (15400-054) were purchased from Gibco® via Invitrogen Life Technologies (NY, USA). Sodium bicarbonate (S-5761), (P4458), L-ascorbic acid (A-8960), L-glutamine (G-5736), !- glycerphosphate (G-9891), dexamethason (D-4902), dimethyl sulfoxide (DMSO) (D-

8418), Doxorubicin hydrochloride (D-1515) were purchased from Sigma-Aldrich (St.

Louis, MO, USA). Tyrphostin AG1024 was from Calbiochem (La Jolla, IL, USA). All culturing dishes, flasks, plates and pipettes were purchased from BD Biosciences (NJ,

USA). Cryovial was purchased from NUNC (Roskilde, Denmark). The 0.22 )m membrane syringe (SLLGH04NL) and bottle-top filter (SCGPT02RE) were purchased from Millipore (MA, USA). Further details can be found at Appendix 1 (C).

3.1.2. Reagents for histochemistry, immunohistochemistry and immunocytochemistry

Positively charged “Ultrafrost” 25 x 75 x 1.0 mm slides (J3800AMNZ), coverslips and mounting media were purchased from Lomb scientific (NSW, Australia). Primary 67

antibodies were commercially purchased and details can be found at Appendix 1 (D).

EnVision™ system with anti-mouse (K4000) and anti-rabbit (K4002) HRP labelled polymer and liquid DAB+ substrate chromogen system (K3468) were purchased from

DAKO (CA, USA). Universal in situ hybridization DAB system (SH-2019-06) and oligonucleotide probe for human Alu DNA (PR-1001-01) was purchased from Biogenex

(CA, USA). All chemicals used for staining were purchased from Sigma-Aldrich (St.

Louis, MO, USA) and details can be found at Appendix 1 (A).

3.1.3. Reagents for molecular detections

Absolute ™ blue QPCR SYBR® green mix plus ROX (AB-4166) was purchased from

ABgene through Thermo Fisher Scientific, Australia. TRI reagent (AM9738) was purchased from Ambion (OH, USA). Amersham ECL western blotting analysis system

(RPN2109) and Hyperfilm ECL were purchased from GE Healthcare, Australia.

SuperScript® III first-strand synthesis system for RT-PCR was purchased from

Invitrogen Life Technologies (NY, USA). All other chemicals used for molecular detection were purchased from Sigma-Aldrich (St. Louis, MO, USA) and details can be found at Appendix 1 (A).

3.1.4. Equipment

Multiplate and RNA/DNA absorbance:

A Sunrise™ plate reader (TECAN, Männedorf, Switzerland) was used for

“Absorbance/Optical Density” readings. The DNA and RNA concentration and quality were determined by absorbance reading at wavelength of 260 and 280 nm with a 68

GeneQuant Pro (Amersham Biosciences, Buckinghamshire, UK) and NanoDrop® ND-

1000 spectrophotometer (Nanodrop technologies, DE, USA).

Flow Cytometry and Microscopy:

The flow cytometry experiment was carried out in a “Cell Lab Quanta™ SC MPL” from

Beckman-Coulter (Brea, CA, USA). For histology and immunocytochemistry assays, the images were viewed using an Olympus upright microscope BX51TRF and images taken using an Olympus universal digital micro-imaging camera DP72 (Olympus,

Tokyo, Japan).

Reverse transcription and qRT-PCR:

A “Hybaid PCR express” thermal cycler machine (Thermo Fisher Scientific, VIC,

Australia) was used for cDNA synthesis. An “Mx3000P QPCR” system manufactured by Stratagene™ (La Jolla, CA, USA) was used to perform qRT-PCR.

Electrophoresis (agarose and SDS-PAGE)

An UVItec gel documentation system and UVIDocMw software (Version 99.03 for

Windows, UVItec Limited, Cambridge, UK) were used for electrophoresis documentation and analysis. A “Mini PROTEAN III Electrophoresis” and a “Mini Trans-

Blot® Electrophoretic Transfer” system (Bio-rad, Hercules, California, USA) were used in western blotting analysis.

Mouse imaging and microtomography:

X-ray radiography was performed with a Faxitron MX 20 (Faxitron X-ray corporation,

Wheeling, IL, USA). Micro-CT scans of mouse bones were performed with a Skyscan

1072 (Skyscan, Kontich, Belgium) scanner and images were analysed with NRecon software (Skyscan, Kontich, Belgium) and CTan software (Skyscan, Kontich, Belgium). 69

Animal tissue processing and histochemistry

Mice tissues were processed using a Shandon Excelsior ES Processor and were embedded into paraffin blocks using a Shandon Histocentre 3 (Thermo Fisher

Scientific, VIC, Australia). A Leica microtome RM2165 (Leica Microsystem GmbH,

Wetzlar, Germany) was used for cutting paraffin embedded specimens. The specimen placed onto slides for further histochemical or immunostaining process.

3.1.5. Mammalian cell lines

Six different commonly used human osteosarcoma cell lines and a mouse fibroblast cell line were used in this project. The human osteosarcoma cell lines Saos-2, U-2 OS,

MG63 and the mouse fibroblast cell line were purchased from American Type Culture

Collection (ATCC, Manassas, USA), while the human osteosarcoma cell lines HOS,

143B and SJSA were kindly provided by Dr. D. Thomas (Peter MacCallum Cancer

Centre, University of Melbourne, Melbourne, Australia)

3.1.6. Animals

Ethical approval was obtained from the UNSW animal ethics committee (ACEC

08/66A). Athymic BALB/c nu/nu (‘nude’) mice were obtained from the Animal

Resources Centre (Perth, Australia).

3

3 70

3.2. Methods

3.2.1. Mammalian cell culture

3.2.1.1. Media and supplements preparation

RPMI1640 powder was reconstituted using Milli-Q water with the addition of 2 g/L of sodium bicarbonate as per manufacturer’s instructions. The reconstituted medium was sterilized by filtration through a 0.22 µm bottle-top filter cup (Millipore) into pre-sterilized

Schott bottles (Schott, Germany) and stored at 4°C until use. Any alteration in the pH of the medium, which contained phenol red, would be reflected by a colour change in the medium.

Supplements including fetal bovine serum, antibiotics and other required supplements as stated in Appendix 1 (C) were aliquoted and stocked at - 20°C after filter sterilized using a 0.22 µm syringe-fit filter unit (Millipore) until use. Supplements were then added to the prepared RPMI1640 medium as required for the different conditions in this project.

3.2.1.2. Cell resuscitation

Cells were thawed in a water bath at 37°C immediate ly after removal from liquid nitrogen storage and transferred onto a 100 mm tissue culture plate in 10 mL of pre- warmed growth medium. The tissue culture plate was incubated overnight under standard conditions at 37°C and regulated with 5% C O2. The medium was replaced the next day with fresh pre-warmed growth medium.

An alternate thawing process was carried out for HOS, where the cell line was more sensitive to DMSO. The thawed cell suspension was transferred into a 15 mL tube and was thoroughly mixed with 10 mL of pre-warmed growth medium gently prior to centrifugation at 500 x g for 10 min. The supernatant was discarded and the cell pellet 71

was resuspended in 10 mL of fresh pre-warmed growth medium and allowed to grow on a 100 mm tissue culture plate until ready for passage.

3.2.1.3. Cell passage

Cell passage was carried out when cell growth reached 80% confluence or above.

Firstly, the growth medium was aspirated and attached cells were rinsed twice with pre- warmed DPBS followed by incubation at 37°C for no m ore than 15 min with 5 ml of pre- warmed trypsin-EDTA solution per 100 mm plate for cell detachment. Trypsin was inactivated by adding 5 mL of serum containing growth medium and transferred into a

15 mL tube. The cell suspension was then centrifuged at 500 x g for 10 min and the supernatant was discarded. The cell pellet was resuspended thoroughly in 5 mL of pre- warmed fresh growth medium.

Concentration and viability of cells were assessed by trypan blue dye (0.5% w/v trypan blue, 0.8% w/v NaCl), where only dead cells with damaged membrane will be stained blue, while healthy live cells will not be stained. Equal amount (100 µL) of filtered trypan blue solution was mixed with the cell suspension and 10 µL of the mixture was transferred onto the counting chamber of a haemocytometer and the viable cells were counted and recorded. The cell suspension was then diluted to the required density before seeding onto a new tissue culture plate and the medium was replaced every third day.

3.2.1.4. Cryostorage

Cells were cultured until confluency reached 80% or more and cell concentration was determined as described in section 3.2.1.3. Cells were resuspended into 2 x 106 cells/mL containing 5% v/v DMSO and 1 mL of cell suspension was transferred into cryovials. The vials were placed into a “1° freezin g container” (Nalgene® Labware, NY, 72

USA) and stored at -80°C for a maximum of two weeks before being transferred into a liquid nitrogen tank for long-term storage.

3.2.1.5. Establishment of a Doxorubicin resistant osteosarcoma cell line

Parental osteosarcoma cell line was cultured (section 3.2.1.3), followed by intermittent exposure to gradually increasing concentration of Doxorubicin [381,382]. Osteosarcoma cells were cultured at each Doxorubicin concentration until the treated cells achieved a growth rate similar to that of the untreated cells, before being advanced to a higher concentration. Approximately 3 - 4 weeks (4 - 8 passages) were required to establish an acceptable growth at each new drug concentration as stepwise increment of drug concentrations produced variants, which adapted to different level of drug concentrations. Variants were continuously cultured in the presence of the drug for at least 4 weeks before cell stocking or usage for experiments.

3.2.2. Molecular Techniques

3.2.2.1. Total ribonucleic acid (RNA) extraction

The procedure of total RNA extraction was based on the TRI Reagent® (Ambion, Inc.,

USA) product manual with minor modifications. Briefly, cells were cultured to 80 - 90% confluence in a 100 mm plate and 1 mL TRI reagent was added directly to the cells after the medium was aspirated. The plate was incubated for 5 min at room temperature and the sample was homogenized with the aid of a pipette. The cell lysate was transferred into a 1.5 mL centrifuge tube and mixed thoroughly with 100 µL of bromochloropropane (BCP) followed by 15 min incubation at room temperature with occasional mixing. The mixture was centrifuged at 12,000 x g for 15 min at 4°C. The upper aqueous layer was transferred to a fresh 1.5 mL centrifuge tube and mixed with 73

1 volume of isopropanol by inverting the tube several times until homogenized. The homogenized sample mixture was incubated for 10 min at room temperature and then centrifuged at 12,000 x g for 10 min at 4°C. The su pernatant was removed and 500 µL of freshly prepared 75% v/v ethanol made with RNase-free water (DEPC treated water) was added gently to wash the pellet, followed by centrifugation at 10,000 x g for 5 min at 4°C. The ethanol was removed and the RNA pellet was allowed to air-dry for 5 min, followed by resuspending in 50 µL of RNase-free water.

All total RNA samples were treated with deoxyribonuclease I (DNase I) to remove any contamination of chromosomal deoxyribonucleic acid (DNA). For every 1 µg of total

RNA sample, 1 µL of 10 X reaction buffer (200 mM Tris-HCl at pH 8.4, 20 mM MgCl2,

500 mM KCl) was added and made up to 9 µL with RNase-free water, followed by the addition of 1 µL of DNase I (1 unit/µL). The sample mixture was then treated for 30 min at 37°C. The DNase I enzymatic reaction was inactiv ated by treating the sample with 1

µL of 25 mM EDTA (pH 8.0) for 10 min at 65°C.

The DNase I treated total RNA samples were purified by mixing thoroughly with 1 volume of TRI Reagent® and the mixture was incubated for 5 min at room temperature.

BCP was added in 1:100 to the volume of TRI and mixed by vortex, followed by incubation for 15 min at room temperature. The mixture was centrifuged at 12,000 x g for 15 min at 4°C, and the upper aqueous layer was transferred into a fresh 1.5 mL centrifuge tube. The sample was mixing thoroughly with 1 volume of isopropanol and incubated for 2 h at -80ºC after, followed by centrifugation at 12,000 x g for 10 min at

4°C. The RNA pellet was washed gently with 500 µL o f freshly prepared 75% v/v ethanol with RNase-free water after the supernatant was removed, followed by centrifugation at 10,000 x g for 5 min at 4°C. The ethanol was removed and the RNA pellet was allowed to air-dry for 5 min, followed by resuspending in 50 µL of RNase- free water. 74

3.2.2.2. Assessment of RNA samples

RNA concentration was determined by optical density measurement at 260 nm and the quality was assessed by the ratio of optical density measurement 260/280 and 260/230 using GeneQuant Pro (Amersham Biosciences, Buckinghamshire, UK). The integrity of

RNA was also confirmed by visual inspection of the 28s and 18s rRNA on agarose gels.

Electrophoresis of RNA sample was carried out on a 1.5% w/v agarose gel with TBE buffer (89 mM Tris-base, 89 mM boric acid, 2 mM EDTA in Milli-Q water) at a constant

90 V for 30 - 45 min. In addition, RNA quantity and integrity were further assessed by using a NanoDrop® ND-1000 spectrophotometer (Nanodrop technologies, DE, USA) and “Bioanalyzer™” assay before microarray experiments.

3.2.2.3. Microarray experiment

One-cycle target RNA labelling, hybridization (Affymetrix™ Fluidics Station 450), scanning (GeneChip® Scanner 3000) and raw data acquisition of the Affymetrix™

GeneChip Human Genome U133 plus 2.0 Arrays were performed by the Clive and

Vera Ramaciotti Centre for Gene Function Analysis (UNSW, Sydney, Australia) following a standard procedure from Affymetrix™.

3.2.2.4. Microarray expression data processing, quality control and normalization

Fluorescence signal intensity data (Affymetrix™ “CEL” data files) and

“present/marginal/absent” call data (Affymetrix™ “CHP” data files) were processed with

GeneSpring™ GX 10.0.2 software (Agilent Technologies Inc., Santa Clara, CA).

Differentially expressed genes were identified according to the guided workflow in the program. Briefly, probe level intensity data were processed according to the “Robust

Multichip Average” (RMA) algorithm for background adjustment, normalization and log transformation of the perfect match values and then normalized to median for each gene prior to further analysis. Normalized data below the 20th percentile were removed. 75

Independent sample (unpaired/Asymptotic computation) t-test with Benjamini-

Hochberg correction was performed to distinguish genes that were statistically significant (p<0.05) between the triplicate experiments of the two human osteosarcoma cell lines. A list of the differentially expressed genes was generated by selecting genes with more than 3-fold expression changes and with “present” call only.

3.2.2.5. Gene ontology and clustering analyses for data mining

Genes identified from the microarray analysis were subjected to gene ontology (GO) classification and then functional annotation clustering by the DAVID Bioinformatics

Resources 2006 (National Institute of Allergy and Infectious Diseases, National

Institute of Health, Frederick, MD 21702; http://david.abcc.ncifcrf.gov/tools.jsp) with the default 13 annotation categories, including GO terms, protein-protein interactions, bio- pathways and gene functional summaries. Unsupervised hierarchical cluster analysis with “Euclidean Distance” was performed to identify specific genes. The gene expression profile identified from the microarray analysis was compared with the gene expression profile of mesenchymal stem cells, adipocytes, chondrocytes, fibroblast and osteoblasts obtained from “Gene Expression Omnibus” (National Centre for

Biotechnology Information – “NCBI”, http://www.ncbi.nlm.nih.gov/geo/) with accession number GSE9451 [383]. A list of differentially expressed genes was obtained by comparing the result acquired from the functional annotation clustering with the expression hierarchical clustering.

3.2.2.6. Generation of complimentary deoxyribonucleic acid

Synthesis of first strand complimentary deoxyribonucleic acid (cDNA) from total RNA was performed in accordance to the product manual of the SuperScript III™ first-strand synthesis system for reverse transcription polymerase chain reaction (RT-PCR) by

Invitrogen Life Technologies (Carlsbad, CA, USA). To summarize, 5 µg of total RNA sample was mixed with 1 µL of 50 µM oligo(dT)20, 1 µL of 10 mM dNTP mix and made 76

up to 10 µL with RNase free water. This sample mixture was heated for 5 min at 65°C and cooled down for a minimum of 1 min in ice. Synthesis buffer containing 2 µL of 10

X RT buffer, 4 µL of 25 mM MgCl2, 2 µL of 0.1 M DTT, 1 µL of RNaseOUT™ (40 unit/µL) and 1 µL of SuperScript™ III reverse transcriptase (200 unit/µL) was mixed with the sample mixture and the reverse transcription reaction was performed for 50 min at 50°C. The reaction was terminated by heating for 5 min at 85°C and then cooled in ice immediately. The sample was treated with 1 µL of RNase H for 20 min at 37°C.

The final cDNA product was stored at -20°C.

3.2.2.7. Dissociation curves analysis for qRT-PCR

The specificity of primers designed for qRT-PCR was confirmed by dissociation curve analysis. Dissociation curve analysis was required after each set of the qRT-PCR experiments that use SYBR green for detection. This analysis is to ensure that the results were representing the correct quantification with no interference.

The qRT-PCR experiment requires stringent conditions to achieve high accuracy. A well designed PCR primer with high specificity should only produce one form of amplicon. Production of multiple amplicons could be a result from cross reactivity of targets or formation of primer dimers and would obstruct the precision of the quantification process. Thus, a specific high quality primer should only display a single homogeneous peak from the dissociation curve analysis of various samples. Re-design of primer is required if more than one peak was identified. The dissociation curve of the amplicon amplified by the GAPDH primers is shown in Figure 3.1, where only one type of product is present at a particular dissociation temperature, and the height of peaks represented the different amount of products formed in each reaction. Higher initial concentration in the sample will results a stronger signal by the formation of more PCR products. 77

Figure 3.1 Dissociation curve analysis of GAPDH qRT-PCR primer. The peaks represented the serial dilution samples tested with the present of GAPDH. The sample with higher dilution was shown with lower peak height.

3.2.2.8. Efficiency of primers for qRT-PCR

Relative gene expression was analysed using the “Fold change = 2 -DDCt ” method [384].

In order to use the formula, the efficiency of primers has to be identified to take into account that the difference between the change in cycle threshold (DCT) of the house keeping gene and the cycle threshold (CT) of the gene of interest was less than 0.1 over a range of dilutions of the cDNA template. The efficiency of each primer was determined in accordance to the “Introduction to quantitative PCR - Methods and application guide” from Stratagene™ (La Jolla, CA, USA) by standard curve analysis

(Figure 3.2), which the primer efficiency was obtained from the slope of the standard curve with R2 (also known as Pearson Correlation Coefficient) 1 0.985 [385]. 78

Figure 3.2 Standard curve analysis of GAPDH qRT-PCR primer. The efficiency of the primer was tested against a set of serially diluted samples. The slope of the curve was used to determine the efficiency of the GAPDH primer for the qRT-PCR reaction.

3.2.2.9. Quantitative real-time polymerase chain reaction

Expression of a gene in cells was determined by qRT-PCR, which quantitatively analyses the mRNA level. Experiments were performed in accordance to the product manual of Absolute™ Blue QPCR SYBR® Green Mix Plus ROX Vial from ABgene®

(Epsom, Surrey, UK). Briefly, each reaction contained 12.5 µL of 2 X reaction mixture,

100nM of forward and reverse primer, 0.1 µL of cDNA sample and made up to 25 µL with PCR grade water. The details of the thermal cycling profile, dissociation curve profile and properties of qRT-PCR primers are shown in Table 3.1. 79

Table 3.1. Machine setting and primers information of qRT-PCR

(A) qRT-PCR thermal cycling profile

Step Temperature Time Number of Cycle Enzyme activation 95°C 15 min 1 cycle Denaturation 95°C 15 sec  Annealing 60°C 30 sec 40 cycles Extension 72°C 30 sec

(B) Dissociation curve profile

Step Temperature Time Number of Cycle Denaturation 95°C 30 sec 1 cycle Starting temperature 60°C 30 sec 1 cycle Melting step 60°C 10 sec 80 cycles

(C) qRT-PCR primers properties

Gene Primer Sequence Product size (bp) Reference sequence (NCBI) ACTB Forward ATCGAGCACGGCATCGTCAC 145 NM_001101 Reverse TCTTCTCGCGGTTGGCCTTG

ALP Forward CCACCACGAGAGTGAACCATGC 134 NM_000478 Reverse CATGAGCTGGTAGGCGATGTCC

DLX5 Forward CTTCCAAGCTCCGTTCCAGACG 140 NM_005221 Reverse AGGTAGGAGAGCAGTAGCCGTG

FGFR2 Forward GCCTCTCTTCAACGGCAGACAC 149 NM_022970 Reverse TTCCGCCATGACCACTTGCC

BGLAP Forward AGTGAAGAGACCCAGGCGCTAC 131 NM_199173 Reverse AGCCGATGTGGTCAGCCAAC

SP7 Forward TGCAACTGGCTCTTCTGCGG 137 NM_152860 Reverse TTGCTCAGGTGGTCGCTTCG

RUNX2 Forward TCCAGAATGCTTCCGCCATGC 100 NM_004348 Reverse GGCTTCCATCAGCGTCAACACC

RUNX3 Forward TCACTCAGCACCACAAGCCAC 104 NM_004350 Reverse AAGGAGCGGTCAAACTGGCG

GAPDH Forward AAGAAGGTGGTGAAGCAGGCG 117 NM_002046 Reverse AGCGTCAAAGGTGGAGGAGTGG 80

3.2.2.10. Protein/DNA array analysis and data mining

Nuclear protein was extracted and combined with labelled oligonucleotides to form protein/DNA complexes. Oligonucleotides were eluted from the complexes and hybridized onto TranSignal™ Protein/DNA Array I (Panomics Inc., CA, USA) according to the product manual. Information for the 54 distinct consensus binding elements on the array is available from Panomics (http://www.panomics.com). Chemiluminescence procedure with Hyperfilm ECL (Amersham Pharmacia Biotech, Uppsala, Sweden) was used for array signal detection. Exposed films were imaged using UVI-Doc Mw (version

99.03) software (UVItec, Cambridge, UK). Information of genes associated with each transcription factor was obtained from the array product manual and the NCBI database (http://www.ncbi.nlm.nih.gov/). Information of genes that are regulated and targeted by a particular transcription factor was obtained from the Cold Spring Harbor

Laboratory, Zhang’s Lab: Computational Biology and Bioinformatics Webpage,

“Transcriptional Regulatory Element Database” (TRED) (http://rulai.cshl.edu/TRED).

3.2.2.11. Protein extraction and sample assessment

Cells were cultured until confluence, washed and scraped in ice-cold PBS using a cell scrapper. The cell mixture was transferred into a pre-chilled 1.5 mL centrifuge tube and centrifuged at 1,500 rpm for 5 min at 4°C. The cell pellets were lysed in a freshly prepared lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% v/v

Triton X-100, 1% w/v sodium deoxycholate, 0.1% w/v SDS in Milli-Q water) containing

1% v/v protease inhibitor and 1% v/v phosphatase inhibitor (50 mM) for 10 min at 4°C with shaking. Solubilised proteins were separated from the cell debris by centrifugation at 14,000 x g for 15 min. Protein concentration was measured with a DC protein assay kit II in a 96-well plate and each sample was prepared in triplicates. The plate was incubated for 15 min at room temperature before reading at 620 nm on a Sunrise™ plate reader to obtain optical density (OD) values. 81

3.2.2.12. SDS PAGE electrophoresis and Western blotting

About 20 µg total protein per sample was loaded in a 10% w/v sodium dodecyl sulfate polyacrylamide gel for electrophoresis (SDS-PAGE) at a constant 150 V for 45-60 min in a Laemmli running buffer (250 mM Glycine, 25 mM Trsi-base, 1% w/v SDS in Milli-Q water). The proteins were then transferred onto a polyvinylidene fluoride (PVDF) membrane in a transfer buffer (40 mM glycine, 50 mM Tris-base, 0.04% w/v SDS, 20% v/v methanol in Milli-Q water) at a constant current of 250 mA for 1 h and blocked with a 5% w/v skim milk phosphate buffered saline solution (PBS) for 1 h at room temperature with gentle shaking. Primary antibody diluted in a 1% w/v BSA/PBS solution was added onto the membrane and incubated overnight with shaking at 4°C.

The dilution factor was 1:200 for IGF-1R antibody and 1:10,000 for anti-!-actin. The membrane was washed in PBST (PBS with 0.05% v/v Tween-20) three times for 5 min each prior to incubation with anti-mouse and anti-rabbit secondary antibody (1:5,000) diluted in a 1% w/v BSA/PBS solution for 1 h with shaking at room temperature.

Subsequently, the membrane was washed in PBST three times for 5 min each.

Detection of the band was performed with chemiluminescence ECL western blotting analysis system on an autoradiography film developed in a dark room with a developing and fixative solution (Ilford™). The intensity of the bands were later analysed with “Image J” software [386].

3.2.3. In vitro assays and drug treatment

3.2.3.1. Detection of alkaline phosphatase (ALP) activity

Cultured cells were washed twice with ice-cold 0.9% w/v NaCl buffered solution and lysed with 0.1% v/v Triton-X in 0.1 M Tris-HCl buffer (pH 9.8) for 10 min at -20°C. Total cell lysate was collected and centrifuged at 10,000 rpm for 10 min at 4oC. The 82

supernatant was collected in pre-chilled tube and total protein was quantified by “DC

Protein Assay” (BioRad Laboratories). ALP activity was assessed by measuring the absorbance at 620 nm at 5 min intervals in room temperature over a period of 1 h in diethanolamine-MgCl2 reaction buffer (1 M Diethanolamine, 0.5 mM MgCl2 in water) contained 3.8 mM p-nitrophenylphosphate as the substrate. ALP activity was expressed as nanomole of N-nitrophenol/mg of protein/min.

3.2.3.2. Detection of calcium deposition and mineralization

Cultured cells were washed with PBS twice and fixed with 3.7% v/v formaldehyde in

PBS for 15 min at room temperature. Excess fixative agent was rinsed off with Milli-Q water.

Alizarin Red S Staining

The cells were stained with Alizarin red S staining solution (40 mM Alizarin red S at pH

4.1) for 20 min at room temperature after the fixation process. The excess staining solution was rinsed off with Milli-Q water. The number of positive orange-red calcium nodules were counted and recorded under light microscopy within 30 min after staining. von Kossa’s Staining

The cells were treated with aqueous sliver nitrate solution (1% w/v sliver nitrate in Milli-

Q water) under UV light for 30 min at room temperature after the fixation process and excess solution was rinsed off with Milli-Q water. The cells were then stained with a sodium thiosulfate solution (5% w/v sodium thiosulfate in Milli-Q water) for 5 min at room temperature and excess staining solution was rinsed off with Milli-Q water. The stained cells were counterstained with nuclear fast red solution (0.1% w/v nuclear fast red, 5% w/v aluminium sulphate in Milli-Q water) for 5 min at room temperature and excess staining solution was rinsed off with Milli-Q water. The stained cells were 83

dehydrated in a series of ethanol and air-dried. The number of positive black mineral nodules formed around the cells were counted and recorded under light microscopy.

3.2.3.3. Crystal violet colorimetric assay

The cells were gently rinsed twice with pre-warmed di-valent PBS (PBS containing 1 mM CaCl2 and 0.5 mM MgCl2) and incubated in excess crystal violet staining solution

(0.5% w/v crystal violet, 20% v/v methanol in Milli-Q water) for 10 min at room temperature. The stained cells were gently rinsed with Milli-Q water at least 3 times to remove unbound stain and then air-dried completely. Equal volume of elution buffer (50 mM sodium citrate, 50% v/v ethanol in Milli-Q water) was added to each sample and incubated for 30 min with shaking at room temperature. The optical density (OD) of crystal violet dye eluted from each sample was measured at 570 nm by a spectrometer

(Sunrise™, Tecan Austria).

3.2.3.4. Characterization of cell growth

Prior to performing proliferation and other functional assays, it is required to determine the optimal seeding density, which allows cell growth over a period of time without reaching confluence. The cells were cultured in a 96-well tissue culture plate at four different concentrations 5 x 103, 2.5 x 103, 1.25 x 103 and 0.625 x 103 cells/well and in triplicates at each cell concentration. The cells were harvested at 24, 48, 72 and 96 h after inoculation and stained with crystal violet solution to determine the growth rate of the cells. Data was log transformed to determine the exponential growth phrase. The doubling time for each cell line was calculated by the following equation.

Equation 1 Doubling time analysis 84

where ln is the natural logarithm, CT1 is the concentration of cell at T1, CT2 is the concentration of cell at T2, T1 is the time point number one and T2 is the time point number 2.

3.2.3.5. Combination drug effectiveness analysis

The effectiveness of drug treatment was determined by analysing the anti-proliferative effects according to the Chou & Talalay Method [387]. The effective dosage of each drug was first identified and then used to determine the range of drug dosage to use in the analysis of the combination effect. The cells were seeded in a 96-well tissue culture plate at optimal density and allowed to settle overnight before drug treatment at various concentrations. The treated cells were stained with crystal violet solution to determine the growth situation at various drug concentrations. The growth effects of experimentally treated cells were compared to the controls (non-treated and vehicle control) using the following equation.

Equation 2 Percentage of relative growth inhibition

where ODTreated is the optical density of a treated group and ODControl is the optical density of a control group.

Both single and combination drug treatment effects were assessed by the automatically computed combination index (CI) on Calcusyn™ software (Biosoft®, Cambridge, UK).

The combination index was defined as the following equation. 85

Equation 3 Combination index analysis

where CIA+B is the combination index for a fixed effect (F) for the combination of drug A and drug B; D AA+B is the concentration of drug A in the combination A+B giving an effect F; D BA+B is the concentration of drug B in the combination A+B giving an effect F;

DA is the concentration of drug A alone giving an effect F DB is the concentration of drug B alone giving an effect F; and  is a parameter with value 0 when A and B are mutually exclusive and 1 for mutually non-exclusive drugs that have independent modes of action [388]. The CI indicates super-additive when < 0.85, sub-additive when >

1.15 and additive when it is between 0.85 - 1.15.

3.2.3.6. Trypan blue exclusion assay

Cells were seeded in a 12-well tissue culture plate at a density of 1 x 105 cells/well and allowed to settle overnight before treatment. After 72 h of drug treatments, the cells were harvested as described in section 3.2.1.3 and stained with 1 volume of trypan blue staining solution. Viable and total cell counts were performed within 5 min after the cells were stained with trypan blue.

3.2.3.7. Clonogenic assay

Cells were seeded in 6-well tissue culture plates at a density of 5 x 102 cells/well and allowed to settle overnight before drug treatment. The experiment was terminated when colonies consist of more than 50 cells were identified in the control (non-treated) groups. The treated cells were rinsed with di-valent PBS and stained with crystal violet solution for 10 min. The stained cells were washed with Milli-Q water for at least 3 times to remove unbound stain and air-dried dry before digital images were taken. 86

“Clono-Counter” software [389] was used to analyse the digital images and the survival fraction of each drug treatment was quantified [390].

3.2.3.8. Flow cytometry and data analysis

Cells were cultured in a 6-well tissue culture plate at a density of 1 x 105 cells/well and allowed to settle overnight before drug treatment for 72 h. The treated cells were harvested as described in section 3.2.1.3 and then fixed in 70% v/v ethanol for 30 min at -20 oC. The fixed cells were rinsed twice with DPBS and stained with 5 µg/mL of propidium iodide in DPBS with 10 µg/mL of RNase A for a minimum of 30 min prior to analyse by a flow cytometer (Cell Lab Quanta™ SC MPL, Beckman-Coulter, USA).

Cell cycles were analysed and determined by WinMDIv2.9™ (Purdue University, West

Lafayette, IN) and Cylchred™ (Cardiff University, Wales, U.K.) software.

3.2.3.9. Immunocytochemistry for detection of protein and apoptosis

Cells were cultured in an 8-well chamber slide at optimal density and allowed to settle overnight before any treatment. At the designated time point, the growth medium was aspirated and each chamber was rinsed twice with DPBS. The cells was fixed with

3.7% v/v formaldehyde in PBS containing 0.1% v/v Triton-100 for 30 min at room temperature and then quenching with 0.3% v/v H2O2 and 50% v/v methanol in Milli-Q water for 10 min at room temperature. Each chamber was washed twice with PBS for 5 min and then 100 µL of primary antibody (1 µg/mL) diluted in 1% BSA/PBS was added for overnight incubation at 4oC. The chambers were removed the next day and the slide was washed three times with PBST for 5 min. The cells was incubated with HRP conjugated polymers (anti-rabbit or anti-mouse) from the EnVision™ system for 60 min at room temperature and washed four times with PBST for 5 min. The reaction was visualized by incubating the slide with DAKO liquid DAB+ chromogen staining buffer

(3,3’-diaminobenzidine in chromogen solution) for 30 min at room temperature. The 87

cells were then counterstained with filtered Harris’s haematoxylin for 3 min, followed by

2 dips in 0.1% v/v acid ethanol and 6 dips in Scott’s blue solution, while the slide was washed with water thoroughly in between these steps. The slide was dehydrated in a series of ethanol and cleared in 3 changes of xylene. The slide was coverslipped using

EUKITT mounting medium and allowed to dry overnight. Under light microscopy, 10 random fields per sample were used to identify and record the positively stained cells.

3.2.3.10. In vitro invasion assay

A solubilised tissue extract Matrigel™ (BD Biosciences, San Jose, California, USA) was allowed to defrost overnight at 4ºC and diluted at 1:10, 1:20 and 1:40 with serum free media before being applied onto polycarbonate filter transwells (6.5mm wide / 8

µM pore size). The transwells were allowed to dry overnight to allow coating of the filter.

The coated transwells were then inserted into a 24-well tissue culture plate (Corning

Costar, Cambridge, MA) for culturing. The cells were cultured in the coated transwells at a density of 5 x 103 cells/well and allowed to settle overnight before drug treatments.

After drug treatments, the transwells were rinsed with DPBS and then the bottom side of the filters were stained in crystal violet solution for 10 min. Cells that migrated to the lower surface of the filters were stained. The transwells were washed to remove unbound stain and air-dried before digital image was taken. Ten random fields were selected from each transwell. The stained cells that penetrated the Matrigel™ were counted and compared between treatments.

3.2.4. In vivo animal model and assays

3.2.4.1. Orthotopic mouse model

All experiments were performed with permission from the Animal Care and Ethics

Committee, University of New South Wales (ACEC 08/66A). Fifteen, 5-week-old female 88

Balb/c nu/nu mice from the Animal Resource Centre (Perth, Australia) were used. The mice were maintained for a week under specific pathogen-free conditions at constant temperature (25 ± 0.5ºC) and humidity (50%) before surgery. Cultured 143B osteosarcoma cells were harvested and quantified as described in section 3.2.1.3 and resuspended in PBS at the designated concentrations of 3 x 106 cells/100 µL.

Intratibial injection (Figure 3.3) was performed when a nude mouse was anaesthetised by inhalation of a mixture of isoflurane (1 - 3%) and oxygen (100%) on a warm pad.

The left leg of the mouse was disinfected by spraying with antiseptic povidone-iodine

(Orion Labs., Balcatta, WA, Australia). A 0.5 cm longitudinal skin incision was made along the left tibia from the distal end of the patella tendon to expose the proximal tibia by flexing the knee over 90º angles. Cortical layer of the tibia was punched through using a 27 1/2 gauge needle 2 mm distal to the patella tendon and 3 x 105 cells/10 µL of 143B osteosarcoma cells in PBS was injected into the tibia at 45 º angles upwards.

The skin incision was closed using EPIGLU (Meyer-Haake GmbH, medical Innovations,

Am Joseph 9, Germany) and the mouse was returned to the cages after full recovery from anaesthesia.

The health of the mice and xenograft tumour growth were monitored twice a week.

Mice were sacrificed at 2, 4 and 6 weeks (n = 5) after the inoculation of tumour cells by the use of carbon dioxide gas under anaesthesia. The mice were euthanized in a carbon dioxide chamber. Tumour sizes were measured by a pair of callipers. 89

Figure 3.3 Surgical procedures for the establishment of an orthotopic mouse model of osteosarcoma. (A) A longitudinal skin incision of 0.5 cm was made on the left leg of the anaesthetized mouse. (B) Intratibial injection of osteosarcoma cells into the tibia. (C) The skin incision was closed with EPIGLU®. (D) The mouse was returned to the cage after full recovery from anaesthesia.

3.2.4.2. X-ray radiography and microtomography imaging

Necropsy was carried out on the mice after sacrificing and x-rays of the legs were taken. The anteroposterior and the lateral planes x-rays were taken with a Faxitron MX

20 (Faxitron X-ray corporation, Wheeling, IL, USA) at 17 kV for 60 sec. The mice were also subjected to microtomography (Micro-CT) analysis. Both legs were removed and immersed in 10% neutral buffered formalin for fixation. Before micro-CT acquisition, the legs were washed with PBS and maintained in PBS during the acquisition process.

Micro-CT images were acquired using the Skyscan 1072 scanner (Skyscan, Kontich,

Belgium) at 65 kV, 100 mA, 10 sec exposure time with a low pass aluminium filter to minimize beam hardening artifacts. The sample was scanned with 0.9o step for a total of 185o with 20X magnification. The tibias were scanned at a pixel resolution of 9.98 90

µm. Averaging was applied on every two frames to produce a resulting pixel resolution of 19.96 µm. Cross sectional images were generated using NRecon software (Skyscan,

Kontich, Belgium) and a ring artefact correction factor of 8.0. Mimics13 software

(Materialise, Leuven, Belgium) was used to generate 3-Dimensional models. 3- dimensional analyses on a total of 20 slices per site were performed using CTan v.1.5.0 software (Skyscan, Kontich, Belgium).

3.2.4.3. Histochemical and immunohistochemical analysis

The lungs and legs were fixed in 10% v/v neutral buffered formalin. The fixed legs were then decalcified in 10% v/v formic acid in 10% v/v neutral buffered formalin for overnight. All tissues were processed with a Shandon Excelsior ES Processor (Thermo

Fisher Scientific, VIC, Australia) and were embedded in paraffin blocks with a Shandon

Histocentre 3 (Thermo Fisher Scientific, VIC, Australia). Sectioning of the paraffin blocks were carried out with a Leica microtome RM2165 (Leica Microsystem GmbH,

Wetzlar, Germany) at 5 µm thickness. Each section was mounted onto a siliconized slide and attachment was achieved by incubated overnight in an oven at 60°C.

Normal Harris haematoxylin and eosin (H&E) staining was performed. In brief, the sections were dewaxed in three changes of xylene for 5 min each, followed by rehydration in two changes of 100% v/v ethanol, two changes of 95% v/v ethanol, once in 70% v/v ethanol, once in 50% v/v ethanol and water for 5 min each. The rehydrated sections were then stained in filtered Harris’s haematoxylin for 3 min, differentiated in acidic alcohol (2 dips) and treated in Scott’s blue solution (10 dips). The sections were thoroughly rinsed with running tap water after these three steps. The stained sections were dehydrated in a series of ethanol (10 dips each). The dehydrated sections were then stained in eosin for 20 sec and rinsed in three changes of 100% v/v ethanol for 5 dips each. The sections were finally cleared in three changes of xylene for 10 dips each and coverslipped using EUKITT mounting medium. 91

For immunohistochemistry, the sections were treated in an antigen retrieval solution

(pH 6.0) (DAKO, CA, USA) for 20 min at 95°C after d ewaxing to re-open the antigen- antibody binding sites. The sections were then treated in a freshly prepared quenching solution (0.3% v/v H2O2, 50% methanol in Milli-Q water) for 10 min to inhibit any endogenous peroxidise activity and rinsed with Milli-Q water, followed by incubating in

PBST for 5 min. The treated sections were bordered with a wax-pen and blocked with

1% w/v BSA/PBS, followed by incubation with primary antibody overnight at 4°C.

Primary antibodies were diluted to 1 µg/mL with 1% w/v BSA/PBS. The sections were washed three times with PBST for 5 min the next day. Horseradish peroxidase (HRP) conjugated polymer of EnVision™ system (anti-mouse or anti-rabbit polymers) was applied to the section and incubated for 60 min at room temperature, followed by washing four times with PBST for 5 min each. The sections were stained with DAKO liquid DAB+ chromogen staining buffer (3,3’-diaminobenzidine in chromogen solution) for 30 min at room temperature to visualize the target protein. The stained sections were counterstained in filtered Harris’s haematoxylin for 3 min, differentiated in acidic alcohol (2 dips) and treated in Scott’s blue solution (10 dips). The sections were thoroughly rinsed with running tap water after these three steps. The counterstained sections were dehydrated in a series of ethanol (10 dips each) and cleared in three changes of xylene for 10 dips each and coverslipped using EUKITT mounting medium.

3.2.4.4. Detection of osteoclastic activity with TRAP staining

The sections were dewaxed and hydrated as in the histochemical process (section

3.2.4.3), followed by staining in a pre-warmed TRAP staining solution (37oC) for 90 min in the dark and then rinsed in water for 5 min. The stained sections were counterstained with Mayer’s haematoxylin for 30 sec and rinsed well in two changes of water for 5 min each. A permanent aqueous mounting medium was used to preserve the stained sections. 92

3.2.4.5. Detection of human Alu DNA with in situ hybridization

Cells of human origin and host cells in the xenograft tumour mass were distinguished by detection of human Alu DNA with in situ hybridization. Experiments were carried out in accordance to the Biogene universal ISH DAB Kit manual. Briefly, the slides were dewaxed and hydrated as described in section 3.2.4.3. Proteinase K digestion and antigen retrieval treatments on the section was carried out as described in the manufacturer manual. Hybridization was carried out overnight at 37°C using human Alu probe diluted at 1:5 with hybridization buffer. A biotinylated anti-Alu probe, streptavidin-

HRP conjugated polymer and DAB chromogen were used for target detection. The sections were counterstained with Mayer’s haematoxylin and preserved in a permanent aqueous mounting medium.

3.2.5. Statistical analysis

Statistical analysis was performed using the PASW Statistics 18 software (SPSS, Inc,

Chicago, IL), unless in some cases statistics results were obtained from the specific analysis program (e.g. Microarray analysis by Genespring™ GX and drug effectiveness analysis by Calcusyn™). Independent t-test or t-test with Bonferroni’s correction following a significant ANOVA was commonly used for comparison of difference between 2 or more samples. Spearman and Pearson correlation coefficient analyses were performed to investigate the relationship between the compared samples.

Statistical values were considered significant when p(2-tailed) < 0.05 93

CHAPTER 4. NEW GENE GROUPS ASSOCIATED WITH DISSIMILAR OSTEOBLASTIC 4 DIFFERENTIATION LINKING TO OSTEOSARCOMAGENESIS

4.1. Background

Osteosarcoma is a malignant mesenchymal tumour and a common primary tumour of bone. It is frequently localized at the distal femur and proximal tibia region where rapid bone growth occurs. This disease mainly occurs in adolescence and the elderly, and it is the second highest cause of cancer-related death after lymphoma in the paediatric age group [52]. Osteosarcoma has complex tumorigenesis, tumour progression and metastatic processes. So far various cytogenetic and genetic abnormalities have been identified in osteosarcoma, including (i) dysregulation of tumour suppressor genes

(TP53, RB1), CDK inhibitors (p16INK4a, p14ARF) and signalling pathways (Wnt, Shh); (ii) activation of oncogenes (MYC, JUN, FOS, ERBB2) and (iii) overexpression of certain types of genes/proteins (MET, MMPs, S100A6) [37,52]. However, our knowledge and understanding of the etiology of osteosarcoma remains limited.

It has been recently suggested that osteosarcoma can be regarded as a differentiation disease, that any disruption along the differentiation of mesenchymal stem cells to osteoblasts will lead to the development of osteosarcoma, thus making the treatment and the study of osteosarcoma very difficult [52,155]. Studies have demonstrated that restoring differentiation and/or overcoming differentiation defects could partially regulate the tumourigenicity of osteosarcoma cells. Hence, this has been exploited as a therapeutic approach combined with current osteosarcoma therapies [154,202,391]. Current 94

studies are focusing on either osteogenic factors/regulators or transcription factors. For example, restoring expression of a well known bone growth mediator, RUNX2, and its downstream transcription factor, Osterix, reduced tumourigenicity and increased anti- tumour properties in osteosarcoma cells [392,393]. However, treatment with another popular bone growth stimuli, “Bone Morphogenetic Proteins” (BMPs), did not assist restoration and/or promotion of terminal osteoblastic differentiation, but rather enhanced tumourigenicity of osteosarcoma [155]. The controversial results further support that osteosarcomagenesis is a complex process and may involve more genes or gene groups, thus more investigation of the relationship between osteoblastic differentiation and osteosarcoma is required.

Human osteosarcoma cell lines have been widely used in osteosarcoma related studies. These cell lines are “osteoblast-like cells” and known to be at different differentiation state, which represents some distinctive characteristics. For example, several studies have shown that the two commonly used osteosarcoma cell lines,

Saos-2 (more-differentiated) and U-2 OS (less-differentiated), expressed different level of osteogenic or osteoblastic factors, such as BMPs and Smads [153-155,159]. Our previous study also demonstrated that the implantation of Saos-2 cells into the thigh muscle of nude mice induced ectopic bone formation, but a xenograft tumour was formed with the implantation of U-2 OS cells [159]. Due to limitations of available technologies, previous studies were limited with a small numbers of genes. 95

4.2. Hypothesis and aim

It is hypothesized that studying the global gene expression of the two human osteosarcoma cell lines with the dissimilar differentiation status would enhance the understanding of the association between osteoblast differentiation and osteosarcoma tumorigenesis. The aim of this study is to identify the essential group of genes that distinguishes the osteogenic characteristics between these two human osteosarcoma cell lines and to provide insight into the genetic attribute in relation to osteosarcoma differentiation and osteogenic properties, advance our understanding towards osteosarcomagenesis, and may have a potential clinical implication in treatment of the disease.

4.3. Methods

Affymetrix™ microarray, real-time PCR, Protein/DNA array and other technologies were employed to study the gene profile of the two human osteosarcoma cell lines,

Saos-2 and U-2 OS. Gene expression analysis was carried out by GeneSpring™ GX

10.0.2 software (Agilent Technologies Inc., Santa Clara, CA) and online tools from

DAVID Bioinformatics Resources 2006 (National Institute of Allergy and Infectious

Diseases, National Institute of Health, Frederick, MD 21702; http://david.abcc.ncifcrf.gov/tools.jsp).

4

4 96

4.4. Results

4.4.1. Osteogenic induction properties of Saos-2 and U-2 OS

ALP activity (section 3.2.3.1) and the degree of calcium deposition and mineralization

(section 3.2.3.2) are markers often used to show osteoblast phenotypic characteristics and bone formation. As shown in Figure 4.1, the ALP activity of C2C12, a mouse myoblast cell line, was stimulated after culturing in the Saos-2 conditioned medium for

6 days. The formation of mineralized nodules in C2C12 was also increased after culturing for 42 days when compared to the normal growth medium and the osteogenic medium. These effects were concentration-dependent. As the concentration of Saos-2 conditioned medium increased from 25% (1:3) to 50% (1:1) their effects were enhanced (p<0.05). However, there were no significant changes observed for the

C2C12 cells cultured in the U-2 OS conditioned media. This result demonstrated that the factors in the conditioned media released by Saos-2 could induce the osteogenic characteristics of C2C12 and mineralizations were enhanced when the concentrations of these factors were increased. However, the factors in the conditioned media released by U-2 OS showed no or little effects to the osteogenic characteristic of

C2C12. These results confirmed that the two human osteosarcoma cell lines had maintained their own osteogenic induction ability. Thus, they are suitable to be used for further gene expression analysis to compare their differences. 97

(continue on next page) 98

(continue from previous page)

Figure 4.1 Osteogenic induction properties of Saos-2 and U-2 OS. Subconfluent mouse myoblast cells, C2C12, were cultured for 6 days in the Saos-2 or U-2 OS conditioned medium, normal growth medium or osteogenic medium for quantitative analysis of alkaline phosphatase activity (A). Cells cultured for 42 days were used for quantitative analysis of calcium deposition (B) and mineralization (C). Different proportion of Saos-2 conditioned media was also tested. Triplicate (A) or duplicate (B and C) experiments with triplicate samples in each were performed and data are shown in mean values with and error bars of standard deviation.

4.4.2. Detection of the quality of extracted RNA

Quality of RNA is an essential factor to ensure accuracy and quality of data obtained from expression analysis. After the extraction process, the quality and quantity of RNA were initially analysed with a NanoDrop® ND-1000 spectrometer (section 3.2.2.2).

RNA was considered as “pure” and suitable to use in the microarray experiment, when the ratio of absorbance of the sample at 260 nM / 280 nM is about 2.0 and the ratio of absorbance at 260 nM / 230 nM is between 2.0 to 2.2. As shown in Table 4.1, all of the six extracted RNA samples prepared for microarray analysis were all of good quality. 99

The quantity and quality of the RNA were analysed with a more stringent analysis before the microarray experiment with an Agilent 2100 Bioanalyzer™. In Figure 4.2, the electropherogram and gel image of all six RNA samples were shown with their concentration and RNA Integrity Number (RIN), where a score of 10 represents the highest RNA integrity with minimal degradation and score of 1 is the lowest integrity [394].

All RNA samples extracted were in excellent condition (score: 9.3-10) and suitable to be used for microarray experiments with both high purity and high integrity.

Table 4.1. RNA sample quality assessment by the ratio of absorbance

RNA Sample Ratio 260/280 Ratio 260/230

Saos-2 Sample 1 1.94 2.14

Saos-2 Sample 2 1.92 2.01

Saos-2 Sample 3 1.94 2.14

U-2 OS Sample 1 2.06 2.16

U-2 OS Sample 2 1.94 1.93

U-2 OS Sample 3 2.00 1.99 100

(continue on next page) 101

(continue from previous page)

Figure 4.2 Quality assessment of the RNA samples by Bioanalyzer™. Electropherogram summary, sample properties and gel image were shown. (S = Saos- 2 sample , U = U-2 OS sample, L = RNA Ladder sample, 1-3 = Sample 1-3) 102

4.4.3. Confirmation of the quality of microarray data

Triplicate experiments for each osteosarcoma cell line were performed in the study of gene expression. The quality of all six microarray data sets was assessed by several different parameters. According to the Affymetrix™ analysis manual, a good microarray experiment with high quality RNA from cell lines from many origins should meet the following criteria for all the different arrays in the comparison [395]: (i) scaling factors do not differ by more than 3- fold changes, (ii) signal background levels are less than 100,

(iii) noise level do not differ by more than 3- fold changes and (iv) 3’/5’ GAPDH ratio are not more than 3. As shown in the table below (Table 4.2), our samples meet all criteria aforementioned.

Table 4.2. Five different criteria were compared amongst the six microarrays and used to confirm the integrity of the gene expression data

RNA Sample Present Call Background Noise 3’/5’ Ratio Scaling % GAPDH Factors

Saos-2 Sample 1 46.8 79.46 3.060 0.94 1.705 Saos-2 Sample 2 46.1 86.32 3.280 0.96 2.104 Saos-2 Sample 3 47.9 53.85 3.660 1.01 2.650

U-2 OS Sample 1 45.7 76.32 2.840 0.90 2.222 U-2 OS Sample 2 46.6 78.25 2.910 0.89 2.503 U-2 OS Sample 3 48.6 71.24 2.650 0.91 2.438

The spike control, box-plot and scatter plot after normalization and transformation processes were used to investigate the consistency of data. As shown in Figure 4.3, comparable spike gene expression, similar distribution of expressed genes and no outliners were found. Thus, no inconsistency or significant variations were identified from the gene expression data amongst the six microarrays used for comparison. 103 104

Figure 4.3 Graphical illustration showing the consistency of the gene expression data between the six samples. (A) Gene expression of the spike control genes and (B) box-and-whisker plot of the gene distribution, amongst the six microarrays. (C) The scatter plot shows the expression of each gene in between the triplicate sample of the two experimental groups. 105

4.4.4. Identification of 75 differentially expressed genes

Obtaining reasonable and scientifically significant genes or gene groups from multiple expressed profiling genes are extremely important steps in making the conducted experiments successful and beneficial. Each microarray chip can detect approximately

38,500 well-characterized human genes. After quality control and normalization, 32,500 genes were available for further analysis.

Firstly, the gene expression profile of Saos-2 was compared with U-2 OS (section

3.2.2.4). As shown in Figure 4.4, 1,968 genes were identified as differentially expressed genes by the selection of genes with fold-change more than 3 and p-value less than 0.05. The distribution of these genes in regards to fold changes was also shown. Then, these differentially expressed genes were subjected to classification analysis (section 3.2.2.5). Gene ontology classification categorized that twelve biological processes were involved in cell and organ development, cell differentiation and cell adhesion, and one molecular function involved in “protein binding” was highly correlated to the selected genes. This was based on the selection of terms with EASE score less than 0.05 (p<0.05) and false discovery rate less than 0.05 (FDR<0.05)

(Figure 4.5 A). Functional annotation clustering was also used for categorization based on thirteen databases available and grouping the differentially expressed genes by a novel algorithm that measures the relationships of the annotations involved and annotation groups were ranked based on the “Enrichment scores”. Four different annotation clusters (Figure 4.5 B) were selected with enrichment score more than 3, in which it contained 629 unique genes (Appendix 2). The first cluster contains genes related to cell or organ development and cell differentiation. The second cluster contains genes related to cell adhesion process and a class of adhesion protein called

“cadherin”. The third cluster contains genes related to negative regulation of cells. The fourth cluster contained genes related to structural patterning process. 106

Figure 4.4 Differentially expressed gene in Saos-2 compared to U-2 OS. (A) Genes were selected with the conditions FC>3 and p<0.05, as shown in the black regions 1,968 genes were identified. (B) The distribution of these genes in regards to fold changes. 107

Figure 4.5 Gene ontology and functional annotation clustering were used to classify 1,968 differentially expressed genes. Both analyses showed that cell and organ development, cell differentiation and cell adhesion, and protein binding were highly related to the selected genes. (A) Gene ontology terms were selected with EASE score less than 0.05 (p<0.05) and false discovery rate less than 0.05 (FDR<0.05). (B) Functional annotation clustering were analysed based on 13 databases. Results were selected with enrichment scores more than 3 and displayed with the terms, percentage of gene involved from the analysis and significance of the term in each cluster. 108

Additionally, the differentially expressed genes between the two human osteosarcoma cells in this study were compared to normal human cells by unsupervised hierarchical clustering analysis of signal intensity, which identified ten clusters of 210 genes with a discrete expression pattern that was similar to osteoblast and chondrocyte (Appendix

3). The 628 genes obtained from the functional annotation clustering analysis were matched with the 210 genes from the hierarchical clustering analysis. Finally, 75 genes

(Table 4.3) were identified in both analyses and denoted as the differentially expressed gene in this study, in which 29 genes were found down-regulated and 46 genes up- regulated in Saos-2 when compared with U-2 OS. The discrete expression patterns compared between these 75 genes with the normal human cell are shown in Figure 4.6.

Among the identified genes, 27 genes (36%) are transcription factors, 15 genes (20%) are well known bone related factors or markers, other genes have known function in cell surface, nervous system and cell developments, and some other cellular functions. 109

Table 4.3. The list of 75 differentially regulated genes in Saos-2 compared to U-2 OS. A total of 27 transcription factors (36%) (underline) and 15 bone related factors (20%) (asterisk*) were found in the gene list.

(A) 29 down-regulated genes in Saos-2 compared to U-2 OS

Expression Gene Title Gene Symbol Entrez UniGene ID Level (FC) Gene

10-50 Laminin, alpha 1 LAMA1 284217 Hs.270364 Spectrin repeat containing, nuclear SYNE1 23345 Hs.12967 envelope 1, mRNA (cDNA clone IMAGE:4830497) Integrin, alpha 2 (CD49B, alpha 2 ITGA2 3673 Hs.482077 subunit of VLA-2 receptor) Keratin 34 KRT34 3885 Hs.296942 Protein tyrosine phosphatase, receptor PTPRF 5792 Hs.272062 type, F Quinolinate phosphoribosyltransferase QPRT 23475 Hs.513484 Discoidin, CUB and LCCL domain DCBLD2 131566 Hs.203691 containing 2

5-10 CD97 molecule CD97 976 Hs.466039 4-aminobutyrate aminotransferase ABAT 18 Hs.336768 Schlafen family member 5 SLFN5 162394 Hs.709347 Caveolin 1, caveolae protein, 22kDa* CAV1* 857 Hs.74034 Tissue factor pathway inhibitor TFPI 7035 Hs.516578 (lipoprotein-associated coagulation inhibitor) RGM domain family, member B RGMB 285704 Hs.526902

3-5 Glutathione peroxidase 1 GPX1 2876 Hs.76686 Arrestin, beta 1 ARRB1 408 Hs.503284 Glutathione S-transferase pi 1 GSTP1 2950 Hs.523836 Leucine rich repeat containing 6 LRRC6 23639 Hs.591865 Mitochondrial ribosomal protein L40 MRPL40 64976 --- A kinase (PRKA) anchor protein 2 /// AKAP2 /// 11217 /// Hs.591908 paralemmin 2 /// PALM2-AKAP2 PALM2 /// 114299 /// readthrough transcript PALM2-AKAP2 445815 RB1-inducible coiled-coil 1 RB1CC1 9821 Hs.196102 v-rel reticuloendotheliosis viral oncogene RELB* 5971 Hs.654402 homolog B Tripartite motif-containing 13 TRIM13 10206 Hs.436922 Pleckstrin homology-like domain, family PHLDA2 7262 --- A, member 2 Aldehyde dehydrogenase 7 family, ALDH7A1 501 Hs.483239 member A1 Nuclear factor of kappa light polypeptide NFKB1* 4790 Hs.654408 gene enhancer in B-cells 1* Tribbles homolog 3 (Drosophila) TRIB3 57761 Hs.516826 PALM2-AKAP2 readthrough transcript PALM2-AKAP2 445815 --- Cadherin 4, type 1, R-caderin (retinal) CDH4 1002 Hs.473231 Heat Shock 70kDa Protein 1A and 1B HSPA1A / 3303 / Hs.274402 HSPA1B 3304

110

(B) 46 up-regulated genes in Saos-2 compared to U-2 OS

Expression Gene Title Gene Symbol Entrez UniGene ID Level (FC) Gene

> 100 CD24 molecule CD24 100133941 Hs.694721 Myocyte enhancer factor 2C MEF2C 4208 Hs.653394

50-100 FAT tumour suppressor homolog 3 FAT3 120114 Hs.98523 (Drosophila) Runt-related transcription factor 2* RUNX2* 860 Hs.535845

10-50 Zic family member 1 (odd-paired ZIC1 7545 Hs.647962 homolog, Drosophila) Sp7 transcription factor* SP7* 121340 Hs.209402 Distal-less homeobox 5* DLX5* 1749 Hs.99348 Protein tyrosine phosphatase, receptor PTPRD 5789 Hs.446083 type, D Scinderin SCIN 85477 Hs.655515 Distal-less homeobox 6* DLX6* 1750 Hs.249196 Integrin-binding sialoprotein (also known IBSP* 3381 Hs.518726 as Bone Sialoprotein II) * Alkaline phosphatase, liver/bone/kidney* ALPL* 249 Hs.75431 Inscuteable homolog (Drosophila) INSC 387755 Hs.591997 Collagen, type X, alpha 1 COL10A1* 1300 Hs.520339 Tetraspanin 2 TSPAN2 10100 Hs.310458 Integrin, alpha 10 ITGA10 8515 Hs.158237 Gap junction protein, beta 2, 26kDa GJB2 2706 Hs.524894 Osteomodulin* OMD* 4958 Hs.94070 Creatine kinase, brain CKB 1152 Hs.173724 Neuronal guanine nucleotide exchange NGEF 25791 Hs.97316 factor Homeobox D4 HOXD4 3233 Hs.591609

5-10 Shroom family member 2 SHROOM2 357 Hs.567236 Bone morphogenetic protein 4* BMP4* 652 Hs.68879 Fibroblast growth factor receptor 2* FGFR2* 2263 Hs.533683 Unc-5 homolog B (C. elegans) UNC5B 219699 Hs.522997 Patched homolog 1 (Drosophila) PTCH1 5727 Hs.494538 Tumour necrosis factor, alpha-induced TNFAIP2 7127 Hs.525607 protein 2 Interleukin 11 IL11 3589 Hs.467304 Collagen, type XII, alpha 1 COL12A1 1303 Hs.101302 Ataxin 1 ATXN1 6310 Hs.434961 Cadherin, EGF LAG seven-pass G-type CELSR1 9620 Hs.252387 receptor 1 (flamingo homolog, Drosophila) SLIT and NTRK-like family, member 6 SLITRK6 84189 Hs.525105 Transient receptor potential cation TRPM8 79054 Hs.366053 channel, subfamily M, member 8 Jagged 2 JAG2 3714 Hs.433445

3-5 Sema domain, immunoglobulin domain SEMA3B 7869 Hs.82222 (Ig), short basic domain, secreted, (semaphorin) 3B Cell adhesion molecule 1 CADM1 23705 Hs.370510 Microtubule-associated protein 2 MAP2 4133 Hs.368281 Fibroblast growth factor receptor 3 FGFR3* 2261 Hs.1420 PDS5, regulator of cohesion PDS5B 23047 Hs.716441 maintenance, homolog B (S. cerevisiae)

(continue on next page) 111

(continue from previous page)

Expression Gene Title Gene Symbol Entrez UniGene ID Level (FC) Gene

3-5 Plexin domain containing 2 PLXDC2 84898 Hs.658134 ST8 alpha-N-acetyl-neuraminide alpha- ST8SIA4 7903 Hs.308628 2,8-sialyltransferase 4 Sushi domain containing 5 SUSD5 26032 Hs.196647 Wingless-type MMTV integration site WNT11* 7481 Hs.108219 family, member 11 Axin 2 AXIN2 8313 Hs.156527 Cytochrome P450, family 39, subfamily CYP39A1 51302 Hs.387367 A, polypeptide 1 Wingless-type MMTV integration site WNT10B* 7480 Hs.91985 family, member 10B

Figure 4.6 The hierarchical clustering analysis showing the average signal intensity of the identified 75 differentially expressed genes in Saos-2 compared to U-2OS after several analyses. Distinctive gene expression patterns were recognized from the comparison between Saos-2 and U-2 OS cell lines and five normal human cells. Clusters (a), (b) and (c) showed that the genes expression in Saos-2 corresponded to the two bone forming cell types, osteoblast and chondrocyte, while gene expressions in U-2 OS genes were similarly expressed as others. 112

4.4.5. Verification of microarray data by qRT-PCR and immunocytochemistry qRT-PCR (section 3.2.2.9) was used to validate the microarray data. Five representative genes ALPL, DLX5, FGFR2, RUNX2 and SP7 from 75 differentially regulated genes and three other genes BGLAP, GAPDH and RUNX3 were selected.

To normalize the qRT-PCR data, a relative expression value was obtained by comparing each gene expression with the β-actin transcript. Similarly, in the microarray, the expression of the selected genes was compared with the β-actin transcript to obtain a relative expression value. The β-actin normalized relative expression ratios of the eight selected genes are presented in Figure 4.7. The results showed that the expression trends of the selected genes were consistent between the microarray and the quantitative RT-PCR data.

Immunocytochemistry was then used to verify protein expression in the cell lines

(section 3.2.3.9). As shown in Figure 4.8, expression of the four proteins was consistence with the mRNA expression analysis by qRT-PCR. Protein expression of alkaline phosphatise, osterix and runt-related transcription factor 2 was higher in Saos-

2 and glyceraldehyde 3-phosphate dehydrogenase was similar between Saos-2 and U-

2 OS. 113

Figure 4.7 Verification of microarray results by qRT-PCR. Relative gene expressions (mean ± standard deviation) of the eight represented genes were compared between microarray analysis and qRT-PCR. Significant gene expression was identified when p<0.05. [ALPL = Alkaline Phosphatase, DLX5 = Distal-less Homeobox 5, FGFR2 = Fibroblast Growth Factor Receptor 2, BGLAP = Bone Gamma- carboxyglutamate Protein (Osteocalcin), SP7 = Sp7 Transcription Factor (Osterix), RUNX2 = Runt-related Transcription Factor 2, RUNX3 = Runt-related Transcription Factor 3, GAPDH = Glyceraldehyde 3-phophate Dehydrogenase] 114

Figure 4.8 Verification of microarray results by immunocytochemistry. Protein expression was consistent to the mRNA expression detected by qRT-PCR. Photos displayed are at 20 times magnification.

4.4.6. Differential transcriptional protein expression in the dissimilar human osteosarcoma cell lines

Transcription factors are important in regulating gene expression in cells for any cellular activities. Therefore, analysis of the biochemical activities of transcription factors in the two human osteosarcoma cell lines provides extra information on their distinctive characteristics.

Analysis showed that a total of 23 out the 54 detectable transcription factors, 19 in

Saos-2 and 23 in U-2 OS, were found to be active in the Protein/DNA array (section

3.2.2.10). Among the 23 active transcription factors, 8 of them showed differential activities; Brn-3, CBF, c-Myb and MEF-1 has a higher activity in Saos-2, while Stat4,

TR, VDR and HSE has a higher activity in U-2 OS (Figure 4.9). These factors were 115

associated with 22 genes from the Affymetrix™ microarray (Table 4.4). The highly regulated gene MEF2C in Saos-2 was categorized as significantly regulated in both microarray and transcription factor activity analyses. This experiment was performed by

Dr. Yan Yu.

Transcription factor target genes were retrieved from the “Transcriptional Regulatory

Element Database” (TRED) for further analysis. The database only contains information for 12 out of the 23 active transcription factors identified, and all together

774 unique transcription factor targeted genes were obtained. Comparison between these genes and the 75 differentially expressed genes identified 11 genes, in which

CELSR1, RUNX2 and UNC5B were found up-regulated and CAV1, GSTP1, HSPA1A,

ITGA2, MRPL40, NFKB1, RELB and TRIB3 were found down-regulated in Saos-2 when compared to U-2 OS.

Transcription factors are transcribed from a gene on chromosome into RNA in the nucleus. The RNA is then translated into protein in the cytoplasm like most of other proteins in eukaryotes. Nuclear localization signals direct these transcription factor proteins to the nucleus during the regulation process and ligand binding is essential for nuclear receptors for its relocation from cytoplasm to the nucleus [396]. In this analysis, transcription factor proteins were extracted and formed a protein-probe complex with the labelled probes with the specific binding site. The un-bound probes were removed.

The labelled probes were then separated from the protein and hybridized to the array for signal detection. The signal intensity directly represents the quantity of the transcription factor proteins in the samples. Thus, investigation of the transcription factors not only provides the biochemical activities of the proteins but could also serve as an additional verification of the protein translation level to the microarray data. 116

Figure 4.9 Differential transcriptional regulation in Saos-2 and U-2 OS. (A) The chemiluminescence images of the TranSignal™ Array I probed with nuclear proteins extracted from the two human osteosarcoma cell lines. (B) The corresponding transcription factors were shown. The differentially active transcription factors were marked on the images.

117

Table 4.4. Expression of genes related to transcription factors. 8 of the 23 differentially active transcription factors identified from the Protein/DNA array between Saos-2 and U-2 OS and their related genes expression in Saos-2 from Affymetrix™ microarray

Transcription Related Gene Title Unigene ID Gene Factor Genes in Regulation Microarray in Saos-2 Brn-3 POU4F1 POU class 4 homeobox 1 Hs.654522 up POU4F2 POU class 4 homeobox 2 Hs.266 up POU4F3 POU class 4 homeobox 3 Hs.553499 down

CBF NFYA nuclear transcription factor Y, alpha Hs.10441 down (CCAAT NFYB nuclear transcription factor Y, beta Hs.84928 up binding NFYC nuclear transcription factor Y, gamma Hs.233458 down factor) c-Myb MYB v-myb myeloblastosis viral oncogene Hs.654446 up homolog (avian)

MEF-1 MEF2A myocyte enhancer factor 2A Hs.268675 up MEF2B myocyte enhancer factor 2B Hs.153629 up MEF2C myocyte enhancer factor 2C Hs.653394 up MEF2D myocyte enhancer factor 2D Hs.314327 down

HSE HSF1 heat shock transcription factor 1 Hs.530227 down (Heat Shock HSF2 heat shock transcription factor 2 Hs.158195 up Consenus HSF2BP heat shock transcription factor 2 Hs.406157 down Elements) HSF4 binding protein Hs.512156 down HSF5 heat shock transcription factor 4 Hs.380061 up HSFX1///HFS heat shock transcription factor family Hs.592255 up X2 member 5 heat shock transcription factor family, X linked 1 /// heat shock transcription factor family, X linked 2

Stat 4 STAT4 Signal transducer and activator of Hs.80642 down transcription 4

TR NR1D1 /// nuclear receptor subfamily 1, group D, Hs.724 down THRA member 1 /// thyroid hormone receptor, alpha Hs.724 down THRA thyroid hormone receptor, alpha Hs.187861 down THRB thyroid hormone receptor, beta

VDR VDR vitamin D (1,25- dihydroxyvitamin D3) Hs.524368 Down receptor

118

4.5. Discussion

Osteosarcoma arises from impaired differentiation of immature osteoblast into more mature types. Evaluation of genetic profile of the disrupted osteoblast differentiation in osteosarcoma was difficult due to the lack of appropriate samples and limited knowledge about the disruption of the osteoblast differentiation process.

The present study initially characterized the osteogenic induction ability of the two human osteosarcoma cell lines, Saos-2 and U-2 OS, using the mouse myoblast cells,

C2C12. Analysis revealed that the ALP activity and the mineralized nodules formation in C2C12 were increased after culturing in the Saos-2 conditioned media compared with the U-2 OS conditioned media, the normal media or the osteogenic media. These results suggest that the two osteosarcoma cell lines, which were dissimilar in tumour differentiation status, maintained difference in their osteogenic induction ability and thus, were feasible to be used in the gene profiling studies.

The osteogenic medium, which contains !-glycerolphosphate, ascorbic acid and dexamethasone, was used previously to induce osteogenic phenotypes, such as alkaline phosphatase and mineralization, in mice adipocyte-derived stem cells and marrow stromal cells [397,398]. In the present study, a minor but not statistically significant increase on alkaline phosphatase and mineralization was found between the osteogenic medium and normal media, suggesting that modifications on the use of osteogenic medium as a positive control of osteogenic induction in C2C12 was required in the future study. In consistent with other studies, this osteogenic medium could be used as osteogenic supportive medium in C2C12 and only a low amount of alkaline phosphatase activity was found in this cell line, when the medium was used alone without any other osteogenic factor involved [399,400]. 119

Our previous study also verified that Saos-2 differentially expressed a profile of osteogenic factors compared with U-2 OS [159], justifying the different osteogenic induction abilities of the conditioned media obtained from both cell lines. In addition, this was also consistent with our previous in vivo study, which demonstrated that injection of osteosarcoma cells into mice thigh muscle induced ectopic bone formation for Saos-2 and xenograft tumour formation for U-2 OS [159].

All of the above data supports the notion that the more-differentiated Saos-2 has distinctive osteogenic induction ability and osteogenicity compared with the less- differentiated U-2 OS. These differences were shown to be consistent in both in vitro and in vivo conditions, as well as in the current passage of cells. Evidence from other studies also supported the osteogenicity of Saos-2 [157,162] and tumourigenicity of U-2

OS [154].

Microarray studies were then applied to investigate the genetic profile of the two dissimilar osteosarcoma cell lines followed by subsequent analysis. One of our key investigations was to compare the differentiated expression profile obtained from our study against normal human cell lines.

The study by Kubo et.al [383] utilized microarray to identify the transcription factors that could be used as specific markers for human mesenchymal cells during the differentiation of human mesenchymal stem cells into adipocytes, chondrocytes, fibroblasts and osteoblasts. Our study employed the same technology as that of Kubo et al., thus the microarray expression results between both studies are directly comparable. This particular comparison assisted in the identification of differentially expressed genes that are important to the osteogenic characteristic of Saos-2 in regards to the differentiated and non-differentiated cell conditions, and eliminated the genes that differed only between the two human osteosarcoma cell lines. 120

Gene ontology analysis suggested that genes involved in organ development, cell binding and adhesion and nervous system signalling are important to differentiate the osteogenic properties between the two human osteosarcoma cell lines. Given that development of osteosarcoma involves dysregulation of the differentiation process, it explicates that the differentially expressed genes are associated with multiple organ developmental processes. Furthermore, cellular organization is an important step during organ developmental processes. It justifies that the differentially expressed genes are also correlated with cellular adhesion processes.

An increasing number of investigations have shown the emerging relationships between the human nervous system and bone, and regarded it as “Neuroskeletal biology” [401]. Important discoveries include the decrease in bone mass resulting from leptin stimulation of !-adrenergic signalling system and the increased expression of the main osteoblast differentiation mediator RUNX2 resulting from stimulation of N-methyl-

D-aspartate signalling system in osteoblasts [402-404]. This bone mass reduction through the nervous system could be reversed with the use of !-adrenergic receptor antagonist

(also known as “!-blocker”) by eliminating the activation of !-adrenergic signalling system from the binding of leptin [405]. The !-adrenergic receptor antagonist has attracted many studies with regards to bone metabolism, and has been reviewed for its potential clinical applications in osteoporosis and fracture healing [406]. Bone abnormality such as osteoporosis is proposed as a neuroskeletal disease [407]. Thus, it is possible that other factors related to the nervous system are involved in the osteoblast differentiation and bone formation, or even in the development of osteosarcoma.

Many studies have also reviewed the correlation of cell adhesion and cadherin molecules to cancer progression and differentiation [408,409]. Expression of an epithelial cell adhesion molecule (C-CAM) has been found to correlate with cell differentiation 121

during development of human prostate but not in prostate cancer cells [410]. Induced expression of C-CAM in a human prostate cancer model have shown anti-tumour effects [411]. Moreover, expression of neuronal cadherin (N-cadherin) was found up- regulated concomitantly with osteoblast differentiation [412], but was down-regulated in osteosarcoma [413].

Amongst the 75 differentially expressed genes in Saos-2, osteoblast related markers such as FGFR2, DLX5, DLX6, MEF2C, WNT10B, RUNX2 and SP7 were identified in our analysis. RUNX2 has been recognized as the major transcription factor controlling osteoblast differentiation. All these osteoblast-related markers contribute to the regulation of bone formation during skeletal development and post-natal life [172,173,414].

Several other factors with known functions in bone formation and development were also detected in our analysis, which includes the bone markers, such as alkaline phosphatase (ALPL) and bone sialoprotein (IBSP), the bone morphogenic protein 4

(BMP4), collagens (COL10A1, COL12A1), osteoclast regulators (NFKB1 and SEMA3B) and a transcription factor (MEF2C) [415-417]. Knockout of CAV1 and NFkB1 genes in mice have also been demonstrated to positively affect bone growth [418,419].

Cell adhesion process and cadherin are one of the significant gene clusters in this study after both gene ontology and functional annotation analysis. Factors such as

CADM1, CD24, CD97, CDH4, CELSR1, FAT3, GJB2, ITGA2, ITGA10, SCIN and

TSPAN2 are different types of cell surface proteins involved in cell binding and adhesion. Scinderin (SCIN), also known as Adseverin, is belongs to the gelsolin family of actin regulatory proteins. Its expression was found in mouse embryos restricted to endochondral bone formation and the development of adult outer renal medulla and intestine [420]. In humans, the only function identified for this protein is to stabilize actin

[421,422]. 122

Gap junction proteins, also known as connexins, play an important role in signal transmission between cells. Gap junction protein alpha 1 (GJB1) expression is found in osteoblasts, osteocytes and chondrocytes, where it mediates osteoblast differentiation and bone formation [423]. However, deficiency of GJB1 only showed incomplete but not totally ablated bone formation [424]. Other gap junction proteins such as gap junction protein !2 (GJB2) may also play a role in transmitting signalling factors during bone formation.

CD24, a cell surface protein, was found to be the most significantly altered gene from our microarray analysis. CD24 is a heavily glycosylated short mucin-like protein, that anchors to the outer surface of the cytoplasmic membrane by glycosyl-phosphatidyl- inositol anchor protein but has no transmembrane domain. CD24 was initially identified as a surface antigen of B-cells and was correlated to the maturation of B-cells. Further studies demonstrated that it also participates in cell proliferation and differentiation, and may be related to carcinogenesis [425]. Expression of CD24 protein was not only detected in normal hemopoietic cells, such as B-lineage cells and mature granulocytes, but also in breast, non-small cell lung and colorectal cancers [426-428]. Although CD24 is suggested as a prognostic marker for these cancers, its biological role with regards to these cancers are not known.

Studies have demonstrated the cell signal transmission and cell-to-cell interaction of

CD24 in B-cells [429,430]. It is also known that mammalian haematopoiesis occur in the skeletal system. However, there is limited knowledge in the understanding of the relationship between the skeletal and the hematopoietic system. One of the evidence demonstrating the relationship between the two systems is that the haematopoietic cytokines granulocyte colony-stimulating factors, which have an important function during haematopoiesis, is constitutively produced by osteoblast cells [431]. Surface protein, such as CD24, may be involved in the communication between haematopoietic 123

cells and osteoblasts, and may serve a role in controlling the osteoblast differentiation or bone formation.

Cell-to-cell communication has always been an important factor in cell regulation.

Several recent reviews have highlighted the importance of communication between cells within the skeletal system and between cells in the skeletal and haematopoietic system [432,433]. Osteoblast differentiation and bone formation are complex cellular processes involving the regulation of many different cells. Although many factors are recognized as important regulators for these processes, other regulatory factors and mechanisms remains unexplored. Further investigations are required to understand the relationship between overexpression of CD24 and disrupted osteoblast differentiation in osteosarcoma.

Another interesting finding from our results is the down-regulation of ARRB1 (!-arrestin

1). As discussed earlier, activation of !-adrenergic signalling system in osteoblasts can reduce bone mass, and can be reversed by !-blocker. The activation of the !- adrenergic signalling system requires a protein complex containing !-adrenergic receptor and activated c-Src. Src recruitment is mediated by !-arrestin 1, which serves as an adaptor protein to bind both the c-Src and agonist-occupied !-adrenergic receptor [434]. Thus, the down-regulation of !-arrestin 1 will hamper the activation of !- adrenergic signalling system and may impair the normal function of osteoblasts.

A large proportion of the differentially expressed genes were found to be related to transcription factors. The activity of transcription factors and its targeted genes were investigated by the use of both Protein/DNA array and Affymetrix™ microarray analysis.

A gene list of the differentially active transcription factors and their targeted binding genes was constructed and matched with the 75 differentially expressed gene list. This analysis identified a total of 12 significant genes, 1 gene of a transcription factor and 11 genes of transcription factor binding targets, in regards to the differentially active 124

transcription factors between Saos-2 and U-2 OS. Genes identified from this analysis would not corresponded directly to the active transcription factors in this stage, as the gene expression of the active transcription factors should occur in the earlier stage.

Thus, it is reasonable that the majority of genes identified from this analysis were the transcription factor binding targets.

Amongst the 12 genes identified from transcription factor analysis, CAV1, MEF2C,

NKFB1, RELB and RUNX2 are related to osteoblast and/or bone formation [418,435,436], but not with the other 7 genes. Reviewing the annotations and functions of the other 7 genes from NCBI database showed that (i) CELSR1, ITGA2 and UNC5B are membrane proteins; (ii) CELSR1, GSTP1 and UNC5B are involved in the development and signalling of the nervous system; (iii) ITGA2 is involved in a cellular binding and adhesion, and extracellular stimulation responding system; (iv) MRPL40 is a ribosomal protein that helps in protein synthesis within the mitochondrion; (v) HSPA1A is a heat response protein and plays a role in the conformation of protein; (vi) TRIB3 is a putative protein kinase that is induced by the transcription factor NF-kappaB and negatively regulates cell survival. Further investigation in these genes is required to understand their roles in osteoblast differentiation and osteosarcoma development.

4.6. Conclusion

In summary, analysis of gene expression profile from 2 osteosarcoma cell lines with dissimilar differentiation status showed for the first time that genes from cell-adhesion and nervous system are expressed differentially between the cells. Moreover, genes involved in osteoblast differentiation were also identified in this study. The current findings provide a basis for further investigation of the interaction between these genes, 125

which may help to advance our understanding in the mechanisms over the control of osteoblast differentiation and development of osteosarcoma. 126

CHAPTER 5. ASSOCIATION OF RAS/RAF/MEK/ERK PATHWAY WITH LUNG METASTASIS IN AN 5 ORTHOTOPIC MOUSE MODEL OF OSTEOSARCOMA

Disclosure statement

This study was partially supported by the project grant (2008) from the Australian

Orthopaedic Association Research Foundation. It was designed by A/Prof. Yang and

Dr. Yu. The tumour cell inoculation surgery, the in vivo model evaluation and the evaluation of immunohistochemistry of IGF-1R downstream regulating factors was performed by Dr. Yu.

5.1. Background

The 5-year survival rate of non-metastatic osteosarcoma patient under the current conventional therapeutic strategy has reached a plateau of 60-70% with no further improvement for the past two decades [2]. Long-term survival for patients with metastatic or recurrent disease is about 30% [7]. Recurrence is most common in the lung and >80% of the metastatic osteosarcoma patient presented with pulmonary metastasis [437]. Therefore, it is important to identify and develop new treatment approaches and modalities to improve the treatment of osteosarcoma.

The discoveries of the major cellular signalling pathways involved in the osteosarcoma progression and/or lung metastasis will help in the identification of potential targets for 127

development of novel treatment. It has recently been reported that the insulin-like growth factor (IGF) system (including ligands, receptors, IGF binding proteins and proteases) plays an important role in the formation and homeostasis of bone [323], and also in Ewing’s sarcoma [324]. Consistent expression of IGF-1R was identified in osteosarcoma tissue samples and cell lines, but the expression of IGF-I and IGF-II was varied [104,105]. Although IGF-I dependent growth was observed in some osteosarcoma cell lines in in vitro survival and proliferation [326], no clear evidence showed that IGF-I and IGF-II functions as a dominant autocrine growth or motility factor in osteosarcoma

[105]. A study showed that lowering the serum level of IGF-1 had no sustained clinical response in osteosarcoma patients [327]. Thus, targeting IGF-1R would be an alternate approach in the treatment of osteosarcoma. Tyrosine kinase inhibitors and specific monoclonal antibodies were demonstrated as an effective therapeutic approach in in vitro and in vivo models of Ewing’s sarcoma and osteosarcoma [379,438].

Interaction of ligands (e.g. IGF-1 and IGF-2) and IGF-1R induces the phosphorylation at the tyrosine kinase domain and leads to the phosphorylation of intracellular proteins, which then activates signalling pathways that regulated cell survival, proliferation and metabolism. Amongst all signalling pathways in the IGF-1R system, three major pathways Ras/Raf/MEK/ERK, Ras/Raf/MEK/p38-MAPK, and Ras/PI3K/Akt/mTOR are important for tumorigenesis, maintenance of a phenotype, and protection from apoptosis [439]. Specific pathway signalling factor inhibitors are commercially available and have been used to study of these pathways in cancers. A study using the

MEK/ERK inhibitor U0126 [350] showed that MEK signalling was correlated with anchorage-independent growth and induced apoptosis in human breast cancer cell lines with constitutively activated MEK signalling pathway [352]. The use of p38-MAPK inhibitor SB202190 identified that p38-MAPK pathway was responsible for the modulation of both P-glycoprotein mediated and unmediated multidrug resistance in human gastric cancer cells [440]. Moreover, the use of PI3K inhibitor LY294200 revealed 128

that PI3K pathway was responsible for the modulation of P-glycoprotein mediated multidrug resistance in mouse leukemic cancer cells [441].

5

5.2. Hypothesis and aim

It is hypothesized that imbalanced (abnormal) regulation of the IGF-1R system and its signalling are associated with tumour progression and lung metastasis of osteosarcoma.

Firstly, this study was aimed to set up an orthotopic mouse model of human osteosarcoma with spontaneous pulmonary metastasis. Secondly, to examine the protein expression of IGF-1R and its downstream signalling factors, Akt, MEK1/2 and p38 MAP kinase together with their activated form at the primary site (tibia) and a metastasis spot (lung) using immunohistochemistry. Lastly, to assess the effect of down-regulation of MEK/ERK activity by the MEK/ERK inhibitor U0126 on in vitro invasion by an osteosarcoma cell line.

5

5.3. Methods

Intratibial injection of human osteosarcoma cells was used to establish the orthotopic mouse model of human osteosarcoma. X-rays, micro CT, histochemistry, immunohistochemistry and in situ hybridization were used for end points characterization. Matrigel™ invasive assay was used to study the invasiveness of osteosarcoma cells. 129

5.4. Results

5.4.1. Tumour growth and metastasis: gross examination

Local tumour growth was detected as early as 7 days post inoculation (Figure 5.1). As shown in Figure 5.2, tumour sizes were increased in a geometric progression from day

7 up to day 42. At day 42 the local tumour had entirely occupied the left hind limbs.

Although some of the animals experienced reduced mobility, no sign of depression was noted in all animals throughout of the study.

Animals were sacrificed at 2, 4 and 6 weeks post tumour cell inoculation and the hind limbs and the lungs were harvested and examined by general observation and palpation (Figure 5.3). Gross examination at 2-week animals revealed that hard nodules formed from the proximal tibia (Figure 5.3 A) with sizes ranging from 2 x 2 x 3 mm to 12 x 6 x 5 mm and no abnormality of the lungs was observed. At 4 weeks the tumour sizes increased in 3 – 5 folds (Figure 5.3 B) and the tumours were encapsulated within a membranous tissue with an abundant blood supply. Multiple white nodules in the lungs were observed (Figure 5.3 C). At 6 weeks the entire left hind limbs of the animals were covered by a huge mass size of 23 x 21 x 20 mm on average and the lumps were still encapsulated with multiple cysts formed underneath the membrane as examined by palpation. More nodules were observed in the lungs from this group of animals (Figure 5.3 D). 130

Figure 5.1 Local tumour growth at the left tibia at week 1 (A), week 3 (B) and week 6 (C) post osteosarcoma cells inoculation.

Figure 5.2 Tumour growth curve of six weeks post inoculation. The bars represent the standard deviation at each time point. 131

Figure 5.3 Gross examination of the tumour at the primary sites and lung. (A) Hard nodule identified from the left tibia in a 2-week post inoculation mouse. (B) An encapsulated tumour with abundant blood supply from a 4-week post inoculation mouse. White nodules were visualized in the lung of a mouse at 4-week (C) and 6- week (D) post inoculation. The photo (D) was taken after formalin fixation.

5.4.2. Radiographic characterizations of the mouse model of osteosarcoma

X-ray Imaging analysis of the primary tumour mass and the local bone resorption was carried out as described in section 3.2.4.2. As shown in Figure 5.4, images of the lateral and anteroposterior X-ray revealed that bone resorption was found at the proximal tibia of the left legs in all 5 cases at 2 weeks post inoculation, but no fracture or deformation of the bone was detected. The x-ray images of the 4-week animals had revealed a more severe condition that the proximal tibias were resorbed with the resorption extending to the knee joints and femurs. Joint deformation and segmental bone defects were presented. The x-ray images of the 6-week animals showed the left hind limbs were further resorbed from the mid shaft of the femurs to that of the tibias with multiple cysts formed in the lumps. 132

Figure 5.4 X-ray images of the hind limbs of the mice at 2, 4 and 6 weeks post inoculation. Lateral (LAT) images of the right (R) and left (L) limbs and anteroposterior (AP) images of the left tibia were taken from two representative animals at each time point. Bone resorption and tumour mass were identified at the proximal tibia of the left limb, where the human osteosarcoma cell was inoculated. 133

Micro-CT analysis (section 3.2.4.2) confirmed that bone resorption at the proximal tibias of the left legs of 2-week animals (Figure 5.5) and no notable abnormal bone formation. The sizes of the tumours at 4 and 6 weeks were too big for micro-CT analysis.

Figure 5.5 Micro-CT images of the hind limbs of the mice at 2 weeks post inoculation. The reconstructed three-dimensional images showed intact right (R) tibias and bone defects in the left (L) tibias (arrows).

5.4.3. Microscopic examination of the primary and the secondary xenograft tumours

Microscopic examination (section 3.2.4.3) revealed xenograft tumour formed at the proximal tibia of the left hind limb in all cases at all time points. At 2 weeks the tumour mass went through the full thickness of the tibial cortex, the marrow cannel and the surrounding soft tissues. Bone resorption was noted by the rough thinner cortex and the presence of multi-nucleus osteoclast cells (Figure 5.6 A). At 4 weeks the tumour had covered the whole tibia and the knee joint. Multiple necrotic areas were present in the tumour mass. The tibia was interrupted and fragmented (Figure 5.6 B). At 6 weeks the entire left hind limb was covered by tumour mass with enlarged areas of necrotic tissues. The bony tissues of the whole limb were fragmented (Figure 5.6 B). At all time the tumour mass was capsulated by membranous tissues. Osteoclast activity tested by 134

TRAP staining was allocated at the margin of the tumour and the rough surface of the cortices (Figure 5.7). Osteoclast activity was present at all time points.

The lung tissues at 2 weeks were sectioned through and microscopic analysed at 25

µm interval. No tumour cell noted at this time point. Multiple tumour mass was noted in the lung of 4 out of 5 mice at 4 weeks and in the lung of all mice at 6 weeks(Figure 5.8).

The human origin of the lung metastasis was tested by in situ hybridization technique

(section 3.2.4.5) for human Alu gene expression (Figure 5.9). 135

Figure 5.6 Histochemical analysis of the hind limbs of the mice with human osteosarcoma cell inoculation. (A) At 2 weeks xenograft tumour formed at the proximal tibia of the left hind limb [L]. The tibial cortex was interrupted and multiple nucleus osteoclast cells were noted at the margin of tumour and bone (solid-arrows). The normal intact right hind limb [R] from the same animal was shown for comparison. (B) At 4 weeks and 6 weeks the bony tissues of the left hind limb were totally fragmented (arrow-heads) and multiple areas of necrotic tissues [N] formed in the membranous tissues encapsulated (dotted-arrows) tumour mass [T]. At 6 weeks a much bigger tumour mass was formed. Images were taken from representative animals. 136

Figure 5.7 TRAP staining of the left tibias of the mice at 2, 4 and 6 weeks post inoculation. Osteoclast activity was found at the margin of the resorbed bone (B) and the tumour (T).

Figure 5.8 Histochemical analysis of the lungs of the mice at 4 and 6 weeks post inoculation. Tumour mass (T) was identified in the lung of the mice. 137

Figure 5.9 Detection of human Alu gene expression in the mice at 4 and 6 weeks post inoculation. In situ hybridization was used to identify the population of the human osteosarcoma cells in the lung.

5.4.4. Detection of the IGF-1R signalling pathway in the mouse model of osteosarcoma

Immunohistochemistry (section 3.2.4.3) was performed to investigate the involvement of IGF-1R signalling in the mouse model of osteosarcoma. Total proteins of IGF-1R,

MEK, Akt and p38-MAPK and phosphorylated MEK, Akt and p38-MAPK were tested

(there was no phosphorylated IGF-1R antibody available). Total protein expression of

IGF-1R, MEK, Akt and p38-MAPK were found in the primary tumour (Figure 5.10 A) and lung metastasis (Figure 5.10 B). The expression of total proteins of MEK, Akt and p38-MAPK were found mainly located in the nucleus of the tumour cells, whereas the expression of IGF-1R was mainly found located on the membrane of the tumour cells.

As shown in Figure 5.10, only the phosphorylated MEK protein was identified in both primary tumour and lung metastasis, but no phosphorylation of Akt and p38-MAPK was 138

detected in the samples. The staining of phosphorylated MEK was found mainly in the nucleus of the tumour cells and the staining intensity increased with time and thus it suggested that the amount of activated MEK protein was increased during the tumour progression. In addition, chamber slide cultured human (section 3.2.3.9) human osteosarcoma 143B cells, which used in the inoculation, showed a similar staining pattern (Figure 5.11) to the findings from the animal tissues.

(continue to the next page) 139

(continue from the previous page)

Figure 5.10 Detection of the IGF-1R signalling proteins and their phosphorylated forms in the mouse model of osteosarcoma. The total protein expression of the three IGF-1R related signalling pathway factors was detected in the primary tumour (A) and lung metastasis (B). Only phosphorylated MEK was tested positive in both primary and lung metastasis and the staining intensity was increased with time. Activation of the other 2 proteins was undetected at any time point. 140

Figure 5.11 Detection of the IGF-1R signalling proteins and their phosphorylated form in the chamber slide cultured 143B osteosarcoma cell line. The total protein expression of the three IGF-1R related signalling pathway factors was detected. Only phosphorylated MEK was test positive but the activation of the Akt and p38 proteins was not detected. 141

Table 5.1. Semi quantification of the total protein expression of the three IGF-1R related signalling pathway factors in the primary tumour, lung metastasis and 143B osteosarcoma cells

Leg Lung 143B

4 Weeks 6 Weeks 4 Weeks 6 Weeks

IGF-1R ++++ ++++ +++

Akt +++ ++++ ++++

Phospho-Akt 0 0 0 0 0

MEK ++++ ++++ ++++

Phospho-MEK + ++ +++ ++++ ++++

p38 ++++ ++++ ++++

Phospho-p38 0 0 0 0 0

Scores are based on the percentage of positive cells detected: 0 = <10%, + = 10-25%, ++ = 25-50%, +++ = 50-75%, and ++++ = >75%.

5.4.5. Regulation of invasion by inhibition of MEK/ERK under in vitro condition

Immunohistochemical analysis from the previous section revealed that MEK/ERK pathway is important and involved in osteosarcoma lung metastasis. The MEK/ERK inhibitor U0126 was used to investigate the effects of MEK/ERK inhibition on the in vitro invasive ability of the human osteosarcoma 143B cell using Matrigel™ invasion assay (section 3.2.3.10).

Dilution of Matrigel™ at 1:10, 1:20 and 1:40 was made with the growth medium.

Increased dilution lead to the decreased in the concentration of Matrigel™ and then resulted in a thinner matrix layer. The thickness of the matrix is related to the resistance to invasion in the in vitro condition. The invasiveness of the cells was measured by identifying the number of cells that penetrated the matrix layer and growth at the bottom of the transwell (Figure 5.12). As shown in Table 5.1, the MEK/ERK inhibitor U0126 treatment significantly reduced the cellular invasion at different 142

concentration of Matrigel™ on the human osteosarcoma 143B cell line when compared to the untreated control.

Table 5.2. Effect of MEK/ERK inhibitor U0126 on invasion by 143B osteosarcoma cells

Matrigel™ dilution U0126 (10 nM) Vehicle (0.06% DMSO) Significance P (2-tailed)

1:10 4 ± 1 23 ± 7 < 0.001

1:20 6 ± 1 44 ± 9 <0.001

1:40 6 ± 2 49 ± 11 <0.001 (Data shown were the means ± standard deviation of cells that penetrated the Matrigel™ and identified at the bottom of the transwell.)

Figure 5.12 Representative images of sample at 1:40 Matrigel™ dilution from the invasion assay. Photo on the left was selected from the vehicle control sample, which showed 9 of 143B cells penetrated Matrigel™ and reached the bottom filter area. Photo on the right was selected from the U0126 treated sample, where only 1 of 143B cells was identified at the bottom area. Thus, cell penetration was dramatically reduced after U0126 treatment.

5.5. Discussion

The low survival rate in metastatic disease and high rate of relapse and metastasis in patients under the current conventional treatments are the factors that make osteosarcoma treatment a challenging task in clinical oncology. Moreover, no further 143

improvement of the clinical situation achieved after the introduction of multi-regime chemotherapy. It is necessary to identify new molecular targets and develop new treatment approaches and modalities to improve the current clinical situation. In the present study, a usable orthotopic mouse model of human osteosarcoma growth and metastasis was successfully established.

Osteosarcoma is similar to many other cancers that deregulation of cell growth leads to the development of tumour. The insulin-like growth factor receptor (IGF-1R) signalling is one of the well studied cell growth regulation pathway and is involved in osteosarcoma pathogenesis and associated with chemoresistance. Ras signalling system is one of the well characterized downstream signalling pathway of IGF-1R and included three major signal transduction pathways, Ras/Raf/MEK/ERK,

Ras/Raf/MEK/p38-MAPK, and Ras/PI3K/Akt/mTOR.

Characterization and analysis on the model showed that only MEK protein was activated in the xenograft primary and metastatic tumours but not other two IGF-1R signalling proteins p-Akt and p-p38. As shown in Figure 5.13, the Ras/Raf/MEK/ERK will be the major pathway remained, if p38 and Akt pathways are not involved. The results are the first to support that the Ras/Raf/MAPK(MEK/ERK) signalling is important for osteosarcoma growth and metastasis. The Ras/Raf/MEK/ERK signalling is involved in the activation of transcription factors in regulating gene expression and cell survival, and is correlated with tumour progression and poor prognosis in patients with different cancers such as breast, colorectal, prostate and melanoma when aberrant activation occurred [442]. Moreover, studies have shown that somatic activation of Ras signalling in mice induced tumour development and metastasis [443,444], and hyperactivation of Ras signalling in epithelial cells enhanced the tumourigenic and metastatic potential [445].

Blocking MEK/ERK function by the MEK/ERK inhibitor U0126 significantly reduced in vitro invasive ability in the osteosarcoma cells. This study had demonstrated a potential 144

clinical implication of MEK/ERK in the prediction of metastasis and development of targeted therapy in osteosarcoma.

Figure 5.13 The three major Ras signalling pathways in IGF-1R.

Human osteosarcoma cell induced tumour mouse models are commonly used by researchers to study osteosarcoma progression and metastasis as well as the identification of the correlated genetic and biomolecular factors [1,272,446]. One of the commonly used cell lines in osteosarcoma study is the human osteosarcoma 143B cell line. A study showed that an orthotopic mouse model of 143B formed primary and spontaneous metastatic tumours efficiently, which makes it a useful model in studying factors involved in the spreading of human osteosarcoma by showing a varied degree of tumour growth at both primary and metastatic sites [253]. The orthotopic inoculation of osteosarcoma cells is a suitable model to represent the aspects of the human osteosarcoma, such that tumour is propagated from the bone and eventually mimicking a spontaneously metastasizing process in patients[1]. In the present study, the orthotopic 143B mouse model was established and confirmed as a reliable model for studying osteosarcoma growth and metastasis, as well as investigating the associated biomolecules. 145

The 143B cell line was a sub-clone of the human osteosarcoma HOS cell line generated via the transformation of k-ras oncogene similar to KRIB [1]. The major pathways involved in the Ras signalling system in gene regulation are the

Ras/Raf/MEK/ERK, Ras/Raf/MEK/p38-MAPK, and Ras/PI3K/Akt/mTOR pathways. The present study verified that only MEK proteins were activated in both primary tumour and pulmonary lung metastasis, whilst Akt and p38 were not activated, regardless to the positive detection of IGF-IR, MEK, Akt and p38 total proteins. These results indicated that activation of MEK/ERK signalling factor in the Ras signalling pathways play an essential role in local tumour development and lung metastasis in this mouse model of osteosarcoma.

Many studies demonstrated that the three signalling pathways in the Ras signalling system play different roles in gene regulation. A study showed that osteosarcoma cell differentiation was regulated differently by the balance between two signalling pathways, where the cell differentiation was promoted by the increased activity of p38-

MAPK pathway and suppressed by the increased activity of MEK/ERK pathway [447].

Furthermore, the hyperactivation of the Ras/Raf/MEK/ERK signalling was required in the epithelial-mesenchymal transition, tumorigenesis and tumour metastasis induced by transforming growth factor !, while the hyperactivation of Ras/PI3K/Akt signalling shown tumorigenesis but not metastasis and opposition effects to transforming growth factor ! induced apoptosis and epithelial-mesenchymal transition [445]. Enhanced invasion effects in N-cadherin transfected breast cancer was demonstrated by the fibroblast growth factor 2 stimulated matrix metalloproteinase MMP-9 expression along with the activation of the Ras/Raf/MEK/ERK signalling [448]. All these findings suggested that the Ras/Raf/MEK/ERK pathway is associated to the osteosarcoma metastatic process. 146

The Ras/Raf/MEK/ERK signal transduction pathway is present in all eukaryotic cells and is one of the well studied Ras effector pathway in terms of its biochemistry and association in disease [303]. This pathway is essential in the regulation of cellular process such as proliferation, differentiation, survival, angiogenesis, adhesion, mobility and transformation, in which the signal transduction to the nucleus starts from the extracellular stimulation of the IGF-1R that subsequently stimulates a cascade of intracellular phosphorylation with Ras, Raf, MEK and ERK, and from the activation of transcription factor such as AP1 that acts on more than 50 substrates in the cytosol [449-

451].

Approximately 30% of human cancers observed with constitutive expression of Ras that is resulted from mutation of the gene itself or amplification induced by other factors

[452,453] and approximately 7% of all cancers reported with mutation in B-Raf [454].

Moreover, the significant correlation between the level of ERK activation and the presence of lymph-node metastases identified from the investigations using head and neck squamous-cell and breast carcinomas, suggested that the signalling transduction by the Ras/Raf/MEK/ERK pathway is critical for cancer metastasis in patients [455,456].

The exact mechanism involved in the activation of the Ras/Raf/MEK/ERK pathway to promote tumour metastasis remains unclear. Alterations in the expression of integrin receptors for extracellular matrix proteins are strongly associated with the acquisition of invasive and/or metastatic properties by human cancer cells. A study showed that the

Ras/Raf/MEK/ERK signalling pathway regulates the expression of individual integrin subunits in a variety of human and mouse cell lines [457]. Expression of the 6- and !3- integrin on cell surface was reduced after pharmacological inhibition of MEK1 in a number of human melanoma and pancreatic carcinoma cell lines. Additionally, a 5 to

20-fold induction of cell surface a6- and b3-integrin expression was found after conditional activation of the Ras/Raf/MEK/ERK pathway in NIH 3T3 cells. 147

Gene knock-out mouse model shown that Raf-1 and BRAF signalling factor of the

Ras/Raf/MEK/ERK pathways was involved in the protection of apoptosis and induction of angiogenesis in endothelial cells, in which angiogenesis is known to be an essential process for tumour development and metastasis [458-460]. Thus, all this evidence indicates that the Ras/Raf/MEK/ERK pathway is involved in tumour metastasis via at least aforementioned mechanisms.

In the present study, MEK/ERK signalling factor was found highly activated in the pulmonary lung metastasis in the orthotopic mouse model of osteosarcoma, it was hypothesised that inhibition of MEK/ERK should translate into a measurable reduction in osteosarcoma invasive/metastatic ability. It was proved correct that the in vitro invasive ability of the 143B osteosarcoma cell was significant reduced after U0126 treatment and compared to the untreated controls.

It is for the first time to show that inhibition of MEK/ERK signalling factor by U0126 reduced in vitro invasion in osteosarcoma. Investigation from other studies has been showed that in vivo tumour invasion and metastasis was effectively inhibited by U0126 in other cancers [350,461,462]. Phase II clinical trials on the use of U0126 has been carried out in colon cancers, in which the host cell malignant behaviour was related to the frequent upstream K-ras and B-raf mutation that continuously transducting activated signals to the MEK/ERK signalling pathway [463]. Thus, it suggested that inhibition of

MEK signalling factor in the Ras/Raf/MEK/ERK pathway is a potential approach to implicate as a treatment strategy of patients with osteosarcoma.

Detection of the human-specific Alu gene was used in the present study to monitor disseminated human cancer cells in a background of large numbers of xenogeneic host

(e.g. mouse, rat or chick) cells. Conventional detection of metastasis by molecular methods such as dot blot hybridization [464], polymerase chain reaction (PCR) [465] and real-time PCR [466], required extraction of genomic DNA from the sample, which is 148

becoming difficult in some cases when the sample size is small. Detection of metastasis in orthotopic mouse model of osteosarcoma is achievable by using in vivo imaging technology [467], but it is limited by the availability of the technology and the possible uncertainty from the use of a luciferase transfected cell lines. In situ hybridization assay that direct detection of human Alu mRNA expression in human osteosarcoma cancer cells located in the mouse xenograft tissue samples is comparatively an easier and more reliable technique. This method is able to provide some extra information of the distribution of human cancer cells in the xenograft tissue samples, which is easily detected by using a simple microscopic technique. Therefore, the present study suggested that in situ hybridization assay in detection of human Alu mRNA is a very valuable tool in studying metastasis in orthotopic animal model of cancers.

5.6. Conclusion

A usable orthotopic mouse model of human osteosarcoma growth and metastasis was established and for the first time confirmed that the Ras/Raf/MEK/ERK pathway is activated in this mouse model of osteosarcoma. Furthermore, the reduction of invasiveness after inhibition of MEK/ERK showed its potential as therapeutic targets for the treatment of metastatic osteosarcoma. 149

CHAPTER 6. IGF-1R TARGETED THERAPY AND ITS ENHANCEMENT OF CHEMOSENSITIVITY IN HUMAN 6 OSTEOSARCOMA CELL LINES

6.1. Background

Surgery and multi-agent chemotherapeutic regimens are currently the mainstay of treatment for primary and/or metastatic osteosarcoma and have achieved 5-year disease-free survival of 70% only in the patients with primary osteosarcoma [2]. Despite the fact that application of chemotherapy has notably improved patient survival, development of drug resistance and toxic side effects associated with the use of chemotherapeutic agents remain a serious problem, and dose intensification did not show significant improvements to the clinical outcomes [2,468]. In addition, it has been shown that radiotherapy brought no significant survival benefit to patients with osteosarcoma [8-10]. Therefore, identification and development of new therapeutic targets to increase the effectiveness of the current therapeutic modalities in osteosarcoma patients are important.

Type I insulin-like growth factor receptor (IGF-1R) and its signalling regulates cell survival, proliferation and metabolism, as well as participate in the development and progression of tumours, including osteosarcomas [33,309,469]. In addition, IGF-1R also plays a role in tumour resistance to chemotherapy and radiotherapy [309]. Thus, IGF-1R and its downstream signalling pathways have become attractive targets in cancer research for improving current cancer treatments [282,328]. 150

Several approaches have been exploited in targeting IGF-1R to block its signalling, including the use of antibodies, specific tyrosine kinase inhibitors, recombinant proteins and siRNAs, and many of these IGF-1R inhibition agents are in different stages of the clinical trials [33,328,333,334]. Tyrphostin AG1024 is a commercially available small molecule IGF-1R tyrosine kinase inhibitor, and it was widely used in studies for investigation of the inhibition of IGF-1R and its signalling [342,345,470,471].

Inhibition of IGF-1R has shown positive effects in regulating tumourigenic and metastatic properties of Ewing’s sarcoma cells [332,378] and also in osteosarcoma cells

[331]. However, further studies are required to investigate whether IGF-1R inhibition together with other conventional therapy, particularly chemotherapy could improve the treatment in osteosarcoma.

6

6.2. Hypothesis and aim

It is hypothesized that targeting IGF-1R alone or in combination with Doxorubicin chemotherapy would promote favourable anti-proliferative effect on osteosarcoma.

Mechanism studies would helped to improve the understanding and verified the potential implication of this IGF-1R targeted combination therapy in osteosarcoma treatment.

This study aimed to investigate the effect of mono-drug therapy with the IGF-1R inhibitor Tyrphostin AG1024 or Doxorubicin, and in combination of both drugs on a panel of 6 human osteosarcoma cell lines and a Doxorubicin resistant osteosarcoma sub-line, followed by investigating the mechanisms of the combination therapy.

6 151

6.3. Methods

Drug treatment effectiveness as analyzed by the Calcusyn™ program and the mechanisms of the treatments were investigated by studying the cytotoxicity, apoptosis, clonogenic survival and cell cycle progression. Doxorubicin resistant cell line was established by stepwise increase of doxorubicin concentration in the growth medium.

6.4. Results

6.4.1. IGF-1R expression in osteosarcoma cell lines

IGF-1R expression was examined in a panel of 6 osteosarcoma cell lines. Total proteins extracted (section 3.2.2.11) from osteosarcoma cells were analysed with western blotting (section 3.2.2.12) and was carried out by Johnathan Kuang from our group. The relative expression level was calculated by comparing net band intensity to the !-actin control. A rabbit polyclonal IgG antibody against the IGF-1R extracellular - subunit and a mouse monoclonal anti-!-actin antibody were used for the protein detection. Analysis showed that IGF-1R expression (Figure 6.1) was positively detected at different levels in all 6 osteosarcoma cell lines used in this study, where

HOS had the highest level of IGF-1R expression and U-2 OS had the lowest level

(MG63 = 0.556, Saos-2 = 0.889, U-2 OS = 0.222, 143B = 0.857, HOS = 1.095, SJSA =

0.905). This result justifies the selected osteosarcoma cell lines in the following IGF-1R targeted therapy experiments. Studies from other researchers confirmed that IGF-1R was expressed in MG-63 and U-2 OS cell lines, thus it was further confirmed in this examination [472,473]. 152

Figure 6.1 Western blot analysis of IGF-1R expression in osteosarcoma cell lines. Relative expression of IGF-1R was assessed by comparing the band net intensity ratio of IGF-1R to !-actin.

6.4.2. Determination of growth condition in osteosarcoma cell lines for drug treatment analysis

Prior to the examination of drug effects, the growth rate (section 3.2.3.4) (Figure 6.2) and seeding concentration of individual osteosarcoma cell lines were analysed to determine the optimum growth condition for the investigation, which cells were allowed to grow under normal condition for 72 h without reaching 100% confluence and remained in linear growth phase. Assessment showed that the cell concentration for all

6 osteosarcoma cells in further drug treatment analysis was found to be 5 x 103 cells per well in a 96-well plate. 153

Figure 6.2 Growth rate of the osteosarcoma cell lines. The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples. ANOVA and post-hoc Bonferroni correction test analysis showed that the doubling time was significantly different (p<0.008) among the six human osteosarcoma cell lines, which were then divided into 3 groups: slow (a), medium (b) and fast (c) growth based on Bonferroni test results in which there was significant difference in growth speed between the individual cell lines from different groups but not within the group.

6.4.3. Anti-proliferation effects of Tyrphostin AG1204 on osteosarcoma cell lines

The anti-proliferation effects of the IGF-1R inhibitor, Tyrphostin AG1024 were evaluated in 6 osteosarcoma cells that had confirmed with IGF-1R expression by crystal violet staining (section 3.2.3.3). Increasing dosages of Tyrphostin AG1024 [2.5,

5, 10 and 20μM] were used to investigate the dose-response of individual cell lines and medium contained 0.03% DMSO (drug-vehicle) was used as an untreated control. All procedures involved in the use of Tyrphostin AG1024 were carried out in low light intensity condition. 154

Dose dependent growth inhibition was observed in all 6 osteosarcoma cell lines after

72 h of drug treatment and the dose effect curves are shown in Figure 6.3. Analysis showed that individual cell lines had different sensitivities to Tyrphostin AG1024 with about 2-fold differences between the highest and the lowest sensitive demonstrated by the IC50 values obtained from the Calcusyn™ drug treatment effects program (Table

6.3 at p.177). Saos-2 was found as the most resistant cell line to Tyrphostin AG1024 with IC50 at 20.21 ± 4.28 µM, while 143B was the most sensitive cell line with IC50 at

10.32 ± 1.43 µM. Correlation analysis showed that there was a negative relationship between the level of IGF-1R expression and the sensitivity to Tyrphostin AG1024 treatment in osteosarcoma cell line (Figure 6.1 and Table 6.3 at p.177) (Pearson, r = -

0.455, p = 0.364; Spearman, r = - 0.314, p = 0.564). However, the relationship did not reach statistically significant level. 155

Figure 6.3 Dose effect curves of the osteosarcoma cell lines from Tyrphostin AG1024 treatment. Curves were taken from one representative experiment and showing the % growth inhibition (Effect) versus drug concentration (Dose) after 72 h of drug treatment. The data conformities were indicated by the linear correlation coefficient (R2) value shown in each curve. 156

6.4.4. Anti-proliferation effects of Doxorubicin on osteosarcoma cell lines

The anti-proliferation effect of Doxorubicin was also evaluated in 6 osteosarcoma cells

(section 3.2.3.3). Increasing dosages of Doxorubicin [2.5, 5, 10, 20, 50 and 100nM] were used to investigate the dose-response of individual cell lines and medium with no addition of drug was used as an untreated control.

Dose dependent growth inhibition was observed in all 6 osteosarcoma cell lines after

72 h of drug treatment and the dose effect curves are shown in Figure 6.4. Analysis showed that individual cell lines had different sensitivities to Doxorubicin with about 4- fold differences between the highest and the lowest sensitive demonstrated by the

IC50 values obtained from the Calcusyn™ drug treatment effects program (Table 6.3 at p.177). MG63 was found as the most resistant cell line to Doxorubicin with IC50 at

56.60 ± 7.49 nM, while 143B was the most sensitive cell line with IC50 at 14.99 ± 4.94 nM. Correlation analysis showed that there was a negative relationship between the level of IGF-1R expression and the sensitivity to Doxorubicin treatment in osteosarcoma cell line (Figure 6.1 and Table 6.3 at p.177) (Pearson, r = - 0.364, p =

0.478; Spearman, r = - 0.371, p = 0.497). However, the relationship did not reach statistically significant level. 157

Figure 6.4 Dose effect curves of the osteosarcoma cell lines from Doxorubicin treatment. Curves were taken from one representative experiment and showing the % growth inhibition (Effect) versus drug concentration (Dose) after 72 h of drug treatment. The data conformities were indicated by the linear correlation coefficient (R2) value shown in each curve. 158

6.4.5. Anti-proliferation effects of the combination treatment with Tyrphostin AG1024 and Doxorubicin on osteosarcoma cell lines

The anti-proliferation effect in the drug combination treatment with Tyrphostin AG1024 and Doxorubicin was also evaluated in the 6 osteosarcoma cells (section 3.2.3.5). To investigate if the combined drug treatment would enhance the anti-proliferative effects, the “Constant ratio two drugs combination - diagonal design” adapted from Chou-

Talalay method [474] was applied for this purpose. A schematic presentation for mono drug treatment and two drug combination treatment are shown in Table 6.1, media only and media with 0.03% DMSO were used as controls in both mono drug and drug combination treatments. All osteosarcoma cells were treated for 72 h before being analysed for its growth situation.

As shown in Figure 6.5, similar to the single drug treatment, dose dependent growth inhibitions were identified after the combination treatment in all 6 osteosarcoma cell lines. Growth inhibition enhancements were identified when a higher percentage of relative inhibition achieved in drug combination treatment compared to mono drug treatments. Enhanced growth inhibitions were found in the drug combination treatment for all 6 osteosarcoma cell lines at the higher specific dose combinations (AG 10 µM /

Dox 10 nM, AG 20 µM / Dox 20 nM). Additionally, all osteosarcoma cell lines except for

SJSA showed enhanced growth inhibitions at the lower specific dose in the combination therapy (AG 2.5 µM / Dox 2.5 nM, AG 5 µM / Dox 5 nM). Moreover, the higher dose of Tyrphostin AG1024 showed a significant growth inhibition in all 6 osteosarcoma cell lines. Collectively, the results also demonstrated that the growth inhibition effect in the combination treatment under current dose ratio was mainly contributed by IGF-1R inhibitor Tyrphostin AG1024. 159

Table 6.1. The “constant ratio two drug combination” drug combination treatment design model.

Tyrphostin AG1024 (AG)

2.5 nM 5 nM 10 nM 20 nM

AG 2.5 nM 2.5 µM + ------Dox 2.5 µM AG 5 nM Doxorubicin 5 µM ------+ ------(Dox) Dox 5 µM AG 10 nM 10 µM ------+ ------Dox 10 µM AG 20 nM 20 µM ------+ Dox 20 µM (The top horizontal grids represented Tyrphostin AG1024 monotherapy and Doxorubicin monotherapy was showed on the left vertical grids. Four different dosage of the two drugs used in combination was illustrated at the diagonal grids.) 160

(continued on the next page) 161

(continued on the next page) 162

(continued on the next page) 163

Figure 6.5 Growth inhibition comparison in osteosarcoma cell lines after mono drug and combination treatment at a specific dosage. The graphs show the percentage of relative inhibition (vertical axis) and the different dosage treatment groups (horizontal axis): Tyrphostin AG1024 monotherapy (light grey), Doxorubicin monotherapy (white) and combination treatment (dark grey). The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples. Significance in cell growth inhibition after combination treatment was determined by ANOVA and Bonferroni test wherever appropriate. *p<0.05, compared with Doxorubicin monotherapy. **p<0.05, compared with Tyrphostin AG1024 monotherapy.

6.4.6. Synergy analysis of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on osteosarcoma cell lines

Synergy analysis was performed by Calcusyn™ drug effects program to assess the enhanced growth inhibition effects statistically by calculation of combinational index (CI) and graphically by isobologram plotting. Combination index denoted that the drug combination effect is super-additive when the median drug effect analysis shows that the CI is smaller than 0.85 under Chou-Talalay experimental strategy, an sub-additive effects when CI is larger than 1.15, and an additive effect when CI is between 0.85 to

1.15. In the isobologram, a super-additive effect is determined when the combination data point falls at each ED (effective dose) level on the lower left of the diagonal line, an additive effect is determined when a data point falls on the diagonal line, and a sub- additive effect is determined when the data point falls on the upper right of the diagonal line.

The significance of the combination effect was determined by ED90, which represented the 90% efficacy of the drug treatment reaction. As shown in Figure 6.6, the combination index at ED90 revealed that super-additive anti-proliferation effect was found on 143B and HOS, additive effect on Saos-2, and sub-additive effect on MG63,

U-2 OS and SJSA after drug combination treatment with Tyrphostin AG1024 and

Doxorubicin. The strongest enhancing effect was achieved on 143B with the CI = 0.54 164

at ED90 after drug combination treatment, while the least enhancing effect was achieved in U-2 OS with highest CI = 2.28 at ED90. Although demonstrated in the previous section that growth inhibition effect was significantly increased at the higher dosage in all 6 osteosarcoma cell line after combination treatment, this analysis showed that 3 out of 6 cell lines did not benefit from the combination treatment.

Furthermore, Figure 6.7 displayed the comparison of the dose effect curves from drug combination treatment and single drug treatment of the 6 osteosarcoma cell lines.

Figure 6.6 Synergy analysis of the osteosarcoma cell lines after combination treatment. The significance of the combination treatment effectiveness was considered based on the CI values at ED90. The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples. ANOVA and Bonferroni correction tests were performed to analyse difference among/between individual cell lines. Two cell lines (U-2 OS and 143B), which had both different CI distribution and significantly statistical CI difference (p<0.05), were chosen for the mechanism studies. 165

(continued on the next page) 166

(continued from the previous page)

Figure 6.7 Dose effect curves and isobologram analysis in the osteosarcoma cell lines after combination treatment. The dose effect curves on the left showed a measure of % relative inhibition following AG1024 monotherapy (green), Doxorubicin monotherapy (blue) and combination therapy (red). The isobolograms on the right showed the required Tyrphostin AG1024 (x-axis) and Doxorubicin (y-axis) doses for a given effect level ED50-isobol (red), ED75-isobol (green) and ED90-isobol (blue). Results displayed were from a representative experiment. 167

As showed in Table 6.3 (p.177), each cell line had different sensitivities to Tyrphostin

AG1024 and Doxorubicin in the combination therapy demonstrated by the IC50 values analysed from the Calcusyn™ drug treatment effects program. 143B was the most sensitive cell line to Tyrphostin AG1024 and MG63 was the most sensitive to

Doxorubicin, while Saos-2 was the most resistant to both Tyrphostin AG1024 and

Doxorubicin in the combination therapy. This result demonstrated that no direct relationship between the effectiveness of the combination treatment and sensitivity to the drugs. Furthermore, the computed analysis also provided “drug reduction index”

(DRI) in all osteosarcoma cell lines, in which DRI is a measure comparing drug dosage at IC50 of the mono drug treatment over the drug combination treatment (DRI>1 is reduction, DRI = 1 is no change, DRI<1 is no reduction). All 6 osteosarcoma cell lines showed different levels of drug reduction in both Tyrphostin AG1024 and Doxorubicin in the combined treatment (Table 6.3 at p.177). However, this DRI was a useful piece of information to the cell lines that showed synergistic and additive effects but not to the antagonistic situation. A smaller level of reduction was demonstrated in Tyrphostin

AG1024 within the osteosarcoma cell lines in synergistic and additive group with the highest reduction of 1.23-fold change in 143B and a higher level of reduction was shown in Doxorubicin with the highest reduction of 5.48-fold change in HOS.

The two osteosarcoma cell lines 143B and U-2 OS were chosen for further analysis in order to understand the anti-proliferation effects from the combination treatment. 143B was having the strongest enhancing effect (synergistic) from the combination treatment, and being the most sensitive to Tyrphostin AG1024 in both single drug and combination drug treatment. U-2 OS was having the least enhancing effect

(antagonistic) based on the CI analysis. 168

6.4.7. Mechanism studies on selected osteosarcoma cell lines from the combination treatment with Tyrphostin AG1024 and Doxorubicin

Studies were carried out to investigate mechanisms involved in the enhanced anti- proliferation effects after the combination drug treatment. Cytotoxicity (trypan blue exclusion assay), cell survival ability (clonogenic assay), cell cycle progression (flow cytometry analysis), and apoptosis induction (immunocytochemical staining) were used to examine the two selected osteosarcoma cell lines after mono or combination treatment with Tyrphostin AG1024 and Doxorubicin.

6.4.7.1. Cytotoxic effect of the combination treatment

Cytotoxicity was determined by analysing the percentage of non-viable cells after treatment with drugs for 72 h (section 3.2.3.6). Drug concentration at 10 µM of

Tyrphostin AG1024, 10 nM of Doxorubicin, or a combination of 10 µM of Tyrphostin

AG1024 and 10 nM of Doxorubicin was used for this analysis based on the fact that these were the minimum level of drugs that had showed growth inhibition in all 6 osteosarcoma cells from the previous section.

As shown in Figure 6.8, 13.55% of 143B and 5.96% of U-2 OS non-viable cells were found after treatment with Tyrphostin AG1024 alone, and 10.19% of 143B and 4.65% of U-2 OS were found after treatment with Doxorubicin alone. Moreover, 24.47% of

143B and 9.56% of U-2 OS non-viable cells were found after drug combination treatment. It was clear that the percentage of non-viable cells was increased after treatment with Tyrphostin AG1024 or Doxorubicin alone compared to the untreated and drug vehicle controls and was further increased after the combination treatment in both

143B and U-2 OS osteosarcoma cell lines. Furthermore, the overall percentage of non- viable cells in 143B was about 2-fold higher than U-2 OS, and it was likely caused by the higher drug sensitivity in 143B. This result supports that the additive or supper- 169

additive anti-tumour effect by the combination therapy with AG1024 and Doxorubicin in selective osteosarcoma cell lines is at least due to induction of more cytotoxicity to the targeted cells.

Statistical analysis showed that the percentage of non-viable cells in both 143B and U-

2 OS osteosarcoma cell lines was significantly (p<0.05) increased for about 2-fold when compared the combination drug treatment to the mono drug treatments.

Moreover, the percentage of non-viable cells was increased for about 3-fold when compared the combination drug treatment to the untreated and drug vehicle controls in both cell lines. The result had verified that the cytotoxicity was increased in the combination drug treatment compared to the mono drug treatments. The potential significance of this study in clinic setting will provide a way to achieve higher treatment efficacy without increasing of dose(s) of toxic drug(s).

Figure 6.8 Cytotoxic effect of the combination treatment in the selected osteosarcoma cell lines. Percentage of non-viable cells in 143B and U-2 OS was significantly (p<0.05) increased from mono drug treatment to combination treatment, this indicated that the cytotoxicity of the combination treatment was increased. The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples. 170

6.4.7.2. Clonogenic inhibition by the combination treatment

Cell survival ability was determined by analysing the number of the colony (equal or more than 50 divided cells from a single osteosarcoma cell) formed after mono and combination drug treatment, which was calculated as the survival fraction in regards to the untreated or vehicle controls (section 3.2.3.7). Osteosarcoma cells were treated with Tyrphostin AG1024 (2.5, 5 and 10 µM), Doxorubicin (2.5, 5 and 10 nM), or a combination of Tyrphostin AG1024/Doxorubicin (2.5 µM / 2.5 nM, 5 µM / 5 nM, 10 µM /

10 nM) until a colony was found in the controls after 10 days.

As shown in Figure 6.9, dose dependent inhibition of colony formation was identified in the mono drug treatment with Doxorubicin but had minor inhibition effect with

Tyrphostin AG1024 on both 143B and U-2 OS. Moreover, dose dependent inhibition of colony formation was also identified in the combination treatment. However, the inhibition effect of mono drug treatment with Doxorubicin was very similar to the combination treatment. Although variation of colony formation was identified, but the overall reduction effects in 143B (p=0.2399) and overall enhancement effects in U-2

OS (p=0.2723) were not statistically significant after comparison of drug combination treatment to the Doxorubicin only treatment. This result indicated that the inhibition effect observed in the combination treatment was dependent upon Doxorubicin but with minimum effect from Tyrphostin AG1024, suggesting the combination therapy had more advantages than mono drug treatment in terms of maintain (e.g. colony inhibition) or improvement (e.g. cytotoxicity) in therapeutic efficacy without addition of individual drug doses (no increment of side-effect). Most of osteosarcoma patients are children, who have a low tolerance to Doxorubicin. Another advantage for the combination therapy would be maintaining treatment efficacy with Tyrphostin AG1024 plus the reduced dose of Doxorubicin. In this way, more children patients will be tolerable to the treatment and benefited. 171

Figure 6.9 Clonogenic survival status in the selected osteosarcoma cell lines after combination treatment. Limited inhibition effects was shown in colony formation from inhibition of IGF-1R alone and a dose dependent effect was shown between the combination therapy and Doxorubicin only treatment in both 143B and U-2 OS. Small reduction of colony formation was identified from 143B after combination treatment but not significant. The inhibition effect from the combination treatment was mostly dependent upon Doxorubicin. The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples.

6.4.7.3. Cell cycle arrest induction by the combination treatment

Uncontrolled tumour growth is associated with deregulated cell cycle progression and thus tumour growth inhibition could be mediated via alteration of cell cycle [475].

Osteosarcoma cells were analysed by flow cytometry to examine the cell cycle distribution after 72 h of mono and combination drug treatment (section 3.2.3.8). Drug concentration at 10 µM of Tyrphostin AG1024, 10nM of Doxorubicin, or a combination of 10 µM of Tyrphostin AG1024 and 10nM of Doxorubicin was used for this analysis based on the fact that these were the minimum level of drugs that had showed growth inhibition in all 6 osteosarcoma cells from the previous section.

As shown in Figure 6.10, higher percentage of cells in the G1-phase were found in

143B after treated with 10µM Tyrphostin AG1024 (82.45%) compared with the vehicle 172

treated control (74.02%). Similarly, higher percentage of U-2 OS cells in the G1-phase were also found after treatment with 10µM Tyrphostin AG1024 (74.81%) compared with the vehicle treated control (62.56%). However, higher percentage of cells in the

G2/M-phase were only found in 143B after treated of 10nM Doxorubicin (61.93%) compared with the untreated control (20.99%), whilst U-2 OS cells in the G2/M-phase were almost unchanged between Doxorubicin treated (21.93%) and untreated control

(22.52%). After combination treatment, 143B showed a dual cell peaks in the G1

(35.95%) and G2/M-phase (32.43%). However, U-2 OS cells only showed a G1 peak, which was slightly increased after combination treatment (69.88%) compared to the vehicle (62.56%) and untreated (63.22%) control.

The result clearly showed that treatment with Tyrphostin AG1024 lead to G1-phase cell cycle arrest in both 143B and U-2 OS, but treatment with Doxorubicin only leads to

G2/M-phase cell cycle arrested in 143B cells but not in U-2 OS. Moreover, an accumulative effect was shown in both osteosarcoma cell lines that 143B had both G1 and G2/M-phase arrested and U-2 OS had only G1-phase arrested after the drug combination treatment. Apart from differential cytotoxicity effect on different osteosarcoma cell lines introduced in the previous section, the combination therapy also induced different cell cycle arrest models (single G1-phase block or blocking at both G1 and G2/M phases) in different osteosarcoma cell lines. All these results provided evidence to explain the different responses between various osteosarcoma cell lines, such as supper-additive, additive or sub-additive effect to the combination therapy. 173

Figure 6.10 Cell cycle distribution analysis in the selected osteosarcoma cell lines after combination treatment. Cells were treated with mono or combined-drugs for 72 hours and stained with PI. Flow cytometry was use to determine the cell cycle progression status after drug treatment. Analysis showed that Doxorubicin had caused G2/M arrest in the 143B but not in U-2 OS, and IGF-1R inhibition have caused G1 arrest in both cells. In the combination therapy, 143 had showed both G1 and G2/M arrest, but U-2 OS only have G1 arrest. The differences showed here may be due to the different sensitivity of the cells towards each drug and also effectiveness of the combination therapy to the 2 cells. Result displayed was from a represented experiment.

6.4.7.4. Apoptosis induction by the combination treatment

Apoptosis is a process of programmed cell death involved a group of enzymes called

“caspases”, which leads to a series of biochemical events causes cell morphological changes and death, and defective of such process resulted in uncontrolled cell proliferation in many disease, such as cancer [476]. There are two apoptosis pathways, mitochondrial and death receptor pathways. Both pathways will eventually result in activation of caspase-3 and lead to apoptosis. Osteosarcoma cells were analysed by immunocytochemical staining (section 3.2.3.9) of the “cleaved caspase-3 protein” 174

(evidence of activation of caspase-3) to examine the apoptosis after 24 h of mono and combination drug treatment. Drug concentration at 40 µM of Tyrphostin AG1024,

100nM of Doxorubicin, or a combination of 40 µM of Tyrphostin AG1024 and 100nM of

Doxorubicin was used for this analysis.

As shown in Figure 6.11, 28.76% of 143B and 13.67% of U-2 OS apoptotic cells were found after treatment with Tyrphostin AG1024 alone, and 27.44% of 143B and 18.54% of U-2 OS were found after treatment with Doxorubicin alone. In contrast, 39.38% of

143B and 26.88% of U-2 OS apoptotic cells were found after drug combination treatment. Results showed that the percentage of apoptotic cells was increased after treatment with Tyrphostin AG1024 or Doxorubicin alone compared to the untreated and drug vehicle controls and was further increased after the combination treatment in both

143B and U-2 OS osteosarcoma cell lines. Furthermore, the apoptotic cells in 143B was higher than U-2 OS in different drug treatment approaches, and it was likely caused by the higher drug sensitivity of 143B.

Statistical analysis showed that the percentage of apoptotic cells was significantly

(p<0.05) increased for roughly 1.4-fold in 143B and 2-fold in U-2 OS when comparing the combination drug treatment to the mono drug treatments. Moreover, the percentage of apoptotic cells was increased for more than 3-fold when comparing the combination drug treatment to the untreated and drug vehicle controls in both cell lines.

The result had verified that apoptosis increment was another advantage for the combination treatment as compared to the mono drug treatments. 175

Figure 6.11 Apoptotic effect of the combination treatment in the selected osteosarcoma cell lines. Percentage of cells with positive staining of cleaved caspase-3 protein was significantly (p<0.05) increased from mono drug treatment to combination treatment, this indicated that the apoptotic effects of the combination treatment was increased. The means and error bars (standard deviation) were derived from triplicate experiments using triplicate samples.

6.4.8. Effectiveness of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on a Doxorubicin- resistance osteosarcoma cell line

There has been a clinical problem for some patients to resist Doxorubicin treatment.

This resistance can be primary or secondary developed after a period of Doxorubicin treatment. To identify a way of overcoming this problem, Doxorubicin-resistant osteosarcoma cell lines were established from the parental cells and used to further the investigation of the effectiveness of the drug combination treatment in osteosarcoma cell lines (section 3.2.1.5). 143B, which was the most drug sensitive cell line from the previous examination, was chosen to produce a Doxorubicin-resistant strain. The

143B-Dox-400 was obtained when parental 143B cells was adapted to 400 nM

Doxorubicin supplemented in the growth medium after 39 passages of culture in a slow

Doxorubicin dose increasing process. The 143B-Dox-400 was continuously cultured 176

and stocked in the presence of 400 nM Doxorubicin to maintain the selective pressure.

However, this selective pressure was removed during the seeding and settlement procedure prior to any study.

6.4.8.1. Anti-proliferation effects of Tyrphostin AG1024 and Doxorubicin mono drug treatment on a Doxorubicin- resistant osteosarcoma cell line

The anti-proliferation effects of Tyrphostin AG1024 and Doxorubicin was evaluated in the self-established 143B-Dox-400 Doxorubicin-resistant osteosarcoma cell line.

Increasing dosages of Tyrphostin AG1024 [2.5, 5, 10 and 20μM] and Doxorubicin [5,

10, 20, 50, 100 and 200 nM] were used to investigate the dose-response of the cell line, medium contained 0.03% DMSO and medium only were used as controls. The dose effective curves (Figure 6.12) showed a dose dependent growth inhibition was observed in 143B-Dox-400. As shown in Table 6.3 (p.177), drug sensitivity of 143B-

Dox-400 was compared to its parental 143B osteosarcoma cell line. The result confirmed that Doxorubicin-resistant strain 143B-Dox-400 had about 120-fold increased in resistance to Doxorubicin and showed a very similar sensitivity to

Tyrphostin AG1024 when compared to the parental 143B osteosarcoma cell line (Table

6.3 at p.177).

Figure 6.12 Dose effect curves on the 143B-Dox-400 Doxorubicin-resistant osteosarcoma cell line. Drug treatment of Tyrphostin AG1024 was shown on the left and Doxorubicin was on the right. Curves were taken from one representative experiment. 177

6.4.8.2. Synergy analysis of drug combination treatment with Tyrphostin AG1024 and Doxorubicin on a Doxorubicin- resistant osteosarcoma cell line

The “Constant ratio two drug combination - diagonal design” adapted from Chou-

Talalay method [474] was again applied for the investigation of the anti-proliferation effect in the drug combination treatment with Tyrphostin AG1024 and Doxorubicin in the 143B-Dox-400 Doxorubicin-resistant osteosarcoma cell line (Table 6.2). The synergy analysis of combinational index (CI = 1.13 ± 0.13) and isobologram plotting

(Figure 6.13) on 143B-Dox-400 showed that an additive effect after drug combination treatment with Tyrphostin AG1024 and Doxorubicin.

Furthermore, as shown in Table 6.3 (p.177), the drug sensitivity of 143B-Dox-400 in the combination treatment was 10.09 ± 3.93 µM with Tyrphostin AG1024 and was 0.34 ±

0.02 nM with Doxorubicin. The drug reduction of 143B-Dox-400 was 1.05-fold with

Tyrphostin AG1024 and >1,000-fold with Doxorubicin in the combination treatment.

This result of the low AG1024 drug reduction is similar to the 143B parental cell line, but the reduction in Doxorubicin was much more elevated in the 143B-Dox-400

Doxorubicin-resistant cell line than the parental cell line, suggesting IGF-1R inhibition was an effective way to overcome Doxorubicin-resistance. This result will have a clinical implication in treatment of Doxorubicin-resistant osteosarcomas using IGF-1R blocker and/or Doxorubicin. 178

Figure 6.13 Dose effect curves and isobologram analysis in 143B-Dox-400 osteosarcoma cell line after combination treatment. Curves were taken from one representative experiment.

Table 6.2. Drug combination treatment design model for Doxorubicin resistant osteosarcoma cell line

Tyrphostin AG1024 (AG)

2.5 nM 5 nM 10 nM 20 nM

AG 2.5 nM 5 µM + ------Dox 5 µM AG 5 nM Doxorubicin (Dox) 10 µM ------+ ------Dox 10 µM AG 10 nM 20 µM ------+ ------Dox 20 µM AG 20 nM 40 µM ------+ Dox 40 µM (The top horizontal grids represented Tyrphostin AG1024 monotherapy and Doxorubicin monotherapy was showed on the left vertical grids. Four different dosage of the two drugs used in combination was illustrated at the diagonal grids.) 179

Table 6.3. Drug sensitivity (IC50 values) of the 6 osteosarcoma cell lines and Doxorubicin resistant cell line. 5.8 5.8 4.43 4.43 2.74 5.48 4.11 18.93 18.93 DRI after >1,000.00 >1,000.00 Treatment Treatment Combination Combination ndex (DRI) indicated indicated (DRI) ndex IC 50 at IC Treatment Treatment 2.99 ± 0.70 0.70 2.99 ± 1.53 3.83 ± 0.91 7.65 ± 1.22 5.48 ± 0.99 3.65 ± 0.02 0.34 ± 15.35 ± 4.00 4.00 15.35 ± Combination Combination Doxorubicin (nM) Doxorubicin Treatment Treatment 56.60 ± 7.49 7.49 56.60 ± 5.92 22.23 ± 4.14 33.86 ± 7.74 30.02 ± 4.94 14.99 ± 42.00 ± 10.79 10.79 42.00 ± IC 50 at Mono Mono 50 at IC 1,808.26 ± 647.26 647.26 ± 1,808.26 1.24 1.24 1.10 2.07 1.07 1.04 1.23 1.05 DRI after Treatment Treatment Combination Combination triplicate experiments with triplicate samples. Drug reduction i reduction Drug samples. triplicate with experiments triplicate IC 50 at IC Treatment Treatment 9.63 ± 0.38 0.38 9.63 ± 0.96 9.88 ± 1.23 8.41 ± 11.48 ± 2.15 2.15 11.48 ± 0.61 11.94 ± 2.23 18.88 ± 3.93 10.09 ± Combination Combination Tyrphostin AG1024 (µM) AG1024 Tyrphostin Mono IC 50 at IC Treatment Treatment 14.22 ± 2.17 2.17 14.22 ± 1.09 13.09 ± 2.02 19.91 ± 4.28 20.21 ± 1.30 14.18 ± 1.43 10.32 ± 4.30 10.62 ± HOS 143B 143B SJSA MG63 MG63 Saos-2 Saos-2 U-2 OS Cell Line Cell Line 143B-Dox-400 143B-Dox-400 Group Group Additive Additive Resistant Resistant Doxorubicin Doxorubicin Sub-additive Sub-additive Super-additive Super-additive (Results were displayed as the mean IC50 ± S.E.M. derived from S.E.M. derived ± mean IC50 as the displayed were (Results the fold of the mean IC50 reduction achieved after the combination treatment compared to the mono drug treatment.) drug treatment.) mono the to compared treatment combination the after achieved reduction IC50 mean the of the fold 180

6.5. Discussion

Search for new therapeutic approaches is required to improve the current multimodality treatments available for osteosarcoma patients. Targeted therapy of IGF-1R inhibition has shown its potential in cancer treatments, but with limited information on its effects in osteosarcoma. This study is the first to show that growth inhibition of osteosarcoma cells by Doxorubicin was enhanced by blocking IGF-1R, suggesting inhibition of IGF-

1R signalling may be helpful in increasing of the sensitivity of osteosarcoma to chemotherapeutic agents. Further investigation showed that enhanced growth inhibition effect was also achieved in chemoresistant osteosarcoma cell by blocking of

IGF-1R signalling and combined with chemotherapy. Results from this study demonstrated that targeting IGF-1R have a potential in clinical application as an effective treatment option with benefit to improve chemotherapy for osteosarcoma patients.

In recent years, studies have shown the pivotal role of IGF-1R signalling in regulation of cell cycle, cellular proliferation and during embryonic development and growth, and later the discovery of its involvement in cancer transformation and development, and especially development of resistance in treatments makes it a potential therapeutic target for cancer treatment [286,288,309]. Studies showed that experimental inhibition of

IGF-1R receptor in cancer cells mediated the proliferation and survival of these cells after treatments [333,345,477]. Evidence from these studies also demonstrated that targeting IGF-1R was an effective therapeutic strategy in dealing human cancer such as breast cancer, prostate cancer, melanoma, mesothelioma and Ewing’s sarcoma

[332,333,345,378,477-479]. Expressions of IGF-1R were also found in primary and metastatic osteosarcoma [104], but there is limited information about the effects of targeting inhibition of IGF-1R in the treatment of osteosarcoma. 181

In this study, Western blot analysis showed that the total protein expression of IGF-1R appeared to be much higher in Saos-2 cells than in U-2 OS cells. Gene expression analysis by microarray study only showed a minor but not significant increase of IGF-

1R in Saos-2 compared to U-2 OS. This result suggested that IGF-1R protein expression in osteosarcoma may under translational control, which was a similar event to p53 in human leukemic blast cells [480]. In addition, IGF-1R protein and gene expression does not associate with patient survival in non-small lung cancer. The IGF-

1R protein expression is related to EGFR expression in squamous cells versus other histologies and correlates with EGFR expression, while its high gene copy number is associated to positive prognostic value [481]. Other studies have demonstrated that activation of IGF1R is an important factor contributed to protect cancer cells from apoptosis induced by chemotherapy and radiation, and the receptor activation also mediates resistance to chemotherapy and radiation [482-484]. Thus, targeting the IGF-1R activation and signaling becomes a potential strategy for osteosarcoma treatment.

Different strategies had been developed to target and inhibiting the IGF-1R receptor and associated downstream pathways, such as neutralizing antibodies, single strand

RNA and small molecule inhibitors [33]. An effective and commercially available synthetic tyrosine kinase inhibitor, Tyrphostin AG1204, was chosen to target the IGF-

1R in the 6 human osteosarcoma cell lines, which exhibited different levels of IGF-1R expression, and to demonstrate its effectiveness in modulating the chemosensitivity of

Doxorubicin (a commonly used chemotherapeutic agent in conventional osteosarcoma treatment [22,24]).

The mode of action of Tyrphostin AG1204 is blocking phosphorylation of IGF-1R intracellular domain. Anti-proliferation effects of Tyrphostin AG1024 in breast cancer cells and malignant mesothelioma cells [345,470,479] was also demonstrated in the different types of human osteosarcoma cell lines used in this investigation with a close 182

range of sensitivity. Results from this study showed that enhanced growth inhibition and increased sensitivity to the drugs in osteosarcoma cells was observed when

Tyrphostin AG1024 was used in combined with Doxorubicin, and this is similar to the investigation carried out in breast cancer cells [345,478]. The analysis by drug effectiveness computer program Calcusyn™ revealed a heterogenic effects of the combination treatment, which half of the tested osteosarcoma cell lines showed sub- additive effects (3/6) and the other half showed additive (1/6) to super-additive (2/6) effects. These results indicate that inhibition of IGF-1R signalling is an effective treatment in osteosarcoma and in some cases enhanced effects can be achieved in combine with chemotherapy.

The effectiveness analysis showed that increased drug sensitivity (additive and super- additive effect) to Tyrphostin AG1024 plus Doxorubicin was found in some human osteosarcoma cell lines, suggesting the combination therapy may be working in selective patients in clinics. Future studies following this discovery will focus on the mechanisms in depth in which how the sensitivity of osteosarcomas to the combination therapy is controlled as well as on establishment of diagnostic methods to identify sensitive osteosarcomas to the combination treatment.

IGF-1R is a tyrosine kinase receptor found to have unique characteristic and functions, and it has a high percentage homology to insulin receptor (IR) (~70%) [286]. Targeting the IGF-1R receptor with tyrosine kinase inhibitor AG1024 at a high dose could cross react with IR receptor and may affect blood sugar level [334,485]. As introduced by the drug company product sheet, AG1024 at current dose 10 µM or lower has no effect on sugar level on cell lines tested. In addition clinical trials using a specific IGF-1R inhibitor suggested that hyperglycemia was mild and easy to be controlled by oral drug administration [486]. Doxorubicin (also known as “Adriamycin”) is anthracycline antibiotics, would cause cardiomyopathy, transient electrocardiographic abnormalities, 183

emesis, alopecia, mucositis, myelosuppression, etc. during applications [2,22,24]. The enhanced drug effectiveness showed after the combination treatment in the present study suggested a beneficial situation in the clinical practice for drug dosage reduction in osteosarcoma patients, which reduced the side effects, such as blood sugar levels and cardiac toxicity that would be induced by the use of IGF-1R blocker and chemotherapeutic agent, respectively.

Primary resistance or development of resistance to chemotherapeutic drugs in osteosarcoma is a major factor that contributes to the result of an unsuccessful treatment. As IGF-1R signalling plays a role in development of resistance in chemotherapy for cancers [286,288,309], a Doxorubicin-resistant osteosarcoma cell line was established to examine the effectiveness of IGF-1R inhibition on Doxorubicin sensitivity. Results showed that growth inhibition of the Doxorubicin-resistant osteosarcoma cell line was achieved by treatment with IGF-1R inhibitor, Tyrphostin

AG1024 in the present or absence of Doxorubicin. IGF-1R inhibition alone or in combination with Doxorubicin was not only effective in Doxorubicin-sensitive osteosarcomas but also in Doxorubicin-resistant counterparts. The combination treatment had an additive effect with significant increment in drug sensitivity to

Doxorubicin. These results indicated that IGF-1R inhibition is one of effective methods to overcome Doxorubicin-resistance. This will have a clinical implication in treatment of

Doxorubicin-resistant osteosarcomas using IGF-1R blocker alone or together with

Doxorubicin. Most of osteosarcoma patients are children, who have a low tolerance to

Doxorubicin. Another advantage for the combination therapy would be maintaining treatment efficacy with Tyrphostin AG1024 plus the reduced dose of Doxorubicin. In this way, more children patients will be tolerable to the treatment and benefited.

The two osteosarcoma cell lines, which showed as the most super-additive (143B) and sub-additive (U-2 OS) effect from the combination treatment analysis, were selected for 184

mechanism studies to investigate the anti-tumour effect from the drug treatments.

Results showed that combination therapy using Tyrphostin AG1024 and Doxorubicin induced more apoptosis and cytotoxicity compared to single drug treatments. 143B was generally more sensitive to the different treatments than U2OS. In addition the combination drug treatment resulted in cell cycle arrest in one (U2OS) or two phases

(143B). These results can explain why the combination drug treatment can achieve super-additive and additive anti-proliferative effect in some sarcoma cell lines tested.

Understanding the mechanism in depth behind the drug sensitivity and identifying methods for selection of sensitive osteosarcomas will be helpful for introducing this combination therapy (IGF-1R blocker plus Doxorubicin) approach to clinic in the future.

Cell cycle arrest is a factor known to contribute in apoptosis. IGF-1R and its downstream signalling is able to promote G1-S transition through PI3K/Akt and

MEK/ERK pathways by increasing cyclin D1 and CDK4 gene expression, leading to retinoblastoma protein phosphorylation, releasing of the transcription factor E2F, and then synthesising of cyclin E [487,488]. In addition, IGF-1R down-regulates the transcription of p27Kip1 or alters its processing and nuclear localization leading to the increase of cyclin D1 level, which also exerts a positive effects on regulation of cell- cycle progression at the G1-S interface [489]. The regulation of other cyclins by IGF-1R may also interfere G2-M transition [293]. Doxorubicin, which caused intercalation of the double helix DNA, would inhibit DNA and RNA synthesis and have specific activity during the S phase of the cell cycle [478].

In the present study, analysis showed that Tyrphostin AG1024 treatment had induced both 143B and U-2 OS cells arrested in G1-phase, which was consistent with the discovery in a prostate cancer cell line (DU145) [490]. Doxorubicin treatment had only induced 143B cells arrested in G2/M-phase cycle, but not in U-2 OS cells. In contrast, the combination drug treatment induced both G1 and G2/M-phase arrests in 143B cells. 185

However, only G1-phase arrest was found in U-2 OS under the combination therapy.

Differences in doxorubicin sensitivity of both cell lines was the main factor contributed to the different cell cycle response after the combination treatments. The common doxorubicin resistance mechanisms but not IGF-1R associated G2M transition in tumor cells may possibly be involved in U-2 OS since under IGF1R inhibition doxorubicin failed to induce G2/M arrest in the cell line. The common doxorubicin resistance mechanism included: (i) overexpression of membrane-associated efflux pump P- glycoprotein mediating multiple drug resistance genes (MDR); (ii) altered expression of topoisomerase II and integrins; and (iii) changes in glutathione levels [491]. Furthermore, epigenetic regulation of p53 in the form of wild-type (U-2 OS) or mutant (143B) may also play a role and leads to the different cell cycle response after the combination treatments [492,493]. The complex process in which U2OS resists the effect of doxorubicin on G2/M cell cycle progression required further investigations.

The cell cycle result suggested that Tyrphostin AG1024 had reduced the mitogenicity of cell and avoid cell entering into S-phase of cell cycle and this effect had no interference to the G2/M blockage induced by Doxorubicin. Certainly, inhibition of IGF-

1R and chemotherapeutic agent appeared to act on different cell cycle checkpoints, thus targeting IGF-1R by Tyrphostin AG1024 may potentiate the effect of Doxorubicin.

Development of chemoresistant is a major problem in osteosarcoma treatment, detection of cell cycle of the tumor cells can be explored as a strategy in monitoring effectiveness, prediction of outcome or reference for making therapeutic decision in the treatment for osteosarcoma patient.

Clonogenic assay or colony formation assay is an alternative in vitro cell survival analysis used to determine the effectiveness of cytotoxic agents and also able to determine cell reproductive death after treatment with radiation treatment [390]. This analysis has been frequently used in cancer research to investigate the proliferation or 186

the reproduction capacity retained in the tumour cells after treatment. Results from the current study showed that a dose dependent clonogenic survival was found in osteosarcoma cell after treatment with Doxorubicin while a much minor effect was shown after treatment with Tyrphostin AG1024. The dose dependent clonogenic survival found in the combination treatment was resulted from the Doxorubicin. This results indicated that the combination therapy had more advantages than mono drug treatment in terms of maintain (e.g. colony inhibition and G2/M arrest) or improvement

(e.g. apoptosis and cytotoxicity) in therapeutic efficacy without addition of individual drug doses (no increment of side-effect).

Although the clonogenic survival in osteosarcoma cells after the combination treatment was not agreed with cytotoxicity and apoptosis analysis, it was not an uncommon event.

Conflicts of the clonogenic survival and apoptosis was also identified in thyroid carcinoma cells and epithelial tumour cells after treatments, and the mechanisms involved in Bcl-2 oncoprotein and p53 tumour suppressor protein was alleged playing a role in this situation [494,495]. The unsusceptible apoptosis effects of evaluated level of

Bcl-2 from p53 mutation and together with the length of the treatment time were contributed to the differences and the conflicts between the two types of assays [496].

Nonetheless, p53 mutation and dysregulation is commonly found in osteosarcoma [54,57], the discrepancy in this analysis is unavoidable in the future studies.

6.6. Conclusion

It is the first time this study showed that IGF-1R targeted therapy enhanced

Doxorubicin chemotherapy in osteosarcoma cells. The cytotoxic effects from the combination treatment with IGF-1R inhibitor and chemotherapeutic agent were enhanced, and the cell cycle analysis indicated that such enhancement was the 187

additive effects obtained from the two drugs. The combination treatment showed similar additive anti-proliferation effect at the Doxorubicin resistant human osteosarcoma cell line. Thus, inhibition of IGF-1R may be useful in clinic to conquer chemoresistance in the future 188

CHAPTER 7. CONCLUSION AND FUTURE 7 DIRECTIONS

7.1. Conclusion

Osteosarcoma is the most common primary malignant bone tumour with a lower occurrence compared to other human cancers and is found to be a clinically and molecularly heterogeneous group of malignancies. Conventional osteosarcoma is the prominent primary bone sarcoma in humans that is characterised by the production of osteoid and/or bone by tumour cells. Osteosarcoma has been regarded as a differentiation disease that any interruption along the osteoblast differentiation process will leads to the development of cancer. Although many factors are known to contribute in osteosarcomagenesis of or participate at different stage of osteoblast differentiation, our knowledge to osteosarcoma progression and molecular pathogenesis and interaction remains limited. Osteosarcoma investigation has been held back by the complexity of the genetic changes and the rarity of samples.

To improve the comprehensive understanding between osteosarcomagenesis and osteoblast differentiation, the gene profiles of the two human osteosarcoma cells with dissimilar differentiation status was investigated in the first part of this study by using

Affymetrix™ gene expression microarray. Our analysis have shown that genes of cell surface proteins and cadherins involved in cellular binding and adhesion, as well as genes in nervous system development and signalling are essential in characterizing the differences between the two human osteosarcoma cells. The !-adrenergic signalling system related ARRB1 (!-arrestin 1), and other nervous system related genes

CELSR1, GSTP1 and UNC5B, as well as the cell surface protein related genes 189

CADM1, CD24, CD97, CDH4, CELSR1, FAT3, GJB2, ITGA2, ITGA10, SCIN and

TSPAN2 were identified in this study for the first time to show their potential involvement in the disruption of osteoblast differentiation that may relate to the development of osteosarcoma.

Osteosarcoma treatment remains a challenging task in clinical oncology and no further major improvement was shown after the introduction of multi-regime chemotherapy and surgery. The low survival rate in metastatic disease, high rate of relapse and metastasis, and resistant to chemotherapy are the major problems in osteosarcoma patients. Thus, it is essential to identify new molecular targets and develop of new treatment approaches and modalities to improve the current clinical situation.

In the second part of this study, a usable orthotopic mouse model of human osteosarcoma growth and metastasis was established to demonstrate the independent relationship between activation of the IGF-1R signalling and malignant potential. The mouse model with consistent primary and spontaneous lung metastasis formed by intratibial injection of 143B, which is a K-ras transformed human osteosarcoma cell line, was established in this study and was verified by gross examination, diagnostic imaging and histochemistry. Further examination of the model with immunostaining technique for biological marker detection confirmed that the total protein expression of

IGF-1R and its downstream signalling factors, Akt, MEK1/2 and p38-MAP kinase at the primary site (tibia) and the metastasis site (lung), but only the activated form of MEK1/2 was found in both sites in the orthotopic mouse model of osteosarcoma. Thus, our results for the first time confirmed that the Ras/Raf/MEK/ERK pathway is activated in this mouse model of osteosarcoma.

The successful establishment of the mouse model by using the K-ras transformed

143B human osteosarcoma cell in this study has again confirmed the involvement of

Ras signalling in tumour development and metastasis. The discovery from this study 190

suggested that the IGF-1R/MEK signalling, in particular Ras/Raf/MEK/ERK pathway may play an important role in the osteosarcoma lung metastasis. The association between Ras/Raf/MEK/ERK signalling and metastatic potential is verified by using osteosarcoma cells. It is also the first time that inhibition of MEK/ERK function by its specific inhibitor U0126 significantly reduced in vitro invasive ability of osteosarcoma cell. This result suggested that MEK and ERK are also potential therapeutic targets for the treatment of metastatic osteosarcoma.

In the third part of this study, the effectiveness of IGF-1R inhibition in osteosarcoma was examined. Drug targeting IGF-1R alone or in combination with chemotherapy was used to treat 6 human osteosarcoma cell lines and 1 self-established Doxorubicin resistant human osteosarcoma cell line. In all 6 osteosarcoma cell lines, the IGF-1R inhibitor Tyrphostin AG1024 effectively inhibited proliferation at high drug concentration and enhanced anti-proliferative effect was found with drug combination treatment. Drug effectiveness analysis showed that additive and super-additive effects were achieved in

3 out of 6 osteosarcoma cell lines with drug combination treatment. It is the first time this study showed that IGF-1R targeted therapy enhanced Doxorubicin chemotherapy in osteosarcoma cells. Mechanism studies on the two selected osteosarcoma cells revealed that Tyrphostin AG1024 is cytotoxic (late apoptosis) to osteosarcoma cells and that cytotoxicity is resulted from IGF-1R inhibition induced apoptosis via perturbation of cell cycle. The cytotoxic effects from the combination treatment with

IGF-1R inhibitor and chemotherapeutic agent were enhanced, and the cell cycle analysis indicated that such enhancement was the additive effects obtained from the two drugs.

The Doxorubicin and Tyrphostin AG1024 combination treatment showed the similar additive anti-proliferation effect was also demonstrated at the Doxorubicin resistant human osteosarcoma cell line, which demonstrated the effectiveness in overcoming 191

the major problem of an unsuccessful treatment in osteosarcoma by the development of resistance to chemotherapeutic drugs. Inhibition of IGF-1R may be useful in clinic to conquer chemoresistance in the future.

The relationship between the IGF-1R signalling and malignant potential, as well as the effectiveness of IGF-1R inhibition in the treatment of native and chemoresistant human osteosarcoma cell lines verified the potential of targeting IGF-1R as a treatment option.

It is clear that IGF-1R inhibition has positive effect and able to enhance Doxorubicin chemosensitivity in growth inhibition of osteosarcoma cells through induction of apoptosis, cytotoxicity and G1-arrest and suggested that targeted therapy of IGF-1R would have a positive impact on the application in osteosarcoma treatments. In addition, the combined chemotherapeutic drug e.g. Doxorubicin has extra anti-tumour effects such as colony inhibition and G2/M-arrest, etc. The increased effectiveness of drug combination is a beneficial perspective in a clinical context that able to reduce the amount of individual drugs required for the patients to achieve the same treatment effects as well as reduce the side effects. 192

7.2. Future studies

Fully understanding the tumorigenesis of osteosarcoma is essential for eradication of the disease. Limitation of available technology and rarity of the disease have barricaded the progress. Thus, research of osteosarcoma is a lengthy and continuous process.

This study is the first to report that cell surface protein and nervous system related genes ARRB1, CADM1, CD24, CD97, CDH4, CELSR1, FAT3, GJB2, GSTP1, ITGA2,

ITGA10, SCIN, TSPAN2 and UNC5B may play a role in between osteoblast differentiation and osteosarcoma development. Individual genes based future investigations such as gene knock down or gene overexpression will reveal their potential links with the development of osteosarcoma.

Investigation and comparison of the gene profile from osteoblastic and non-osteoblastic osteosarcoma patient samples can help to advance and validate the results obtained from this study in relation to the disruption of osteoblast differentiation, which can subsequently improve the clinical diagnosis and treatment in osteosarcoma and bone defect diseases.

The IGF-1R/MEK signalling, in particular the Ras/Raf/MEK/ERK signalling pathway was shown to play an important role in the osteosarcoma lung metastasis and potentially mediating metastatic activity in the mouse model of human osteosarcoma.

However, the exact mechanism is not fully understood. Future studies will be concentrated on the mechanism relating to lung metastasis by osteosarcoma cells as well as the potential of MEK or ERK targeted therapy in the mouse model of human osteosarcoma. Furthermore, some significant genes identified from our studies, for example, IGF-1R and its signalling pathways will be detected in a cohort of clinical specimens to verify prospective candidates for the progression of osteosarcoma. 193

IGF-1R targeted therapy has been shown as an effective method in the treatment of cancers such as breast cancer, prostate cancer and Ewing’s sarcoma, but limited investigation in osteosarcoma. This study has shown that IGF-1R targeted therapy is effective in osteosarcoma cells and also in chemoresistant osteosarcoma cells. It enhanced the chemosensitivity by inducing apoptosis and cell cycle arrest. Further study will be focused on the effects of the combination treatment in vivo in animal model, so as to identify safety dose range for potential clinical trials of IGF-1R targeted therapy and/or chemotherapy in osteosarcoma patients.

Other future studies may also include the mechanism in depth behind the drug sensitivity, identifying methods for selection of sensitive osteosarcomas for the combination therapy, as well as p53 mutation and dysregulation in osteosarcoma. 194

CHAPTER 8. APPENDIXES 78

Appendix 1: List of materials used and solution recipes.

(A) List of chemicals, reagents and experimental systems

Brand Name Catalogue Number ABgene Absolute™ Blue QPCR SYBR® Green Mix Plus ROX AB-4166

Ambion (OH, USA) TRI Reagent AM9738

Amersham ECL Western Blotting Analysis System RPN2109

Biogenex Oligonucleotide Probe for Human Alu DNA PR-1001-01 Universal ISH DAB Kit SH-2019-06

Biorad 30% Acrylamide/Bis Solution (37.5:1) 161-0158 Agarose 161-3101 !-Mercaptoethanol 161-0710 DC protein assay kit II 500-0122 Ethidium bromide 161-0433 TEMED 161-0800

DAKO Antigen retrieval solution (pH 6.0), 10X S2369 EnVision+ System- HRP Labelled Polymer - Anti-Mouse K4000 - Anti-Rabbit K4002 Liquid DAB+ substrate chormogen system K3468

ICN Triton-X 100 194854

Ilford Hypam Fixer 1758285 Phenisol X-ray Developer 1757635

Invitrogen Life Tech. SuperScript® III First-Strand Synthesis System for RT- 18080-051 PCR

Orion Labs Pty. Ltd. Riodine™ - povidone-iodine

Scharlau Silver Nitrate PL0050

Sigma-Aldrich 2-methoxyethanol (EGME) 360503 Alizarin Red S A-5533 Aluminum sulphate A-0843 Ammonium hydroxide 338818 BCP (1-Bromo-3-chloropropane) B-9673 Boric acid B-7901 195

Brand Name Catalogue Number Sigma-Aldrich Calcium chloride C-5080 Chloroform C-2432 Crystal violet C-3886 Diethanolamine D-0681 Diethyl pyrocarbonate (DEPC) D-5758 Ethyl alcohol (Ethanol) E-7023 Ethylenediaminetetraacetic acid (EDTA) E-5134 EUKITT® quick-hardening mount medium for microscopy 03989 Fast red violet-LB salt F-1625 Glacial acetic acid solution 33209 Glycine G-7126 Hydrochloric acid H-1758 Hydrogen peroxide solution (30%) 18312 Isopropanol I-9516 Magnesium chloride M-2670 Napthol AS-MX phosphate N-4875 Nuclear fast red N-8002 p-nitrophenylphosphate S-9042 Potassium chloride P-9333 Protease inhibitor P-8340 Sodium acetate S-8750 Sodium acetate trihydrate S-7670 Sodium azide S-8032 Sodium bicarbonate S-5761 Sodium citrate tribasic C-0909 Sodium orthovanadate (phosphatase inhibitor) S-6508 Sodium thiosulfate S-1648 Sodium tartrate dibasic dihydrate S-4797 TRIS (hydroxymethyl)aminomethane 252859 Trypan blue T-6146 Tween-20 P-1379

Univar Formaldehyde (38%) 230 (Ajax Finechem) Sodium chloride 465 Sodium phosphate dibasic 391 196

(B) Preparation of solutions

Solution Name Recipe 0.1 M Tris-HCl Buffer (pH 9.8) Dissolve 3.03 g TRIS in Milli-Q water, adjust pH to 9.8 with HCl and make up to 250 mL with Milli-Q

0.5% Trypan blue: Dissolve 0.5 g Trypan Blue and 0.8 g NaCl in 100 mL of Milli-Q water, sterilized by 0.22 )m filter and store at 4oC. Warm to room temperature before usage

0.9% NaCl washing buffer Dissolve 9 g NaCl in 1 L of Milli-Q water

1% BSA/PBS Dissolve 1 g of BSA in 100 mL of PBS and make 1 mL aliquots before storing at – 20 oC

10% Formalin (3.7% buffered Mix 1 L formaldehyde with 1 L of 10X PBS and make up to 10 L formaldehyde solution) with Milli-Q water

Acidic Alcohol Mix 1 mL of HCl (36% v/v) with 999 mL of 70% ethanol

Alizarin Red S staining solution Dissolve 1.37 g of Alizarin Red S powder in Milli-Q water, adjust (40 mM) pH to 4.1 with NH4OH and make up to 100 mL with Milli-Q water

Aqueous silver nitrate solution Dissolve 1 g of silver nitrate in 100 mL of Milli-Q water (1%)

Crystal violet elution buffer Dissolve 10 g of sodium citrate in of Milli-Q water, adjust pH to 4.2 with HCl and make up to 200 mL with Milli-Q, mix with 1 volume of 100% ethanol before usage

Crystal violet staining solution Dissolve 0.5 g crystal violet powder in 20 mL methanol and make up to 100 mL with Milli-Q water

Decalcification buffer Mix 1 L formaldehyde with 1 L of 10X PBS and 1 L of formic acid, then make up to 10 L with Milli-Q water

Diethanolamine buffer solution Dissolve 50 mg MgCl2 and 100 mg sodium azide in 48.5 mL diethanolamine, then mix with Milli-Q water, adjust pH to 9.8 with HCl and make up to 500mL with Milli-Q

2+ 2+ Di-valent (Mg /Ca ) PBS Dissolve 147 mg of CaCl2 and 50 mg MgCl2 in 1 L PBS

Eosin Y solution Mix 200 mL of eosin Y (1% w/v solution) with 600 mL of 95% ethanol and 4 mL of glacial acetic acid

Harris’s haematoxylin solution Dissolve 2.5 g of haematoxylin in 25 mL absolute alcohol. Add 50 g potassium alum in 500 mL Milli-Q water and dissolve with heat applied. Mix the two solution and remove from heat source, followed by carefully adding1.25 mercuric oxide and 20 mL glacial acetic acid

Lysis buffer (RIPA) 10mM Tris-HCl pH7.4, 150mM NaCl, 5mM EDTA, 1% v/v Triton X- 100, 1% Sodium deoxycholate and 0.1% SDS

Nuclear fast red solution (0.1%) Dissolve 5 g aluminum sulfate in 100 mL and then dissolve 0.1 g of nuclear fast red powder with heating. Cool the solution to room temperature and sterilize with 0.22 µm syringe filter 197

Solution Name Recipe PBST 0.05% v/v: Mix 500 )L Tween-20 in 1 L of PBS, stored at room temperature

Phosphate buffered saline Dissolve 8 g NaCl, 0.2 g KCl, 1.15 g Na2HPO4 and 0.2 g KH2PO4 in (PBS): 1 L of Milli-Q water, stored at room temperature

Quenching solution Mix 1 mL of 30% H2O2 solution with 50 mL of methanol and make up to 100 mL with Milli-Q water

SDS PAGE running buffer Dissolve 30.3 g Tris-base, 144 g glycine and 10 g SDS in 1 L of (10x, for western blot): Milli-Q water

SDS PAGE sample buffer Dissolve 1 g SDS, 0.93 g DTT and 1.2 mg bromophenol blue, in the (5x, for western blot): mixture of 3 mL of 1.0 M Tris-Cl (pH 6.8) and 5 mL glycerol, then make up to 10 mL Milli-Q water. Aliquot and store at -20oC

Sodium thiosulfate solution (5%) Dissolve 5 g of sodium thiosulfate in 100 m L of Milli-Q water

Scott’s blue Dissolve 2 g of sodium bicarbonate and 10 g of magnesium sulphate (or 20g MgSO4•7H2O) in 1 L of Milli-Q water

TBE electrophoresis buffer Dissolve 10.8 g Tris-base and 5.5 g boric acid in 800 mL of Milli-Q water, mixed with 4 mL 0.5 M EDTA (pH 8.0) and make up to 1 L

TRAP staining buffer Dissolve 6.8 g sodium acetate trihydrate and 5.8 g sodium tartrate dibasic dehydrate in water, adjust pH to 5.0 with acetic acid and made up to 500 mL with Milli-Q water

TRAP staining solution Dissolve 5 mg Napthol AS-MX in 250 µL 2-methoxyethanol and mixed with 50 mL TRAP staining buffer, followed by addition of 30 mg fast red violet-LB salt

Transfer buffer (for western Dissolve 3.03 g Tris-base and 14.4 g glycine, add in 200 mL blot): methanol, followed by make up to 1 litre with Milli-Q water

Western blotting blocking buffer Dissolve 5 g of skim milk powder in PBS solution (5% w/v): 198

(C) List of cell culture reagents and consumables

Supplier Name Catalog Stock Conc. / Remark Number Becton All culturing vessels and pipettes Dickinson (BD)

Calbiochem Tyrphostin AG1024 121767 1 mg/µL (- 20 oC)

Gibco® Antibiotic-Antimycotic solution 15240-062 5 mL aliquot (- 20 oC) Dulbeco Phosphate buffered solution 21600-051 (DPBS) Foetal bovine serum (FBS) 10099-141 50 mL aliquot (- 20 oC) Penicillin / Streptomycin 15140-122 5 mL aliquot (- 20 oC) Roswell Park Memorial Institute 1640 31800-014 (RPMI1640) Media Trypsin-ethylenediaminetetraacetic 15400-054 5 mL aliquot (- 20 oC) acid (Trypsin-EDTA)

NUNC Cryovials 377267

Sigma !-glycerphosphate G-9891 Dexamethason D-4902 Dimethyl sulfoxide (DMSO) D-8418 Doxorubicin hydrochloride D-1515 HEPES H4034 L-asorbic acid A-8960 L-glutamine G-5736 200 mM (- 20 oC) Sodium bicarbonate S-5761 199

(D) List of antibodies

Supplier Name Catalog Number Remark Abnova Osterix (Transcription factor SP7) H00121340-M01 Mouse monoclonal IgG

Calbiochem Rabbit IgG N101

Cell Akt (C67E7) 4691 Rabbit monoclonal IgG Signalling Cleaved capase-3 (Asp 175) 9661 Rabbit polyclonal MEK 1/2 9122 Rabbit polyclonal p38 MAPK 9212 Rabbit polyclonal Phospho-Akt (Ser473) 9271 Rabbit polyclonal Phospho-MEK 1/2 (Ser217/221) 9121 Rabbit polyclonal Phospho-p38 MAPK 4631 Rabbit monoclonal IgG

DAKO Mouse IgG X0931

Santa Cruz Alkaline phosphatase SC-137215 Mouse monoclonal IgG Donkey anti-rabbit IgG-HRP SC-2317 Secondary Antibody Glyceraldehyde 3-phosphate SC-32233 Mouse monoclonal IgG dehydrogenase Goat anti-mouse IgG-HRP SC-2302 Secondary Antibody IGF-1R (1H7) SC-461 Mouse monoclonal IgG IGF-1R! (C-20) SC-713 Rabbit polyclonal IgG pTyr (PY99) SC-7020 Mouse monoclonal IgG Runt-related transcription factor 2 SC-10758 Rabbit polyclonal IgG

Sigma Actin-! A1978 Mouse monoclonal IgG 200

Appendix 2: List of 629 genes identified after functional annotation analysis

(A) The up-regulated 330 genes found in Saos-2 compared with U-2 OS cell line

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 216379_x_at 422.42 CD24 CD24 molecule Hs.694721 201387_s_at 182.79 UCHL1 ubiquitin carboxyl-terminal esterase L1 Hs.518731 (ubiquitin thiolesterase) 218559_s_at 153.12 MAFB v-maf musculoaponeurotic fibrosarcoma Hs.712609 oncogene homolog B (avian) 221730_at 141.09 COL5A2 collagen, type V, alpha 2 Hs.445827 205229_s_at 129.18 COCH coagulation factor C homolog, cochlin (Limulus Hs.21016 polyphemus) 206796_at 112.52 WISP1 WNT1 inducible signalling pathway protein 1 Hs.492974 209199_s_at 104.68 MEF2C myocyte enhancer factor 2C Hs.653394 236029_at 91.40 FAT3 FAT tumour suppressor homolog 3 Hs.98523 (Drosophila) 201744_s_at 87.96 LUM lumican Hs.406475 221841_s_at 80.62 KLF4 Kruppel-like factor 4 (gut) Hs.376206 236859_at 77.92 RUNX2 runt-related transcription factor 2 Hs.535845 228640_at 65.20 PCDH7 protocadherin 7 Hs.479439 213943_at 63.79 TWIST1 twist homolog 1 (Drosophila) Hs.66744 207147_at 57.45 DLX2 distal-less homeobox 2 Hs.419 213416_at 53.08 ITGA4 integrin, alpha 4 (antigen CD49D, alpha 4 Hs.694732 subunit of VLA-4 receptor) 209343_at 51.84 EFHD1 EF-hand domain family, member D1 Hs.516769 225639_at 51.57 SKAP2 src kinase associated phosphoprotein 2 Hs.200770 203637_s_at 51.00 MID1 midline 1 (Opitz/BBB syndrome) Hs.27695 203186_s_at 50.27 S100A4 S100 calcium binding protein A4 Hs.654444 207039_at 49.42 CDKN2A cyclin-dependent kinase inhibitor 2A Hs.512599 (melanoma, p16, inhibits CDK4) 206373_at 45.89 ZIC1 Zic family member 1 (odd-paired homolog, Hs.647962 Drosophila) 221541_at 45.62 CRISPLD2 cysteine-rich secretory protein LCCL domain Hs.513779 containing 2 227812_at 45.10 TNFRSF19 tumour necrosis factor receptor superfamily, Hs.149168 member 19 1552340_at 44.21 SP7 Sp7 transcription factor Hs.209402 203139_at 35.03 DAPK1 death-associated protein kinase 1 Hs.380277 213707_s_at 32.84 DLX5 distal-less homeobox 5 Hs.99348 214043_at 31.71 PTPRD protein tyrosine phosphatase, receptor type, D Hs.446083 1552365_at 31.41 SCIN scinderin Hs.655515 242940_x_at 30.02 DLX6 distal-less homeobox 6 Hs.249196 203498_at 29.57 RCAN2 regulator of calcineurin 2 Hs.440168 242138_at 29.05 DLX1 distal-less homeobox 1 Hs.407015 236028_at 28.52 IBSP integrin-binding sialoprotein Hs.518726 201

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 215783_s_at 28.28 ALPL alkaline phosphatase, liver/bone/kidney Hs.75431 226766_at 28.11 ROBO2 roundabout, axon guidance receptor, homolog Hs.13305 2 (Drosophila) 218934_s_at 27.50 HSPB7 heat shock 27kDa protein family, member 7 Hs.502612 (cardiovascular) 232054_at 26.46 PCDH20 protocadherin 20 Hs.391781 218469_at 25.97 GREM1 gremlin 1, cysteine knot superfamily, homolog Hs.40098 (Xenopus laevis) 238877_at 25.51 EYA4 eyes absent homolog 4 (Drosophila) Hs.596680 237056_at 25.34 INSC inscuteable homolog (Drosophila) Hs.591997 204602_at 24.34 DKK1 dickkopf homolog 1 (Xenopus laevis) Hs.40499 229354_at 22.99 AHRR /// aryl-hydrocarbon receptor repressor /// Hs.50823 PDCD6 programmed cell death 6 217428_s_at 22.31 COL10A1 collagen, type X, alpha 1 Hs.520339 204179_at 21.74 MB myoglobin Hs.517586 227236_at 21.55 TSPAN2 tetraspanin 2 Hs.310458 204465_s_at 21.46 INA internexin neuronal intermediate filament Hs.500916 protein, alpha 212489_at 21.09 COL5A1 collagen, type V, alpha 1 Hs.210283 212843_at 20.97 NCAM1 neural cell adhesion molecule 1 Hs.503878 223642_at 20.89 ZIC2 Zic family member 2 (odd-paired homolog, Hs.653700 Drosophila) 214930_at 20.63 SLITRK5 SLIT and NTRK-like family, member 5 Hs.591208 203231_s_at 20.04 ATXN1 ataxin 1 Hs.434961 229435_at 19.98 GLIS3 GLIS family zinc finger 3 Hs.162125 236313_at 19.33 CDKN2B cyclin-dependent kinase inhibitor 2B (p15, Hs.72901 inhibits CDK4) 213664_at 19.25 SLC1A1 solute carrier family 1 (neuronal/epithelial high Hs.444915 affinity glutamate transporter, system Xag), member 1 203603_s_at 19.11 ZEB2 zinc finger E-box binding homeobox 2 Hs.34871 203820_s_at 19.10 IGF2BP3 insulin-like growth factor 2 mRNA binding Hs.700696 protein 3 209539_at 18.61 ARHGEF6 Rac/Cdc42 guanine nucleotide exchange Hs.522795 factor (GEF) 6 210239_at 17.72 IRX5 iroquois homeobox 5 Hs.435730 229225_at 17.66 NRP2 neuropilin 2 Hs.471200 202311_s_at 17.41 COL1A1 collagen, type I, alpha 1 Hs.172928 206766_at 17.23 ITGA10 integrin, alpha 10 Hs.158237 227474_at 15.90 LOC654433 CDNA clone IMAGE:4826696 Hs.656660 205413_at 15.60 MPPED2 metallophosphoesterase domain containing 2 Hs.289795 212486_s_at 15.39 FYN FYN oncogene related to SRC, FGR, YES Hs.390567 229797_at 15.32 MCOLN3 mucolipin 3 Hs.535239 220921_at 15.31 SPANXB1/B2 SPANX family, member B1 /// SPANX family, Hs.711489 /F1 member B2 /// SPANX family, member F1 213844_at 15.21 HOXA5 homeobox A5 Hs.655218 230895_at 15.09 HAPLN1 hyaluronan and proteoglycan link protein 1 Hs.2799 202

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 227646_at 14.59 EBF1 early B-cell factor 1 Hs.573143 242128_at 14.47 OTX2 orthodenticle homeobox 2 Hs.288655 224862_at 13.99 GNAQ GTP-binding protein alpha q (GQ) Hs.269782 231863_at 13.85 ING3 inhibitor of growth family, member 3 Hs.489811 217728_at 13.69 S100A6 S100 calcium binding protein A6 Hs.275243 231579_s_at 13.45 TIMP2 TIMP metallopeptidase inhibitor 2 Hs.633514 218502_s_at 13.28 TRPS1 trichorhinophalangeal syndrome I Hs.657018 225664_at 12.83 COL12A1 collagen, type XII, alpha 1 Hs.101302 212636_at 12.81 QKI quaking homolog, KH domain RNA binding Hs.510324 (mouse) 206693_at 12.70 IL7 interleukin 7 Hs.591873 202668_at 12.59 EFNB2 ephrin-B2 Hs.149239 223278_at 12.27 GJB2 gap junction protein, beta 2, 26kDa Hs.524894 206140_at 12.02 LHX2 LIM homeobox 2 Hs.696425 230730_at 11.73 SGCD sarcoglycan, delta (35kDa dystrophin- Hs.387207 associated glycoprotein) 47550_at 11.20 LZTS1 leucine zipper, putative tumour suppressor 1 Hs.521432 33767_at 11.15 NEFH neurofilament, heavy polypeptide Hs.198760 227526_at 11.11 CDON Cdon homolog (mouse) Hs.38034 205907_s_at 10.79 OMD osteomodulin Hs.94070 204159_at 10.67 CDKN2C cyclin-dependent kinase inhibitor 2C (p18, Hs.716664 inhibits CDK4) 200884_at 10.51 CKB creatine kinase, brain Hs.173724 204198_s_at 10.50 RUNX3 runt-related transcription factor 3 Hs.170019 214608_s_at 10.47 EYA1 eyes absent homolog 1 (Drosophila) Hs.491997 209505_at 10.46 NR2F1 Chicken ovalbumin upstream promoter Hs.519445 transcription factor (COUP-TF) 227240_at 10.42 NGEF neuronal guanine nucleotide exchange factor Hs.97316 206460_at 10.35 AJAP1 adherens junctions associated protein 1 Hs.25924 205522_at 10.33 HOXD4 homeobox D4 Hs.591609 1555564_a_at 10.23 CFI complement factor I Hs.312485 206545_at 10.16 CD28 CD28 molecule Hs.591629 216005_at 10.09 TNC Tenascin Hs.143250 201427_s_at 10.08 SEPP1 selenoprotein P, plasma, 1 Hs.714733 206670_s_at 9.95 GAD1 glutamate decarboxylase 1 (brain, 67kDa) Hs.420036 204967_at 9.90 SHROOM2 shroom family member 2 Hs.567236 202149_at 9.51 NEDD9 neural precursor cell expressed, Hs.37982 developmentally down-regulated 9 221571_at 9.32 TRAF3 TNF receptor-associated factor 3 Hs.510528 214708_at 9.20 SNTB1 syntrophin, beta 1 (dystrophin-associated Hs.46701 protein A1, 59kDa, basic component 1) 224734_at 8.97 HMGB1 high-mobility group box 1 Hs.434102 204491_at 8.91 PDE4D phosphodiesterase 4D, cAMP-specific Hs.117545 (phosphodiesterase E3 dunce homolog, Drosophila) 204225_at 8.66 HDAC4 histone deacetylase 4 Hs.20516 203

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 228783_at 8.56 BVES blood vessel epicardial substance Hs.221660 223253_at 8.10 EPDR1 ependymin related protein 1 (zebrafish) Hs.563491 211518_s_at 8.01 BMP4 bone morphogenetic protein 4 Hs.68879 226016_at 7.77 CD47 CD47 molecule Hs.446414 229084_at 7.71 CNTN4 contactin 4 Hs.298705 201124_at 7.62 ITGB5 integrin, beta 5 Hs.536663 203638_s_at 7.41 FGFR2 fibroblast growth factor receptor 2 Hs.533683 204983_s_at 7.38 GPC4 glypican 4 Hs.58367 226899_at 7.37 UNC5B unc-5 homolog B (C. elegans) Hs.522997 207163_s_at 7.35 AKT1 v-akt murine thymoma viral oncogene Hs.525622 homolog 1 212778_at 7.35 PACS2 phosphofurin acidic cluster sorting protein 2 Hs.525626 209815_at 7.33 PTCH1 patched homolog 1 (Drosophila) Hs.494538 213093_at 7.21 PRKCA protein kinase C, alpha Hs.531704 220922_s_at 7.20 SPANXA1/A2 sperm protein associated with the nucleus, X- Hs.558533 /B1/B2/C/F1 linked, family member A1/A2/B1/B2/C/F1 202510_s_at 7.18 TNFAIP2 tumour necrosis factor, alpha-induced protein Hs.525607 2 209286_at 7.03 CDC42EP3 CDC42 effector protein (Rho GTPase binding) Hs.369574 3 203222_s_at 6.83 TLE1 transducin-like enhancer of split 1 (E(sp1) Hs.197320 homolog, Drosophila) 206924_at 6.81 IL11 interleukin 11 Hs.467304 205574_x_at 6.75 BMP1 bone morphogenetic protein 1 Hs.1274 203685_at 6.73 BCL2 B-cell CLL/lymphoma 2 Hs.150749 202481_at 6.73 DHRS3 dehydrogenase/reductase (SDR family) Hs.289347 member 3 210512_s_at 6.49 VEGFA vascular endothelial growth factor A Hs.73793 214651_s_at 6.40 HOXA9 homeobox A9 Hs.659350 204796_at 6.38 EML1 echinoderm microtubule associated protein Hs.12451 like 1 206103_at 6.22 RAC3 ras-related C3 botulinum toxin substrate 3 (rho Hs.45002 family, small GTP binding protein Rac3) 238075_at 6.18 CHEK1 Checkpoint kinase Chk1 (CHK1) Hs.24529 206117_at 6.15 TPM1 tropomyosin 1 (alpha) Hs.133892 212535_at 6.11 MEF2A myocyte enhancer factor 2A Hs.268675 201449_at 6.10 TIA1 TIA1 cytotoxic granule-associated RNA Hs.413123 binding protein 202959_at 6.08 MUT methylmalonyl Coenzyme A mutase Hs.485527 231793_s_at 6.07 CAMK2D calcium/calmodulin-dependent protein kinase Hs.144114 II delta 209031_at 6.06 CADM1 cell adhesion molecule 1 Hs.370510 203019_x_at 5.87 SSX2IP synovial sarcoma, X breakpoint 2 interacting Hs.22587 protein 226022_at 5.86 SASH1 SAM and SH3 domain containing 1 Hs.193133 212406_s_at 5.81 PCMTD2 protein-L-isoaspartate (D-aspartate) O- Hs.473317 methyltransferase domain containing 2 204

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 226492_at 5.80 SEMA6D sema domain, transmembrane domain (TM), Hs.511265 and cytoplasmic domain, (semaphorin) 6D 205110_s_at 5.78 FGF13 fibroblast growth factor 13 Hs.6540 213668_s_at 5.77 SOX4 SRY (sex determining region Y)-box 4 Hs.699195 204236_at 5.74 FLI1 Friend leukemia virus integration 1 Hs.504281 244680_at 5.74 GLRB glycine receptor, beta Hs.32973 229511_at 5.73 SMARCE1 SWI/SNF related, matrix associated, actin Hs.696086 dependent regulator of chromatin, subfamily e, member 1 41660_at 5.66 CELSR1 cadherin, EGF LAG seven-pass G-type Hs.252387 receptor 1 (flamingo homolog, Drosophila) 207149_at 5.63 CDH12 cadherin 12, type 2 (N-cadherin 2) Hs.113684 235057_at 5.59 ITCH itchy E3 ubiquitin protein ligase homolog Hs.632272 (mouse)/Atrophin-1 interacting protein 4 (AIP4) 211137_s_at 5.53 ATP2C1 ATPase, Ca++ transporting, type 2C, member Hs.584884 1 203485_at 5.45 RTN1 reticulon 1 Hs.368626 230226_s_at 5.44 JARID1A jumonji, AT rich interactive domain 1A Hs.76272 232481_s_at 5.44 SLITRK6 SLIT and NTRK-like family, member 6 Hs.525105 243483_at 5.37 TRPM8 transient receptor potential cation channel, Hs.366053 subfamily M, member 8 202984_s_at 5.36 BAG5 BCL2-associated athanogene 5 Hs.5443 203845_at 5.29 KAT2B K(lysine) acetyltransferase 2B Hs.533055 203817_at 5.27 GUCY1B3 guanylate cyclase 1, soluble, beta 3 Hs.77890 229273_at 5.26 SALL1 sal-like 1 (Drosophila) Hs.135787 212310_at 5.26 MIA3 melanoma inhibitory activity family, member 3 Hs.118474 209555_s_at 5.21 CD36 CD36 molecule (thrombospondin receptor) Hs.120949 207397_s_at 5.21 HOXD13 homeobox D13 Hs.152414 213139_at 5.21 SNAI2 snail homolog 2 (Drosophila) Hs.360174 223912_s_at 5.19 CLN8 ceroid-lipofuscinosis, neuronal 8 (epilepsy, Hs.127675 progressive with mental retardation) 209784_s_at 5.17 JAG2 jagged 2 Hs.433445 211022_s_at 5.14 ATRX alpha thalassemia/mental retardation Hs.533526 syndrome X-linked (RAD54 homolog, S. cerevisiae) 208838_at 5.13 CAND1 cullin-associated and neddylation-dissociated Hs.546407 1 226318_at 5.13 TBRG1 transforming growth factor beta regulator 1 Hs.436410 204897_at 5.11 PTGER4 prostaglandin E receptor 4 (subtype EP4) Hs.199248 202554_s_at 5.10 GSTM3 glutathione S-transferase mu 3 (brain) Hs.2006 213005_s_at 5.07 KANK1 KN motif and ankyrin repeat domains 1 Hs.306764 226045_at 5.01 FRS2 fibroblast growth factor receptor substrate 2 Hs.593446 1554251_at 4.98 HP1BP3 heterochromatin protein 1, binding protein 3 Hs.142442 206544_x_at 4.98 SMARCA2 SWI/SNF related, matrix associated, actin Hs.298990 dependent regulator of chromatin, subfamily a, member 2 204011_at 4.96 SPRY2 sprouty homolog 2 (Drosophila) Hs.18676 205

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 203071_at 4.95 SEMA3B sema domain, immunoglobulin domain (Ig), Hs.82222 short basic domain, secreted, (semaphorin) 3B 1569171_a_at 4.94 FXR1 Fragile X mental retardation protein 1 homolog Hs.478407 FXR1 210645_s_at 4.89 TTC3 tetratricopeptide repeat domain 3 Hs.368214 230047_at 4.85 FLJ32810 hypothetical protein FLJ32810 Hs.709625 209090_s_at 4.84 SH3GLB1 SH3-domain GRB2-like endophilin B1 Hs.136309 1552455_at 4.80 PRUNE2 prune homolog 2 (Drosophila) Hs.262857 212751_at 4.75 UBE2N ubiquitin-conjugating enzyme E2N (UBC13 Hs.524630 homolog, yeast) 229800_at 4.74 DCLK1 KIAA0369 gene Hs.507755 244128_x_at 4.71 GLIS1 GLIS family zinc finger 1 Hs.306691 212148_at 4.68 PBX1 pre-B-cell leukemia homeobox 1 Hs.654412 212387_at 4.66 TCF4 transcription factor 4 Hs.644653 219032_x_at 4.63 OPN3 opsin 3 Hs.409081 212037_at 4.62 PNN pinin, desmosome associated protein Hs.409965 235626_at 4.61 CAMK1D calcium/calmodulin-dependent protein kinase Hs.659517 ID 219532_at 4.60 ELOVL4 elongation of very long chain fatty acids Hs.101915 (FEN1/Elo2, SUR4/Elo3, yeast)-like 4 208920_at 4.59 SRI sorcin Hs.489040 222394_at 4.59 PDCD6IP programmed cell death 6 interacting protein Hs.475896 203037_s_at 4.57 MTSS1 metastasis suppressor 1 Hs.336994 200816_s_at 4.56 PAFAH1B1 platelet-activating factor acetylhydrolase, Hs.77318 isoform Ib, alpha subunit 45kDa 1561985_at 4.55 C14orf39 chromosome 14 open reading frame 39 Hs.335754 226112_at 4.54 SGCB sarcoglycan, beta (43kDa dystrophin- Hs.438953 associated glycoprotein) 202512_s_at 4.52 ATG5 ATG5 autophagy related 5 homolog (S. Hs.486063 cerevisiae) 204454_at 4.48 LDOC1 leucine zipper, down-regulated in cancer 1 Hs.45231 209888_s_at 4.48 MYL1 myosin, light chain 1, alkali; skeletal, fast Hs.187338 209082_s_at 4.48 COL18A1 collagen, type XVIII, alpha 1 Hs.517356 212418_at 4.48 ELF1 E74-like factor 1 (ets domain transcription Hs.135646 factor) 202146_at 4.47 IFRD1 interferon-related developmental regulator 1 Hs.7879 240084_at 4.46 CBX2 chromobox homolog 2 (Pc class homolog, Hs.368410 Drosophila) 1554411_at 4.46 CTNNB1 catenin (cadherin-associated protein), beta 1, Hs.476018 88kDa 225595_at 4.46 CREBZF CREB/ATF bZIP transcription factor Hs.535319 213704_at 4.46 RABGGTB Rab geranylgeranyltransferase, beta subunit Hs.78948 209033_s_at 4.44 DYRK1A dual-specificity tyrosine-(Y)-phosphorylation Hs.368240 regulated kinase 1A 227349_at 4.44 HELLS helicase, lymphoid-specific Hs.655830 224645_at 4.42 EIF4EBP2 eukaryotic translation initiation factor 4E Hs.695953 binding protein 2 206

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 225540_at 4.42 MAP2 microtubule-associated protein 2 Hs.368281 214163_at 4.42 HSPB11 heat shock protein family B (small), member Hs.525462 11 203502_at 4.42 BPGM 2,3-bisphosphoglycerate mutase Hs.198365 214277_at 4.42 COX11 COX11 homolog, cytochrome c oxidase Hs.591171 assembly protein (yeast) 225031_at 4.40 CHD6 chromodomain helicase DNA binding protein 6 Hs.716668 204379_s_at 4.36 FGFR3 fibroblast growth factor receptor 3 Hs.1420 225115_at 4.36 HIPK2 homeodomain interacting protein kinase 2 Hs.397465 205698_s_at 4.31 MAP2K6 mitogen-activated protein kinase kinase 6 Hs.463978 207480_s_at 4.27 MEIS2 Meis homeobox 2 Hs.510989 1554018_at 4.27 GPNMB glycoprotein (transmembrane) nmb Hs.190495 219427_at 4.26 FAT4 FAT tumour suppressor homolog 4 Hs.702217 (Drosophila) 238020_at 4.26 PSMC2 MSS1 mRNA for mammalian suppressor of Hs.437366 sgv1 210135_s_at 4.18 SHOX2 short stature homeobox 2 Hs.55967 239769_at 4.18 CDH11 Cadherin-11 Hs.116471 216493_s_at 4.17 IGF2BP3 /// insulin-like growth factor 2 mRNA binding Hs.700696 LOC645468 protein 3 /// similar to putative RNA binding protein KOC 231775_at 4.15 TNFRSF10A tumour necrosis factor receptor superfamily, Hs.591834 member 10a 223472_at 4.12 WHSC1 Wolf-Hirschhorn syndrome candidate 1 Hs.113876 226342_at 4.11 SPTBN1 spectrin, beta, non-erythrocytic 1 Hs.503178 225313_at 4.07 C20orf177 chromosome 20 open reading frame 177 Hs.504920 210302_s_at 4.06 MAB21L2 mab-21-like 2 (C. elegans) Hs.584852 223089_at 4.05 VEZT vezatin, adherens junctions transmembrane Hs.24135 protein 205280_at 4.05 GLRB glycine receptor, beta Hs.32973 204742_s_at 4.04 PDS5B PDS5, regulator of cohesion maintenance, Hs.716441 homolog B (S. cerevisiae) 215030_at 4.04 GRSF1 G-rich RNA sequence binding factor 1 Hs.309763 227276_at 4.02 PLXDC2 plexin domain containing 2 Hs.658134 215245_x_at 4.02 FMR1 fragile X mental retardation 1 Hs.103183 221031_s_at 4.00 APOLD1 apolipoprotein L domain containing 1 225527_at 3.99 CEBPG CCAAT/enhancer binding protein (C/EBP), Hs.429666 gamma 201667_at 3.97 GJA1 gap junction protein, alpha 1, 43kDa Hs.74471 228153_at 3.95 RNF144B ring finger protein 144B Hs.148741 200762_at 3.92 DPYSL2 dihydropyrimidinase-like 2 Hs.173381 203166_at 3.91 CFDP1 craniofacial development protein 1 Hs.461361 217838_s_at 3.89 EVL Enah/Vasp-like Hs.125867 205498_at 3.87 GHR growth hormone receptor Hs.125180 214724_at 3.86 DIXDC1 DIX domain containing 1 Hs.655626 230836_at 3.84 ST8SIA4 ST8 alpha-N-acetyl-neuraminide alpha-2,8- Hs.308628 sialyltransferase 4 207

Probe Set ID Fold Gene Gene Title UniGene ID change Symbol 1567107_s_at 3.82 TPM4 tropomyosin 4 Hs.631618 210510_s_at 3.80 NRP1 neuropilin 1 Hs.131704 203489_at 3.74 SIVA1 SIVA1, apoptosis-inducing factor Hs.112058 221583_s_at 3.73 KCNMA1 potassium large conductance calcium- Hs.144795 activated channel, subfamily M, alpha member 1 229173_at 3.73 KIAA1715 KIAA1715 Hs.209561 204753_s_at 3.72 HLF hepatic leukemia factor Hs.196952 219351_at 3.70 SEDLP /// spondyloepiphyseal dysplasia, late, Hs.446620 TRAPPC2 /// pseudogene /// trafficking protein particle ZNF547 complex 2 /// zinc finger protein 547 204963_at 3.67 SSPN sarcospan (K-ras oncogene-associated gene) Hs.183428 205066_s_at 3.64 ENPP1 ectonucleotide Hs.527295 pyrophosphatase/phosphodiesterase 1 206837_at 3.63 ALX1 ALX homeobox 1 Hs.41683 225669_at 3.60 IFNAR1 interferon (alpha, beta and omega) receptor 1 Hs.529400 226275_at 3.60 MXD1 MAX dimerization protein 1 Hs.468908 218718_at 3.58 PDGFC platelet derived growth factor C Hs.570855 211864_s_at 3.57 MYOF myoferlin Hs.602086 222853_at 3.57 FLRT3 fibronectin leucine rich transmembrane protein Hs.41296 3 210657_s_at 3.57 Sep-04 septin 4 Hs.287518 212308_at 3.56 CLASP2 cytoplasmic linker associated protein 2 Hs.108614 202971_s_at 3.56 DYRK2 dual-specificity tyrosine-(Y)-phosphorylation Hs.173135 regulated kinase 2 214954_at 3.55 SUSD5 sushi domain containing 5 Hs.196647 231807_at 3.55 KIAA1217 KIAA1217 Hs.445885 206737_at 3.55 WNT11 wingless-type MMTV integration site family, Hs.108219 member 11 203689_s_at 3.54 FMR1 fragile X mental retardation 1 Hs.103183 201969_at 3.52 NASP nuclear autoantigenic sperm protein (histone- Hs.319334 binding) 222696_at 3.51 AXIN2 axin 2 Hs.156527 222456_s_at 3.50 LIMA1 LIM domain and actin binding 1 Hs.525419 200609_s_at 3.49 WDR1 WD repeat domain 1 Hs.128548 220102_at 3.49 FOXL2 forkhead box L2 Hs.289292 215199_at 3.47 CALD1 caldesmon 1 Hs.490203 228438_at 3.47 LOC100132891 PREDICTED: Homo sapiens hypothetical Hs.710132 protein LOC100132891 (LOC100132891), mRNA 226508_at 3.46 PHC3 polyhomeotic homolog 3 (Drosophila) Hs.529592 231778_at 3.46 DLX3 distal-less homeobox 3 Hs.134194 225022_at 3.45 GOPC golgi associated PDZ and coiled-coil motif Hs.191539 containing 209717_at 3.44 EVI5 ecotropic viral integration site 5 Hs.656836 218660_at 3.44 DYSF dysferlin, limb girdle muscular dystrophy 2B Hs.252180 (autosomal recessive) 236924_at 3.43 GLMN glomulin, FKBP associated protein Hs.666341 208

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 213792_s_at 3.41 INSR insulin receptor Hs.465744 232020_at 3.40 SMURF2 SMAD specific E3 ubiquitin protein ligase 2 Hs.705442 1556037_s_at 3.40 HHIP hedgehog interacting protein Hs.507991 220432_s_at 3.39 CYP39A1 cytochrome P450, family 39, subfamily A, Hs.387367 polypeptide 1 206163_at 3.38 MAB21L1 mab-21-like 1 (C. elegans) Hs.584776 236356_at 3.37 NDUFS1 NADH dehydrogenase (ubiquinone) Fe-S Hs.471207 protein 1, 75kDa (NADH-coenzyme Q reductase) 201944_at 3.36 HEXB hexosaminidase B (beta polypeptide) Hs.69293 225144_at 3.35 BMPR2 bone morphogenetic protein receptor, type II Hs.471119 (serine/threonine kinase) 224967_at 3.35 UGCG UDP-glucose ceramide glucosyltransferase Hs.593014 217367_s_at 3.35 ZHX3 zinc fingers and homeoboxes 3 Hs.380133 226808_at 3.32 ZNF862 zinc finger protein 862 Hs.301277 209124_at 3.30 MYD88 myeloid differentiation primary response Hs.82116 gene (88) 213720_s_at 3.30 SMARCA4 SWI/SNF related, matrix associated, actin Hs.327527 dependent regulator of chromatin, subfamily a, member 4 230100_x_at 3.30 PAK1 p21 protein (Cdc42/Rac)-activated kinase 1 Hs.435714 206213_at 3.29 WNT10B wingless-type MMTV integration site family, Hs.91985 member 10B 219737_s_at 3.29 PCDH9 protocadherin 9 Hs.654709 206847_s_at 3.29 HOXA7 homeobox A7 Hs.660918 230056_at 3.27 BPTF bromodomain PHD finger transcription factor Hs.444200 212249_at 3.27 PIK3R1 phosphoinositide-3-kinase, regulatory Hs.132225 subunit 1 (alpha) 201579_at 3.25 FAT1 FAT tumour suppressor homolog 1 Hs.481371 (Drosophila) 217732_s_at 3.25 ITM2B integral membrane protein 2B Hs.643683 223339_at 3.25 ATPIF1 ATPase inhibitory factor 1 Hs.590908 206706_at 3.23 NTF3 neurotrophin 3 Hs.99171 203726_s_at 3.23 LAMA3 laminin, alpha 3 Hs.436367 235980_at 3.22 PIK3CA HBV preS1-transactivated protein 3 binding Hs.553498 protein 1 (PS1TP3BP1) 202950_at 3.22 CRYZ crystallin, zeta (quinone reductase) Hs.83114 206271_at 3.22 TLR3 toll-like receptor 3 Hs.657724 217849_s_at 3.21 CDC42BPB CDC42 binding protein kinase beta (DMPK- Hs.654634 like) 225532_at 3.20 CABLES1 Cdk5 and Abl enzyme substrate 1 Hs.11108 225093_at 3.18 UTRN utrophin Hs.133135 225780_at 3.17 RSC1A1 regulatory solute carrier protein, family 1, Hs.145049 member 1 218911_at 3.17 YEATS4 YEATS domain containing 4 Hs.4029 209210_s_at 3.16 FERMT2 fermitin family homolog 2 (Drosophila) Hs.509343 206724_at 3.16 CBX4 chromobox homolog 4 (Pc class homolog, Hs.714363 Drosophila) 209

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 204557_s_at 3.14 DZIP1 DAZ interacting protein 1 Hs.656580 203499_at 3.12 EPHA2 EPH receptor A2 Hs.171596 229394_s_at 3.11 GRLF1 glucocorticoid receptor DNA binding factor 1 Hs.509447 204653_at 3.11 TFAP2A transcription factor AP-2 alpha (activating Hs.519880 enhancer binding protein 2 alpha) 220559_at 3.11 EN1 engrailed homeobox 1 Hs.271977 212116_at 3.11 TRIM27 tripartite motif-containing 27 Hs.440382 205367_at 3.10 SH2B2 SH2B adaptor protein 2 Hs.489448 228184_at 3.08 DISP1 dispatched homolog 1 (Drosophila) Hs.528817 203723_at 3.07 ITPKB inositol 1,4,5-trisphosphate 3-kinase B Hs.528087 212248_at 3.05 MTDH Metadherin, mRNA (cDNA clone Hs.377155 IMAGE:4124124) /// Hs.594085 225601_at 3.03 HMGB3 high-mobility group box 3 Hs.19114 202822_at 3.03 LPP LIM domain containing preferred Hs.444362 translocation partner in lipoma 206940_s_at 3.03 LOC100131317 similar to hCG1781072 /// POU class 4 Hs.654522 /// POU4F1 homeobox 1 225250_at 3.02 STIM2 stromal interaction molecule 2 Hs.135763 203989_x_at 3.01 F2R coagulation factor II (thrombin) receptor Hs.482562 214121_x_at 3.01 PDLIM7 PDZ and LIM domain 7 (enigma) Hs.533040 218476_at 3.00 POMT1 protein-O-mannosyltransferase 1 Hs.522449 210

(B) The down-regulated 299 genes found in Saos-2 compared with U-2 OS cell line

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 214079_at 519.07 DHRS2 dehydrogenase/reductase (SDR family) Hs.272499 member 2 200799_at 350.06 HSPA1A /// heat shock 70kDa protein 1A /// heat shock Hs.274402 HSPA1B 70kDa protein 1B 33323_r_at 174.58 SFN stratifin Hs.523718 201841_s_at 164.60 HSPB1 heat shock 27kDa protein 1 Hs.520973 39729_at 152.05 PRDX2 peroxiredoxin 2 Hs.432121 208861_s_at 110.15 ATRX alpha thalassemia/mental retardation syndrome Hs.533526 X-linked (RAD54 homolog, S. cerevisiae) 213558_at 107.67 PCLO piccolo (presynaptic cytomatrix protein) Hs.12376 1569110_x_at 93.71 LOC728613 programmed cell death 6 pseudogene Hs.379186 210538_s_at 79.08 BIRC3 baculoviral IAP repeat-containing 3 Hs.127799 205828_at 78.80 MMP3 matrix metallopeptidase 3 (stromelysin 1, Hs.375129 progelatinase) 211964_at 76.18 COL4A2 collagen, type IV, alpha 2 Hs.508716 209550_at 59.63 NDN necdin homolog (mouse) Hs.50130 209242_at 55.73 PEG3 paternally expressed 3 Hs.716512 203132_at 53.52 RB1 retinoblastoma 1 Hs.408528 203828_s_at 49.81 IL32 interleukin 32 Hs.943 202833_s_at 49.76 SERPINA1 serpin peptidase inhibitor, clade A (alpha-1 Hs.525557 antiproteinase, antitrypsin), member 1 204971_at 49.72 CSTA cystatin A (stefin A) Hs.518198 227404_s_at 48.21 EGR1 Putative zinc finger protein mRNA, 3' flank Hs.708393 209278_s_at 44.33 TFPI2 tissue factor pathway inhibitor 2 Hs.438231 226482_s_at 41.95 hCG_20857 /// thiosulfate sulfurtransferase KAT, putative /// RP11- KAT protein 544M22.4 225612_s_at 40.70 B3GNT5 UDP-GlcNAc:betaGal beta-1,3-N- Hs.208267 acetylglucosaminyltransferase 5 240259_at 38.81 FLRT2 CDNA FLJ51243 complete cds, highly similar to Hs.533710 Leucine-rich repeat transmembrane protein FLRT2 precursor 222891_s_at 37.83 BCL11A B-cell CLL/lymphoma 11A (zinc finger protein) Hs.370549 227048_at 35.24 LAMA1 laminin, alpha 1 Hs.270364 210550_s_at 34.95 RASGRF1 Ras protein-specific guanine nucleotide- Hs.591111 releasing factor 1 1570516_s_at 31.66 OR51B5 olfactory receptor, family 51, subfamily B, Hs.690458 member 5 204337_at 31.46 RGS4 regulator of G-protein signalling 4 Hs.386726 231798_at 31.21 NOG noggin Hs.248201 204733_at 31.17 KLK6 kallikrein-related peptidase 6 Hs.79361 209306_s_at 28.50 SWAP70 SWAP-70 protein Hs.153026 1557094_at 26.08 LOC653110 hypothetical LOC653110 202267_at 25.61 LAMC2 laminin, gamma 2 Hs.591484 211

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 209301_at 24.33 CA2 carbonic anhydrase II Hs.155097 214023_x_at 24.24 TUBB2B tubulin, beta 2B Hs.300701 206018_at 23.10 FOXG1 forkhead box G1 Hs.695962 209291_at 22.77 ID4 inhibitor of DNA binding 4, dominant negative Hs.519601 helix-loop-helix protein 213201_s_at 22.53 TNNT1 troponin T type 1 (skeletal, slow) Hs.631558 205463_s_at 21.70 PDGFA platelet-derived growth factor alpha polypeptide Hs.535898 207717_s_at 21.57 PKP2 plakophilin 2 Hs.164384 36711_at 20.41 MAFF v-maf musculoaponeurotic fibrosarcoma Hs.517617 oncogene homolog F (avian) 206295_at 20.14 IL18 interleukin 18 (interferon-gamma-inducing Hs.83077 factor) 201860_s_at 19.94 PLAT plasminogen activator, tissue Hs.491582 232027_at 19.63 SYNE1 Spectrin repeat containing, nuclear envelope 1, Hs.12967 mRNA (cDNA clone IMAGE:4830497) 206079_at 19.05 CHML choroideremia-like (Rab escort protein 2) Hs.654545 227314_at 18.78 ITGA2 integrin, alpha 2 (CD49B, alpha 2 subunit of Hs.482077 VLA-2 receptor) 207536_s_at 17.68 TNFRSF9 tumour necrosis factor receptor superfamily, Hs.654459 member 9 206343_s_at 17.46 NRG1 neuregulin 1 Hs.453951 206023_at 17.43 NMU neuromedin U Hs.418367 232226_at 16.22 LRRC4C leucine rich repeat containing 4C Hs.135736 205431_s_at 16.16 BMP5 bone morphogenetic protein 5 Hs.296648 202626_s_at 15.68 LYN v-yes-1 Yamaguchi sarcoma viral related Hs.699154 oncogene homolog 205681_at 15.55 BCL2A1 BCL2-related protein A1 Hs.227817 205206_at 15.51 KAL1 Kallmann syndrome 1 sequence Hs.521869 228882_at 15.20 TUB tubby homolog (mouse) Hs.568986 201288_at 15.00 ARHGDIB Rho GDP dissociation inhibitor (GDI) beta Hs.504877 206084_at 14.72 PTPRR protein tyrosine phosphatase, receptor type, R Hs.506076 206067_s_at 14.43 WT1 Wilms tumour 1 Hs.591980 205780_at 14.13 BIK BCL2-interacting killer (apoptosis-inducing) Hs.475055 218084_x_at 13.99 FXYD5 FXYD domain containing ion transport regulator Hs.333418 5 206969_at 13.98 KRT34 keratin 34 Hs.296942 229724_at 13.87 GABRB3 gamma-aminobutyric acid (GABA) A receptor, Hs.302352 beta 3 234469_at 13.65 OR51B4 olfactory receptor, family 51, subfamily B, Hs.679499 member 4 203650_at 13.13 PROCR protein C receptor, endothelial (EPCR) Hs.647450 205418_at 12.45 FES feline sarcoma oncogene Hs.7636 222062_at 12.34 IL27RA interleukin 27 receptor, alpha Hs.132781 235683_at 12.30 SESN3 sestrin 3 Hs.659934 200637_s_at 12.09 PTPRF protein tyrosine phosphatase, receptor type, F Hs.272062 204044_at 12.01 QPRT quinolinate phosphoribosyltransferase Hs.513484 213865_at 11.75 DCBLD2 discoidin, CUB and LCCL domain containing 2 Hs.203691 212

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 230788_at 10.90 GCNT2 glucosaminyl (N-acetyl) transferase 2, I- Hs.519884 branching enzyme (I blood group) 221581_s_at 10.57 LAT2 linker for activation of T cells family, member 2 Hs.647049 227758_at 10.49 RERG RAS-like, estrogen-regulated, growth inhibitor Hs.199487 218950_at 10.30 ARAP3 ArfGAP with RhoGAP domain, ankyrin repeat Hs.25277 and PH domain 3 209604_s_at 10.16 GATA3 GATA binding protein 3 Hs.524134 235467_s_at 9.82 KCNC4 potassium voltage-gated channel, Shaw-related Hs.153521 subfamily, member 4 218678_at 9.82 NES nestin Hs.527971 206290_s_at 9.81 RGS7 regulator of G-protein signalling 7 Hs.655739 222549_at 9.75 CLDN1 claudin 1 Hs.439060 214157_at 9.64 GNAS GNAS complex locus Hs.125898 238127_at 9.64 FLJ41484 hypothetical LOC650669 Hs.715832 205114_s_at 9.40 CCL3 /// chemokine (C-C motif) ligand 3/ligand 3-like Hs.514107 CCL3L1 /// 1/ligand 3-like 3 /// similar to C-C motif CCL3L3 /// chemokine 3-like 1 precursor (Small-inducible LOC728830 cytokine A3-like 1) (Tonsillar lymphocyte LD78 beta protein) (LD78-beta(1-70)) (G0/G1 switch regulatory protein 19-2) (G0S19-2 protein) (PAT 464.2) 205157_s_at 9.00 KRT17 keratin 17 Hs.2785 208510_s_at 8.99 PPARG peroxisome proliferator-activated receptor Hs.162646 gamma 209270_at 8.98 LAMB3 laminin, beta 3 Hs.497636 223392_s_at 8.96 TSHZ3 teashirt zinc finger homeobox 3 Hs.278436 201596_x_at 8.90 KRT18 keratin 18 Hs.406013 208378_x_at 8.73 FGF5 fibroblast growth factor 5 Hs.37055 243541_at 8.73 IL31RA interleukin 31 receptor A Hs.55378 1552721_a_at 8.47 FGF1 fibroblast growth factor 1 (acidic) Hs.483635 224374_s_at 8.15 EMILIN2 elastin microfibril interfacer 2 Hs.532815 239443_at 7.92 PCDHB6 protocadherin beta 6 Hs.283085 228176_at 7.91 S1PR3 sphingosine-1-phosphate receptor 3 Hs.585118 202644_s_at 7.77 TNFAIP3 tumour necrosis factor, alpha-induced protein 3 Hs.211600 209706_at 7.77 NKX3-1 NK3 homeobox 1 Hs.55999 238633_at 7.76 EPC1 EPC1/ASXL2a fusion protein Hs.167805 222838_at 7.72 SLAMF7 SLAM family member 7 Hs.517265 202910_s_at 7.72 CD97 CD97 molecule Hs.466039 217109_at 7.65 MUC4 mucin 4, cell surface associated Hs.369646 209459_s_at 7.63 ABAT 4-aminobutyrate aminotransferase Hs.336768 223629_at 7.51 PCDHB5 protocadherin beta 5 Hs.119693 204057_at 7.44 IRF8 interferon regulatory factor 8 Hs.137427 218829_s_at 7.44 CHD7 chromodomain helicase DNA binding protein 7 Hs.20395 218309_at 7.43 CAMK2N1 calcium/calmodulin-dependent protein kinase II Hs.197922 inhibitor 1 228494_at 7.39 PPP1R9A protein phosphatase 1, regulatory (inhibitor) Hs.21816 subunit 9A 213

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 244565_at 7.14 HMX2 H6 family homeobox 2 Hs.444756 223000_s_at 7.10 F11R F11 receptor Hs.517293 201236_s_at 7.09 BTG2 BTG family, member 2 Hs.519162 215785_s_at 7.05 CYFIP2 cytoplasmic FMR1 interacting protein 2 Hs.519702 203510_at 7.05 MET met proto-oncogene (hepatocyte growth factor Hs.132966 receptor) 1557078_at 7.04 SLFN5 schlafen family member 5 Hs.709347 201042_at 7.01 TGM2 transglutaminase 2 (C polypeptide, protein- Hs.517033 glutamine-gamma-glutamyltransferase) 218796_at 7.00 FERMT1 fermitin family homolog 1 (Drosophila) Hs.472054 205992_s_at 7.00 IL15 interleukin 15 Hs.654378 227198_at 6.92 AFF3 AF4/FMR2 family, member 3 Hs.444414 213843_x_at 6.92 SLC6A8 solute carrier family 6 (neurotransmitter Hs.540696 transporter, creatine), member 8 205637_s_at 6.88 SH3GL3 SH3-domain GRB2-like 3 Hs.270055 222590_s_at 6.81 NLK nemo-like kinase Hs.208759 220446_s_at 6.81 CHST4 carbohydrate (N-acetylglucosamine 6-O) Hs.251383 sulfotransferase 4 203821_at 6.80 HBEGF heparin-binding EGF-like growth factor Hs.799 219257_s_at 6.76 SPHK1 sphingosine kinase 1 Hs.68061 202284_s_at 6.72 CDKN1A cyclin-dependent kinase inhibitor 1A (p21, Cip1) Hs.370771 203065_s_at 6.62 CAV1 caveolin 1, caveolae protein, 22kDa Hs.74034 218541_s_at 6.38 C8orf4 chromosome 8 open reading frame 4 Hs.591849 205659_at 6.38 HDAC9 histone deacetylase 9 Hs.196054 1553613_s_at 6.36 FOXC1 forkhead box C1 Hs.348883 206915_at 6.34 NKX2-2 NK2 homeobox 2 Hs.516922 205153_s_at 6.21 CD40 CD40 molecule, TNF receptor superfamily Hs.472860 member 5 219199_at 6.18 AFF4 AF4/FMR2 family, member 4 Hs.519313 238432_at 5.95 FLJ35776 hypothetical LOC649446 Hs.19872 216841_s_at 5.87 SOD2 superoxide dismutase 2, mitochondrial Hs.487046 1554168_a_at 5.84 SH3KBP1 SH3-domain kinase binding protein 1 Hs.595766 1561197_at 5.84 LOC442028 hypothetical LOC442028 Hs.611545 206857_s_at 5.80 FKBP1B FK506 binding protein 1B, 12.6 kDa Hs.709461 212096_s_at 5.78 MTUS1 mitochondrial tumour suppressor 1 Hs.7946 210038_at 5.78 PRKCQ protein kinase C, theta Hs.498570 213603_s_at 5.78 RAC2 ras-related C3 botulinum toxin substrate 2 (rho Hs.517601 family, small GTP binding protein Rac2) 204421_s_at 5.69 FGF2 fibroblast growth factor 2 (basic) Hs.284244 202157_s_at 5.69 CUGBP2 CUG triplet repeat, RNA binding protein 2 Hs.309288 209035_at 5.65 MDK midkine (neurite growth-promoting factor 2) Hs.82045 203936_s_at 5.59 MMP9 matrix metallopeptidase 9 (gelatinase B, 92kDa Hs.297413 gelatinase, 92kDa type IV collagenase) 201951_at 5.57 ALCAM activated leukocyte cell adhesion molecule Hs.591293 201688_s_at 5.56 TPD52 tumour protein D52 Hs.368433 214

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 211343_s_at 5.50 COL13A1 collagen, type XIII, alpha 1 Hs.695934 206170_at 5.46 ADRB2 adrenergic, beta-2-, receptor, surface Hs.591251 243000_at 5.45 CDK6 cyclin-dependent kinase 6 Hs.119882 227870_at 5.45 IGDCC4 immunoglobulin superfamily, DCC subclass, Hs.591101 member 4 209676_at 5.42 TFPI tissue factor pathway inhibitor (lipoprotein- Hs.516578 associated coagulation inhibitor) 226435_at 5.42 PAPLN papilin, proteoglycan-like sulfated glycoprotein Hs.655583 228181_at 5.31 SLC30A1 solute carrier family 30 (zinc transporter), Hs.519469 member 1 202037_s_at 5.28 SFRP1 secreted frizzled-related protein 1 Hs.713546 205579_at 5.27 HRH1 histamine receptor H1 Hs.1570 241994_at 5.23 XDH xanthine dehydrogenase Hs.250 211177_s_at 5.19 TXNRD2 thioredoxin reductase 2 Hs.443430 227339_at 5.19 RGMB RGM domain family, member B Hs.526902 203148_s_at 5.17 TRIM14 tripartite motif-containing 14 Hs.575631 213135_at 5.13 TIAM1 T-cell lymphoma invasion and metastasis 1 Hs.517228 213620_s_at 5.10 ICAM2 intercellular adhesion molecule 2 Hs.431460 238677_at 5.10 WDR36 WD repeat domain 36 Hs.533237 203786_s_at 5.01 TPD52L1 tumour protein D52-like 1 Hs.591347 205627_at 5.00 CDA cytidine deaminase Hs.466910 200736_s_at 4.91 GPX1 glutathione peroxidase 1 Hs.76686 200894_s_at 4.89 FKBP4 FK506 binding protein 4, 59kDa Hs.524183 203608_at 4.89 ALDH5A1 aldehyde dehydrogenase 5 family, member A1 Hs.371723 201578_at 4.78 PODXL podocalyxin-like Hs.715710 218332_at 4.78 BEX1 brain expressed, X-linked 1 Hs.334370 222557_at 4.78 STMN3 stathmin-like 3 Hs.639609 207180_s_at 4.78 HTATIP2 HIV-1 Tat interactive protein 2, 30kDa Hs.90753 231725_at 4.73 PCDHB2 protocadherin beta 2 Hs.533023 227923_at 4.72 SHANK3 SH3 and multiple ankyrin repeat domains 3 Hs.149035 213825_at 4.71 OLIG2 oligodendrocyte lineage transcription factor 2 Hs.176977 226029_at 4.71 VANGL2 vang-like 2 (van gogh, Drosophila) Hs.99477 207535_s_at 4.69 NFKB2 nuclear factor of kappa light polypeptide gene Hs.73090 enhancer in B-cells 2 (p49/p100) 204781_s_at 4.61 FAS Fas (TNF receptor superfamily, member 6) Hs.244139 222912_at 4.55 ARRB1 arrestin, beta 1 Hs.503284 203394_s_at 4.53 HES1 hairy and enhancer of split 1, (Drosophila) Hs.250666 207233_s_at 4.52 MITF microphthalmia-associated transcription factor Hs.166017 44783_s_at 4.51 HEY1 hairy/enhancer-of-split related with YRPW motif Hs.234434 1 221558_s_at 4.47 LEF1 lymphoid enhancer-binding factor 1 Hs.715742 203074_at 4.47 ANXA8 /// annexin A8 /// annexin A8-like 1 /// annexin A8- Hs.705389 ANXA8L1 /// like 2 ANXA8L2 205935_at 4.45 FOXF1 forkhead box F1 Hs.155591 215

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 205291_at 4.44 IL2RB interleukin 2 receptor, beta Hs.474787 1555351_s_at 4.44 PPHLN1 periphilin 1 Hs.444157 1557458_s_at 4.42 SHB Src homology 2 domain containing adaptor Hs.521482 protein B 214913_at 4.41 ADAMTS3 ADAM metallopeptidase with thrombospondin Hs.590919 type 1 motif, 3 209765_at 4.40 ADAM19 ADAM metallopeptidase domain 19 (meltrin Hs.483944 beta) 206777_s_at 4.36 CRYBB2 /// crystallin, beta B2 /// crystallin, beta B2 Hs.571835 CRYBB2P1 pseudogene 1 208358_s_at 4.36 UGT8 UDP glycosyltransferase 8 Hs.144197 227948_at 4.35 FGD4 FYVE, RhoGEF and PH domain containing 4 Hs.117835 207111_at 4.28 EMR1 egf-like module containing, mucin-like, hormone Hs.2375 receptor-like 1 54037_at 4.28 HPS4 Hermansky-Pudlak syndrome 4 Hs.474436 228863_at 4.26 PCDH17 protocadherin 17 Hs.106511 201656_at 4.24 ITGA6 integrin, alpha 6 Hs.133397 225293_at 4.24 COL27A1 collagen, type XXVII, alpha 1 Hs.494892 201474_s_at 4.21 ITGA3 integrin, alpha 3 (antigen CD49C, alpha 3 Hs.265829 subunit of VLA-3 receptor) 200824_at 4.20 GSTP1 glutathione S-transferase pi 1 Hs.523836 227828_s_at 4.17 FAM176A family with sequence similarity 176, member A Hs.302346 206483_at 4.09 LRRC6 leucine rich repeat containing 6 Hs.591865 206028_s_at 4.07 MERTK c-mer proto-oncogene tyrosine kinase Hs.306178 1555764_s_at 4.06 TIMM10 translocase of inner mitochondrial membrane Hs.235750 10 homolog (yeast) 221933_at 4.05 NLGN4X neuroligin 4, X-linked Hs.21107 203152_at 4.03 MRPL40 mitochondrial ribosomal protein L40 Hs.431307 223435_s_at 4.02 PCDHA1/A10/A protocadherin alpha Hs.199343 11/A12/A13/A2/ 1/10/11/12/13/2/3/4/5/6/7/8/9/alpha subfamily A3/A4/A5/A6/A C, 1/alpha subfamily C, 2 7/A8/A9/AC1/A C2 229975_at 4.02 BMPR1B bone morphogenetic protein receptor, type IB Hs.598475 205266_at 4.01 LIF leukemia inhibitory factor (cholinergic Hs.2250 differentiation factor) 37966_at 4.01 PARVB parvin, beta Hs.475074 227718_at 4.00 PURB purine-rich element binding protein B Hs.349150 210102_at 3.99 VWA5A von Willebrand factor A domain containing 5A Hs.152944 204880_at 3.98 MGMT O-6-methylguanine-DNA methyltransferase Hs.501522 1555137_a_at 3.96 FGD6 FYVE, RhoGEF and PH domain containing 6 Hs.506381 207446_at 3.93 TLR6 toll-like receptor 6 Hs.662185 202345_s_at 3.93 FABP5 /// fatty acid binding protein 5 (psoriasis- Hs.632112 FABP5L2 /// associated) /// fatty acid binding protein 5-like 2 FABP5L7 /// fatty acid binding protein 5-like 7 216

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 215812_s_at 3.93 LOC653562 /// hypothetical LOC653562 /// solute carrier family Hs.540696 SLC6A10P /// 6 (neurotransmitter transporter, creatine), SLC6A8 member 10 (pseudogene) /// solute carrier family 6 (neurotransmitter transporter, creatine), member 8 223349_s_at 3.93 BOK BCL2-related ovarian killer Hs.293753 213623_at 3.92 KIF3A kinesin family member 3A Hs.43670 202986_at 3.92 ARNT2 aryl-hydrocarbon receptor nuclear translocator Hs.459070 2 1554576_a_at 3.92 ETV4 ets variant 4 Hs.434059 205891_at 3.89 ADORA2B adenosine A2b receptor Hs.167046 203659_s_at 3.88 TRIM13 tripartite motif-containing 13 Hs.436922 210151_s_at 3.88 DYRK3 dual-specificity tyrosine-(Y)-phosphorylation Hs.164267 regulated kinase 3 204540_at 3.87 EEF1A2 eukaryotic translation elongation factor 1 alpha Hs.433839 2 202759_s_at 3.87 AKAP2 /// A kinase (PRKA) anchor protein 2 /// Hs.591908 PALM2 /// paralemmin 2 /// PALM2-AKAP2 readthrough PALM2-AKAP2 transcript 208966_x_at 3.81 IFI16 interferon, gamma-inducible protein 16 Hs.380250 212445_s_at 3.72 NEDD4L neural precursor cell expressed, Hs.185677 developmentally down-regulated 4-like 203313_s_at 3.70 TGIF1 TGFB-induced factor homeobox 1 Hs.373550 202033_s_at 3.68 RB1CC1 RB1-inducible coiled-coil 1 Hs.196102 222802_at 3.68 EDN1 endothelin 1 Hs.511899 205205_at 3.68 RELB v-rel reticuloendotheliosis viral oncogene Hs.654402 homolog B 205257_s_at 3.68 AMPH amphiphysin Hs.592182 209803_s_at 3.67 PHLDA2 pleckstrin homology-like domain, family A, Hs.154036 member 2 234486_at 3.65 OR51B2 olfactory receptor, family 51, subfamily B, Hs.680163 member 2 205709_s_at 3.64 CDS1 CDP-diacylglycerol synthase (phosphatidate Hs.654899 cytidylyltransferase) 1 235165_at 3.64 PARD6B par-6 partitioning defective 6 homolog beta (C. Hs.589848 elegans) 204388_s_at 3.63 MAOA monoamine oxidase A Hs.183109 209939_x_at 3.62 CFLAR CASP8 and FADD-like apoptosis regulator Hs.390736 203413_at 3.62 NELL2 NEL-like 2 (chicken) Hs.505326 1564856_s_at 3.61 LOC727924 /// hypothetical LOC727924 /// olfactory receptor, Hs.525666 OR4N4 family 4, subfamily N, member 4 206038_s_at 3.59 NR2C2 nuclear receptor subfamily 2, group C, member Hs.555973 2 212285_s_at 3.59 AGRN agrin Hs.273330 210312_s_at 3.55 IFT20 intraflagellar transport 20 homolog Hs.705431 (Chlamydomonas) 202628_s_at 3.54 SERPINE1 serpin peptidase inhibitor, clade E (nexin, Hs.414795 plasminogen activator inhibitor type 1), member 1 205586_x_at 3.53 VGF VGF nerve growth factor inducible Hs.587325 217

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 223796_at 3.53 CNTNAP3 /// contactin associated protein-like 3 /// contactin Hs.604441 CNTNAP3B /// associated protein-like 3B /// similar to cell LOC643827 /// recognition molecule CASPR3 /// hypothetical LOC653355 /// LOC653355 /// similar to cell recognition RP11-138L21.1 molecule CASPR3 203695_s_at 3.51 DFNA5 deafness, autosomal dominant 5 Hs.520708 229400_at 3.51 HOXD10 homeobox D10 Hs.123070 208950_s_at 3.49 ALDH7A1 aldehyde dehydrogenase 7 family, member A1 Hs.483239 208460_at 3.48 GJC1 gap junction protein, gamma 1, 45kDa Hs.532593 204584_at 3.48 L1CAM L1 cell adhesion molecule Hs.522818 203409_at 3.46 DDB2 damage-specific DNA binding protein 2, 48kDa Hs.700338 201161_s_at 3.45 CSDA cold shock domain protein A Hs.221889 1558163_at 3.45 PEX13 peroxisomal biogenesis factor 13 Hs.161377 213419_at 3.43 APBB2 amyloid beta (A4) precursor protein-binding, Hs.479602 family B, member 2 215073_s_at 3.43 NR2F2 nuclear receptor subfamily 2, group F, member Hs.701977 2 204319_s_at 3.41 RGS10 regulator of G-protein signalling 10 Hs.501200 210426_x_at 3.41 RORA RAR-related orphan receptor A Hs.695914 211965_at 3.37 ZFP36L1 zinc finger protein 36, C3H type-like 1 Hs.85155 209239_at 3.37 NFKB1 nuclear factor of kappa light polypeptide gene Hs.654408 enhancer in B-cells 1 226206_at 3.36 MAFK v-maf musculoaponeurotic fibrosarcoma Hs.520612 oncogene homolog K (avian) 203105_s_at 3.36 DNM1L dynamin 1-like Hs.556296 218145_at 3.35 TRIB3 tribbles homolog 3 (Drosophila) Hs.516826 202875_s_at 3.34 PBX2 pre-B-cell leukemia homeobox 2 Hs.509545 230163_at 3.34 LOC143381 CDNA FLJ31546 fis, clone NT2RI2000974 Hs.388347 41037_at 3.33 TEAD4 TEA domain family member 4 Hs.94865 1557905_s_at 3.33 CD44 CD44 molecule (Indian blood group) Hs.502328 203921_at 3.32 CHST2 carbohydrate (N-acetylglucosamine-6-O) Hs.8786 sulfotransferase 2 204126_s_at 3.30 CDC45L CDC45 cell division cycle 45-like (S. cerevisiae) Hs.474217 223907_s_at 3.29 PINX1 PIN2-interacting protein 1 Hs.490991 220512_at 3.29 DLC1 deleted in liver cancer 1 Hs.134296 202508_s_at 3.29 SNAP25 synaptosomal-associated protein, 25kDa Hs.167317 217941_s_at 3.29 ERBB2IP erbb2 interacting protein Hs.591774 229103_at 3.27 WNT3 wingless-type MMTV integration site family, Hs.445884 member 3 205603_s_at 3.27 DIAPH2 diaphanous homolog 2 (Drosophila) Hs.226483 235745_at 3.26 ERN1 endoplasmic reticulum to nucleus signalling 1 Hs.700027 202760_s_at 3.26 PALM2-AKAP2 PALM2-AKAP2 readthrough transcript 205696_s_at 3.26 GFRA1 GDNF family receptor alpha 1 Hs.591913 36936_at 3.25 TSTA3 tissue specific transplantation antigen P35B Hs.404119 210236_at 3.22 PPFIA1 protein tyrosine phosphatase, receptor type, f Hs.530749 polypeptide (PTPRF), interacting protein (liprin), alpha 1 218

Probe Set ID Fold Gene Symbol Gene Title UniGene ID change 210405_x_at 3.22 TNFRSF10B tumour necrosis factor receptor superfamily, Hs.521456 member 10b 200790_at 3.22 ODC1 ornithine decarboxylase 1 Hs.467701 213698_at 3.22 LOC10013063 similar to ZMYM6 protein /// zinc finger, MYM- Hs.533986 3 /// ZMYM6 type 6 235833_at 3.21 PPAT Amidophosphoribosyltransferase Hs.331420 206866_at 3.19 CDH4 cadherin 4, type 1, R-cadherin (retinal) Hs.473231 226120_at 3.17 TTC8 tetratricopeptide repeat domain 8 Hs.303055 228762_at 3.17 LFNG LFNG O-fucosylpeptide 3-beta-N- Hs.159142 acetylglucosaminyltransferase 202751_at 3.16 TFIP11 tuftelin interacting protein 11 Hs.20225 210609_s_at 3.16 TP53I3 tumour protein p53 inducible protein 3 Hs.50649 205599_at 3.15 TRAF1 TNF receptor-associated factor 1 Hs.531251 202890_at 3.15 MAP7 microtubule-associated protein 7 Hs.486548 201207_at 3.13 TNFAIP1 tumour necrosis factor, alpha-induced protein 1 Hs.76090 (endothelial) 207390_s_at 3.12 SMTN smoothelin Hs.149098 219825_at 3.12 CYP26B1 cytochrome P450, family 26, subfamily B, Hs.91546 polypeptide 1 219688_at 3.06 BBS7 Bardet-Biedl syndrome 7 Hs.591694 218847_at 3.06 IGF2BP2 insulin-like growth factor 2 mRNA binding Hs.35354 protein 2 204905_s_at 3.05 EEF1E1 eukaryotic translation elongation factor 1 Hs.716509 epsilon 1 222351_at 3.05 PPP2R1B protein phosphatase 2 (formerly 2A), regulatory Hs.584790 subunit A, beta isoform 202431_s_at 3.05 MYC v-myc myelocytomatosis viral oncogene Hs.202453 homolog (avian) 221319_at 3.04 PCDHB8 protocadherin beta 8 Hs.287793 228201_at 3.02 ARL13B ADP-ribosylation factor-like 13B Hs.533086 222830_at 3.01 GRHL1 grainyhead-like 1 (Drosophila) Hs.418493 219969_at 3.01 CXorf15 chromosome X open reading frame 15 Hs.555961 219

Appendix 3: The unsupervised hierarchical clustering analysis of gene expression in Saos-2 and U-2 OS compared with five normal human cells*. 220 221

*The gene expression profiles of Saos-2 and U-2 OS from this study was compared with the gene expression profiles of five human cells obtained from “Gene Expression Omnibus” (National Centre for Biotechnology Information – “NCBI”, http://www.ncbi.nlm.nih.gov/geo/) with accession number GSE9451 in the unsupervised hierarchical clustering analysis by GeneSpring™ GX 10.0.2 software. The ten clusters identified in this analysis were denoted with roman numerals from I to X. 222 CHAPTER9 9. REFERENCES

8 : 5 @ 5.QQJ$8%I:J6VJQ$`:` Q VQ:`HQI:IQRVC11 .]QJ :JVQ%IV : :11J I1HV7HC1J1H:C`VCV0:JHV:JR:]]C1H:G1C1 7`Q`R`%$ V 1J$8'8:JHV` V 8C1J8(JHQC85 R ^ _ 8 V``:`125:CIV`1J1 83R=%0:J :JRJVQ:R=%0:J HQIG1J: 1QJH.VIQ .V`:]7`Q`Q VQ$VJ1H:`HQI:8%``8 (]1J8(JHQC8 5R ^ _ 8 2]:J$CV`788QJVG1QCQ$7:JR].71QCQ$779I]C1H: 1QJ`Q`JQ0VCQ VQGC: 1HQ VQ:`HQI: `V: IVJ 8 VR87]Q .V V  5 R ^ _ 8 :7RVJ85Q:J$88 VQ:`HQI:78:1H2H1VJHV:JRC1J1H:C9I]C1H: 1QJ8(` .Q]8C1J8:Q` .I8 5R ^ _ 8 1`:GVCCQ <5 `Q11 52:0:$V 238 9J V`J: 1QJ:C Q VQ:`HQI: 1JH1RVJHV ]: V`J 1J H.1CR`VJ :JR :RQCVHVJ 5I1RRCV:$V:JRVCRV`C7]V`QJ8 J 8'8:JHV` 5 R^ _ 8 1`:GVCCQ<5 `Q11 52:0:$V238 VQ:`HQI:1JH1RVJHV:JR%`010:C`: V``QI R8:JHV` 5  R ^ _ 8 :$1QC1 5 81:1J 5 V`V% : 5 V  :C8QQ`]`Q$JQ1Q VQ:`HQI:7JV1 .V`:]V% 1H:]]`Q:H.8#QJV :``Q1%`:J ]C:J 852 R^ _ 8 3JRV`QJ8 ``VH 10JVQ``:R1Q .V`:]7`Q`QV Q:`HQI: .: `V]QJR QH.VIQ .V`:]78:7QC1J8 '`QH8 5 R ^_ 8 :H.:@7)5 @:H.V02952QCQ07V0)5V :C8)VQ:R=%0:J H.VIQ .V`:]7:JRCQH:C`:R1Q .V`:]7`Q`.1$.R $`:RVQ VQ:`HQI:`Q`V6 `VI1 1V8:7QC1J8'`QH8 5 R ^_ 8 V<:JV7 5:`@<57QCRGV`$295V :C8 :R1Q .V`:]7`Q`CQH:CHQJ `QCQ`Q VQ:`HQI:8 J 8'8 :R1: 8(JHQC8 #1QC8'.7 8 5 R ^ _ 8 :7Q C1J1H IVR1H:C 1J`Q`I: 1QJ :JR QQC  `Q` .V:C .7 C101J$ R :7QC1J1H8HQI8 :7Q Q%JR: 1QJ `Q` VR1H:C R%H: 1QJ :JR VV:`H. ^ _   %JV   `H1 VR   :J a6 30:1C:GCV ``QI7 . ]7LL1118I:7QHC1J1H8HQIL.V:C .LIVR1H:CL9 8 8 8%VH@V` 5 7VG.:`R  5VGV` C8 ( VQ :`HQI:7 01:$JQ 1 5 %`V: IVJ  :JR '`Q$JQ 1 8 <1RR7 2.`10V` 2:`HQI: 9J1 1: 10V  VG`%:`7   `H1 VR   :J a6 30:1C:GCV ``QI7 . ]7LL:`HQI:.VC]8Q`$LQ VQ:`HQI:8. IC8 8 Q.VJ7 388 #QJV7 QJ0VJ 1QJ:C Q VQ :`HQI:8 3 C: Q` 7VJV 1H :JR 7 Q$VJV 1H 1J JHQCQ$7 :JR :VI: QCQ$7  3]`1C   `H1 VR   :J a6 30:1C:GCV ``QI7 . ]7LL: C:$VJV 1HQJHQCQ$78Q`$L %IQ`LQJ0 VQ9 8. IC8 8 2H.:=Q11H< 5 1 QCQ$1H:C %7]1J$ Q` #QJV %%IQ`  ^Q`CR V:C.(`$:J1<: 1QJ J V`J: 1QJ:C 1 QCQ$1H:C C: 1`1H: 1QJQ`%%IQ` _5<882QG1J5 R1 Q`8 52]`1J$V`RV`C:$77V`I:J8  8 CCV1J 521V$:C 78  VQ:`HQI:7 3J: QI1H :JR 1 QCQ$1H :`1:J 8 I8'8 C1J8 ': .QC8  5 R  ^ _  8 %.V:: C1J1H:C '`:H 1HV 5%1RVC1JV  1J(JHQCQ$7 #QJV :JHV` R V` 1QJ 8 ^ _:: 1QJ:C QI]`V.VJ10V:JHV`:V 1Q`@5JH8Q` :.1J$ QJ5 VJJ7C0:J1:8. ]7LL1118JHHJ8Q`$8  8 JJV@1J$52]:J1V`2257QQRI:J3837 VI`Q` .V%`$1H:C :$1J$Q`I%H%CQ@VCV :C:`HQI:8C1J8 (` .Q]8 5 R^ _ 223

 8 2H.1:` <58VCC<581J$.:I5V :C5VR8$.V RI1J% V]VR1: `1HHQJ%C 8Q%`VR8 5<1]]1JHQ  1CC1:I1C@1J7.1C:RVC].1:5238 R 8  8 VRV`I:J 5 8V`J .:C 5 1CGV` 5 :] 8 .V %C 1R1H1]C1J:`7 :J:$VIVJ  Q`  VQ:`HQI:8 %``8 `V: 8(] 1QJ(JHQC85 R ^ _ 8 :J1HV@57: QJ15 QVJ .:C95V :C8 :JR 1I:$1J$1J .VCQH:C :$1J$Q`]`1I:`7I:C1$J:J  I%H%CQ@VCV :CJVQ]C:I7 V]Q` Q` .V :R1QCQ$71:$JQ 1HJHQCQ$77`Q%]8:R1QCQ$75 R  ^ _ 8 %`].73 89I:$1J$GQJV %IQ`1J .V 8:JHV` 5 R ^ _ 8 1 1$ 5 81H@CV 5 `1VG:  5 V  :C8  VQ:`HQI:7 3 %C 1R1H1]C1J:`7 3]]`Q:H. Q 1:$JQ1 :JR `V: IVJ 8"I8#:I8$.71H1:J 5R^_ 8 <:IQ%`V%65 `1H.V 5.1]Q75 V  :C8 VHVJ :R0:JHV1J .VI:J:$VIVJ Q`QV Q:`HQI::JR `Q` .HQI1J$ .V`:]V% 1H `: V$1V8&6]V` V08 "J 1H:JHV`.V`8 5 R ^ _ 8 :``CV581VC:H@228%``VJ  `: V$1VQ`H.VIQ .V`:]71JQ VQ:`HQI:8)J 8(` .Q]85 R ^ _  8 :`%C:JR: 735 VJRV`QJ 5 Q.JQJ 35

8 3`:@15H.1R:35C1I%`: 5V :C89J0QC0VIVJ Q` .V`V 1JQGC: QI:$VJV1J]`1I:`7Q VQ:`HQI::JR Q .V`GQJV:JRQ`  1%V %IQ`8C1J8(` .Q]8 5 R ^ _ 8 8VCH.1 35 VVHV 35 8VJ@Q 35 V  :C8

8 `V`C 5:$JV` 83R73RQ%GCVRVR$VR1Q`R1J %IQ`1$VJV18:: 8V08:JHV`5 R ^_ 8 :J$ 5 <1:J$ <5 2H.VCC:JRV` C5 :$JV` 57`1$Q`1:R1 3 8 HR`Q 1JR%HVR Q V:`HQI: `Q`I: 1QJ 1J `:J$VJ1H I1HV7 HQQ]V`: 101 7 11 . HR=%J :JR .V `QCV Q` VJRQ$VJQ% HR`Q8 :JHV`V8 5 R   ^ _ 8 % .V` 5 CQI1 Q1@1 5 2H.%GV`   5:$JV` 8 HR`Q V6]`V1QJ 1JR%HV GQJV %IQ` 1J `:J$VJ1H I1HV8(JHQ$VJV5 R ^ _ 8 `:JH.135:C

8 :1 `: 35 QGV`  5 V1JGV`$ 3757V`:R  8

8 3@: %@: 5:R: 5CQ@:15V :C8 `G8V6]`V1QJ1HQ``VC: VR11 .1JH`V:VR%`010:CQ`]: 1VJ 11 . Q VQ:`HQI:8:JHV` 5 R^_ 8 C1C]: `1H@2 57V11J$V`C 5C1J$ 25 V  :C8 C1J1HQ]: .QCQ$1H :J:C71 Q`  RLJV% 1II%JQV6]`V1QJ :IQJ$0:`1Q%.1 QCQ$1H%G 7]V:JR$`:RVQ`Q VQ:`HQI:8/QR8$: .QC85 R ^_ 8 .Q% 5 :JR:CC <5 8`Q .I:J 3 5 V  :C8 V`RLJV% V6]`V1QJ 1J Q VQ:`HQI: 1JH`V:V `1@ Q` C%J$ IV : :1:JRH:JGV:QH1: VR11 .$VJV:I]C1`1H: 1QJ8'8$VR1: `8:VI: QC8(JHQC8 5 R^_ 8 .QI: 75 71Q`R:JQ 5 2:JRV` 5 81V`I:JJ 258:@V` <8 3GVJHV Q`  LJV% $VJV V6]`V1QJ 1J Q VQ:`HQI::JR@VCV :C 11J$:`HQI:8C1J8:JHV`V8 5 R ^_ 8 QJ$25 V``V`5 V:% 5V :C8V V6]`V1QJ:JR:`HQI: %IQ`1$VJ1H1 78:JHV`V8 5  R   ^ _ 8 V``:H1J1 5 1 VJ

 8 Q Q05:IVC5 10:`R7 5V :C83I]C1`1H: 1QJQ`3]`1I:V^ 9_1J.%I:JQ VQ:`HQI:8 2VJV.`QIQQIV:JHV` 5 R ^ _  8 C: Q.5C: Q.8 21$J:C1J$: .1:7:JR2 VIVCC21$J:C1J$V 1Q`@8C1J8:JHV`V85R  ^ _  8 <%Q 5.VJ 5VJ$<5V :C8J 1$J:C1J$:JR.%I:JR1V:V71.: :`V V. .V`:]V% 1H1I]C1H: 1QJ8 3:G8)J0V 8 5 R^ _  8 7C:395C:`VJ 7789J010Q:J:C71Q`J 1$J:C1J$1JGQJV8&JRQH`1JQCQ$7 5 R ^ _  8 8QR1JV5CQII828J 21$J:C1J$:JR VQGC: Q$VJV18V08&JRQH`8/V :G811Q`R8 5R ^ _ 8 7C: 395C:`VJ 7 78 :JQJ1H:C J  1$J:C1J$ 1J Q VQGC:  1 `V_%1`VR `Q` Q VQHC:  R1``V`VJ 1: 1QJ8 "JJ8:88"H:R80H18 5 R^ _ 8 :` I:JJ83J H:JQJQ`H.V `: 1J$Q VQGC: Q$VJV18`VJRVCC+1QC8 5 R ^ _ 8 :7RQJ 5V7`%]359.1@:1:35V :C87 Q]C:I1H:JRLQ`J%HCV:`:HH%I%C: 1QJQ` .VGV :RH:J VJ1J ]`Q V1J1:``V_%VJ V0VJ 1J.%I:JQ VQ:`HQI:8)J 8'8:JHV`5 R^_ 8 91:7:C5$:1:5C%`QR:5V :C87 Q]C:I1H:JRLQ`J%HCV:` :1J1J$Q`GV :RH: VJ1J1:QH1: VR11 . C%J$IV : :18C1J8&6]8/V : :15  R  ^_ 8 91:QC517Q.15:1:75V :C8`V_%VJ GV :RH: VJ1J:GJQ`I:C1 1V1JGQJV:JRQ` R 1%V %IQ`8']J8 '8:JHV`V8 5 R ^ _  8 Q:J$85C%GQ 5V:CV7 5V :C8 6]`V1QJQ`<<`VHV] Q`R`VC: VR]`Q V1J ^<  _1:JQ0VCI:`@V` `Q`R1V:V]`Q$`V1QJ1J.1$.R$`:RVQ VQ:`HQI:8)J 8'8:JHV` 5 R^_  8 Q:J$85C%GQ 5V:CV7 5V :C81H@@Q]`1J.1G1 1J0:1QJ:JRIQ 1C1 7Q`2:QRQ VQ:`HQI:HVCC G7IQR%C: 1J$ .VJ RGV :RH: VJ1J]: .1:78:JHV`V8 5 R  ^_  8 QJ$5`1$.  52.:7 8%I:J VCQIV`:V:JR1 `V$%C: 1QJ8/1H`QG1QC8/QC8+1QC8V08 5 R  ^_  8 7`:J$V`5`1$.  52.:7 8 VCQIV`:V1JH:JHV`:JR:$1J$8`1 8V08(JHQC8:VI: QC85 R ^_  8 :J5Q1JR`V 58VJ.: :` 58QI:J 57%1CCQ%<8 VCQIV`:V :H 101 7:JR.%I:J VCQIV`:V `V0V`V `:JH`1] :VI 3V6]`V1QJ1JQ`  1%V %IQ`7HQ``VC: 1QJ11 .$`:RV5.1 QCQ$75:JR]`QC1`V`: 10V :H 101 78:JHV`V8 5 R ^ _ 8 2H.VVC52H.:V`V`C5 :%H.35V :C83C V`J: 10VCVJ$ .VJ1J$Q` VCQIV`V1:QH1: VR11 .H.`QIQQI:C 1J :G1C1 71JQ VQ:`HQI:8(JHQ$VJV5  R ^_ 8 3%V75%`:C1R.:`852H.1:` <258% CV`78 VCQIV`:V:H 101 71J@VCV :C:`HQI:8"JJ80%`$8(JHQC8 5  R ^ _ 8 C:JV` 735 %:J$ 5  V`Q 5 V  :C8 3GVJHV Q` : VCQIV`V I:1J VJ:JHV IVH.:J1I : : `:0Q`:GCV ]`Q$JQ 1H`:H Q`1J]: 1VJ 11 .Q VQ:`HQI:8:JHV`V8 5 R ^_ 8 2:J$1Q`$1<57QGG1735<%H:`VCC1 5V :C8`VVJHVQ` VCQIV`:V:H 101 71JR1``V`VJ I%H%CQ@VCV :C %IQ` .1 Q 7]V:JRHQ``VC: 1QJ11 .:$$`V10VJV8)J 8'8:JHV` 5 R ^_ 8 C1RQ 35 2H.JV1RV`R2 QH@ 5 :%] I:JJ C5 QVJV` 38 VCQIV`:V :H 101 7 1J GVJ1$J GQJV %IQ` :JR %IQ`RC1@VCV1QJ8$: .QC8V8$`:H 8 5 R ^ _  8 :1C:@75VC`I:J87 Q@VCV :CH.:J$V1JHVCC `:J`Q`I: 1QJ:JR %IQ`1$VJV18%``8(]1J82VJV 8 1V085R ^_  8

8 <%Q 5 2.:``` C35 .VJ 5 V 5<%% 8 23  V6]`V1QJ :JR `%JH 1QJ 1J .%I:J Q VQ:`HQI:8 C1J8 (` .Q]8 5 R ^ _ 8 V`7%$1J: 95L%1$CV7 8:` 16IV :CCQ]`Q V1J:V:JR %IQ`IV : :18:JHV`/V : :1V08 5 R ^ _ 8 .:@`:GQ` 1 25 :JR:C 5 : 25 :JR:C 35.:@`:GQ` 1 8 V$%C: 1QJ Q` I: `16 IV :CCQ]`Q V1J:V7 :J Q0V`01V18/QC8VCC8+1QH.VI8 5 R ^_ 8 8=Q`JC:JRC5C: I:`@C5V V`VJ25 V  :C8 : `16 IV :CCQ]`Q V1J:V ]:` 1H1]: V 1J Q VQ:`HQI: 1J0:1QJ8'80%`$8V8 5 R ^ _ 8 C1RQ 35 % %I1 5 9@1 C5 V  :C8 0V`V6]`V1QJ Q` I: `16 IV :CCQ]`Q V1J:V ^_R  HQ``VC: V 11 . IV : : 1H ]Q VJH7 Q` ]QJ :JVQ% :JR R.7R`Q67:I1JQ_%1JQC1JV RQ61RV ^R3L_R1JR%HVR `:J]C:J :GCVQ VQ:`HQI:1J`: 8:JHV`3V 8 5 R ^ _  8 H.1GQ`1 5 1.1R: 5 :$::@: 5 V  :C8 9JH`V:VR V6]`V1QJ Q` IVIG`:JVR 7]V I: `16 IV :CCQ]`Q V1J:VR1HQ``VC: VR11 .]QQ`]`Q$JQ11J]: 1VJ 11 .Q VQ:`HQI:8)J 8'8(JHQC8 5R ^ _  8 V``:H1J1 5 8VJ:1 25 QJ 1HVCC1 5  V  :C8 QCVQ`R :JR1  1%V1J.1G1 Q` 9R1J.%I:J Q VQ:`HQI:7`1JR1J$1J]: 1VJ `QCCQ1VR`Q`R 7V:`8"H :(` .Q]80H:JR8 5 R ^_  8 Q`JVGVH@5<:IGV`  5V 1 ``V`V 58VJ:`R88VJV`1H1:C:JRRV `1IVJ :C1J`C%VJHVQ` 1%V1J.1G1 Q`Q` IV :CCQ]`Q V1J:VR^ 9R_1J %IQ`]`Q$`V1QJ8+1QH.1I1V 5 R ^ _  8 :JRV`H:]]VCCVJ 5 :J :IIV 52 `%7` 28 .V `QCV Q`  H.VIQ@1JV :JR .V1` `VHV] Q` 1J H:JHV`8 :JHV`3V 8 5 R^ _  8 QJ$ 5CQ`<8 `:JC: 1J$ :J :J :$QJ1  Q` H.VIQ@1JV `VHV] Q`  7 ``QI GVJH. Q GVR1RV8 C1J8 :JHV`V85 R ^ _  8 V`11JQ Q 5 :0:CCQJ1 75

 8 3JRV`QJ52%$:IQ QC5Q``15% 5%J  88QJVR1JR%H1J$:$VJ ^893_``QIH%C %`VR.%I:J 2:2RQ VQ:`HQI:8+QJV/1JV`8 5 R ^ _  8 1: 25 Q` . <<5CCV1JV`I:J 28 3 J%RV IQ%V IQRVC Q` .%I:J Q VQ:`HQI: C%J$ IV : :V `Q` V0:C%: 1J$JV1 .V`:]V% 1H `: V$1V8C1J8&6]8/V : :1 5^ _  8 %:J 5 VJRQ`` 5 2<: @Q1@1 5 V  :C8  VQGC: 1H :JR  VQC7 1H %I:J  VQ:`HQI: H:J GV 2 %R1VR1 .:V1VJQ$`:` Q%VQRVC`QR%H1J$2]QJ :JVQ%V : :V8:JHV`)J0V 8 5 R ^ _  8 :  5 @ 5 QJ `V`: C75.QQJ$ 8 3 JQ0VC Q` .Q Q]1H I%`1JV IQRVC ]`Q01RV 1J1$.  1J Q HVCC%C:`:JRIQCVH%C:`H.:`:H V`1 1HHQJ `1G% 1J$ Q.%I:JQ VQ:`HQI:8C1J8&6]8/V : :15 R  ^ _  8 @ 5: 5.QQJ$8QIIQJC7%VRIQ%VIQRVCQ`Q VQ:`HQI:8`1 8V08(JHQC8:VI: QC8 5R ^ _  8 2 V1J 725 <1:J 850:J1=JVJ3 5 V  :C8  HQJ `QC Q` Q`$:J1<: 1QJ5 :VIGC7 :JR :H 101 7 Q` .V `V$%C: Q`7I:H.1JV`7`Q`@VCV :C$VJVV6]`V1QJ8(JHQ$VJV5 R ^_  8 CQG:7:.1 5C`QJVJGV`$ 8 1J1`V01V17 `:JH`1] 1QJ:C V$%C: 1QJ 1J V0VCQ]IVJ  Q` 8QJV8 &JRQH`1JQCQ$7 5R ^ _  8 81CC1:`R 5Q`:J 35.1 CV75V :C8 `:JH`1] 1QJ:C`Q`1C1J$Q`%I:J VGC: 1``V`VJ 1: 1QJ8'8VCC8 +1QH.VI8 5 R^_  8 QJ$ 5 1:J$ 25 H:J%  5 V  :C8 35: `:JH`1] 1QJ:CIQR%C: Q`Q`IVVJH.7I:C VIHVCC R1``V`VJ 1: 1QJ80H1VJHV 5 R ^ _  8 :` I:JJ8 `:JH`1] 1QJ:CJV 1Q`@HQJ `QCC1J$@VCV :CRV0VCQ]IVJ 8%``8(]1J82VJV 81V08 5 R ^ _  8 :`1V8 `:JH`1] 1QJ`:H Q`HQJ `QCC1J$Q VQGC: Q$VJV18"`H.8+1QH.VI8+1Q].78 5 R ^ _  8 :@:.1I:C5.Q%5C%J@VC75 V  :C8 .V Q0VC 1JH 1J$V`RQJ :1J1J$ `:JH`1] 1QJ :H Q`  V`16 9 V_%1`VR`Q` VQGC: 1``V`VJ 1: 1QJ:JR8QJVQ`I: 1QJ8VCC 5 R ^_  8 CQIQ`1 8 V$%C: 1QJQ`Q VQGC: R1``V`VJ 1: 1QJG7 `:JH`1] 1QJ`:H Q`8'8VCC8+1QH.VI8 5R  ^ _  8 1:Q 75 1:J$57V5 V  :C8 QQ]V`: 10V 1J V`:H 1QJ GV 1VVJ :H 10: 1J$ `:JH`1] 1QJ `:H Q`  :JR %J6LG`: 1I%C: VQ VQGC: R]VH1`1HQ VQH:CH1J$VJVV6]`V1QJ8'8+1QC8.VI8 5 R  ^ _  8 :7 5 7%Q 5 7:``V R8V:C <5:J$ 8 J LGV :RH: VJ1J 1$J:C1J$ 1J IVVJH.7I:C ]`Q$VJ1 Q` HQJ `QC Q VQGC: :JRH.QJR`QH7 VR1``V`VJ 1: 1QJR%`1J$0V` VG`: V@VCV Q$VJV11V08VCC 5  R  ^ _  8 :I:$%H.1352:@:IQ QC51J:I1<: Q 5C: %GVC5:@:J1.128 V$%C: 1QJQ`Q VQGC: R1``V`VJ 1: 1QJ IVR1: VRG785Q H.:JRL8':]:JVV1VJ :C0H1VJHVV01V15 R ^ _  8 .:%R.:`7< 5Q`IV1 V`35`%@:C381``V`VJ 1:C7`Q1 .:H Q`QJ `QCQ`8QJVQ`I: 1QJ .`Q%$.  VQ]`Q$VJ1 Q`1``V`VJ 1: 1QJ8+QJV5R^_  8 `J1 <871$J:C1J$1J .VRV0VCQ]1J$VJRQH.QJR`:C@VCV QJ87 Q@1JV2`Q1 .#:H Q`8V08 5 R ^ _  8 .CVJ58%VCVJ<35Q` @:I]38VR$V.Q$1$J:C1J$1J@VCV :CRV0VCQ]IVJ 8+1` .1V`VH V88 5  R ^ _  8 8QJJV  51H@ 8%I:J:H% VI7VCQ1RCV%@VI1:1Q`$:J1

 8 .QI:5C:J:`:8 ]1$VJV 1HQR1`1H: 1QJ1J VQ$VJ1H1``V`VJ 1: 1QJ:JR `:J`Q`I: 1QJ8'8VCC8 +1QH.VI8 5 R ^ _  8 .QI: 5 Q.JQJ23521I35 V  :C8 V`I1J:C Q VQGC:  R1``V`VJ 1: 1QJ5 IVR1: VR G7 `%J6 :JR ] C91R1`%] VR1JQ VQ:`HQI:8'8VCC+1QC8 5  R ^_  8 71GG5C%@V@Q075 V1 . 5V :C82 VIRC1@VHVCC1JGQJV:`HQI:79I]C1H: 1QJ`Q` %IQ`1$VJV18 :VQ]C:1: 5 R ^ _  8 V7: 5CV0V`8J 1$J:CC1J$1J VIHVCC:JRH:JHV`8:: %`V5 R ^ _  8 1:J$ 5%18VRV$V.Q$1$J:C1J$1JRV0VCQ]IVJ :JRH:JHV`81V08VCC 5 R ^ _  8 8QCQ58C:JHQ5VR1J:5V :C8Q H.1$J:CC1J$1JH:JHV` VIHVCC8 C1J8`:JC8(JHQC85R  ^ _  8 C: Q.8V 1Q`@1J$Q` 575Q H.58:JRVR$V.Q$1$J:C1J$]: .1:7R%`1J$H:`H1JQ$VJV18 0 VIVCCV01V15R ^ _  8 VJ$<52.:```C35 :J$5V :C8 V$%C: 1QJQ`Q VQ$VJ1HR1``V`VJ 1: 1QJR%`1J$@VCV :CRV0VCQ]IVJ 8 #`QJ 8+1QH185R^ _  8 :`]1Q <5 7C:R% 5 7QC

8 %`].7 8 7VJV V6]`V1QJ  %R1V %1J$ I1H`Q:``:77 ]`1JH1CV5 ]`QGCVI :JR ]`Q]VH 8 "R0:JHV 1J $.71QCQ$7&R%H: 1QJ 5 R ^_ 8 1J@VC 5 2V$`:0V 5 2%R:` 5 V  :C8 1$. `VQC% 1QJ :J:C71 Q` 3 HQ]7 J%IGV` 0:`1: 1QJ %1J$ HQI]:`: 10V$VJQI1H.7G`1R1<: 1QJ QI1H`Q:``:78:: 82VJV 85 R^ _ 8 71GQJ75%V25"]`1IV`Q`$VJQIVH1VJHV852%JRV`C:JR53721J:%V`3QH1: V59JH8  8 :`H:3<5 QIV`Q 5`:$.1H1283J:C71Q`1H`Q:``:7 6]V`1IVJ Q`7VJV 6]`V1QJ`Q`1C1J$8"I8'8 (G V 827JVHQC8 5 R ^ _  8

 8 Q$. 5 `VI]V 75 :V1 :%I:J %IQ` VCC 31JV5 1J :%I:J %IQ` HVCC )J 1 `Q5 8 Q$.5 R1 Q`8  5 CVJ%I`V7V1Q`@:JR

 8 8V`C1J 5 2:I1R 5 QJ .1JVJ1R :Q 5 V  :C8V0VCQ]IVJ Q`:JQ0VC]QJ :JVQ%IV : :1IQRVCQ` .%I:JQ VQ:`HQI: `:J]C:J VRQ` .Q Q]1H:CC71J Q .VGQJVQ`: .7I1HI1HV8:JHV`V8 5 R  ^ _  8 .VJ 5 <1 5 1QJ$ C5 V  :C8 :JHV` $VJV .V`:]7 G7 R1`VH  %IQ` 1J=VH 1QJ Q` : JQJ01`:C  0VH Q` VJHQR1J$: .7I1R1JV@1J:V$VJV:%I82VJV.V`8 5  R  ^ _  8 C1I 5 210:J:JR.:I 5 2 :0`Q]Q%CQ 95:CC:H@ C8 3R=%0:J  V``VH  Q` : C  C1$:JR ^<_ $VJVR `:JR%HVR6VJQ$VJ1HHVCCC1JV1J:I%`1JVHQCQJH:JHV`IQRVC8'80%`$8V8 5 R ^_  8 81CC1:%35 R775V`VI:J5V :C8%I:J1J V``V`QJ7:]`QR%H 1QJ1J:JV1C7V :GC1.VRHVCCC1JV5 7 8"J 1I1H`QG8"$VJ .VIQ .V`85R ^ _  8 %V5V1 5 %5V :C8 .VV``VH Q`3R 7R1 3QJ .VV6]`V1QJQ`0:H%C:`VJRQ .VC1:C$`Q1 . `:H Q`1JQ VQ:`HQI:RGV:`1J$J%RVI1HV8.1JVVR2V`I:J'Q%`J:CQ`C1J1H:C(JHQCQ$7 5 R ^ _  8 2.15.VJ35:15V :C 8  :GC1.IVJ :JRH.:`:H V`1<: 1QJQ`HVCC%GC1JV11 ..1$.:JRCQ1IV : : 1H ]Q VJ 1:CRV`10VR``QI.%I:JQ VQ:`HQI:8.1J8/VR8'8^&J$C8&R8_ 5 R ^ _  8 .%57%Q5% 5.VJ38  :GC1.IVJ Q`J%RVI1HVIQRVCQ`.%I:JQ VQ:`HQI:HVCC7 11 . R1``V`VJ ]Q VJ 1:CQ`IV : :18.1JVVR2V`I:J'Q%`J:CQ`C1J1H:C(JHQCQ$7 5 R ^ _  8 2% 5 <%Q 5 V 85 V  :C8  :GC1.IVJ  :JR H.:`:H V`1<: 1QJ Q` JV1 .1$.C7 IV : : 1H .%I:J Q VQ:`HQI:HVCCC1JV8C1J8&6]8/V : :1 5 R ^ _  8 .V "IV`1H:J 7]V VCC %C %`V QCCVH 1QJ ^"_ +1QCQ$1H:C VQ%`HV VJ V`8 30:1C:GCV ``QI7 . ]7LL1118: HH8Q`$L8  8 C1JV` 5 C1J

 8 1CCV`85VV8 7]V91J%C1JRC1@V$`Q1 .`:H Q``VHV] Q`:: .V`:]V% 1H :`$V 1JH:JHV`8:JHV`V8 5 R ^ _  8 8`QR 52:I:J135:0:G 89J.1G1 1QJQ` .V 7]V99J%C1JRC1@V7`Q1 .:H Q` VHV] Q` 6]`V1QJ:JR 21$J:C1J$7Q0VC2 `: V$1V`Q`3J 1IV : : 1H .V`:]78+1QH.VI8$.:`I:HQC8 5R ^_  8 3R:I 5 ]:G57:``V G  5:`R82 `%H %`V:JR`%JH 1QJQ` .V 7]V1J%C1JRC1@V$`Q1 .`:H Q` `VHV] Q`8VCC8/QC831`V0H18 5 R ^_  8 <1 51CCV`  8 QCV Q` .V :H 10: 1QJ CQ]] 7`Q1JV 1J `V$%C: 1QJ Q` .V 1J%C1JRC1@V $`Q1 . `:H Q` 9 `VHV] Q`R 7`Q1JV@1J:V8'8+1QC8.VI8 5 R ^ _  8 8:V`$: 5V`%<<15 V1C81J1R V01V17 .V97R VHV] Q`1J:JHV`81QCQ$78)J 8'8:JHV` 5 R ^_  8 1 QC: 75 QQC 53HH1C1 8 7]Q$C7HVI1H V``VH  Q` 1J%C1JRC1@V $`Q1 . `:H Q`R 1J I1HV C:H@1J$ 1J%C1J `VHV] Q`8'8C1J8)J0V 8 5  R ^ _  8 8:.`57`QJV`88 .V97R`VHV] Q`:JR1 HQJ `1G% 1QJ QIV : : 1H %IQ`$`Q1 .RJQ0VC:]]`Q:H.V Q .V1J.1G1 1QJQ`97R `%JH 1QJ82`Q1 .#:H Q`5R^ _  8 3CC1QJ35H9J 7`V5H3`RCV5:G1G8 .V1J%C1JRC1@V$`Q1 .`:H Q` 7]V`VHV] Q`:JRHQCQ`VH :C JVQ]C:1:71J1$. 1J Q1J0:1QJ8:%I8$: .QC8 5 R ^ _  8 3R.:I153`:_5%@. :`89J%C1JRC1@V$`Q1 .`:H Q`R9:61::]: .1:7`Q`H:JHV`H.VIQ]`V0VJ 1QJ8 C1J8:JHV`V85 R ^ _  8 QJV <5CVIIQJ 89J%C1JRC1@V$`Q1 .`:H Q`:JR .V1`G1JR1J$]`Q V1J7G1QCQ$1H:C:H 1QJ8&JRQH`8 V08 5R^ _  8 1VRV`J.Q`V` < 5 7:`1J1 735 ::I 35 V  :C8 3 JV1 ]`Q$V`1QR 7JR`QIV `V0V:C .:  $VJQ Q61H  `V %]`VV .VQI: Q `Q].:618:: %`V5 R^ _  8 2:I:J1335:@:`25

 8 7QCCQG 351C.VCI25:` V`5CVCCV72<8 QCV`Q` :`@1J:V1JH:JHV`7 .V`:]V% 1H]Q VJ 1:CQ` :`$V 1J$ .V :`L CL C1$J:C `:JR%H 1QJ]: .1:780VI1J8(JHQC85 R ^ _  8 1J 5 3R:H.1 5 :I:IQ Q 5 V  :C8 7VJV 1H GCQ@:RV Q` .V 1J%C1JRC1@V $`Q1 . `:H Q`R9 `VH] Q` : ]`QI11J$ `: V$7`Q`.%I:J]:JH`V: 1HH:JHV`8:JHV`V8 5 R ^_  8 VJ1HQ`` 5 3IG`QV 5 Q]]QC: 5 %G1J 8 9J%C1JRC1@V $`Q1 . `:H Q`R :JR 1  `VHV] Q` IVR1: V .V :% QH`1JV]`QC1`V`: 1QJQ`.%I:JQ0:`1:JH:`H1JQI:HVCCC1JV83:G8)J0V 8 5 R ^ _  8 ::3 5V:` . C5<1

8 8:.`57`QJV`88 .V1J%C1JC1@V$`Q1 .`:H Q`R`VHV] Q`^97R _::R`%$ :`$V 7JQ0VC:]]`Q:.V Q H:JHV` .V`:]782`Q1 .:Q`I8)2#V85 R ^_ 8 VV`H.I1   5 V V1 35 8`QQ@Q0V` 5 V  :C8 2]VH1`1H 7`Q1JV @1J:V 1J.1G1 Q` `V$%C: V .%I:J Q VQ:`HQI:HVCC1J01 `Q8C1J8(` .Q]8 5 R ^ _ 8 8VJ1J1 25 :J:`: 5 8:CR1J1 5 V  :C8 9J.1G1 1QJ Q` 9J%C1JRC1@V $`Q1 . `:H Q` 9 `VHV] Q` 1JH`V:V .V :J 1 %IQ`:H 101 7Q`RQ6Q`%G1H1J:JR01JH`1 1JV:$:1J  11J$:`HQI:HVCC8C1J8:JHV`V8 5 R  ^_ 8 QH.V V` 35 1VRVI:JJ 5 VCC:1VCC 758`V1 V`25:H:%C:7821CVJH1J$Q` .V97 $VJV VJ.:JHVVJ1 101 7 Q3RR:I:$1J$:$VJ 1JGQ . 11CRR 7]V:JRI% :J .%I:J]`Q : VH:JHV`8 :JHV`2VJV.V`85 R^ _ 8 V`Q.: 2 5:C%@: 8 97R `VHV] Q` 1J.1G1 Q` 1J HC1J1H:C `1:C R :`C7 CVQJ8 '8 /:II:`7 2C:JR +1QC8 :VQ]C:1:5 R ^ _  8 CC`1H.357`:735 :I35V :C89J%C1JRC1@V$`Q1 .`:H Q`9`VHV] Q`]`1I:`7 `%H %`V7HQI]:`1QJ11 . 1J%C1J`VHV] Q`%$$V  `%H %`:CRV V`I1J:J  .: RV`1JV`%JH 1QJ:C]VH1`1H1 78&/+(^&%`8 /QC8+1QC8 (`$:J8_'8 5 R ^ _  8 Q`I:JJ 57:`H1:R H.V0V``1: 8 8CQH@1J$ .V 1J%C1JRC1@V $`Q1 . `:H Q`R9 `VHV] Q` : :  `: V$7 `Q` :`$V 1J$H:JHV`1`%$11HQ08QR:7:`$V 5R ^ _  8 2:H.RV0521J$. 5%=1 :R:I:$%H.15VV8Q1JR`V$%C: 1QJQ`1J%C1J`VHV] Q`G7:J 1GQR1V:$:1J  .V 7]V 9 1J%C1JRC1@V $`Q1 . `:H Q` `VHV] Q`7 1I]C1H: 1QJ `Q` :J 1R1J%C1JRC1@V $`Q1 . `:H Q` .V`:]7 1J G`V: H:JHV`8:JHV`V8 5 R^ _  8 8%` `%I 5 .% 5 <% 5 V  :C8 3 `%CC7 .%I:J IQJQHCQJ:C :J 1GQR7 Q .V 1J%C1JRC1@V $`Q1 . `:H Q` 9 `VHV] Q` GCQH@ C1$:JRRRV]VJRVJ  1$J:C1J$ :JR 1J.1G1  .%I:J %IQ` $`Q1 . 1J 010Q8 :JHV` V8 5 R ^_  8 7:`H1:R H.V0V``1: 5 V:`QJ 35 :` 1 35 V  :C8 9J 010Q :J 1 %IQ` :H 101 7 Q` R3  R3 JQ0VC5 ]Q VJ 5:JRVCVH 10V1J.1G1 Q`Q` .V97R @1J:V:JHV`VCC 5^_ 8 :C%@:5:`GQJ1 5

 8 %@:<:1: 5 Q$%H.1C5%`:@:I15V.:`:81 Q$VJR:H 10: VR]`Q V1JLV6 `:HVCC%C:`1$J:CR`V$%C: VR @1J:V @1J:V ^ C_ 1J.1G1 Q` `V Q`V :JQ1@1 VJ1 101 7 1J .%I:J G`V:  H:JHV` HVCC C1JV 11 . : HQJ 1 % 10VC7:H 10: VRV6 `:HVCC%C:`R`V$%C: VR@1J:V^ C_]: .1:78/QC8:JHV`.V`85R ^_  8 2.1.52.1V. 5 % 5 V  :C8 3C].:R QI: 1JV 1J:H 10: V 9CL3@  :JR C 1$J:C1J$ ]: .1:7 1J .%I:JC%J$:RVJQH:`H1JQI:3  HVCC7V``VH QJIV : :18#QQR.VI8Q61HQC8 5 R ^ _  8 Q]@1J 35 `Q1V  5:J$ 8 ``VH Q` 7]V1J%C1JRC1@V$`Q1 .`:H Q``VHV] Q` :`$V VR .V`:]7QJ H.VIQ .V`:]71J.%I:JH:JHV`:JR .VJIVH.:J1I1J0QC0VR8'8:JHV`V8C1J8(JHQC8%GC1.QJC1JV7 VG598 LRR R^_  8 %JRV`51VCVJ 5:@1 52%CC10:J853CI:J88]]Q` %J1 1V`Q`1I]`Q01J$ .V .V`:]V% 1H`: 1Q`Q` ]: 1VJ 11 .:`HQI:8:JHV (JHQC8 5 R ^ _  8 C:]V`8571C 5;QJR 57VG.:` 531:R:38Q0VC `V: IVJ  `: V$1V`Q`Q`  1%V:`HQI:8`1 8 V08(JHQC8 VI: QC8 5 R ^ _  8 VI:C3521V$VC 5:`R 5V :C8:JHV` : 1 1H8$8:JHV`'8C1J8 5 R^ _  8 7:`H1: 5 VI:C35:`R 5 V  :C5 &CQG:C :JHV` #:H   #1$%`V  8  5 3 C:J :5 737 3IV`1H:J :JHV`2QH1V 78  8 :@:`:C52:@:VR: 5@%I%`:C83J%]R: VQJQ0V`HQI1J$ RIVR1: VRI%C 1R`%$`V1 :JHV1JH:JHV` H.VIQ .V`:]788%``8$.:`I *V85 R ^ _  8 <%_I:J138VH.:J1IQ`R`%$`V1 :JHV1JH:JHV`H.VIQ .V`:]78+VR1H:C$`1JH1]CV:JR$`:H 1HV5  R ^ _  8 3J$VC1J1359V<<151VGGQ5V :C8 V0V`:CQ`R$C7HQ]`Q V1JRIVR1: VRI%C 1R`%$`V1 :JHV1J.%I: :`HQI: 2R23L6 HVCCG7JQJ V`Q1R:C:J 1R1J`C:II: Q`7R`%$8(JHQC8V]85 R  ^ _  8 8%``V1JR5.V`J1H@75 1J1JC:JR59C:J59C:J83J 1VJV 3 Q .V 7]V1J%C1JRC1@V$`Q1 .`:H Q` `VHV] Q`%]]`VV %IQ`$`Q1 .:JR]`V0VJ 1J0:1QJG7`: ]`Q : VH:JHV`HVCC1J010Q8$`QH8:: C8 $H:R8.H188.8$8 5  R  ^ _  8 .V`J1H@751<5 :J57:J259C:J8 `V: IVJ Q`.%I:JG`V: H:JHV`HVCC11 .:J 1VJV 3 Q .V 7]V  1J%C1JRC1@V $`Q1 . `:H Q` `VHV] Q` 1J.1G1  HVCC $`Q1 .5 %]]`VV %IQ`1$VJV15 :C V` .V IV : : 1H]Q VJ 1:C5:JR]`QCQJ$%`010:C1J010Q8:JHV`&VJV1.V`8 5 R ^_  8

 8 `:6CV` 8 7`Q1JV @1J:V : :`$V  1J H:JHV` .V`:]7R%HHVV :JR `:1C%`V8 (JHQC8 V8 5  R ^_  8 QR`1$%V< 5 :`: V 5 8:JR`V 5 V  :C8 QIG1J1J$ H.VIQ .V`:]7 :JR :`$V VR .V`:]1V 1J IV : : 1H HQCQ`VH :CH:JHV`8Q`CR'8&: `QVJ V`QC85 R ^ _  8 VJ 5 CQ`H.1J 85 .% L25 V  :C8 ]1RV`I:C $`Q1 . `:H Q` `VHV] Q` GCQH@:RV 1J HQIG1J: 1QJ 11 . HQJ0VJ 1QJ:CH.VIQ .V`:]71J.1G1 Q`  1%V:`HQI:HVCC$`Q1 .1J01 `Q:JR1J010Q8C1J8:JHV`V8 5 R ^ _  8 :` 1J 325 :H@1J Q. 5 :` 1J 5 V  :C89J%C1JRC1@V $`Q1 . `:H Q` 9 `VHV] Q` ]: .1:7 1J.1G1 1QJ G7 3 5:CQJVQ`1JHQIG1J: 1QJ11 .CI: 1J1G5RQ6Q`%G1H1JQ`01JH`1 1JV1:JQ0VC .V`:]V% 1H:]]`Q:H. 1J 11J$ %IQ`8C1J8:JHV`V85 R ^ _  8 2HQ C:JR1C5:J:`:51HQCV 175V :C83J 1 %IQ`:H 101 7Q` .V1J%C1JRC1@V$`Q1 .`:H Q`R9`VHV] Q` @1J:V1J.1G1 Q`R3  1JI%H%CQ@VCV :C %IQ`8:JHV`V8 5 R ^ _  8 8%` `%I57 Q1@185<%R11$8$G`Q$: 1QJQ`$16:RQVJQ HQJ`V``V1 :JHV Q:J 1R8&#R :`$V VR .V`:]7 1J 8&# `V]QJ10V %IQ` HVCC8 1J $`QHVVR1J$ Q` .V . $JJ%:C +VV 1J$ Q` .V $IV`1H:J $QH1: 1QJ`Q`:JHV`VV:`H.8 8

 8 .1JJV775CQ]VJ759::HQJ <5`QH@Q]8C: 1H:R.V`VJ  `QI:CHVCC``QI .VGQJVI:``Q1Q` HQIIQJC7%VR `:1JQ`1JG`VRI1HV70:`1: 1QJ1J71VCR5$`Q1 .:JRR1``V`VJ 1: 1QJ8'8VCC821QH.VI8 5 R ^ _  8 C1I5

8 .%IJ:`J1C]:25

 8 : 5C.:H.1$1:J<5.QQJ$8R%J1H`1 1H:C`Q` .V]`Q$`V1QJQ`Q VQ:`HQI:7]`QQ`1J:J Q` .Q Q]1H]QJ :JVQ%C7IV : :1<1J$IQRVC8+QC8:JHV`V8 5 R ^ _  8 2.1IQ 5: %I%`:259G:`:$125V :C82]VH1`1H1J.1G1 Q`Q` CRIVR1: VRH`QR :C@GV 1VVJ C:JR ] 3CR%`1J$R1``V`VJ 1: 1QJQ`.%I:JQ VQ:`HQI:HVCC8'8VCCQII%J8.1$J:C85R^ _  8 2%7:I:C52.:]1`Q957% I:J5:<:J 8831$J:C1J$]: .1:7CV:R1J$ QIV : :11HQJ `QCCVRG7R H:R.V`1J:JR .V7`VHV] Q`8:JHV`VCC5R^_  8 CQCG 357Q`C1H@ 8V0VCQ]IVJ Q`97R9 9J.1G1 Q`1JVR1: `1H2:`HQI:8%``8(JHQC8V]85 R ^ _  8  V1C 5CQCH.8QJ`V``1J$]VH1`1H1 7QJ .V%G1_%1 Q% :`L C1$J:CC1J$]: .1:782`8'8:JHV` 5  R ^_  8 VCCGQ`H@5C:`::`1RV5:`:1 8 .V 3]`Q V1J :@VHVJ `V :$V8:: 8V08+QC8VCC21QC8 5 R ^_  8 CQ .Q 5 :CH:IQJ1H: 25 :H.R:H%:C 25 V  :C8 32 I% : 1QJ :JR HCQJ:C1 7 :J:C71 1J H.1CR`VJ 11 . =%0VJ1CVI7VCQIQJQH7 1HCV%@VI1:^<_8V%@VI1:^2:1J$ Q@V_5R ^ _  8 2 1`V1:C <5CQ]VH@7C5V.1JH.125V :C8< 5 325:JR  I% : 1QJ1JVCRV`C7]: 1VJ 11 .:H% V I7VCQ1RCV%@VI1:82CQQR 5 R ^_  8 7:`JV 5:`:1 87%1C 7:H.:`$VR78R :`1:.%I:JQJHQ$VJV8:JHV`VCC 5R ^_  8 210:`:I:J 25 :J$ 5 %Q0Q 75:CGQJ 8 7]V`V6]`V1QJ Q` I1 Q$VJ:H 10: VR ]`Q V1J @1J:V 1J .%I:JG`V: H:JHV`8'8C1J88J0V 8 5 R ^ _  8 3CG:JVCC5QRQJ7R2V`0: 5 Q=Q5V :C83H 10: VRV6 `:HVCC%C:`1$J:CR`V$%C: VR@1J:V7:QH1: 1QJ11 . V]1RV`I:C $`Q1 . `:H Q` `VHV] Q`L `:J`Q`I1J$ $`Q1 .`:H Q`:C].:V6]`V1QJ1J.V:R:JRJVH@ _%:IQ%H:`H1JQI::JR1J.1G1 1QJG7:J 1RV]1RV`I:C$`Q1 .`:H Q``VHV] Q` `V: IVJ 8:JHV`V8 5 R ^_  8 QQR5.V`11J@15VJV :J:@Q 5V :C89JR%H 1QJQ`GV :R9J V$`1J7VJV 6]`V1QJG72% :1JVR 3H 10: 1QJ Q` .V :R V$%C: VR :`R CR 6 `:HVCC%C:` 21$J:CR V$%C: VR C1J:V 21$J:C1J$ : .1:78 +QC8 VCC821QC85 R ^_  8 Q=JQ1@1<51IIV`358VH@ 5V :C8 JRQ .VC1:C:]Q] Q11J8`:`RRV`1H1VJ I1HV8 :: 8&VJV 8 5  R ^ _  8 %V`5<%H@V 5.1CQVH.V35V :C8 C@1J:V:H 101 71JQ JVHV:`7`Q` :`R`%JH 1QJ86+2(^6%`8 +QC821QC8(`$:J8_'85 R ^_  8 3C:01 35 QQR 5 `:% Q 5 2 %]:H@ 75.V`V. 38 QCV Q` :` 1J 0:H%C:` ]`Q VH 1QJ ``QI R1 1JH  :]Q] Q 1H 1I%C18.H1VJHV5 R ^_  8 2:1R:@58Q%RQ 53GRQ%JV 5V :C8 6 `:HVCC%C:`H:CH1%I]`QIQ V .VI1$`: 1QJQ`G`V: H:JHV`HVCC .`Q%$. .V:H 10: 1QJQ` .VH:CH1%IVJ1J$`VHV] Q`866]8VCCV8 5 R ^ _  8 VJ:@QJ$@:352% .1].QJ$H.:1 89J0QC0VIVJ Q`9C:JR CL]: .1:71J.V]: QH7 V$`Q1 .`:H Q`R 1JR%HVRH.QC:J$1QH:`H1JQI:HVCC1J0:1QJ8Q`CR'8J&: `QV V`QC8 5 R ^_  8 V.5 Q% . 5 %G1J: 5V :C8C 32L8 3I% : 1QJ : %:JR CL:H 10: 1QJ:G1QI:`@V``Q`  CL1J.1G1 Q` .V`:]71JHQCQ`VH :CH:JHV`8+QC8:JHV`1.V`8 5 R ^ _  8 .1:R:351`H.Q11 < 35`7 :C 78 V$1QJ:CRVC10V`7Q`:J:RVJQ01`%0VH Q`HQJ :1J1J$ .V H.V`1H.1: HQC1H7 Q1JVRV:I1J:V$VJV Q]`Q01RVCQH:C:H 10: 1QJQ` R`C%Q`QH7 Q1JV Q%]]`V .V$`Q1 .Q` HQCQJH:`H1JQI:IV : : 1H QC10V`8 %I8&VJV1.V`8 5 R ^ _  8 C1I 5 % 5 CQ0:C@1 C5Q1@1 <8 V_%1`VIVJ  `Q` ]VH1`1H ]`Q V:V 1J H:JHV` HVCC 1J `:0:: 1QJ : `V0V:CVRG7:JQ0VCVI1_%:J 1 : 10V RG:VR::78VCC 5 R ^ _  8 1=C `: 35 VCCQ` 5 :J<:`VCC: 75 V  :C8 3 _%:J 1 : 10V :J:C71 Q` `: VRC1I1 1J$  V] 1J .V IV : : 1H H:H:RV%1J$.%I:JR]VH1`1H`V:CR 1IV]QC7IV`:VH.:1J`V:H 1QJ8:JHV`V8 5  R  ^_  8 7QIV5VCC1J$5L%V95V :C8%JH 1QJ:C1I:$1J$Q`I%C 1R`%$`V1 :JHV1J:JQ` .Q Q]1HIQRVCQ` QV Q:`HQI:%1J$ IR HRV :I1G186%`8'8:%HC8+VR8+QC88I:$1J$5 R ^ _  8 RV2:1J 3%G:1J2QIV`.:%VJ5CV H.V`82Q` R 1%V:`HQI:7:J%]R: V86%`8'8.%`$8(JHQC8 5  R^ _  8 .1 J15%VJ25`Q .V`QV325QCC:@5:H:%C:78 .V 7]V1J%C1JRC1@V$`Q1 .`:H Q``VHV] Q` ]: .1:78C1J8:JHV`V85  R  ^ _ 242

 8 V% H. 5:$$1Q`VCC:<5VJ85V :C8 7`Q1JV@1J:V1J.1G1 Q`37V6V` :J 1CV%@:VI1HV``VH QJ 2  R`V1 :J 8H`R3GCV6]`V1J$HVCC:JRRVH`V:V3C ].Q].Q`7C: 1QJ82`8'8:JHV` 5  R  ^_  8 :`:]:JHV:5Q:HV:J%58%R1% 5V :C8%:C :`$V 1J$Q`97R :JR7 1J.1G1 ]`QC1`V`: 1QJ1J .1$.R$`:RV$C1QI:HVCC:JR1JR%HV`:R1QVJ1 101 71JCRV6]`V1J$HVCC8'8:V%`QQJHQC8 5 R  ^ _  8

 8 :=@%I:` 5:I%J: 8 %C 1]CV ]: .1:7 :`V 1J0QC0VR 1J R`%$ `V1 :JHV Q RQ6Q`%G1H1J 1J :J Q VQ:`HQI:HVCCC1JV8$J 1H:JHV`8*`%$ 5 R ^ _  8 %:J5.%5%:J$5V :C89J V``V`QJRαVJ.:JHVVJ1 101 7Q`.%I:JQ VQ:`HQI:R2HVCC Q RQ6Q`%G1H1J G7 ] RRV]VJRVJ  :]Q] Q18 $H : $.:`I:HQC8 .1J8  5   R  ^97 8L=8  R  8 8 86_^ _  8 1CCV`<5C:VC5VCQJ 5V :C8] `%JH 1QJ::HVCCH7HCVHQJ `QC]`Q V1J1JQV Q:`HQI:8+QC8VCC8 21QC85 R ^ _  8