Novel Therapies for High-Risk Leukaemia in Children

Mawar Murni Karsa

A thesis submitted to the University of New South Wales in fulfilment of the requirements for the degree of Doctor of Philosophy

Children’s Cancer Institute Australia and School of Women’s and Children’s Health Faculty of Medicine The University of New South Wales

September 2018

THESIS DISSERTATION

ORIGINALITY 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 ……………………………………………......

6 September 2018 Date ……………………………………………......

‘I hereby grant the University of New South Wales or its agents 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 …………………………………….…

6 September 2018 Dated …………………………….………….

iii

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 ………………………………….

6 September 2018 Dated …………………………………..

ACKNOWLEDGEMENTS

Firstly, I would like to thank my supervisors, Dr. Michelle Henderson and Dr. Klaartje Somers for the opportunity to do my PhD in the Experimental Therapeutics program. I could not have done this project without their endless support, guidance and understanding over the last four years. They have been great mentors and their passion for science have shown me how exciting this field can be. I also thank my co- supervisors, Associate Professor Rosemary Sutton and Professor Richard Lock for their helpful advice over the past years which have proven to be invaluable in the completion of this thesis. I thank my review panel members, Dr. Jamie Fletcher, Dr. Karen Mackenzie and Professor Glenn Marshall for always looking out for me throughout this journey and providing helpful suggestions for the progress of the project.

I thank Dr. Anna Mariana, Dr. Tim Failes and Dr. Greg Arndt at the Australian Cancer Research Foundation Drug Discovery Centre for helping is with the drug screening. I would like to acknowledge Dr. Andrew Gifford and Dr. Russell Pickford for their expertise and help for the in vivo toxicity and compound stability part of the project, respectively. I also thank Dr. Chelsea Mayoh and Dr. MoonSun Jung for their help with microarray analyses. Thank you for your time and contribution to this project.

I would also like to extend my appreciation to Dr. Amanda Philp for her endless support throughout the years. A huge thank you to colleagues from the MLL research group past and present for sharing their knowledge, never-ending help and encouragement, and friendship. I thank my PhD mates for making this journey much more enjoyable. And thank you to all Children’s Cancer Institute Australia staff especially the Molecular Diagnostics and Experimental Therapeutics groups for the fun, engaging conversations, advice and encouragements during the years.

Finally, I wish to express my deepest gratitude to my parents and sisters for their love, understanding, patience and unwavering support. A special thanks to my sister Sari for the sanity checks, the laughter and the great food! I am so lucky to have such an amazing family whom have kept me going and have made this PhD possible. Thank you for believing in me, I couldn’t have done this without you.

ABSTRACT

Despite remarkable improvements being made in the treatment of childhood acute lymphoblastic leukaemia (ALL), prognosis remains dismal for certain subgroups of high-risk patients including infants with leukaemia harbouring rearrangement of the Mixed Lineage Leukaemia (MLL/KMT2A) gene. The poor clinical outcome linked to the aggressive disease and the limitations of treatment intensification warrants development of more effective, targeted therapeutics. The approach of drug-repurposing, whereby an approved drug may be applied to target a disease other than that for which it was originally intended, is one that is gaining popularity due to the potential to avoid the rising cost and lengthy process of the traditional drug discovery pathway.

To identify novel candidates for high-risk leukaemia, a library of approved drugs and pharmacologically active compounds, was screened against ALL cell lines with or without MLL gene rearrangement, using a cell-based viability assay. The screen identified two MLL-selective bioactive compounds. The purinergic receptor (P2Y) agonist and guanylate cyclase inhibitor, 2-chloroadenosine triphosphate, showed in vitro efficacy against MLL-rearranged (MLL-r) ALL patient-derived xenografts (PDX) whereby sensitivity was associated with decreased expression of several P2Y receptors including P2RY14, which was additionally found to be lowly expressed in MLL-r patients compared to patients without MLL gene rearrangement in a paediatric ALL patient cohort. The second MLL-selective candidate, SID7969543 which targets Steroidogenic Factor-1 (SF-1/NR5A1), showed activity against a subset of MLL-r and CALM-AF10 leukaemia cell lines and synergized with etoposide or cytarabine in vitro.

A subsequent secondary screen aimed to select more potent and clinically applicable compounds identified two FDA-approved drugs, auranofin and disulfiram, which revealed a common ROS-mediated mechanism in potently inhibiting the viability of high-risk leukaemia cell lines and PDX in vitro. Preclinical testing of drug combinations revealed auranofin in combination with cytarabine, a drug currently used in paediatric ALL therapy, demonstrated potential in delaying leukaemia growth in an aggressive MLL-r ALL PDX mouse model. Auranofin also demonstrated synergy with disulfiram in vitro which could be promising for future studies.

In conclusion, this work identified MLL-selective compounds that uncovered potential new targetable pathways in MLL-r leukaemia for further investigation and possible future therapeutic exploitation. The two FDA-approved drugs identified highlighted the therapeutic potential of targeting the ROS pathway for high-risk leukaemia and demonstrated promising clinical utility for these patient subgroups.

vii

CONFERENCES AND AWARDS

Conferences . Karsa, M., Somers, K., Mariana, A., Failes, T., Arndt, A., Haber, M., Norris, M., Sutton, R., Lock R. and Henderson, M. (2017). Repositioning existing drugs as novel therapeutics for high risk leukaemia in children. 22nd Congress of European Hematology Association, Madrid, Spain. Poster presentation.

. Karsa, M., Mariana, A., Failes, T., Arndt, G. M., Sutton, R., Lock, R. and Henderson, M. (2016). Repositioning existing drugs as novel MLL‐rearranged leukaemia therapies. CTx Higher Degree Research Symposium, Melbourne, Australia. Oral and poster presentations.

. Karsa, M., Mariana, A., Failes, T., Arndt, G. M., Sutton, R., Lock, R., Norris, M., Haber, M., and Henderson, M. (2016). Identification of molecules with selective inhibition towards MLL-rearranged leukaemia. 10th Biennial Childhood Leukemia Symposium, Athens, Greece. (Also presented at 6th New Directions in Leukaemia Research, Noosa, Australia). Poster presentations.

. Karsa, M., Middlemiss, S., Richmond, J., Haber, M., Norris, M., Sutton, R., Lock, R. and Henderson, M. (2015). Assessing an in vitro model of cultured patient- derived xenografts for predicting treatment response in vivo. Lowy Symposium, Sydney, Australia. (Also presented at the CTx Higher Degree Research Symposium, Melbourne, Australia). Poster presentations.

Awards . Children’s Cancer Institute Top-Up Scholarship (2014 – 2018) . Cancer Therapeutics CRC Top-Up PhD Scholarship (2015 – 2017) . The 22nd Congress of European Hematology Association (EHA) Travel Grant (2017)

viii

TABLE OF CONTENTS

THESIS DISSERTATION ...... i ORIGINALITY STATEMENT ...... ii COPYRIGHT STATEMENT ...... iii AUTHENTICITY STATEMENT ...... iv ACKNOWLEDGEMENTS ...... v ABSTRACT ...... vi CONFERENCES AND AWARDS ...... viii TABLE OF CONTENTS ...... ix LIST OF FIGURES ...... xiv LIST OF TABLES ...... xviii ABBREVIATIONS AND ACRONYMS ...... xx CHAPTER 1: INTRODUCTION ...... 1 1.1 Childhood acute lymphoblastic leukaemia ...... 1

1.2 Clinical features of childhood acute lymphoblastic leukaemia ...... 1

1.2.1 Immunophenotype ...... 3 1.2.2 Cytogenetics ...... 3 1.3 Treatment of acute lymphoblastic leukaemia ...... 4

1.3.1 Risk factors ...... 5 1.3.2 Treatment protocol ...... 7 1.3.3 Side effects ...... 9 1.4 High-risk acute lymphoblastic leukaemia subtypes ...... 10

1.4.1 MLL-rearranged leukaemia ...... 12 1.4.2 Philadelphia-positive acute lymphoblastic leukaemia ...... 17 1.4.3 Philadelphia-like acute lymphoblastic leukaemia ...... 19 1.4.4 Early T-cell precursor acute lymphoblastic leukaemia ...... 21 1.5 Conventional drug discovery pipeline ...... 22

1.6 Drug repurposing ...... 26

1.6.1 Introduction to drug repurposing ...... 26 1.6.2 Strategies of drug repurposing ...... 27

1.6.3 Methods in discovering potential drugs for novel activity ...... 29 1.6.4 Considerations for drug repurposing...... 32 1.6.5 Drugs repurposed for leukaemia ...... 36 1.7 Summary and thesis perspectives ...... 39

CHAPTER 2: MATERIALS AND METHODS ...... 41 2.1 Reagents and equipment ...... 41

2.1.1 Tissue culture ...... 41 2.1.2 Cytotoxic drugs ...... 42 2.1.3 High-throughput screening ...... 42 2.1.4 Protein isolation and western blot analysis ...... 43 2.1.5 Flow cytometry ...... 44 2.1.6 Microsomal stability assay ...... 44 2.1.7 Cell lines ...... 44 2.1.8 Patient-derived xenografts ...... 46 2.1.9 In vivo studies ...... 47 2.2 Methods ...... 47

2.2.1 Cell culture ...... 47 2.2.1.1 Maintenance of suspension cells ...... 47 2.2.1.2 Maintenance of adherent cells ...... 48 2.2.1.3 Maintenance of patient-derived xenograft cells ...... 48 2.2.2 Trypan blue exclusion assay ...... 48 2.2.3 High-throughput screening ...... 48 2.2.3.1 Primary HTS assay ...... 49 2.2.3.2 Secondary HTS assay: Strategy I ...... 50 2.2.3.3 Secondary HTS assay: Strategy II ...... 50 2.2.4 Cytotoxicity assay ...... 51 2.2.4.1 Single agent cytotoxicity assay ...... 51 2.2.4.2 Fixed-ratio combination cytotoxicity assay ...... 52 2.2.5 Apoptosis detection ...... 53 2.2.6 Microsomal stability assay ...... 54 2.2.7 Reactive oxygen species assay ...... 55 2.2.8 Isolation and quantitation of total cellular protein ...... 55 2.2.9 Western Blot Analysis ...... 55

2.2.10 Gene expression ...... 56 2.2.11 In vivo mouse studies ...... 57 2.2.11.1 Determining maximum tolerated dose in NOD/SCID mice ...... 57 2.2.11.2 Determining hepatoxicity in auranofin-treated NOD/SCID mice ...... 57 2.2.11.3 Determining hepatoxicity in disulfiram-treated NOD/SCID mice ...... 58 2.2.11.4 Engraftment of human leukaemia cells into NOD/SCID mice ...... 58 2.2.11.5 Peripheral blood monitoring of leukaemia engraftment ...... 59 2.2.11.6 Drug treatment of NOD/SCID mice ...... 59 2.2.11.7 Assessment of in vivo drug efficacy ...... 60 2.2.11.8 Assessment of molecular response to drugs in vivo ...... 60 2.2.12 Statistical analyses ...... 61 CHAPTER 3: HIGH-THROUGHPUT SCREENING TO IDENTIFY NOVEL THERAPIES FOR HIGH-RISK LEUKAEMIA ...... 62 3.1 Introduction ...... 62

3.2 Results ...... 65

3.2.1 Primary screening of FDA-approved drugs and pharmacologically active compounds for activity against leukaemia cell lines ...... 65 3.2.1.1 Optimization of assay conditions for high-throughput screening ...... 65 3.2.1.2 Results of primary screen ...... 67 3.2.2 Secondary screen Strategy I: Identification of compounds selective against MLL-rearranged leukaemia ...... 67 3.2.3 Secondary screen Strategy II: Identification of potent novel inhibitors of high-risk leukaemia ...... 77 3.3 Discussion ...... 81

CHAPTER 4: CHARACTERIZATION OF NOVEL CANDIDATE MLL- SELECTIVE INHIBITORS ...... 88 4.1 Introduction ...... 88

4.2 Results ...... 90

4.2.1 Characterization of 2-chloroadenosine triphosphate as a selective inhibitor of MLL-rearranged leukaemia cells ...... 90 4.2.1.1 Selectivity of 2-chloroadenosine triphosphate towards MLL-rearranged leukaemia ...... 90

4.2.1.2 Determination of synergy between conventional chemotherapeutics and 2- chloroadenosine triphosphate in vitro ...... 96 4.2.1.3 Effect of 2-chloroadenosine triphosphate on infant MLL-rearranged ALL patient-derived xenograft cells in vitro...... 102 4.2.1.4 Markers of response to 2-chloroadenosine triphosphate in infant MLL- rearranged ALL patient-derived xenografts and patient samples ...... 104 4.2.1.5 Effect of 2-chloroadenosine triphosphate on normal peripheral blood mononuclear cells ...... 110 4.2.1.6 Comparison of activity between 2-chloroadenosine triphosphate and its parent compound, 2-chloroadenosine ...... 110 4.2.1.7 In vivo stability of 2-chloroadenosine triphosphate ...... 113 4.2.2 Characterization of SID7969543 as a selective inhibitor of MLL- rearranged leukaemia cells ...... 115 4.2.2.1 Selectivity of SID7969543 towards MLL-rearranged leukaemia ...... 115 4.2.2.2 Determination of synergy between conventional chemotherapeutics with SID7969543 in vitro ...... 120 4.2.2.3 Activity of SID7969543 in infant MLL-rearranged ALL patient-derived xenografts in vitro ...... 120 4.2.2.4 Toxicity of SID7969543 in normal peripheral blood mononuclear cells ...... 120 4.2.2.5 In vivo stability of SID7969543 ...... 123 4.2.2.6 Level of SF-1/NR5A1 in leukaemia patients ...... 123 4.3 Discussion ...... 125

CHAPTER 5: CHARACTERIZATION OF INDUCERS OF REACTIVE OXYGEN SPECIES, AURANOFIN AND DISULFIRAM, AS POTENTIAL THERAPEUTICS FOR HIGH-RISK LEUKAEMIA ...... 132 5.1 Introduction ...... 132

5.2 Results ...... 136

5.2.1 Auranofin and disulfiram affect viability of MLL-rearranged and other high-risk leukaemia cells in vitro ...... 136 5.2.2 Auranofin and disulfiram induce apoptosis in high-risk leukaemia cells 141 5.2.2.1 Impact of copper on disulfiram response ...... 143 5.2.2.2 Confirmation of apoptosis induction ...... 146 5.2.3 Auranofin and disulfiram kill leukaemia cells in vitro by increasing intracellular ROS levels ...... 148

xii

5.2.3.1 Treatment of auranofin or disulfiram is accompanied by induction of oxidative stress pathway ...... 150 5.2.4 Auranofin and disulfiram/Cu potently affect viability of high-risk leukaemia patient-derived xenografts in vitro ...... 155 5.2.4.1 Auranofin activity in MLL-rearranged ALL patient derived xenografts ...... 158 5.2.5 Determining in vivo efficacy of auranofin using patient-derived xenograft models of MLL-rearranged leukaemia ...... 160 5.2.5.1 Determining the maximum tolerated dose of auranofin in NOD/SCID mice 161 5.2.5.2 In vivo efficacy of auranofin in a patient-derived xenograft model for high-risk MLL-rearranged leukaemia ...... 164 5.2.5.3 Cytarabine potentiates auranofin in vitro ...... 166 5.2.5.4 Prioritizing PDX for auranofin/cytarabine in vivo combination testing ...... 168 5.2.5.5 Efficacy of combined treatment with auranofin and cytarabine in a patient- derived xenograft model for MLL-rearranged leukaemia ...... 171 5.2.6 Determining in vivo efficacy of disulfiram/Cu using patient-derived xenograft models of MLL-rearranged leukaemia ...... 173 5.2.6.1 Determining the maximum tolerated dose of disulfiram/Cu in NOD/SCID mice 173 5.2.6.2 In vivo efficacy testing of disulfiram/Cu in a patient-derived xenograft model for MLL-rearranged leukaemia ...... 175 5.2.6.3 Potentiation of disulfiram/Cu in high-risk leukaemia ...... 177 5.3 Discussion ...... 181

CHAPTER 6: CONCLUSIONS AND FUTURE DIRECTIONS ...... 191 APPENDIX ...... 200 REFERENCES ...... 218

xiii

LIST OF FIGURES

Figure 1.1: Overall survival in children with acute lymphoblastic leukaemia...... 2 Figure 1.2: Frequency of genetic alterations of paediatric acute lymphoblastic leukaemia...... 2 Figure 1.3: Event-free survival of paediatric ALL patients according to subtypes...... 11 Figure 1.4: MLL-interacting proteins and frequency of MLL translocation in infant and paediatric leukaemia patients...... 13 Figure 1.5: Schematic diagram of phases of drug discovery and development...... 25 Figure 1.6: Potential avenues of drug repurposing...... 28 Figure 3.1: Optimization of cell density and resazurin incubation time for PER-485 and CEM cell lines for high-throughput screening in 384-well format...... 66 Figure 3.2: Diagram representing the result of the primary screen of a library of approved drugs and pharmacologically active compounds...... 68 Figure 3.3: Diagram representing the result of the primary screen of a library of approved drugs and pharmacologically active compounds in terms of differential activity towards PER-485 and CEM cells...... 69 Figure 3.4: Characterization of twelve MLL-selective candidate compounds in a full- dose response screen...... 71 Figure 3.5: Viability of cell line panel at 10 μM concentration of each of the seven short-listed compounds...... 73 Figure 3.6: Viability of MLL-rearranged, CALM-AF10 and MLL-wild-type leukaemia cell lines at 10 µM compound concentration...... 75 Figure 3.7: Summary of high-throughput screening of a library of approved drugs and pharmacologically active compounds for identification of MLL-selective candidate compounds (Strategy I)...... 76 Figure 3.8: Characterization of 28 potent compounds in a full dose response screen. ... 79 Figure 3.9: Summary of high-throughput screening of a library of approved drugs and pharmacologically active compounds to identify potent compounds for high-risk leukaemia (Strategy II)...... 80 Figure 4.1: Chemical structure of 2-chloroadenosine triphosphate...... 91 Figure 4.2: Cytotoxicity of 2-chloroadenosine triphosphate across a range of MLL- rearranged, CALM-AF10 and MLL-wild-type leukaemia cell lines...... 91

xiv

Figure 4.3: Comparison of viability between cell lines treated with 10 µM 2- chloroadenosine triphosphate...... 93 Figure 4.4: Apoptosis in PER-485 cells over 72-hour treatment course of 2- chloroadenosine triphosphate...... 95 Figure 4.5: Cytotoxicity of five conventional chemotherapeutics currently used in paediatric ALL therapy, against a leukaemia cell line panel...... 98 Figure 4.6: Drug response profile of leukaemia cell line panel...... 99 Figure 4.7: Cytotoxicity of 2-chloroadenosine triphosphate in infant MLL-rearranged ALL patient-derived xenograft cells in vitro...... 103 Figure 4.8: Expression levels of P1 and P2 receptors in MLL-rearranged leukaemia and MLL-wild-type leukaemia patients...... 107 Figure 4.9: Expression level of GUCY1A3 in MLL-rearranged leukaemia and MLL- wild-type leukaemia patients...... 109 Figure 4.10: Cytotoxicity of 2-chloroadenosine triphosphate in normal peripheral blood mononuclear cells...... 111 Figure 4.11: Prediction of metabolic stability of 2-chloroadenosine triphosphate using a mouse microsomal stability assay...... 114 Figure 4.12: Chemical structure of SID7969543...... 116 Figure 4.13: Cytotoxicity of SID7969543 across a range of MLL-rearranged, CALM- AF10 and MLL-wild-type leukaemia cell lines...... 116 Figure 4.14: Viability of cell line panel and comparison of viability between cell types at 10 µM SID7969543...... 118 Figure 4.15: Apoptosis of PER-485 cells over 72-hour treatment course of SID7969543...... 119 Figure 4.16: Cytotoxicity of SID7969543 in infant MLL-rearranged patient derived xenograft cells in vitro ...... 122 Figure 4.17: Cytotoxicity of SID7969543 in normal peripheral blood mononuclear cells...... 122 Figure 4.18: Prediction of metabolic stability of SID7969543 in a mouse microsomal stability assay...... 124 Figure 4.19: Expression levels of of SF-1/NR5A1 in MLL-rearranged leukaemia and MLL-wild-type leukaemia patients...... 124 Figure 5.1: Chemical structures of auranofin and disulfiram...... 137

Figure 5.2: Cytotoxicity of auranofin and disulfiram...... 138 Figure 5.3: Auranofin and disulfiram have potent activity against high-risk leukaemia cells in vitro...... 140 Figure 5.4: Auranofin and disulfiram induce apoptosis in sensitive high-risk leukaemia cell lines...... 142 Figure 5.5: Copper increases efficacy of disulfiram in high-risk leukaemia cell lines. 144 Figure 5.6: Comparison of efficacy of disulfiram/Cu between PER-485 and non- malignant cell line MRC-5...... 145 Figure 5.7: Confirmation of apoptosis induction in high-risk leukaemia cell lines...... 147 Figure 5.8: Auranofin and disulfiram increase intracellular ROS...... 149 Figure 5.9: Auranofin induces higher expression of proteins in the oxidative stress pathway and pre-treatment with NAC prevents auranofin-induced ROS production and rescues cells from apoptosis...... 151 Figure 5.10: Disulfiram induces higher expression of proteins in the oxidative stress pathway and pre-treatment with NAC prevents disulfiram-induced ROS production and rescues cells from apoptosis...... 154 Figure 5.11: Auranofin and disulfiram/Cu affect viability of high-risk leukaemia PDXs in vitro...... 157 Figure 5.12: Nrf2 and γH2AX expression in MLL-6 and MLL-7 cells treated with auranofin in vitro...... 159 Figure 5.13: Morphologic appearance of Glisson’s capsule in livers of mice in control and treatment groups in MTD study...... 162 Figure 5.14: Morphologic appearance of Glisson’s capsule and hepatocytes in livers of mice in control and 5 mg/kg treatment groups of MTD study...... 163 Figure 5.15: In vivo efficacy testing of auranofin in MLL-rearranged ALL xenograft, MLL-6...... 165 Figure 5.16: Immunoblots of Nrf2 and γH2AX in MLL-6 and MLL-7 in vivo treated cells...... 165 Figure 5.17: Immunoblots of γH2AX levels in PER-485 and REH cells treated with auranofin and cytarabine...... 167 Figure 5.18: Combination assays of auranofin and cytarabine in MLL-rearranged patient-derived xenografts...... 169

xvi

Figure 5.19: Immunoblots following auranofin and cytarabine combination treatment in MLL-rearranged patient-derived xenografts...... 170 Figure 5.20: In vivo efficacy testing of auranofin and cytarabine combination in a MLL- rearranged ALL xenograft...... 172 Figure 5.21: In vivo efficacy testing of disulfiram/Cu in MLL-rearranged ALL patient xenograft models...... 176 Figure 5.22: Combination assays of disulfiram and other drugs in vitro and analysis using Bliss Independence methodology...... 179

xvii

LIST OF TABLES

Table 1.1: Risk factors in paediatric acute lymphoblastic leukaemia ...... 6 Table 1.2: Comparison between experimental and in silico approaches in identifying potential drugs for repurposing...... 33 Table 2.1: Characteristics of cell line panel...... 45 Table 2.2: Characteristics of patient-derived xenograft panel...... 46 Table 2.3: Interpretation of Combination Index values from CalcuSyn program...... 52 Table 2.4: Treatment protocols for in vivo studies...... 60 Table 2.5: Treatment protocols for study for molecular response in vivo...... 61 Table 3.1: Cytotoxicity of seven shortlisted compounds in a high-risk leukaemia cell line panel...... 74 Table 4.1: Cytotoxicity of 2-chloroadenosine triphosphate in a panel of 25 cell lines... 92

Table 4.2: IC50 values for chemotherapeutics against a leukaemia cell line panel...... 99 Table 4.3: Calculated combination index and drug reduction index for 2- chloroadenosine triphosphate in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and MOLM-13 cells...... 101 Table 4.4: Translocations present in panel of infant MLL-rearranged ALL PDXs...... 103

Table 4.5: Correlation analyses of expression of P1 and P2Y receptors against IC50 for 2-chloroadenosine triphosphate in MLL-rearranged patient-derived xenografts...... 105 Table 4.6: P-values from unpaired t-tests of mRNA expression of P1 and P2Y receptors in MLL-rearranged leukaemia versus MLL-wild-type leukaemia patients based on expression microarray data of a paediatric ALL patient cohort...... 106 Table 4.7: Correlation analyses of expression of genes encoding guanylate cyclase (GC) activators, GC family 1 and 2 against 2-chloroadenosine triphosphate IC50 for MLL- rearranged patient-derived xenografts...... 108 Table 4.8: P-values from unpaired t-tests of mRNA expression of genes encoding guanylate cyclase (GC) activators, GC family 1 in MLL-rearranged leukaemia versus MLL-wild-type leukaemia patients based on expression microarray data of a paediatric ALL patient cohort...... 109 Table 4.9: Comparison of cytotoxicity of 2-chloroadenosine-triphosphate and 2- chloroadenosine...... 112 Table 4.10: Cytotoxicity of SID7969543 in a panel of 25 cell lines...... 117

xviii

Table 4.11: Calculated combination index and drug reduction index for SID7969543 in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and MOLM-13 cells...... 121 Table 5.1: Cytotoxicity of auranofin and disulfiram against a panel of leukaemia, solid tumour and non-malignant cell lines...... 139 Table 5.2: Cytotoxicity of auranofin and disulfiram/Cu in high-risk paediatric leukaemia patient-derived xenograft panel...... 156 Table 5.3: Calculated combination index and drug reduction index for auranofin in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and PER-490 cells...... 167

Table 5.4: Fold decrease of auranofin IC50 for MLL-rearranged patient derived xenografts in auranofin and cytarabine combination assays...... 169 Table 5.5: Level of alkaline phosphatase and alanine aminotransferase in leukaemia- free NOD/SCID mice during toxicity testing with disulfiram/Cu...... 174 Table 5.6: Calculated combination index (CI) and excess over Bliss (EOB) values for disulfiram in combination with currently used drugs in the treatment of paediatric ALL and auranofin at ED75 in PER-485 cells...... 180

xix

ABBREVIATIONS AND ACRONYMS

2-CADO 2-chloroadenosine 2-Cl-ATP 2-chloroadenosine triphosphate ACC adrenocortical carcinoma ADMET absorption, distribution, metabolism, excrement and toxicity ALDH aldehyde dehydrogenase ALL acute lymphoblastic leukaemia ALP alkaline phosphatase ALT alanine aminotransferase AML acute myeloid leukaemia APL acute promyelocytic leukaemia ATP adenosine triphosphate AUR auranofin B-ALL B-cell ALL BCR breakpoint cluster region BMP-2 bone morphogenetic protein 2 cAMP cyclic adenosine monophosphate CD cluster of differentiation cGMP cyclic guanosine monophosphate CI combination index CLL chronic lymphocytic leukaemia CML chronic myelogenous leukaemia CNS central nervous system Cu copper DCF-DA 2',7'-dichlorodihydrofluorescein diacetate DDC diethyldithiocarbamate DFS disease-free survival DMSO dimethyl sulfoxide DOT1L Disruptor of Telomeric Silencing 1-like protein DRI drug reduction index DSF disulfiram ED75 effective dose 75 (concentration causing 75% decrease in viability) EFS event-free survival EOB Excess over Bliss ETP early T-cell precursor FDA Food and Drug Administration FLT3 Fms-like receptor tyrosine kinase-3 GC guanylate cyclase GCLC glutamate-cysteine ligase catalytic GCLM glutamate-cysteine ligase modifier GSH glutathione GSSG glutathione disulphide

H2AX H2A histone family, member X H3K4 histone H3 at lysine 4 H3K79 histone H3 at lysine 79 HDAC histone deacetylase HMOX1 haem oxygenase 1 HOX homeobox HSC hematopoietic stem cell HSCT hematopoietic stem cell transplantation HTS high-throughput screening i.p intraperitoneal KEAP1 Kelch-like ECH-associated protein 1 LGD leukaemia growth delay MEK mitogen-activated protein kinase kinase/extracellular signal–regulated kinases kinase MGMT O6-methylguanine-DNA methyltransferase MLL Mixed Lineage Leukaemia MLL-r MLL-rearranged MLL-wt MLL-wild-type MRD minimal residual disease MSC mesenchymal cell MTD maximum tolerated dose NAC N-acetyl cysteine NCI National Cancer Institute NIH National Institutes of Health NOD/SCID non-obese diabetic/severe combined immunodeficient Nrf2 Nuclear factor (erythroid-derived 2)-like 2 PARP Poly(ADP-ribose) polymerase PB peripheral blood PBMC peripheral blood mononuclear cells PCR polymerase chain reaction PDX patient-derived xenografts Ph Philadelphia chromosome Ph+ Philadelphia chromosome positive RA rheumatoid arthritis redox reduction/oxidation RFU relative fluorescence unit ROS reactive oxygen species SF-1 Steroidogenic Factor-1 SID SID7969543 T-ALL T-cell ALL TKI tyrosine kinase inhibitor TrxR thioredoxin reductase γH2AX phosphorylated H2AX

xxi

CHAPTER 1: INTRODUCTION

1.1 Childhood acute lymphoblastic leukaemia

Acute lymphoblastic leukaemia (ALL) is a haematological disease that arises from proliferation of malignant progenitor cells of lymphoid origin (Woo et al., 2014). It is the most prevalent paediatric cancer, accounting for 25% of all malignancies in children below the age of 15 (Pui, 2010) and approximately 80% of all childhood leukaemias. In children under 15, the incidence of ALL is around 41 in a million individuals (Medina- Sanson, 2016), with peak incidence occurring between the age of 2 and 5 years (Margolin et al., 2011). Over the past five decades, clinical outcome of this once fatal disease has had an incredible improvement due to development of systemic and intrathecal chemotherapy, risk stratification and risk-directed therapy (Pui, 2010). Figure 1.1 shows the overall survival rates of children will ALL treated under the Children’s Cancer Group and Children’s Oncology Group regimens over the period of 40 years (Hunger and Mullighan, 2015). Survival has increased from below 10% in the 1960s to above 90% in recent years (Hunger et al., 2012). Despite recent advancements in the treatment of childhood ALL, up to 20% of patients relapse due to resistant leukaemic cells or transformation of a pre-malignant clone (Bailey et al., 2008), which result in poor prognosis with overall survival rate of only 30 – 50% (Mullighan et al., 2008; Parker et al., 2010). This high-risk group, consisting of patients with several high-risk subtypes of leukaemia with very poor prognosis, needs more effective and targeted treatments as the currently used chemotherapeutics have reached their limits of effectiveness and safety.

1.2 Clinical features of childhood acute lymphoblastic leukaemia

Presenting symptoms of ALL patients include thrombocytopenia, causing easy bruising, bone pain that may be induced by leukaemic infiltration of the periosteum (membrane lining the outer surface of bones), bone infarction and expansion of the marrow cavity by leukaemic cells (Lanzkowsky, 2010). Anaemia caused by tumour competition for erythrocyte production in the marrow, pallor and fatigue from anaemia, infection caused

Figure 1.1: Overall survival in children with acute lymphoblastic leukaemia. Overall survival in children with acute lymphoblastic leukaemia enrolled in the Children’s Cancer Group and Children’s Oncology Group clinical trials between 1968 – 2009. Adapted from Hunger and Mullighan (2015).

Figure 1.2: Frequency of genetic alterations of paediatric acute lymphoblastic leukaemia. Genetic lesions present in B-ALL are indicated in blue and T-ALL in pink. Adapted from Mullighan (2012).

2 by neutropenia, cytokines released from leukaemic cells or from infection are also common symptoms of ALL (Redaelli et al., 2005; Hunger and Mullighan, 2015).

ALL was historically classified based on leukaemia cell morphology according to the French-American-British (FAB) system proposed in 1976 and reviewed in 1985, which requires examination of peripheral blood and bone marrow smears (Meija-Arangure, 2016), and a diagnosis was established upon bone marrow replacement by 25% or more leukaemic lymphoid blasts (Onciu and Pui, 2012). Presently, immunophenotype and cytogenetic status are additionally determined to allow sub-classification and patient risk stratification.

1.2.1 Immunophenotype

Immunophenotyping, the determination of expression of individual cluster of differentiation (CD) markers, is used to further classify ALL into B- or T-lineage ALL. The CD system identifies cell surface molecules such as ligands, receptors, adhesion and migration molecules, or cytokine receptors, in which the lineage-specific expression profile indicates the differentiation and maturation stages of either B or T cells (Behm, 2012). B-ALL represents approximately 85% of childhood ALL (Lemmers et al., 2000) and typically expresses TdT, CD34, and HLA-DR, CD19 and CD79a. CD expression of CD10, CD22, CD24, and CD20 further categorizes B-ALL into precursor B-ALL, while mature B leukaemia lacks the TdT and CD34 expression, highly expresses CD20 and produces cell surface light chain-restricted immunoglobulin molecules (Onciu and Pui, 2012). T-ALL comprises only 15% of paediatric ALL (Muffly and Larson, 2012), is commonly positive for TdT, variably expresses CD1a, CD2, CD3, CD4, CD5, CD7, and CD8, and may express CD10 (Meija-Arangure, 2016).

1.2.2 Cytogenetics

Approximately 3 in 4 childhood ALL patients carry chromosomal abnormalities (Mullighan, 2012a), including aneuploidy and chromosomal translocations which further sub-classify ALL into groups with different risks of relapse. Aneuploidy, or an abnormal number of chromosomes, is known to correlate with patient prognosis.

Hypoploidy, characterized by less than 46 chromosomes, is associated with poor outcome, particularly the rare near ploidy, where 23 to 29 chromosomes is associated with a very high risk classification with reported 5-year event-free survival (EFS) and 5-year-overall survival of only 40% (Harrison, 2000; Moorman et al., 2010). Up to 30% of paediatric ALL patients exhibit high hyperdiploidy, characterized by chromosome number above 50, which is associated with good outcome. A good prognosis is linked to the gain of chromosome X, 4, 6, 10, 14, 17, 18, and 21, attributed to higher chemosensitivity possibly due to an increased tendency of the blasts to undergo apoptosis (Paulsson et al., 2010; Onciu and Pui, 2012). This group has a reported 6-year EFS of 80 – 99% (Dastugue et al., 2013). Genetic alterations also play a significant role in the cause of ALL and are used as prognostic factors. Genetic changes can affect regulatory processes in leukaemic cells causing unlimited capacity for self-renewal, subverting controls of normal proliferation, blocking cell differentiation and promoting resistance to apoptosis (Pui, 2009). A graphical representation of estimated gene alteration frequencies in paediatric ALL is shown in Figure 1.2. Chromosome rearrangements such as t(12;21) ETV6-RUNX1 (TEL-AML1) and t(1;19) TCF3-PBX1 (E2A-PBX1) in B-ALL, and TAL1 rearrangement, NOTCH1 mutation and SIL-TAL in T-ALL are associated with good patient outcome (Rubnitz and Pui, 1997; Camos and Colomer, 2006; Moorman, 2012). By contrast, t(9;22) BCR-ABL1 (Philadelphia chromosome) and Mixed Lineage Leukaemia (MLL) gene rearrangement at 11q23 in B- ALL and TLX3/HOX11L2 and TLX1/HOX11 MYC rearrangements in T-ALL, correspond to poor prognosis (Szczepanski et al., 2010; Woo et al., 2014; Pui et al., 2015). Recently identified lesions in genes such as IKZF1, which encodes transcriptional regulator of B lymphoid development, and CRLF2, involved in B-cell precursor proliferation and survival, are also found to be associated with poor prognosis (Mullighan et al., 2009; Roberts and Mullighan, 2015; Sutton et al., 2018).

1.3 Treatment of acute lymphoblastic leukaemia

The application of risk-directed therapy has been a major contributor to improvements in the treatment of paediatric ALL. Patients are stratified into risk groups according to clinical and biological features to ensure high-risk patients receive an appropriate

4 intensive treatment protocol, whilst the lower risk cohort can be given less intense therapy, thereby reducing treatment toxicities in these patients.

1.3.1 Risk factors

Besides morphological and genetic presentation discussed above, basic clinical features are also used to stratify patients and determine the treatment regimen for paediatric ALL patients (Cooper and Brown, 2015). The National Cancer Institute (NCI) uniform age and white cell count (WCC) criteria, published in 1996, are still used in some countries to determine patient risk and have helped to improve ALL treatments over the years. Standard risk patients are characterized as having WCC of equal or less than 50,000/μL and age between 1 and 9.99 years, and high risk patients as having WCC of more than 50,000/μL or an age of 10 and above (Smith et al, 1996). Another important risk factor is early response to treatment, which was shown to be the most powerful determinant of prognosis (Borowitz et al., 2008). Response was previously measured using microscopic examination of bone marrow sample during remission to determine the clearance rate of leukaemic cells. Detection of residual leukaemic cells or minimal residual disease (MRD) is presently determined by methods with higher specificity and sensitivity, such as flow cytometry and polymerase chain reaction (PCR)-based techniques, which also allow monitoring of the leukaemic cell population throughout the treatment period. MRD has been shown repeatedly to be a significant and independent prognostic factor in paediatric ALL (Steinherz et al., 1996; Basso et al., 2009; Campana, 2010). Table 1.1 shows commonly used risk factors in childhood ALL, which includes earlier mentioned determinants, as well as additional features such as central nervous system (CNS) involvement of the leukaemia.

Table 1.1: Risk factors in paediatric acute lymphoblastic leukaemia

Risk factor Favourable Unfavourable

Age 1 to 9 10 and above White cell count ≤ 50,000/µL > 50,000/µL Immunophenotype Precursor B-ALL T-ALL CNS leukaemia Absent Overt Testicular enlargement Absent Massive Genetic abnormalities Hyperdiploidy Hypodiploidy > 50 chromosomes < 44 chromosomes DNA Index >1.16 - 1.6 DNA Index <1.16 Genetic abnormalities for ETV-RUNX1 fusion BCR-ABL1 fusion (Ph+) B-ALL E2A-PBX1 fusion MLL gene rearrangement CRLF2 gene rearrangement IKZF1 gene alteration Genetic abnormalities for TAL1 rearrangement TLX3/HOX11L2 rearrangement T-ALL NOTCH1 mutation TLX1/HOX11 rearrangement SIL-TAL fusion MYC rearrangement Prednisone response Good Poor MRD level after < 0.01% ≥ 1% induction therapy

Combinations of these factors are used to classify patients mainly into three treatment groups: standard/low, intermediate and high. Standard risk: precursor B-ALL patients with favourable features listed in the table; high risk: patients with at least one high risk feature including MLL translocation, Ph+ or early T-cell precursor subtypes, poor prednisone response, induction treatment failure or >1% MRD at the end of induction treatment; intermediate risk: patients not classified as standard/low or high risk. CNS: central nervous system; MRD: minimal residual disease; Ph+: Philadelphia chromosome positive

1.3.2 Treatment protocol

The current chemotherapy protocol consists of four components of treatment; remission induction, intensification or consolidation, maintenance and CNS-directed therapy. The aim of the remission induction phase is to eradicate 99% or higher of the initial leukaemia cells and to restore normal performance status and haematopoiesis with absolute granulocyte count <0.5x109/L and platelet count of >100x109/L (Pui et al., 2008; Pui, 2012). The phase usually lasts 4 – 6 weeks and has a complete remission rate of approximately 95 – 98% (Cooper and Brown, 2015; Medina-Sanson, 2016). It typically includes the administration of a glucocorticoid such as prednisone, prednisolone or dexamethasone, vincristine, asparaginase and often an anthracycline such as doxorubicin or daunorubicin. After restoration of normal haematopoiesis, patients undergo the second block of therapy which is consolidation treatment. This phase aims to accelerate the reduction of leukaemia cells and eliminate residual disease, as well as minimize drug resistance by including drugs not used in the first block of therapy such as mercaptopurine, methotrexate, etoposide and cytarabine, which in turn increase long-term benefit (Pui and Evans, 2006; Cooper and Brown, 2015). The consolidation phase lasts around 6 – 9 months but could vary for high-risk patients receiving more aggressive and longer regimens. The protocols for consolidation therapy also vary among the risk groups but have proven to be beneficial (Pizzo and Poplack, 2002; Pui et al., 2008; Pui, 2010; Redaelli et al., 2005).

In maintenance therapy, patients generally receive 1.5 – 2.5 years of treatment with methotrexate and mercaptopurine to reduce residual leukaemic cells and maintain remission by suppressing re-emergence of drug-resistant clones (Stanulla and Schrappe, 2009). In order to prevent significant numbers of patients from relapsing, a prolonged maintenance phase is preferred (Schrappe et al., 2003), and it has been shown to benefit all forms of childhood ALL (Childhood ALL Collaborative Group, 1996; Redaelli et al., 2005). The other treatment element is CNS-directed therapy, which starts during the remission induction phase, and continues into the maintenance phase. The CNS acts as a sanctuary site for leukaemia cells, which become protected from systemic drugs by the blood-brain barrier (Pizzo and Poplack, 2002; Medina-Sanson, 2016). Conventionally, patients undergo cranial irradiation, but due to neurotoxicity and

7 second neoplasms, this has been largely replaced by intrathecal and systemic chemotherapy of methotrexate alone or in combination with other drugs such as cytarabine and hydrocortisone (Pui and Evans, 2006; Stanulla and Schrappe, 2009; Pui, 2012). Patients at high risk of CNS relapse such as those with T-ALL and t(1;19) (TCF3-PBX1), receive a more intensive protocol. The use of CNS-directed therapy has been shown to reduce the rate of CNS relapse from 50% in the 1960s to less than 5% over the past decades (Stanulla and Schrappe, 2009).

Another therapeutic option for childhood ALL is hematopoietic stem cell transplantation (HSCT). Due to intensive therapy associated with the transplantation process and its high risk of side effects, including transplant-related morbidity and graft-versus-host disease, transplantation is reserved for selected very high risk patients whose cancer could not be cured with chemotherapy or those for whom HSCT has a significant potential survival advantage over standard therapy (Pui, 2010; Leung and Pulsipher, 2012). Indications of candidates for HSCT are slow early response including remission induction treatment failure, high MRD at week 10 – 12, and unfavourable cytogenetic status such as Philadelphia chromosome positive (Ph+) or MLL gene rearrangement (Vrooman and Silverman, 2016). An optimal donor is a human leukocyte antigen (HLA)-matched family donor, however allografts from matched unrelated donors can be comparable to sibling donors (Pui and Evans, 1998; Harned and Gaynon, 2008). HSCT was previously shown to improve outcome for Ph+ ALL patients with very poor response to remission induction therapy and infants with hypodiploidy (Pui et al., 2002; Redaelli et al., 2005). However recent data reported no benefit of HSCT in outcome for hypodiploid ALL or MLL-rearranged ALL in infants, and with new therapeutic options such as tyrosine kinase inhibitors, investigators are not recommending transplantation in first remission for Ph+ ALL unless necessary (Pui, 2010; Vrooman and Silverman, 2016). Nevertheless, benefit of HSCT was evident in children with high-risk T-ALL compared to those who received chemotherapy only with 10-year overall survival of 59% and 35%, respectively (Schrappe et al., 2012).

1.3.3 Side effects

Despite the fact that the current improvements in protocols dramatically increased the survival of childhood ALL patients over the past decades, the chemotherapeutics used in these treatments also cause short- and long-term side effects in these children. Common toxicities associated with currently used drugs are hepatotoxicity, cardiotoxicity, nephrotoxicity and pancreatitis (Cooper and Brown, 2015). Some of the drugs have also been reported to cause hypersensitivity reactions, the clinical pattern of which can vary from mild and moderate, with symptoms such as skin rash, urticaria, nausea and diarrhea, to severe, with symptoms such as chest pain, angina pectoris, seizures and anaphylaxis that may be life threatening. These reactions mostly occur within minutes to days after drug administration (Shepherd, 2003; Ruggiero et al., 2013).

Studies of long-term effects of childhood survivors of ALL have grown steadily over the years through comprehensive data collection, particularly by consortia with large numbers of participating institutions. Second neoplasms particularly skin cancers (basal cell carcinoma) and cancers involving the CNS (meningioma) are prevalent in the survivors, with cumulative incidence at 30 years of approximately 17% in survivors who undergone cranial radiation and 7% for non-irradiated individuals (Robison, 2011). As incidence is strongly associated with radiation therapy, its elimination in recent protocols will reduce subsequent cancers in future cohorts (Mody et al., 2008; Robison, 2011). However, the replacement of radiation with intensive systemic and intrathecal treatment leads to a different set of complications. Anthracyclines such as doxorubicin are believed to produce oxygen-free radicals that damage cardiac myocytes, which over time increases cardiac wall stress and decreases myocardial contractility (Adams and Lipschultz, 2005). Cardiac-related effects, including cardiomyopathy, congestive heart failure, coronary artery disease, myocardial infarction and cardiac arrest, are found to be dose-related, and as a result, doses have been lowered in contemporary protocols (Ness et al., 2011). Neurocognitive functions such as working memory, information processing speed and fine motor functioning were reported to be significantly impaired in childhood ALL survivors (Iyer et al., 2015) and likely to be related to intrathecal methotrexate, corticosteroids and vincristine, which have been associated with acute

9 neurotoxicities such as paralysis, behavioural changes and peripheral neuropathy, respectively (Hill et al., 1998; Ness et al., 2011; Iyer et al., 2015). Alkylating agents such as cyclophosphamide have been shown to impact fertility of both male and female cancer survivors who underwent chemotherapy at a young age, affecting the cells, organs and hormones involved in the reproduction system (Schwartz, 1999; Thomas- Teinturier et al., 2015). Due to the gonadal injury, female survivors were less likely to become pregnant and male survivors were less likely to sire a pregnancy (Green et al., 2009; Green et al., 2010; Hudson, 2010). Survivors also have been shown to have abnormal physical growth rate, higher prevalence of chronic fatigue, increased obesity, mental health problems and poor general health compared to their normal siblings or unrelated individuals (Mody et al., 2008; Robison, 2011; Hamre et al., 2013).

It is clear that side effects related to the treatment of ALL impact quality of life of the survivors, especially for patients who received treatments during infancy and childhood period. With greater number of children and adolescents cured of ALL, it is increasingly important to have a strategy in reducing these sequelae. Follow up data and long-term surveillance of survivors allow early intervention and assistance for higher risk individuals such as frequent cancer screenings, hormone replacement therapy, fertility preservation, special education needs, psychological support and counselling to increase their quality of life. It also allows better understanding of their incidence and enables health professionals to refine future treatments to still maintain high cure rate but minimize adverse effects (Hudson, 2011; Robison, 2011). Additionally, it highlights the critical need to develop novel therapies with low toxicities.

1.4 High-risk acute lymphoblastic leukaemia subtypes

Patients with high-risk leukaemia require intensified therapy due to their poor prognosis. New therapeutic strategies are needed for these patients as their outcome is still relatively poor despite intensified therapies (Figure 1.3). However, there have been some recent successes due to the advancement in treatment protocols and the use of novel targeted therapies. Here, four high-risk subtypes of ALL, MLL-rearranged leukaemia, Ph+ ALL, Ph-like ALL and early T-cell precursor (ETP) ALL, are of special interest due of their frequency in paediatric ALL and the severity of the disease subtype.

Figure 1.3: Event-free survival of paediatric ALL patients according to subtypes. (A) Kaplan-Meier curves estimate event-free survival of patients enrolled on the Dutch DCOG and German COALL clinical trials according to disease subtypes. MLL- rearranged ALL, BCR-ABL1-positive (Ph+) ALL and BCR-ABL1-like (Ph-like) ALL patients are shown here with worse survival rates compared to subtypes associated with a good prognosis, such as patients with ETV-RUNX1 translocation. This figure is taken from Boer and den Boer (2017). (B) Kaplan-Meier curves estimate event-free survival of patients with typical T-ALL versus ETP ALL treated on St Jude protocol. ETP ALL patients are shown to have worse prognosis compared to typical T-ALL patients. This figure is taken from Coustan-Smith et al. (2009).

1.4.1 MLL-rearranged leukaemia

Disease background

Rearrangements of the MLL gene, also named Lysine [K]-specific Methyltransferase 2A or KMT2A, on chromosome 11 region q23 occur in up to 10% of childhood ALL patients (Slany, 2009). The rearrangement is also present, at higher frequency of 15 – 23%, in paediatric patients with acute myeloid leukaemia (AML), a haematopoietic malignancy arising from the clonal transformation of myeloid cell precursors with 5- year disease-free survival (DFS) of 65% (Rubnitz et al., 2012). However, in infants younger than 1 year of age suffering from leukaemia, approximately 70% of cases have MLL gene rearrangement (Slany, 2009). The chromosomal change is shown to be an independent prognostic indicator in both MLL-rearranged (MLL-r) ALL and AML (Chowdhury and Brady, 2008), and associated with poor prognosis in both diseases (Pui et al., 2008; Balgobind et al., 2011).

The MLL gene encodes a 500 kDa nuclear protein containing an N-terminal binding domain, a C-terminal SET (Suppressor of variegation, Enhancer of zeste, Trithorax) domain with intrinsic histone methyltransferase activity that specifically methylates histone H3 at lysine 4 (H3K4), and several conserved domains such as plant homeodomain (PHD) finger motifs and a bromodomain (BD) (Zhang et al., 2012). The evolutionarily-conserved H3K4 methylation function of the MLL gene product plays a fundamental role in gene activation (Ansari and Mandal, 2010). For full methyltransferase activity, the MLL protein associates with a complex of proteins including WDR5 which assemble around the SET domain to prepare the chromatin for efficient transcription. As the chromatin unwinds, WDR5 protein will recognize the H3K4, and RBBP5 and ASH2L proteins stabilize the active form of MLL for the methylation process (Slany, 2009) (Figure 1.4A). MLL is structurally and functionally homologous to the Drosophila melanogaster protein trithorax, which is involved in epigenetic regulation of developmental genes such as the homeobox (HOX) family genes (Winters and Bernt, 2017). The MLL protein has been found to be important in maintaining an epigenetic state to allow transcriptional control of HOX genes that

Figure 1.4: MLL-interacting proteins and frequency of MLL translocation in infant and paediatric leukaemia patients. (A) MLL structure and MLL-interacting proteins (MLLn: MLL N terminus; MLLc: MLL C terminus); (B) frequency of MLL translocation in infant and paediatric leukaemia patients according to disease type. Numbers in the pie chart represent percentages of patients with a specific translocation based on a worldwide study of MLL-rearanged leukaemia patients. Various translocation partners are indicated. These figures have been taken partially from Winters and Bernt (2017) and Meyer et al. (2018), respectively.

13 regulate haematopoietic differentiation. Reciprocal translocation of the MLL gene with one of a number of different partner genes results in the expression of two fusion proteins, one containing the MLL N-terminus fused to the C terminal region of the partner protein and a reciprocal fusion between the C-terminus of MLL and the N terminus of the fusion partner. The expression of various MLL fusion allele leads to dysregulated expression of the HOX family genes such as HOXA7 and HOXA9, and the HOX cofactor MEIS1, contributing to the stem cell characteristic of MLL-r leukaemogenesis by maintaining self-renewal properties and survival advantages of those cells (Balgobind et al., 2011; Grembecka et al., 2012; Marschalek, 2016; Winters and Bernt, 2017). This distinct gene expression signature of MLL-r leukaemia driven by aberrant function of MLL fusion proteins indicates a unique subset that could be targeted therapeutically (Armstrong et al., 2002).

An international multi-centre genomic study of the MLL recombinome has shown the extent of heterogeneity of MLL-r leukaemia, with the finding of 94 translocation partner genes in 2345 paediatric and adult MLL-r leukaemia patients around the world. However, nine predominant specific gene fusions account for more than 90% of the recombinations, with the most common being MLL-AF4 t(4;11) (MLL-AFF1 or MLLT2), MLL-AF9 t(9;11) (MLL-MLLT3), MLL-ENL t(11;19) (MLL-MLLT1), MLL- AF10 t(10;11) (MLL-MLLT10), MLL-ELL t(11;19)(q23,p13.1) and MLL-AF6 t(6;11) (MLL-MLLT4) (Meyer et al., 2018). The frequency of each fusion varies with age (infant vs. older children) and disease type (ALL vs. AML) (Figure 1.4B). Among the prevalent fusion proteins, MLL-AF4 and MLL-ENL are most commonly associated with ALL, while the others are predominantly associated with an AML disease phenotype (Marschalek, 2010a; Meyer et al., 2018).

MLL fusion partners have been shown to be crucial in leukaemogenesis. The most common fusion partners belong to protein network involved in histone binding, interacting with Disruptor of Telomeric Silencing 1-like (DOT1L), a histone methyltransferase, and the positive transcription elongation factor (pTEFb), a heterodimeric protein complex of CyclinT1/2 and CDK9 that phosphorylates RNA polymerase II for efficient transcriptional elongation (Marschalek, 2010b). The fusion proteins recruit DOT1L to aberrant gene locations, enhancing the methylation of

H3K79, leading to aberrant expression of leukaemogenic HOXA genes and their cofactors, required for development and maintenance of MLL-fusion-mediated leukaemogenesis (Bernt et al., 2011). In several animal models, lack of DOT1L resulted in a complete loss of H3K79 methylation, which in turn downregulated HOXA9 and MEIS1 genes and reversed transformation in MLL-rearranged cells (Chen and Armstrong, 2015; Wang et al., 2016). Malignant transformation of leukaemia cells dependent on DOT1L has also been reported in cells carrying the CALM-AF10 fusion gene (Chen et al., 2013). The fusion between the Clathrin Assembly Lymphoid Myeloid (CALM) gene, also located on chromosome 11q23 as the MLL gene, and a known MLL partner gene, AF10 (MLLT10), located on chromosome 10p12, results in an aggressive leukaemia subtype that confers poor prognosis and is hard to treat (Imamura et al., 2002; Caudell and Aplan, 2008; Greif et al., 2008; Savage et al., 2010). The shared MLL–like aetiology was also demonstrated in the aberrant expression of MEIS1 and HOX cluster genes (Caudell and Aplan, 2008), further emphasizing the significance of DOT1L and HOX family genes as common driver of MLL-r and CALM-AF10 leukaemogenesis.

Therapeutic strategies

In contrast to the improvements in survival rates for paediatric leukaemia patients without MLL gene rearrangement, barely 50% of infants with MLL rearrangement survive five years after diagnosis (Hilden et al., 2006; Slany, 2009; Kotecha et al., 2014). To improve the survival of MLL-r leukaemia patients, a hybrid regimen based on the standard protocol of ALL and AML, including a more intensive chemotherapy course, was tested in 482 infants with MLL-r leukaemia enrolled on the Interfant-99 clinical trial over a period of six years. Despite the more aggressive treatment, survival rates did not significantly increase (Pieters et al., 2007). The role of HSCT continues to be under debate, with a few studies showing no benefit in transplantation for MLL-r patients (Pui et al., 2002; Winters and Bernt, 2017). However in the Interfant-99 cohort, very poor prognosis infants younger than 6 months who underwent HSCT displayed an increased survival of 59% compared to 22% in those receiving chemotherapy only (Mann et al., 2010). The heterogeneity of the disease as demonstrated by the diverse patient outcome in clinical trials (Pui et al., 2003), and poor survival rates associated

15 with the disease indicate the need for new targeted therapies. Current conventional therapies for infant ALL have been extensively modified and intensified, however the maximum survival rate has been reached at the cost of toxicities (Hilden et al., 2006; Pieters et al., 2007).

Several MLL-selective drugs were recently developed through targeted drug design and high-throughput screening of novel compounds, including a highly potent small molecule DOT1L inhibitor, EPZ5676 (Pinometostat), which selectively kills cells bearing MLL or CALM-AF10 translocations (Daigle et al., 2013). The compound subsequently entered phase I clinical trial testing for paediatric and adult patients with relapsed/refractory MLL-rearranged leukaemia. However the limited clinical responses reported so far and continued disease progression in the adult clinical trial (Stein and Tallman, 2015) suggested emergence of treatment resistance in these patients (Campbell et al., 2017). High expression of Fms-like receptor tyrosine kinase-3 (FLT3) is found be very common in MLL-r leukaemia patients (Armstrong et al., 2002) and prognostic of outcome (Meshinchi et al., 2001), and hence FLT3 inhibitors have been suggested as a therapeutic approach. However, although the FLT3 inhibitor midostaurin showed potential in a phase-I/II trial of adult AML (Balgobind et al., 2011), in paediatric patients, the drug demonstrated limited single-agent activity and was recommended to be used in combination with established chemotherapies (Zwaan et al., 2015). Inhibition of histone deacetylase (HDAC), a chromatin-modifying enzyme, was highlighted by gene expression profiling studies as a potential target in MLL-r ALL (Stumpel et al., 2012) and has been shown to inhibit MLL-r cells in vivo (Stubbs et al., 2015) and in one MLL-r ALL adult patient (Burbury et al., 2011). The HDAC inhibitor vorinostat entered clinical trial for patients with refractory or relapsed MLL-r leukaemia, aged 1 – 21, together with proteasome inhibitor, bortezomib, which itself was shown to be active in preclinical studies of MLL-r leukaemia (Cruickshank et al., 2017), in combination with established chemotherapy agents. However, the clinical trial was recently terminated, due to slow patient accrual (NCT02419755). Nevertheless, another HDAC inhibitor, panobinostat, showed promising results in an adult MLL-r patient and in a high-risk paediatric AML clinical trial (Burbury et al., 2011; Bug et al., 2017). The drug is currently under investigation in various leukaemia subtypes and further analysis will determine its efficacy in MLL-r paediatric patients.

Besides chemotherapy, blinatumomab, a bispecific T-cell engager (BiTE) monoclonal antibody, has been shown, as a single agent, to be an effective therapeutic option for relapsed and refractory ALL in children (von Stackelberg et al., 2016) and adults (Kantarjian et al., 2017). The antibody construct binds to CD3-positive T cells and CD19-positive B cells, inducing cell lysis by directing the T cells to target commonly prevalent CD19-positive ALL blasts (Kantarjian et al., 2017). In a case report of an infant with MLL-r ALL, treatment with the antibody demonstrated a lineage switch to AML and the patient achieved complete remission with myeloid-directed chemotherapy (Rayes et al., 2016). Blinatumomab-induced lineage switch has also been reported in adult leukaemia patients with MLL-rearrangement (Duffner et al., 2016; Haddox et al., 2017), however it is currently unknown whether this patient cohort is at increased risk for lineage switch. Therefore a lineage independent marker, such as quantitative PCR for MLL-AF4, is recommended for patients receiving blinatumomab treatment (Duffner et al., 2016). A few months ago, a phase II clinical trial was approved to study the feasibility, safety and efficacy of this antibody for infants with MLL-r ALL in combination with the current Interfant-06 backbone (EudraCT: 2016-004674-17). Nevertheless, a current lack of highly efficacious treatment for this heterogenous disease warrants development of therapeutic agents with better specificity towards each MLL-r sub-group.

1.4.2 Philadelphia-positive acute lymphoblastic leukaemia

Disease background

Approximately 3 – 5% of children with ALL harbour the t(9;22) Philadelphia chromosome, whereby a reciprocal translocation occurs between chromosomes 9 and 22, joining the 3’ portion of the tyrosine kinase (ABL1) proto-oncogene on the long arm of chromosome 9 to the breakpoint cluster region (BCR) gene on chromosome 22 (Harrison, 2000). Two transcripts have been described as the most common in leukaemia, a major 210 kDa protein (p210), predominantly found in chronic myelogenous leukaemia (CML), and a smaller 190 kDa transcript (p190), which 90% of childhood Ph+ B-ALL patients express (Mishra et al, 2006; Teitell and Pandolfi, 2009).

The loss of the N terminus of the ABL1, which negatively regulates ABL1 kinase activity has been shown to be a major contributor to leukaemogenesis (Hantschel, 2012; Bernt and Hunger, 2014). The BCR-ABL1 fusion protein has a constitutively active tyrosine kinase activity which affects several signaling pathways leading to haematopoietic cell transformation, contributing to leukaemia cell self-renewal, proliferation and growth, as well as reducing apoptosis and differentiation (Advani and Pendergast, 2002; Koo, 2011; Mullighan, 2012b; Forghieri et al., 2015). Besides BCR- ABL1 fusion gene, approximately 70 – 80% Ph+ ALL patients also displayed somatic mutations of the IKZF1 gene, which encodes early transcription factor Ikaros, impairing lymphoid maturation (Bernt and Hunger, 2014). Deletion of part or all of the IKZF1 gene results in alterations such as haploinsufficiency or expression of dominant negative isoforms that cooperate with BCR-ABL1, contributing to HSC-like properties of leukaemic cells such as enhanced self-renewal and promoting resistance to therapy (Mullighan, 2012b; Roberts and Mullighan, 2015). In adult Ph+ ALL patients, IKZF1 alterations are shown to be negatively associated with DFS (Martinelli et al., 2009).

Therapeutic strategies

Ph+ ALL patients were previously reported to have very poor prognosis with high chemotherapy resistance, and therefore relied heavily on HSCT as therapy, with one study showing that the 5-year DFS was 65% for patients undergoing HSCT compared to only 25% for patients receiving chemotherapy alone (Pui, 1995; Arico et al., 2000; Bernt and Hunger, 2014). The development of tyrosine kinase inhibitor (TKI), imatinib mesylate, which inhibits BCR-ABL1 and other aberrant tyrosine kinases has revolutionized the treatment and outcome of these patients (Pui, 2010; Forghieri et al., 2015). The drug, approved by the United States of America (USA) Food and Drug Administration (FDA) in 2000 for the therapy of CML patients, acts as a competitive inhibitor to the adenosine triphosphate (ATP) binding site on the kinase domain, preventing the oncogenic protein from switching to its activated form (Koo, 2011; Hantschel, 2012). Administration of imatinib in addition to intensive chemotherapy resulted in a 5-year DFS of 70% compared to 65% in patients receiving sibling donor HSCT and 59% in patients receiving unrelated donor HSCT. The follow-up study showed no advantage for HSCT and confirmed the good outcome achieved through

18 administration of imatinib together with intensive chemotherapy (Schultz et al., 2009; Schultz et al., 2014).

However, resistance to TKI has been reported, which was identified to arise from mutation in the BCR-ABL1 ATP-binding domain, preventing the binding of imatinib, and overexpression of BCR-ABL1 protein (Mishra et al., 2006). Fortunately, most mutant BCR-ABL1 proteins are still sensitive to second generation TKIs, dasatinib and nilotinib. Dasatinib was shown to have greater potency than imatinib, with activity in relapsed or resistant Ph+ ALL, and the ability to penetrate the CNS (Koo, 2011). In adult CML patients, dasatinib was superior to imatinib, providing 2-year DFS of 64% compared to 46% for imatinib (Kantarjian et al., 2012). In adult Ph+ ALL, 96% of patients receiving dasatinib in combination with chemotherapy achieved remission and the estimated 2-year survival rate was 64% (Ravandi et al., 2010). A follow-up study demonstrated long-term survival was achievable and treatment could be better tailored to the patient to avoid toxicities (Ravandi et al., 2015). Another second generation TKI, nilotinib has also been shown to have activity in imatinib-resistant ALL harbouring a BCR-ABL1 mutation (Koo, 2011), and had a 72% 2-year overall survival rate in Ph+ ALL subjects aged between 17 – 71 who received nilotinib in addition to other chemotherapy agents or HSCT (Kim et al., 2015). These studies demonstrate the advantages of including TKIs into standard regimens for paediatric Ph+ ALL, thereby producing comparable survival rates to HSCT, which has been shown to cause toxicities and long-term impacts from chronic graft-versus-host disease. However, optimization of therapy and further understanding of the mechanism of TKIs are still required to decrease side effects and facilitate development of improved targeted therapies.

1.4.3 Philadelphia-like acute lymphoblastic leukaemia

Disease background

In a large genome-wide study of high-risk leukaemia using single-nucleotide– polymorphism microarrays and transcriptional profiling, a new disease subtype was identified as exhibiting a gene expression profile similar to Ph+ ALL but lacking the BCR-ABL1 fusion protein (Mullighan et al., 2009). This subtype is more common in

19 adolescent and young adults with a frequency of more than 25% compared to only 10 – 15% in younger children (Roberts and Mullighan, 2015). Philadelphia-like (Ph-like) ALL patients frequently have altered IKZF1 gene similar to Ph+ ALL, which was also shown to be prognostic of outcome in BCR-ABL1-negative ALL patients. A retrospective study reported approximately 50% of IKZF1-mutated patients relapsed compared to 25% of patients with wild-type IKZF1 (Mullighan et al., 2009). Besides this alteration, approximately 1 in 2 Ph-like ALL patients also harbour abnormalities that dysregulate expression of the cytokine receptor gene CRLF2, which are associated with activating mutations in the Janus kinase genes, JAK1 and JAK2, suggesting co- expression of CRLF2 and JAK1/2 mutations are important in lymphoid transformation (Mullighan, 2012a; Heatley et al., 2017). Non-Down syndrome childhood ALL patients with dysregulated CRLF2 expression were shown to have inferior outcome compared to other patients (Ensor et al., 2011; Roberts et al., 2013). Ph-like ALL patients have also been shown to have ABL-activating lesions that are structurally similar to BCR-ABL1 such as the ETV6-ABL1 and NUP214-ABL1 aberrations on other transcription factors such as TCF3, EBF1, PAX5 or VPREB1 (den Boer et al., 2009; Boer and den Boer, 2017). Leukaemogenesis in a further subset of patients was suggested to result from tyrosine kinase signaling activation due to fusions involving various other tyrosine kinase genes (being similar to BCR-ABL1) or from overexpression of CRLF2 (Boer and den Boer, 2017).

Therapeutic strategies

Studies have shown that when treated under standard protocols, Ph-like ALL patients have a higher relapse rate (37%) than those with non-Ph-like ALL (16%). These patients also have unfavourable 5-year DFS of 59.5% compared to 84.4% for other precursor-B ALL (den Boer et al., 2009). The presence of ABL-activating lesions and JAK mutations, confers sensitivity to the TKIs, imatinib and dasatinib, and the dual JAK1/2 inhibitor, ruxolitinib preclinically. Preliminary clinical data demonstrated potential for these inhibitors in Ph-like patients, however larger clinical trials of the inhibitors in combination with standard chemotherapy, are still ongoing to confirm their efficacy (Roberts et al., 2014b, Inaba et al., 2017; Tasian et al., 2017; Wells et al., 2017).

1.4.4 Early T-cell precursor acute lymphoblastic leukaemia

Disease background

Early T-cell precursor (ETP) ALL is a form of T-ALL, characterized by early arrest in T cell differentiation at a stage similar to HSC and myeloid progenitors (Haydu and Ferrando, 2013). This ALL subgroup has a distinct gene expression profile of cytoplasmic CD3+, absence of CD1a and CD8, reduction in CD5, and expression of at least one myeloid or stem cell marker such as CD13 or CD34 (Coustan-Smith et al., 2009). Whole genome sequencing of ETP ALL cases demonstrated the heterogeneity of the disease, with 67% patients having activating mutations in genes regulating cytokine receptor and RAS signaling, such as KRAS, FLT3, JAK1 and BRAF3, as well as loss of function or dominant negative translocations, deletions and mutations in haematopoietic and lymphoid development regulators such as RUNX1, IKZF1 and ETV6 (Zhang et al., 2012, Roberts and Mullighan, 2015). The expression profile resembled HSC and granulocyte macrophage lineages, as well as a stem-like signature associated with poor outcome in acute myeloid leukaemia and a signature associated with inferior outcome in IKZF1-mutated high-risk B-progenitor ALL, all consistent with its aggressive, poorly differentiated stem cell-like leukaemia characteristics (Zhang et al., 2012).

Therapeutic strategies

Despite transplantation, ETP ALL patients have extremely poor prognosis. Comparison of ETP and typical ALL patients in two patient cohorts revealed higher rates of remission failure or haematological relapse in ETP ALL patients with 72% compared to 10% at 10 years in children enrolled at the St Jude Children’s Research Hospital, and 57% compared to 14% at 2 years in patients treated in the AIEOP trial (Coustan-Smith et al., 2009). This finding has been confirmed by others in different paediatric patient cohorts, as well as adolescents and adults (Ma et al., 2012; Neumann et al., 2012; Jain et al. 2016) High-dose dexamethasone is being tested in this patient cohort due to improved survival seen in T-ALL patients enrolled on the AIEOP/BFM ALL 2000 study (Pui 2012). Given that ETP ALL patients displayed JAK mutations, the efficacy of JAK inhibitor ruxolitinib was assessed preclinically, showing a decreased peripheral

21 and splenic blast count in vivo, which could potentially extend to patients (Maude et al., 2015). Myeloid-directed therapies have also been suggested to improve patient outcome due to the myeloid-like expression signature of the disease. AML protocols have been tested in a limited number of patients. In a study of adult ETP ALL patients, three patients were treated with an AML-like protocol, however it was not stated whether it resulted in better outcome compared to patient treated with the German Acute Lymphoblastic Leukemia Multicenter Study Group (GMALL)-like protocol (Neumann et al., 2013). In one paediatric case, the ETP ALL patient failed to respond to the AML protocol (Ma et al., 2012). To date the lack of effective therapy indicates a need for novel therapeutics in improving survival of these patients.

1.5 Conventional drug discovery pipeline

The process of drug discovery typically begins with a drug screen consisting of an appropriate assay and readout. Due to advances in drug screening technology and equipment over the past two decades, huge numbers of drugs can now be tested simultaneously through a high-throughput technology. For example, robotics and automation enable 10,000 compounds from chemical libraries to be screened per day in a 96-well microplate format. The technology has recently evolved to use miniaturized assay formats with 384- and 1536-well microplates, which reduces assay sample amount up to 20-fold, and allows more than 100,000 compounds to be tested per day in ultra-high-throughput screening (Mayr and Bojanic, 2009). With assays that are robust, reliable reproducible, simple and relevant (Young, 2017), and improved chemical libraries focused on compounds with drug- and lead-like properties such as stability and bioavailability, high-throughput screening (HTS) is a major driver of drug discovery.

Two categories of assays utilized in drug discovery are target-based biochemical assays and phenotype-based cellular assays. Biochemical assays measure the function of a purified target such as enzyme inhibition, protein activities and receptor-ligand binding (Sundberg, 2000; Hughes et al., 2011), whereas cell-based screening measures functionally and biologically relevant cellular activities such as viability, proliferation and apoptosis, and hence has been frequently used in cancer drug discovery (Zang et al., 2012). In a cell-based HTS, the large number of initial hits typically obtained is

22 filtered and refined by a broader array of tests against a larger cell panel to validate and confirm activity of hit compounds in relevant tumour cells. By the end of testing, the small number of hits left, called leads, will be characterized and optimized through an extensive series of in vitro assays such as cytotoxicity, proliferation and migration assays, and gene expression and transcriptional profiling, to determine mechanism of action, genomic and proteomic effects in target cells, pharmacokinetics and pharmacodynamics of the compound and potential synergy with other drugs (Figure 1.5). It has been shown that many newly identified drugs have low clinical potency, thus limiting their clinical application as a single agent (Pui, 2010). Synergy with either currently used drugs in patient therapy or a second compound targeting a different signaling pathway or protein can enhance efficacy, reduce concentration of individual drugs and hence increase the chance of successful advancement into the clinic (Sun et al., 2016). Structural and chemical assessment of leads such as impurity identification and evaluation of solubility and stability will also be established to determine the need for physicochemical optimization.

Once mechanism profiling and chemical/structural optimization of a candidate are completed, the lead candidate will undergo a program of pre-clinical testing to determine efficacy in relevant pre-clinical animal models, establish in vivo absorption, distribution, metabolism, excrement and toxicity (ADMET) data and select an appropriate biomarker or clinical endpoint for clinical efficacy assessment (Li and Jones, 2012) (Figure 1.5). The selection of animal model depends on the type of disease and the type of drug selected. Testing of cancer drugs typically employs mouse models such as patient-derived xenograft models, where patient-derived cells are injected subcutaneously or intravenously to transplant mice with tumour material, or genetically engineered mouse models, whereby mice are engineered to carry specific human oncogenes (transgenic) or remove tumour suppressor genes (knockout) (Wartha et al., 2014). For ADMET data, route of administration, bioavailability, clearance and distribution, influenced by previous in vitro pharmacokinetic and pharmacodynamics data, are determined to formulate initial prescription, dose and maintenance treatment plan (Doogue and Polasek, 2013). Once therapeutic potential, safety and mechanism of drug potency are established, drug candidates will enter the first of three phases of clinical trials (Figure 1.5). Phase I is a

‘first in human’ investigation performed in a small group of patients with advanced disease that is resistant to current therapy, with the goal being to assess the safety, tolerability, pharmacokinetics and pharmacodynamics of the drug in humans. This phase is designed to determine maximal tolerated dose and severity of adverse effects/toxicities but can also give an indication of potential efficacy (Jeha, 2012). A review of phase I studies found the overall response rate in 460 trials was only 4.4% for a novel single agent, 11.7% for a combination of novel drugs and 16.4% for a combination with an FDA-approved drug (Horstmann et al., 2005), indicating the complexity of performing a clinical trial and the low accuracy of prediction of drug activity in patients based on preclinical testing. If a drug candidate is deemed to be tolerable and promising, a phase II trial will be undertaken to assess preliminary efficacy in 40 – 60 patients. If a drug is shown to have sufficient efficacy, results will be validated and confirmed in a phase III study to determine its advantage compared to standard therapy (Jeha, 2012). If the drug demonstrates promising efficacy at the end of all three phases, an Investigational New Drug (IND) application will be submitted to the regulatory body, which then reviews its safety and efficacy. If enough evidence is observed, a New Drug Application (NDA) is put forward and if approved, the drug could be marketed (Dickson and Gagnon, 2004).

Overall, this conventional approach of drug discovery has been shown to yield many new therapeutics over the past few decades, such as imatinib for the treatment of Ph+ ALL, and improve the survival of patients with some malignancies. However, with the rising costs and the long drug development and approval process, more and more efforts are being made towards repurposing existing drugs for new diseases, which is discussed in the next section.

Drug discovery and hit refinement - High-throughput screening of chemical libraries - Broad array of tests in other cell panels to select leads

Lead optimization

- Medicinal chemistry years

- Combination assays

6 -

- Gene expression and proteomics 3 - In vitro pharmacokinetics and pharmacodynamics

Pre-clinical development - Animal testing - ADMET

- Biomarker

Clinical trials

years

- Phase I: First time in man - - Phase II: Proof of principle 5

- Phase III: Proof of concept in larger patient cohort

years

Review from regulatory body

2 -

- New Drug Application (NDA) application and approval 1

Figure 1.5: Schematic diagram of phases of drug discovery and development. Hits obtained from a drug screen will undergo a series of assays to identify lead compounds, which are then characterized and optimized to determine their mechanism of action and achieve optimal physicochemical attributes. This is followed by preclinical testing to establish efficacy in animal models and gather ADMET data. A compound with therapeutic potential will enter three phases of clinical trials and if deemed promising, NDA will be submitted for approval by a regulatory body. ADMET: absorption, distribution, metabolism, excrement and toxicity.

1.6 Drug repurposing

1.6.1 Introduction to drug repurposing

As described previously, the traditional drug discovery pipeline typically begins with de novo identification of a new chemical entity, followed by its validation, development, manufacturing and clinical application. Advancing novel therapeutic agents requires extensive testing in vitro and in preclinical animal models followed by clinical trials in order to meet necessary requirements regarding safety and efficacy (Sukhai et al., 2011). Despite the fact that this approach has resulted in advancements of therapies for a wide range of diseases, it has recently been shown to be inefficient and uneconomical. The development of a pharmaceutical drug containing a novel active ingredient is estimated to cost around US$1.78 billion (Paul et al., 2010) and takes an average of 14 years from discovery to market in the USA and European Union countries (Pammolli et al., 2011). Despite billions of dollars being invested in research and development by pharmaceutical companies and governments worldwide, the proportion of drugs approved for clinical use by regulatory bodies has remained stagnant over the past 40 years (Shim and Liu, 2014) with figures ranging from 3 to 8% (Williams, 2011).

Drug repositioning or repurposing is one of the new therapeutic approaches that overcome some of the hurdles of the traditional approach in the drug discovery process. The terminology refers to the identification of new therapeutics from a pool of existing drugs for application in the treatment of diseases other than those for which they were intended. This process typically starts with drugs that have been tested in humans with an acceptable level of safety and tolerability to avoid derailment due to toxicities not predicted by preclinical studies (Strittmatter, 2014) potentially resulting in more rapid drug approvals. Repositioning approved drugs is thought to be the most efficient approach to therapeutic development as these drugs have established formulation and manufacturing methods, extensive ADMET data, clinical trial safety endpoint and post- marketing surveillance safety (Li and Jones, 2012). Through bypassing time-consuming development, optimization and ADMET steps, the drug discovery timeline could be reduced to 3 – 12 years (Ashburn and Thor, 2004) compared to the conventional drug discovery pipeline with a timeline of 9 – 15 years (Figure 1.5).

Even though limited by the number of existing drugs and their derivatives, drug repurposing has found success in heavily invested, ‘more profitable’ diseases such as cancer and HIV infection, and shown to be one of the few viable options in discovering new therapeutics and drug class for very rare, orphan diseases, or neglected diseases such as Plasmodium spp. infection (Ekins et al., 2011). One of the earliest repurposed drugs for cancer therapy is thalidomide, a sedative developed by German company Grunenthal in 1957 to alleviate morning sickness in pregnant women. Shortly after being introduced in the market, the drug was found to cause birth defects if taken during the first trimester of pregnancy (Ekins et al., 2011), thus immediately withdrawn by manufacturers. However recently, the drug was found to inhibit angiogenesis and possess anti-cancer activity in several types of cancers. It received the FDA approval for the treatment of multiple myeloma in combination with dexamethasone in 2006 (Shim and Liu, 2014).

1.6.2 Strategies of drug repurposing

After a decade of experience in repurposing drugs, several strategies have been put forward to enhance the efficiency of this approach. Li and Jones (2012) summarized six avenues for drug repositioning (Figure 1.6). Path 1 describes the identification of drugs through serendipitous observations, such as in the case of thalidomide. Three years after its ban, a clinician at the Hadassah University Hospital, Israel administered the drug to a critically ill patient with erythema nodosum leprosum in an attempt to relieve his pain. He found that thalidomide was effective in controlling the symptoms of the disease. In depth research and clinical trials that followed showed that thalidomide’s activity against leprosy is through inhibition of tumour necrosis factor alpha. The drug has since received FDA approval and is widely used in treatment of erythema nodosum leprosum (Gan et al., 2011). In path 2 and 3, repositioning opportunities are detected through traditional drug discovery strategies and HTS where existing drugs inhibit another disease phenotype or target, respectively. As an example of path 2, nelfinavir, a competitive inhibitor of human immunodeficiency virus (HIV) aspartyl protease and FDA-approved drug for the treatment of HIV infection (Moyle et al., 1998), was recently reported to have anti-cancer activity in in vivo models for several cancers

Figure 1.6: Potential avenues of drug repurposing. Six potential avenues of drug repurposing. This figure is taken from Li and Jones (2012).

28 including multiple myeloma, non-small cell lung cancer and breast cancer. To date, it has entered more than 20 phase I or II clinical trials for various cancers (Shim and Liu, 2014). Taking path 3, imatinib, originally developed for treatment of CML by inhibiting fusion protein BCR-ABL, was also found to inhibit v-kit oncogene homolog (KIT) and platelet-derived growth factor receptors (PDGFRs). This mutation drives proliferation of gastrointestinal stromal tumour (GIST). Imatinib later received FDA approval in 2008 for GIST treatment (Druker, 2004).

Repositioning can also occur when the target protein or target pathway is discovered to have other roles (path 4 and 5, respectively). Duloxetine, a serotonin and norepinephrine reuptake inhibitor, was originally developed to treat depression. However the serotonin and norepinephrine signaling pathways were also found to be involved in spinal cord activation of the external urethral sphincter in 1998, which led to marketing of the drug for stress urinary incontinence (SUI) (Schuessler, 2006). The drug was then approved in Europe for SUI in 2004, as well as fibromyalgia in 2008 and chronic musculoskeletal pain in 2010 when the same signaling pathways were identified to be key in both diseases (Li and Jones, 2012). In path 6, side effects observed in clinical trials leads to repositioning candidates. The most famous drugs of this category are Sildenafil (Viagra) and Minoxidil which were both developed for hypertension but repurposed for erectile dysfunction and hair loss respectively (Bradley, 2005).

1.6.3 Methods in discovering potential drugs for novel activity

Two approaches are typically taken to discover novel activities. The first strategy encompasses the experimental approach, usually entailing HTS of existing FDA- approved drugs and pharmacologically active compounds. The second approach is based on computational or in silico analyses, which involves hypothesizing drug suitability and efficacy for a specific type of disease through available information on the molecules and targets such as compatibilities and similarities in chemical structures and targets.

Experimental approach

The most widely used experimental strategy to identify novel drug molecules is based on phenotypic screening, whereby drugs are added to cells of interest and effect of the drug is determined. Drugs with the ability to induce the desired effect, such as cell death or reduction in proliferation, will then be further analyzed. With the advancement in HTS, drug libraries containing hundreds or thousands of drug molecules can be efficiently screened within a few days. The screening can be done in a ‘blinded’ manner whereby pharmacology and biological data are not taken into account, giving a flexibility for wider disease application. In a period of 10 years between 1999 and 2009, such screening methods identified approximately 34% of FDA-approved compounds (McCabe et al., 2015). A number of chemical libraries are available from both commercial and public sectors, such as Prestwick by French Prestwick Chemical Co., composed of 1120 off-patent medicines whereby more than 85% are clinical pharmaceutics (Gan et al., 2011) and the National Institute of Health (NIH) Chemical Genomic Centre (NCGC) Pharmaceutical Collection (NPC) library consisting 2400 small molecules approved for clinical use in the USA, European Union, Japan and Canada (Shim and Liu, 2014).

Target-based screening is an alternative approach in which researchers utilize the knowledge about the disease pathogenesis and screen libraries that contain disease- or target/molecule-specific compounds. Drug repurposing for cancers with known targets such as tyrosine kinase for CML and non-small cell lung carcinoma where mutations cause increased receptor tyrosine kinase activity leading to cell proliferation (Paul and Mukhopadhyay, 2004), could be achieved by screening available tyrosine kinase inhibitor libraries. For brain cancer, there are several specialized drug libraries for compounds capable of crossing the blood-brain barrier such as the CNS- Penetrant Compound Library (MedChem Express), where bioactive CNS-penetrating molecules targeting kinases, ion channels and others were assembled in a screenable format. The added advantage of target-based screening is its higher efficacy and lower likelihood of toxicity as aberrations are more likely to be cancer specific (Gan et al., 2011).

In silico analysis

Another popular approach to identify potential repurposing candidates is based on the use of in silico analysis whereby bioinformatics software is used for analyzing gene expression profiles, database mining for drug side effects and docking studies (Sukhai et al., 2011). One way to detect repositioning opportunities is through a similarity-based approach, which measures similarities in chemical space, genomic space or clinical knowledge space (Bisgin et al., 2012). The most well-known in silico analysis method is based on the matching of gene expression signatures between diseases and comparing molecular origin profiles of different diseases to pinpoint shared aetiology. Examination of relevant changes in expression based on drug-target protein binding profiles might reveal shared targets, potentially leading to treatments targeting the aberrant biological system (Wu et al., 2012). Bioinformaticians could also further construct disease- specific signaling pathways that predict potential drug efficacy (Jin and Wong 2014). A successful example of this approach was reported for AML. The gene expression signature of the disease was determined by comparing differentiation-related genes such as the autosomal chronic granulomatous disease-associated gene, NCF1, in bone marrow samples from patients and healthy donors. A gene-expression-based small molecule screen was performed with an AML cell line, HL-60 found epidermal growth factor receptor (EGFR) kinase inhibitor, DAPH1, to induce morphologic, biochemical and functional changes that indicate myeloid maturation (Stegmaier et al., 2004). However, since DAPH1 had not been clinically developed, studies were performed on an FDA-approved EGRF inhibitor, geftinib and it was discovered to induce myeloid differentiation and inhibit proliferation of AML cells (Stegmaier et al., 2005). The subsequent preclinical testing led to the advancement of geftinib into clinical trial for relapsed or refractory AML patients (NCT00130702).

Molecular docking, a study of drug target against target molecules with similar chemical structure, is also a popular in silico method in finding repurposing candidates. Databases of two- and three-dimensional structures of compounds are available to determine structural similarities for predicting drug reactivity in specific diseases. This strategy goes with the premise that structurally similar molecules could have the pharmacophore features that determine bioactivity, allowing virtual screening of

31 chemical libraries against targets of interest or active sites (Ekins et al., 2011; Li and Jones, 2012).

Both the experimental and in silico approaches to identify repurposing opportunities have unique advantages and disadvantages. The experimental approach has a lower rate of false results and is easy to validate, however is time and labour consuming. In silico analysis on the other hand is quick and labour efficient, and initially does not require the development of a screening assay. It does however require a huge amount of background information on the disease, drug structure, molecular profiles, has a higher rate of false results (Shim and Liu, 2014). Table 1.2 compares the two approaches.

1.6.4 Considerations for drug repurposing

The most important advantages of drug repurposing comprise lowering the cost of drug development and facilitating more rapid progression to clinical trial testing. Besides these, it also offers the opportunity to position less toxic or non-cancer drugs for anti- cancer therapy that could potentially increase the patient’s quality of life (Shim and Liu, 2014). Despite these benefits, there are several issues that should be taken into consideration when using this avenue.

Some of the limitations of drug repurposing are similar to those of the traditional drug discovery pipeline. HTS of drug libraries could result in false positive and negative results, however this could be minimized by introducing secondary screens and follow- up experiments, the use of more experimental replicates and including more than one cell line (or counter screening) during testing (Li and Jones, 2012). Strict statistical analyses should also be performed to reduce any inaccuracies or misleading results. A challenge associated with the in silico method of drug repurposing include integration of multiple data profiles from different experiments. For example, microarray data are not measured in absolute units and improper handling of gene profiles from different experimental settings will not correctly analyze similarities between the biological states of the experiments (Lander, 1999). Also, cells from different diseases can respond very differently to the same treatment. RNA-sequencing technology could overcome

Table 1.2: Comparison between experimental and in silico approaches in identifying potential drugs for repurposing. Approach Advantages Disadvantages Experimental - Low rate of false positives - Time and labour-consuming - Hits could be easily - Requires screening assay validated development

In silico - Time and labour efficient - High rate of false results - Does not require - Requires structural development of screening information of target proteins assays and information on drug- induced phenotype

33 these problems as it can detect RNA levels over a wider dynamic range (Iorio et al.,2013). To ensure successful drug repurposing, publicly available microarray or even RNA-sequencing data should be integrated with knowledge of drug mechanism of action and the relationship between genes and signaling pathways. A similar approach should also be taken when using other databases such as sequencing data to ensure accurate interpretations of germline and somatic mutations of specific diseases in determining functional disease targets, as these databases only map aberrations toknown pathways but are unable to accurately interpret the results relevantly to the disease (Li and Jones, 2012).

Despite the advantage of reducing the time of drug development process, researchers still have to endure the rigorous regulations in advancing the repurposed candidates into the clinic, especially for phase III clinical trials which are cost- and time-consuming compared to phase I and II trials. The average cost per patient for a phase II and phase III study is US$36,000 and US$47,000, respectively (Novac, 2013). Ideally, the candidates would be selected from a group of drugs with characterized pharmacokinetic and pharmacodynamic data, and well-defined disease targets, which would reduce the risk for complications down the pipeline and increase the chance for successful fast- tracking of the compound (Gupta et al., 2013). Another proposed strategy to increase the chance of successful repurposing entails the simultaneous testing of these compounds in several proof-of-concept/phase II studies in a few diseases so that the most promising indication will be brought forward to phase III trials (Novac, 2014). This strategy could be carried out in a collaborative consortium, and could potentially save time and cost, as well as extract maximum potential from the drug candidates and reduce failure rates.

Intellectual property (IP) protection is another caveat associated with drug repurposing. Patents protect drugs for up to 20 years in the USA and Europe, from the filing date (McCabe et al., 2015). If the compound is still under protection, the interested party wanting to repurpose the drug can negotiate acquisition or licensing of the drug with the party holding the composition-of-matter (COM) IP (Ashburn and Thor, 2004). From a commercial perspective, pharmaceutical companies are hesitant to develop ‘old’ drugs for repurposing as the compound cannot be protected against competitors under IP laws.

However, off-patent repurposing candidates could be protected through method-of-use (MOU) patents containing claims focused on the method of use (Shim and Liu, 2014). An example is nitroxoline, an off-patent drug discovered to have anti-growth and anti- angiogenic activities in bladder cancer, which is now protected under MOU due to the discovery of its anti-cancer properties (Shim and Liu, 2014). Besides the option of obtaining an MOU, companies could also either reformulate generic drugs with newly- found activities to increase efficacy, determine new drug combination or obtain exclusive marketing approval for new countries to prevent competition (Ashburn and Thor, 2004). The Orphan Drug Act also assists with patenting issues by offering pharmaceutical companies incentives to develop novel treatments for rare diseases, such as funding and the ability to perform small scale (limited number of patients) clinical trials when assessing an orphan drug application, thereby encouraging research towards drug repurposing while maintaining market exclusivity for 7 to 10 years (McCabe et al., 2015).

Finally, despite the rigorous and continued optimization of the drug repurposing process, failures will still occur and could be attributed to many reasons, which include inter-patient and intra-tumour heterogeneity resulting in low efficacy among patient groups, the development of treatment-resistant disease, and the occurrence of adverse side effects when given in combination with other drugs (Li and Jones, 2012). However, these pitfalls are equally applicable to the conventional drug discovery. There are ample examples of unsuccessful repurposing such as the rejection by the FDA of the drug combination of bupropion, approved for treatment of depression, and naltrexone, previously approved for opioid addiction, for the treatment of appetite regulation and energy expenditure in obesity due to potential adverse cardiovascular effects (Plodkowski et al., 2009; Caveney et al., 2011).

Even with these limitations, drug repurposing offers an alternative to the time consuming and expensive exercise of de novo drug development. With an estimated 10,000 commercially available drugs (Chong and Sullivan, 2007) and databases of chemical structures, transcription profiles and other data for in silico analyses, which would grow each year as drug approval increases, these could certainly provide expanding opportunities for repositioning drugs for a variety of diseases.

1.6.5 Drugs repurposed for leukaemia

Drugs that have been discovered by repurposing approaches for leukaemia and have progressed into clinical trials include old chemotherapeutics, drugs withdrawn from the market due to toxicities, and drugs with off-target effects previously approved for other diseases.

Bortezomib for acute lymphoblastic leukaemia

Bortezomib, a proteasome inhibitor, was originally developed for the treatment of multiple myeloma (MM), a malignant disease of the plasma cell. The inhibitor works in the ubiquitin–proteasome pathway by blocking the 26S proteasome unit involved in the degradation of abnormal or misfolded proteins (Kouroukis et al., 2014). In MM cells, the FDA-approved drug induces apoptosis, alters bone marrow microenvironment by disrupting multiple signaling pathways involved in adhesion and cell survival, and reverses drug resistance of these cells (Richardson et al., 2005; Field-Smith et al., 2006; Palumbo and Anderson, 2011). Because increased proteasomal activity is associated with carcinogenesis, the compound was tested in several other malignancies and found to be potent in ALL (Koyama et al., 2014; Takahashi et al., 2017).

Recently, the Therapeutic Advances in Childhood Leukaemia and Lymphoma (TACL) consortium reported bortezomib in combination with vincristine, dexamethasone, pegylated asparaginase and doxorubicin as having acceptable toxicity and remarkable activity in children with relapsed B-ALL who previously failed 2 – 3 regimens (NCT00440726) (Messinger et al., 2012). The result was confirmed in another patient cohort and intravenous treatment with bortezomib, in combination with the same chemotherapeutics used by the TACL consortium, was found to be a suitable option for childhood relapsed and refractory ALL (Bertaina et al., 2017). The combination is currently being trialed in a larger group of patients (EudraCT: 2012-000810-12). Besides B-ALL, bortezomib-based therapy has also been tested in Ph+ ALL and AML patients, however the regimen did not meet preset minimum response criteria in order to allow treatment continuation in the relapsed, refractory or secondary AML childhood patient cohort (Horton et al., 2014; Zhao et al., 2015). Nevertheless, bortezomib is

36 currently being tested in combination with HDAC inhibitor, panobinostat in a clinical trial for childhood T-ALL (NCT02518750). It could potentially be an effective treatment for this disease as the drug combination was reported to show synergy and demonstrated encouraging activity in patients with peripheral T-cell lymphoma (Tan et al., 2015).

Arsenic trioxide for acute promyelocytic leukaemia

Arsenic derivatives were used in the late 1800s till early 1900s for the treatment of tuberculosis and syphilis. Studies also report effects in several CML patients in the USA (Kwong and Todd, 1997). Due to the side effects of these compounds, which include cardiovascular problems, developmental abnormalities and several types of cancers, their use was abandoned almost a century ago (Gan et al., 2011). However in the 1970s, medical professionals in China tested intravenous administration of arsenic trioxide (ATO) in large scale clinical trials of several types of cancers including more than 1000 patients and found it to be efficacious, particularly in acute promyelocytic leukaemia (APL) with up to 84% of patients achieving complete remission and 28% surviving more than 10 years (Zhang et al., 2001; Zhu et al., 2002; Gan et al., 2011). Subsequent smaller clinical trials in the US, Europe and Japan confirmed its efficacy in APL patients. In combination with all-trans retinoic acid (ATRA), which acts by inducing differentiation commitment of neoplastic progenitor cells and promyelocytes, ATO induces partial differentiation and apoptosis with caspase activation in leukaemic cells (Shen et al., 1997; Zhou et al., 2007; Matthews et al., 2011). A recent study reported a 7-year survival rate of more than 90% in a long-term follow-up data of 117 adult patients enrolled in the APL-07 trial treated with ATRA-ATO combination, drawing the conclusions that this drug combination provides excellent results as first-line therapy for APL (Zhu et al., 2016).

Ribavirin for acute myeloid leukaemia

Ribavirin, a purine-nucleotide analogue with activity against RNA and DNA viruses was discovered more than 40 years ago by Witkowski and colleagues (Borden and Culjkovic- Kraljacic, 2010). This anti-viral drug, also known as Virazole was approved

37 for the treatment of severe respiratory syncytial virus infection, but also used in several other virus infections such as Lassa fever, influenza A and B (Te et al., 2007). Ribavirin, in conjunction with IFN-α, is used as standard care of hepatitis C therapy. Recently, it was found that ribavirin targets the eukaryotic translation initiation factor, eIF4E, an oncogene overexpressed in approximately 30% of cancers including chronic lymphomas, myelogenous leukaemia, and M4 and M5 subtypes of AML (Topisirovic, 2003; Borden and Culjkovic-Kraljacic, 2010). The oncogenic properties of eIF4E are directly linked to its ability to bind 7-methyl guanosine of the 5’ mRNA (Kentsis et al., 2004), enhancing nuclear mRNA export and increasing translation of transcripts essential for proliferation, survival and metastases (Assouline et al., 2009; Assouline et al., 2015). Ribavirin competes with eIF4E/mRNA for the functional site, suppressing eIF4E-mediated oncogenic transformation (Kentsis et al., 2004).

A small proof-of-principle clinical trial with ribavirin (NCT00559091) was carried out in adult M4/M5 AML patients with relapsed or refractory disease, or patients unable to undergo chemotherapy. Reduction of eIF4E levels was associated with good prognosis (Assouline et al., 2009). The study also reported that clinical response was associated with relocalization of nuclear eIF4E to the cytoplasm by ribavirin. A follow-up trial for ribavirin combined with low-dose cytarabine demonstrated that this combination increased time to relapse or treatment failure compared to monotherapy (Assouline et al., 2015). A new AML trial is underway to test the efficacy of ribavirin in combination with vismodegib, a Hedgehog pathway inhibitor (NCT02073838).

Auranofin for chronic lymphocytic leukaemia

Auranofin is a drug approved by the FDA for the treatment of rheumatoid arthritis (RA) in 1985, and marketed under the brand name of Ridaura® by the Smith Kline and French Laboratories (Mirabelli et al., 1986). However, the development of highly efficacious targeted therapies for RA that are based on the inhibition of pro- inflammatory cytokine signaling (such as the tumour necrosis factor (TNF)-α inhibitor infliximab and interleukin (IL)-1 inhibitor anakinra), has resulted in discontinuation of auranofin for the treatment of the disease (Doan et al., 2005). Auranofin recently gained a lot of traction for repurposing for the treatment of several cancers including chronic

38 lymphocytic leukaemia (CLL), parasitic and bacterial infections (Bulman et al., 2015; Capparelli et al., 2016) as well as neurodegenerative diseases such as Parkinson’s and Alzheimer’s diseases (Madeira et al., 2013). The main mechanism of auranofin is postulated to be through inhibiting enzymes involved in the cellular reduction/oxidation (redox) processes such as thioredoxin reductase and glutathione transferases. Reduced levels of these enzymes results in increased levels of reactive oxygen species (ROS), which in turn causes oxidative stress followed by intrinsic apoptosis. Cancer cells particularly rely upon redox enzymes, which results in a selectivity of auranofin towards cancer cells compared to the healthy cells (Roder and Thomson, 2015).

Only two years after the identification of its preclinical activity against CLL cells, auranofin entered into a phase I/ II clinical trial for adult CLL (NCT01419691) (Weir et al., 2012; Shen et al., 2013; Fiskus et al., 2014). However, based on a preliminary report, the best response reported was stable disease, and increases in the levels of ROS and apoptosis measured in patient blood samples were transient and reverted to baseline or below by day 7 (Saba et al., 2013). Auranofin was not pursued any further after the clinical trial ended due to the limited clinical activity in patients, as well as the development of more promising treatments for CLL such as a Bruton tyrosine kinase inhibitor, ibrutinib and phosphatidylinositol-3-kinase inhibitor, idelalisib (Vitale et al., 2017). Nevertheless, auranofin, in combination with other drugs, is currently in adult clinical trials for the treatment of other cancers including advanced or recurrent non- small cell lung cancer or small cell lung cancer (NCT01737502) (Polley et al., 2016) and recurrent glioblastoma (NCT02770378), indicating its potential as a cancer therapeutic.

1.7 Summary and thesis perspectives

In summary, ALL is the most common paediatric cancer. Advancement in the treatment of ALL, risk stratification and application of risk-directed therapy has resulted in an improvement of survival rate of this once fatal disease such that it now exceeds 90%. Despite this development, patients with high-risk leukaemia subtypes, particularly MLL-r leukaemia, Ph+ ALL, Ph-like ALL and ETP ALL, still have dismal prognosis. The overall aim of this study is to identify novel compounds for the treatment of high-

39 risk leukaemia, with a special focus on MLL-r leukaemia, which predominantly affects infants. More aggressive treatment regimens have not been highly beneficial as the survival rate remains barely 50%, thus warranting development of new therapeutics with better specificity and lower toxicity to minimize life-long side effects.

Drug repurposing is potentially a highly efficient strategy for advancing the drug discovery pipeline, allowing the identification of new therapeutics from a collection of existing drugs for the treatment of diseases other than those for which they were originally intended. With established formulation and extensive pharmacology and clinical safety data, new drug candidates could be ready for clinical trials considerably faster than those taken through the traditional approach. Several publications have reported the identification of FDA-approved drugs for the treatment of various types of leukaemias through drug repurposing. Some of these drugs such as bortezomib, successfully advanced into clinical trials for relapsed and refractory ALL and were incorporated into patient treatment protocol (Bertaina et al., 2017). Recently, a HTS found a HDAC inhibitor, romidepsin, in combination with a key component of leukaemia treatment, cytarabine, to be efficacious in preclinical models of MLL-r infant ALL (Cruickshank et al., 2017), indicating the potential of phenotypic screening of old drugs in identifying new therapeutics for this very high-risk subset of ALL.

Therefore, the first aim of this study was to identify novel compounds from a HTS of approved drugs and pharmacologically active compounds for the treatment of high-risk leukaemia. The second aim was to characterize hit compounds by investigating their potency and activity against a panel of high-risk leukaemia, solid tumour and non- malignant cell lines. This aim involved elucidation of the selectivity of the compounds, their mechanism of action in killing leukaemia cells and their potential synergy with currently used drugs in ALL treatment. Finally, the study advanced towards testing the compounds in patient-derived xenografts to address the third aim, which was to determine compound efficacy in highly relevant preclinical models. Collectively, these studies provide insight into the drug repurposing process and have identified novel therapeutics for high-risk leukaemia, which will increase our understanding of the disease and it is hoped ultimately lead to improved treatment of this childhood malignancy.

CHAPTER 2: MATERIALS AND METHODS

2.1 Reagents and equipment

2.1.1 Tissue culture

Roswell Park Memorial Institute-1640 (RPMI-1640), Dulbecco’s modified Eagle medium (DMEM), Minimum essential medium-α (MEM-α), Dulbecco’s phosphate buffered saline (PBS), trypsin solution (0.25% (w/v) trypsin in Hank’s solution), Penicillin (10,000U/mL), Streptomycin (10,000 U/mL), L-glutamine (29.2 mg/mL) liquid (P/S/G), MEM Non-Essential Amino Acid (100X) and MEM Sodium Pyruvate Solution (100X) were purchased from Invitrogen Life Technologies (Sydney, NSW, Australia). Quality Biologicals Serum Free 60 medium (QBSF-60) was purchased from Quality Biologicals (Gaithersburg, MD, USA). FMS-like tyrosine kinase-3 (Flt-3) ligand was purchased from Amgen (Thousand Oaks, CA, USA). Trypan Blue was purchased from ICN Biomedicals Inc. (Aurora, OH, USA). Fetal bovine serum (FBS) was purchased from Thermo Trace (Noble Park, VIC, Australia). Other reagents were purchased from manufacturers as follows: methylene blue from Sigma-Aldrich (Castle Hill, NSW, Australia); ethanol, methanol, isopropanol and sodium dodecyl sulphate (SDS) from Ajax Finechem (Sydney, NSW, Australia).

Tissue culture flasks and 96-well plates (flat- and U-bottom) were purchased at Greiner Bio-One (Fickenhausen, Germany). A Neubauer haemocytometer from Dutec Diagnostics (Sydney, NSW, Australia), was used for cell counting. All cell cultures were maintained in humidified incubators at 37ºC and 5% carbon dioxide (CO2). Cell culture procedures were performed in a Biological Safety Cabinet Class II (AES Environment Pty Ltd, Sydney, NSW, Australia). Cells were collected using centrifuges from Centrifuges Sigma Laborzentrifugen GmbH, model 3-10 from Germany and visualized using an inverted microscope from Olympus Optical Company (Tokyo, Japan).

2.1.2 Cytotoxic drugs

Cytotoxic drugs were purchased from manufacturers as follows: vincristine, etoposide, auranofin and disulfiram from Sigma-Aldrich (Castle Hill, NSW, Australia). Cytarabine, daunorubicin and mitoxantrone were purchased from Clifford Hallam Healthcare Pty Ltd (Eastern Creek, NSW, Australia). Twelve short-listed compounds were purchased as follows: 2-chloroadenosine triphosphate sodium salt from Santa Cruz Biotechnology (TX, USA), butyl β-carboline-3-carboxylate (β-CCB), N- methylhistamine dihydrochloride, amthamine dihydrobromide, iodophenpropit dihydrobromide and SL 327 from Tocris Bioscience (Bristol, United Kingdom), dihydroergotamine tartrate and oxethazaine from Prestwick Chemical PC SAS (llkirch- Graffenstaden, France), U0126, Ro 90-7501, SB 205384 and SID7969543 from Sigma- Aldrich (Castle Hill, NSW, Australia)

Absorbance for cytotoxicity assays with the resazurin reagent was determined using the Bio-Rad Microplate Reader Benchmark from Bio-Rad (Sydney, NSW, Australia).

2.1.3 High-throughput screening

Chemical libraries used in the screening were Prestwick Chemical Library (http://www.prestwickchemical.com/prestwick-chemical-library.html) containing 1280 molecules of which 95% are approved drugs (Prestwick Chemical PC SAS, llkirch- Graffenstaden, France), LOPAC®1280 library (https://www.sigmaaldrich.com/life- science/cell-biology/bioactive-small-molecules/lopac1280-navigator.html) consisting of 1280 pharmacologically active compounds (Sigma-Aldrich, Castle Hill, NSW, Australia), Tocriscreen Plus library (https://www.tocris.com/products/tocriscreen- plus_5840) containing 1280 biologically active compounds (Tocris Bioscience, Bristol, United Kingdom) and Selleck Inhibitor Library (http://www.selleckchem.com/screening/inhibitor-library.html) containing 1927 biologically active small molecule inhibitors (Selleck Chemicals, TX, USA).

The high-throughput screens were performed at the Australian Cancer Research Foundation Drug Discovery Centre for Childhood Cancer at Children’s Cancer Institute

(Randwick, NSW, Australia). 384-well black, clear-bottom tissue culture treated plates were purchased from Sigma-Aldrich (Castle Hill, NSW, Australia). Manipulations were performed using a Hamilton Microlab STAR liquid handling robot (V&P Scientific Inc., San Diego, CA, USA) and test compounds were added using a pintool (V&P Scientific Inc., San Diego, CA, USA). Resazurin reagent was added to the cells to determine cell viability using an automated Multidrop Combi Reagent Dispenser (Thermo Fisher Scientific, Scoresby, VIC, Australia). Cell viability was measured using a fluorimeter (excitation: 530 nm, emission: 590 nm) (Molecular Devices, Sunnyvale, CA, USA).

2.1.4 Protein isolation and western blot analysis

Reagents and equipment used for protein isolation and western blot analyses were purchased as follows: Ponceau S, bovine serum albumin (BSA), protease inhibitor cocktail, phosphatase inhibitor PhosSTOP™ dithiothreitol, Trizma base (Tris), glycerol, Bromophenol blue and Tween®-20 from Sigma-Aldrich (Castle Hill, NSW, Australia); sodium dodecyl sulphate (SDS) from Ajax Finechem (Sydney, NSW, Australia); Nonidet P-40 from Fluka (Buchs, Switzerland) skim milk powder (Coles, Sydney, NSW, Australia); Precision Plus Protein™ Dual Colour Standards, Pierce™ BCA Protein Assay Kit from Thermo Fisher Scientific (Scoresby, VIC, Australia); 4 – 20% Tris-HCl gradient polyacrylamide gels, nitrocellulose membranes, Mini-PROTEAN® Tetra Vertical Electrophoresis Cell, Clarity™ Western ECL Blotting Substrates from Bio-Rad (Sydney, NSW, Australia); 3 mm Chr blotting paper from Whatman (Maidstone, UK); and β-mercaptoethanol from Invitrogen Life Technologies (Sydney, NSW, Australia). Microcentrifuges (Model 5415D) from Eppendorf (Hamburg, Germany) were used to extract protein. Images of protein were captured using the ChemiDocTM Imaging System and analyzed using Image Lab 5.2.1 (Bio-Rad Laboratories, Sydney, NSW, Australia). The following primary antibodies (Abs) were used: and anti-β-actin Sigma-Aldrich (Castle Hill, NSW, AUS), anti-Nrf2 (Merck KGaA, Darmstadt, Germany), and Phospho-Histone H2A.X, cleaved PARP, PARP, HMOX1, MGMT, ALDH1A1 (Cell Signaling Technology, Inc., MA, USA). Secondary antibodies used were HRP-conjugated anti-mouse IgG and anti-rabbit IgG and Cy3- labelled goat anti-mouse IgG (Amersham Biosciences, Piscataway, NJ, USA).

2.1.5 Flow cytometry

The following reagents were purchased as follows: PE Annexin V Apoptosis Detection Kit from BD Biosciences (Australia), 2′,7′-Dichlorofluorescin diacetate (DCFDA) from Sigma-Aldrich (Castle Hill, NSW, Australia), MitoSOX™ Red Mitochondrial Superoxide Indicator from Thermo Fisher Scientific (Scoresby, VIC, Australia). A FACSCalibur flow cytometer (Becton-Dickinson,(Rockville, MD, USA) was used to determine cellular reactive oxygen species level. Apoptosis assay was performed using the BD FACSCanto™ (BD Biosciences, Australia). Flow cytometry analyses for animal work were performed using BD FACSCanto™ and FACSDiva™ software (BD Biosciences, Australia).

2.1.6 Microsomal stability assay

The reagents were purchased as follows: The NADPH Regenerating System Solution A and B and liver microsomes BD Biosciences (Australia). Chromatography was performed on the Accela system and mass spectroscopy analysis was performed using a Quantum Access mass spectrometer (ThermoFisher Scientific, Scoresby, VIC, Australia) at the University of New South Wales. For the first compound, 2- chloroadenosine triphosphate, separation was performed using a ZIC-HILIC 2.1 x 100mm column (Merck, KGaA, Darmstadt, Germany), while the second compound, SID7969543, using a BEH C18 UHPLC Column 50 x 2.1mm (Waters, MA, USA) with a CTC PAL autosampler. Data was processed and chromatograms integrated automatically using XCalibur software (ThermoFisher Scientific, Scoresby, VIC, Australia).

2.1.7 Cell lines

Characteristics of the cell lines used in this thesis are listed in Table 2.1. Cell identities were validated and regularly monitored for mycoplasma. All experiments were conducted with mycoplasma-free cells.

Table 2.1: Characteristics of cell line panel. Cell Line Cytogenetics Disease Supplier PER-485 t(4;11) PER-490 t(4;11) Kindly provided by UR Kees, Telethon Kids Institute, University of Western PER-703A t(1;11) Infant ALL Australia, WA, Australia PER-785A t(4;11) PER-826A Complex t(11;19) RS4;11 t(4;11) Pre-B cell ALL ATCC SEMK2 t(4;11) Pre-B cell childhood ALL ATCC KOPN-8 t(11;19) Infant pre-B cell ALL DSMZ MV4;11 t(4;11) Childhood AML ATCC MOLM-13 t(9;11) AML Kindly provided by R D'Andrea, IMVS, SA, Australia THP-1 t(9;11) Infant AML Kindly provided by W Jessup, Centre for Vascular Research, NSW, Australia CCRF-CEM - T-cell ALL ATCC REH - Pre-B cell ALL ATCC Jurkat - Childhood T-cell ALL ATCC U937 CALM-AF10 AML ATCC KP-MO-TS CALM-AF10 AML Kindly provided by T Imamura, Kyoto Prefectural University of Medicine, Japan KELLY - Neuroblastoma ECACC BE(2)-C - Neuroblastoma Kindly provided by J Biedler, Memorial Sloan-Kettering Cancer Centre, NY, USA Kindly provided by the laboratory of Georgia Chenevix-Trench, QIMR Berghofer HEY - Ovarian carcinoma Medical Research Institute, QLD, Australia Endometrioid ovarian Kindly provided by A de Fazio; Westmead Millennium Institute for Medical 27/87 - cancer Research, NSW, Australia MCF-7 - Breast adenocarcinoma ATCC H460 - Lung carcinoma ATCC LNCaP - Prostate carcinoma ATCC MRC5 - Lung fibroblast ATCC WI-38 - Normal lung ATCC American Type Culture Collection (ATCC), Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ), European Collection of Authenticated Cell Cultures (ECACC).

2.1.8 Patient-derived xenografts

All patient-derived xenograft cells were originally derived from patient bone marrow or peripheral blood samples, and had been subsequently engrafted and serially passaged into non-obese diabetic/severe combined immunodeficient (NOD/SCID) or NOD/SCID gamma (NSG) mice (Lock et al. (2002; Liem et al., 2004; Richmond et al., 2015). Xenograft cells were harvested from the spleens of engrafted mice, purified by Ficoll density gradient separation and cryopreserved in 10% DMSO/90% FBS freezing media. Patient characteristics for the xenografts used in this thesis are listed in Table 2.2.

Table 2.2: Characteristics of patient-derived xenograft panel. Disease Xenograft Subtype Cytogenetics status at Age (y) Sex biopsy MLL-2 t(4;11), MLL-AFF1 - <1 M MLL-5 t(10;11), MLL-MLLT10 Diagnosis <1 M MLL-6 t(11;19), MLL-MLLT1 - <1 M MLL-r ALL MLL-7 t(4;11), MLL-AFF1 - <1 M MLL-8 t(11;19), MLL-MLLT1 Diagnosis <1 F MLL-14 t(11;19), MLL-MLLT1 Diagnosis <1 F ALL-2 Normal Relapse 5.5 F t(17;19), TCF3-HLF, ALL-7 Pre-B ALL Diagnosis 7.4 M biphenotypic ALL-19 Normal Relapse 16.2 M ALL-8 Normal Relapse 12.8 M T-ALL ALL-31 del(6)(q21),del(11q23) Diagnosis 10.2 M ALL-4 t(9;22), BCR-ABL1 Diagnosis 8.9 M Philadelphia ALL-55 Diagnosis 14.5 M + ALL t(9;22)(q34;q11.2), BCR- ALL-56 ABL1 Diagnosis 10 M Philadelphia TGT-052 JAK1 V658F mutation Diagnosis 15 M -like ALL

2.1.9 In vivo studies

Five to seven week old female NOD/SCID (NOD/LtSz-scid/scid) mice were purchased from Australian BioResources (Moss Vale, NSW, Australia). Prior to any experimental work, mice were acclimatized for a minimum of one week in a pathogen-free environment. All experimental work involving NOD/SCID mice was approved by the University of New South Wales Animal Care and Ethics Committee (ACEC 15/129B). The reagents and equipment were purchased as follows: Mini-Collect® EDTA and lithium heparin tubes from Greiner Bio-One (Germany); Insulin syringes (1 mL and 0.5 mL, 29 and 27 gauge x 1/2”) and needles (23 gauge x 1 1/4”) from Livingstone International (Rosebery, NSW, Australia); FACS lysing solution from (BD Bioscience, Australia); phycoerythrin (PE)-conjugated anti-human CD45, allophycocyanin (APC)- conjugated anti-human CD19, and allophycocyanin (APC)-conjugated anti-mouse from Australian Bioresearch (WA, Australia). Kolliphor® EL and 10% neutral buffered formalin were purchased from Sigma-Aldrich (Castle Hill, NSW, Australia).

For biochemistry analyses, activity levels of liver enzymes were measured using the comprehensive diagnostic profile rotors through the VetScan VS2 Chemistry Analyzer, both purchased from Abaxis Inc. (Union City, CA, USA).

2.2 Methods

2.2.1 Cell culture

2.2.1.1 Maintenance of suspension cells

Patient-derived cell lines, PER-485, PER-490, PER-703A, PER-785A and PER-826A were cultured in RPMI with the addition of 20% of fetal calf serum (vol/vol). Other leukaemia cell lines were cultured in RPMI supplemented with 10% of fetal calf serum (vol/vol). Both were supplemented with 2 mM L-glutamine, MEM Non-Essential Amino Acids, MEM Sodium Pyruvate Solution. All were maintained by passaging every 2 – 3 days at a cell density of approximately 2 x 105 cells/mL

2.2.1.2 Maintenance of adherent cells

HEY, 27/87, H460 and LNCaP cells were cultured in RPMI media. KELLY, BE(2)C, MCF-7 and WI-38 were cultured in DMEM media. MRC-5 cells were cultured in α- MEM media. All were supplemented with 10% fetal calf serum. When the cells reached 70 – 80% confluence, they were split 1:5 – 1:10 twice per week, using trypsin.

2.2.1.3 Maintenance of patient-derived xenograft cells

For ex vivo culture of xenograft cells, which had previously been expanded in NOD/SCID mice and isolated from the spleens, the cells were retrieved from cryostorage and thawed in a water bath at 37 °C. Cells were transferred to a 15 mL falcon tube and washed twice with pre-warmed complete RPMI media supplemented with P/S/G to remove traces of DMSO from the freezing medium. Cells were resuspended in QBSF-60 media supplemented with 20 ng/mL FLT-3.

2.2.2 Trypan blue exclusion assay

Cells were mixed with trypan blue at a ratio of 1:1 trypan blue exclusion assay was performed to determine cell number and viability. Cell mixture was loaded onto the haematocytometer counting chamber and visualized using a light microscope. Four grids were counted and number of viable (unstained) and dead (stained blue) cells were noted. Cell density was calculated by multiplying the average viable cell number by two (dilution factor) and by 104 to obtain a value of cells/mL.

2.2.3 High-throughput screening

Prior to screening, 96-well master plates containing 3707 compounds in individual well, dissolved in DMSO were generated by Compounds Australia (Griffith University, QLD, Australia). Upon receipt, plates were stored in -80ºC freezer until used for high- throughput screening (HTS).

2.2.3.1 Primary HTS assay

PER-485 and CCRF-CEM cells were prepared in RPMI media on the day of the screening at their optimal seeding density. 50 μL of cell suspension were immediately seeded into 384-well black, clear-bottom tissue culture treated plates. Test compounds were added to the assay plates using a pintool for a final concentration of 5 μM. Single measurement of compound activity against the negative control has been previously reported by Somers et al. (2016) and Cruickshank et al. (2017). Plates were incubated for 72 hours at 37ºC, 5% CO2. High-throughput assays were performed over a 72-hour treatment course to identify quick-acting compounds. This treatment course has been previously used in other successful high-throughput screens (Somers et al., 2016; Cruickshank et al., 2017). Resazurin was then added using the Multidrop Combi Reagent Dispenser at 5 μL/well. The resazurin quantitatively measures cell viability and proliferation through an oxidation-reduction reaction. Viable and proliferating cells will take up the reagent, causing it to change from the non-fluorescent, blue, oxidized form to the fluorescent, pink, reduced form. A time zero fluorescence read was measured using a plate fluorimeter with excitation at 530 nm and emission at 590 nm. Plates were incubated for 7 hours at 37ºC, 5% CO2 before the final fluorescence was measured.

The following controls were included per assay plate:

1) Blank: RPMI media 2) Positive control: 10 μM cytarabine, which causes 100% cell death in both PER- 485 and CEM cells 3) Negative control: 5 μM DMSO, which causes minimal cell death in both PER- 485 and CEM cells

The difference in Relative Fluorescence Units (RFU) was calculated for each well by subtracting the time zero fluorescence read from the 7-hour fluorescence read. Percent cell viability for each test compound was calculated based on the negative control (set as 100% viability). Percent cell viability was calculated as follows:

% cell viability = [(RFUsample 7 hour – RFUsample 0 hour) / (RFUnegative control 7 hour – RFUnegative control 0 hour)] × 100

The Z-factor was calculated for each assay plate according to Zhang et al. (1999) as follows:

Z-factor = 1 – 3[(SDnegative control + SDpositive control) / (Mnegative control - Mpositive control)],

Where, SD = standard deviation of the controls (7 h – 0 h) and M = mean of the controls (7 h – 0 h)

2.2.3.2 Secondary HTS assay: Strategy I

In Strategy I, ‘hits’ were defined as compounds having 30% or greater growth inhibition in PER-485 compared to CEM cells (viability of CEM – viability of PER-485 ≥30%). 503 compounds identified to have this criterion were further validated in both cell lines in a secondary screen at 5 µM, 1 µM and 0.25 µM concentrations in triplicates. HTS was performed with the same protocol and time course as the primary HTS described in the previous Section 2.2.3.1.

2.2.3.3 Secondary HTS assay: Strategy II

In Strategy II, test compounds from the primary screen that resulted in ≥ 90% reduction in cell viability for both PER-485 and CEM cells at 5 µM were selected for the secondary HTS in a 96-well format. 184 compounds identified to have this criterion were tested on four cell lines: PER-485, CEM, KELLY and MRC-5 at 2, 1, 0.5 and 0.25 µM. Cells were seeded at their respective optimal seeding density into 96-well plates and compounds were added immediately after. Plates were incubated for 72 hours at

37ºC, 5% CO2. Resazurin was then added at 10 μL/well. Fluorescence was read and calculated as described in Section 2.2.3.1.

2.2.4 Cytotoxicity assay

2.2.4.1 Single agent cytotoxicity assay

For leukaemia, solid tumours and non-malignant cell lines, cytotoxicity assays were performed in 96-well, flat-bottom tissue culture treated plates. Leukaemia cells were seeded at their respective optimal cell density on the day assays were performed, while solid tumours and non-malignant cell lines were seeded at 37ºC, 5% CO2 the day before cytotoxic drugs or compounds were added (approximately 16 hours). The cytotoxic drugs or compounds were serially diluted over a range of concentrations in the respective culture medium for each cell line and added in triplicates wells. Plates were incubated for 72 hours at 37ºC, 5% CO2. Resazurin was then added at 20 μL/well or 10 μL/well for suspension (leukaemia) or adherent (solid tumours and non-malignant cells) cell lines, respectively. Following a 6-hour incubation, absorbance was determined using a multiplate reader at an excitation wavelength of 570 nm and emission wavelength of 595 nm.

For xenograft cells and peripheral blood mononuclear cells (PBMCs), cytotoxicity assays were performed in 96-well, U-bottom tissue culture treated plates. Cells were seeded at their respective optimal cell density and incubated for 3 hours at 37ºC, 5%

CO2 to allow cells to equilibrate before cytotoxic drugs or compounds were added. The cytotoxic drugs or compounds were serially diluted and added to duplicate wells as described earlier. Plates were incubated for 48 hours at 37ºC, 5% CO2. Resazurin was then added at 20 μL/well and following an overnight incubation (approximately 16 hours) at 37ºC, 5% CO2, absorbance was determined as described for cell lines.

Percentage survival was calculated as follows:

% survival = [Absorbance of treated wells]avg/[Absorbance of control wells]avg x 100

Data were analyzed with GraphPad Prism Software and the inhibitory concentration resulting in 50% reduction of cell survival relative to control (IC50) values were determined.

2.2.4.2 Fixed-ratio combination cytotoxicity assay

Cells were seeded in 96-well, flat-bottom tissue culture treated plates at their respective optimal cell densities. Cytotoxic drugs or compounds were serially diluted in the appropriate culture medium at a fixed ratio over IC10 – IC90 concentration range. Cytotoxic drug alone, compound alone or a combination of cytotoxic drug and compound were added to the cells in triplicate wells. Plates were incubated for 72 hours at 37ºC, 5% CO2. Resazurin was then added at 20 μL/well and incubated for 6 hours at

37ºC, 5% CO2. Absorbance was determined similarly as described in Section 2.2.4.1.

The average absorbance value for each treatment at each concentration was calculated and entered into the CalcuSyn program. Combination Index (CI) and Drug Reduction Index (DRI) were each generated by the program. CI at effective dose 75 (ED75; the concentration causing 75% decrease in viability) was used to interpret synergism, additivism or antagonism based on Table 2.3. For synergistic combinations, the DRI at ED75 was used to determine how many fold the dose of each component drug can be reduced to achieve a similar effect as if that drug were used alone (Chou, 2006).

Table 2.3: Interpretation of Combination Index values from CalcuSyn program. Range of Combination Index Description <0.1 Very strong synergism 0.1 – 0.2 Strong synergism 0.3 – 0.7 Synergism 0.7 – 0.85 Moderate synergism 0.85 – 0.9 Slight synergism 0.90 – 1.10 Additive 1.10 – 1.20 Slight antagonism 1.20 – 1.45 Moderate antagonism 1.45 – 3.3 Antagonism 3.3 – 10 Strong antagonism >10 Very strong antagonism

(Adapted from Chou, TC; 2006)

For disulfiram, the Bliss Independence model was also used to determine synergistic, additive or antagonistic effects between drugs. In the model, the additive effect of two compounds (A and B) at a concentration is first predicted by the following calculation:

Bliss Independence effect = (FaA + FaB) – (FaA × FaB) where FaA and FaB are the fraction of cells affected by compound A alone and compound B alone, respectively.

Next, to determine the effect of the compound combination (C), the excess (deviation) from the calculated Bliss Independence effect is calculated as:

Excess over Bliss (EOB) = (experimental FaC) – (calculated Bliss Independence FaC) where experimental FaC is the fraction of cells affected by the combination of compounds A and B, and calculated Bliss Independence FaC is the calculated additive effect of A and B. The excess (deviation) from Bliss Independence was calculated at ED75, where synergy is defined by positive EOB values, additive effect by EOB=0 and antagonism by negative EOB values.

2.2.5 Apoptosis detection

Apoptosis was measured by detection of Annexin V binding to the cellular membrane by flow cytometry. Cells were harvested at specific time points post cytotoxic drug or compound treatment, washed with PBS once and Annexin binding buffer once, followed by staining with AnnexinV and 7-Aminoactinomycin D (7-AAD) for 5 minutes in the dark. Samples were analyzed by flow cytometry using a FACSCantoTM. Compensation parameters were established for overlapped fluorophore signals. A minimum of 10,000 events was acquired for each sample. Data were analyzed using FlowJo and apoptosis levels were determined by the percentage of cells stained with AnnexinV and 7-AAD.

2.2.6 Microsomal stability assay

Microsomal stability assays were performed for 2-chloroadenosine triphosphate and SID7969543 to predict compound in vivo stability by determining compound half-life in vitro in liver microsomes. Stock compounds were firstly made to a concentration of 0.5 mM in DMSO. The compounds were then warmed at 37ºC for 5 min in purified water with 20% potassium phosphate (0.5 M, pH 7.4), 5% NADPH Regenerating System Solution A, and 1% NADPH Regenerating System Solution B, at a final compound concentration of 1 μM. Liver microsomes were added to the compound solution at a final concentration of 0.5 mg/mL. A proportion of the mixture was taken out into acetonitrile (1:1) and put on ice for zero time point. The rest of the mixture was incubated at 37ºC and a proportion was taken out and added into acetonitrile (1:1) at 5, 15, 30 and 90 minute time points. All samples were then centrifuged at 21,000 x g for 3 minutes at 4ºC. The supernatants were analyzed by mass spectrometry at the University of New South Wales (Randwick, NSW, Australia) by Dr. Russell Pickford as follows. Samples were injected using a CTC PAL autosampler and chromatography was performed on an Accela system at 400 uL/min. Column eluates were directed into the Heated Electrospray Source of a Quantum Access mass spectrometer. Source conditions and selected reaction monitoring (SRM) transitions were optimized before each analysis using syringe infusion of a standard prior to liquid chromatography-mass spectrometry analyses.

For 2-chloroadenosine triphosphate, separation was performed on a ZIC-HILIC 2.1 x 100 mm column using a 10 minute gradient of 0.1% formic acid in water vs acetonitrile. Analysis was performed in the negative mode. SRM transitions selected were 540>159 at 39V, 540>442 at 22V, 540>273 at 29V. For SID7969543, separation was performed on a BEH C18 UHPLC Column 50 x 2.1 mm using a 7 minute gradient of 0.1% formic acid in water vs acetonitrile. Analysis was performed in the negative mode. SRM transitions selected were 453>302 at 13V, 453>274 at 27V, 453>188 at 40V.

Data was processed and chromatograms integrated automatically using XCalibur software. Compound stability in liver microsomes was calculated by plotting the area under the curve against time over the value of 0 min time point. Results presented are

54 the mean ± SEM of 2 independent experiments. Half-life values were generated by analyzing the curves using one phase exponential decay.

2.2.7 Reactive oxygen species assay

Reactive oxygen species (ROS) levels were determined using flow cytometry. Cells were treated with respective drugs, stained with either DCFDA for one hour or MitoSOX™ Red Mitochondrial Superoxide Indicator for 15 minutes in the dark, harvested at specific time points and washed with PBS. Samples were analyzed using flow cytometry using a FACSCalibur. Compensation parameters were established for overlapped flurophores signals. A minimum of 10,000 events was acquired for each sample. ROS levels were analyzed using FlowJo and determined by the percentage of cells stained with DCFDA.

2.2.8 Isolation and quantitation of total cellular protein

Total cell lysates were isolated from leukaemia cell lines and xenograft cells. Cells were collected by centrifugation, washed with ice-cold PBS and resuspended in cold RIPA buffer (PBS with 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with protease and phosphatase inhibitor cocktail. Cell lysates were incubated on ice for 30 minutes, vortexed every 10 minutes and centrifuged at 21,000 x g for 10 – 20 minutes at 4°C. Protein concentrations were quantified using the Pierce™ BCA Protein Assay Kit according to manufacturer’s instructions.

2.2.9 Western Blot Analysis

For each cell line sample, 15 μg protein (PDX, 50 μg) were mixed with 6x loading buffer (300mM Tris, pH 6.8, 35% glycerol, 0.05% bromophenol blue, 6% β- mercaptoethanol, 9% SDS). Samples were loaded onto a pre-cast 4 – 20% Tris-HCl gradient gel together with protein ladder. Gels were eletrophoresed for 90 minutes at 100 V in a Mini-PROTEAN® Tetra Vertical Electrophoresis Cell system using running buffer (25 mM Tris, pH 8.3, 192 mM glycine and 0.1% SDS). The separated proteins were electro-transferred to a nitrocellulose membrane for 2 hours at 200 mA using the

55 same system in transfer buffer (0.025mM Tris/glycine, pH 8.2 and 20% (v/v) methanol). The membranes were briefly stained with Ponceau S (0.1% w/v Ponceau S in 5% acetic acid (20 mM Tris pH 7.5, 500 mM NaCl) to confirm transfer efficiency and equal loading. Membranes were blocked in 5% (w/v) skim milk in Tris buffered saline (10 mM Tris-HCl pH 8.0, 150 mM NaCl) with 0.05% (v/v) Tween-20 (TBS-T) to prevent non-specific binding of antibodies. Membranes were probed overnight (approximately 16 hours) at 4ºC with primary antibody according to manufacturer’s instructions.

Membranes were washed with TBS-T before being probed with an appropriate secondary antibody according to manufacturer’s instructions for 1 hour at room temperature. Membranes were washed with TBS-T and developed using Clarity™ Western ECL Blotting Substrates according to the manufacturer’s instructions and proteins were visualized using the ChemiDocTM Imaging System. Results were analyzed using Image Lab 5.2.1 software.

2.2.10 Gene expression

A publicly available gene expression microarray dataset was used to determine expression of genes in MLL-rearranged ALL patient-derived xenografts. Raw data were available in the microarray database Gene Expression Omnibus with GEO Series accession number GSE52991. The dataset used Illumina microarray platform (Suryani et al., 2014). The analyses of these data were performed by Dr. Chelsea Mayoh at Children’s Cancer Institute (Randwick, NSW, Australia). Another publicly available gene expression microarray dataset was used to determine expression of several genes between infants with MLL-rearranged ALL and infants with ALL. Raw data were available in the same microarray database (http://www.ncbi.nlm.nih.gov/geo/: accession number GSE19475) (Stam et al., 2010). The dataset used Affymetrix microarray platform and was normalized using the Robust Multichip Average algorithm. The analyses of these data were performed by Dr. MoonSun Jung at Children’s Cancer Institute (Randwick, NSW, Australia).

2.2.11 In vivo mouse studies

2.2.11.1 Determining maximum tolerated dose in NOD/SCID mice

To determine the maximum tolerable dose (MTD) of auranofin, 3 mice were used for each treatment group. 6 – 8 week old NOD/SCID mice were injected intraperitoneally at increasing concentrations starting with either vehicle control (10%, DMSO, 30% cremophor, 60% saline), or auranofin at 2.5, 5, 7.5 and 10 mg/kg for 5 consecutive days, over 3 weeks (3 cycles: 5 days on, 2 days off). For disulfiram, the drug was given orally at increasing concentrations starting with vehicle control (PBS), 100, 150 and 200 mg/kg for 5 consecutive days, over 4 weeks (4 cycles: 5 days on, 2 days off). Another group of mice was also administered similar concentrations of disulfiram, in addition to 1.5 mg/kg copper, orally prior to disulfiram.

Mice were checked on treatment days and weighed weekly to monitor toxicity-related weight loss. Mice were also monitored for signs of toxicity such as decreased activity, ruffled coat and hunched posture. Three weeks after treatment ended, mice were humanely sacrificed by CO2 asphyxiation and post mortems were performed to check for any internal abnormalities.

2.2.11.2 Determining hepatoxicity in auranofin-treated NOD/SCID mice

Liver samples of one mouse from each treatment group were taken upon sacrifice and fixed for 24 hours in 10% neutral buffered formalin before bein transferred to 70% ethanol. Samples were embedded in paraffin, sectioned and stained with haematoxylin and eosin at the Garvan Institute of Medical Research Histopathology Facility at The Kinghorn Cancer Centre (Sydney, Australia).

Additional special stains (Perls Prussian Blue, Masson trichrome, PAS, DiPAS, Orcein, Rhodanine and Reticulin) and immunohistochemical staining (cytokeratin) were performed at the Department of Anatomical Pathology, Prince of Wales Hospital (Randwick, NSW, Australia) on the livers from the control group and 5 mg/kg group.

All analyses were performed by a paediatric pathologist (Dr. Andrew Gifford) at the Sydney Children's Hospital (Randwick, NSW, Australia).

2.2.11.3 Determining hepatoxicity in disulfiram-treated NOD/SCID mice

To monitor disulfiram-mediated hepatoxicity, levels of liver enzymes, alkaline phosphatase (ALP) and alanine aminotransferase (ALT), were tested at 1, 3 and 5 weeks post treatment in mice treated with disulfiram plus copper. For this analysis, mice were first warmed to enhance tail vein dilation by placing the cage 30 cm to an infrared lamp for approximately 15 minutes. Mice were placed in a Perspex restraint inside a biological safety cabinet and a minimum of 100 μL of blood from all mice from each treatment group were pooled into lithium heparin tubes. 100 μL of the pooled blood was then pipetted into the well of the Comprehensive Diagnostic Profile rotor. The rotor was inserted into the VetScan VS2 Chemistry Analyzer and levels of enzymes were automatically generated once analyses completed. 2.2.11.4 Engraftment of human leukaemia cells into NOD/SCID mice

Patient-derived xenografts (PDX) are paediatric ALL patient-derived cells that have been serially passaged in immunodeficient mice, as previously described (Lock et al., 2002). Cells previously cryopreserved were resuspended in PBS to dilute cryopreserving agent. Cells were centrifuged and resuspended at a concentration of 1 – 2.5 x 106 cells/ mL to prepare the desired cell number per 100 μL for inoculation into each mouse, and placed on ice.

Mice were warmed by placing the cage 30 cm to an infrared lamp for approximately 15 minutes until tail veins were dilated. Mice were placed in a Perspex restraint inside a biological safety cabinet and 100 μL of cell suspension was injected using a 23 gauge x 1/2” insulin syringe via the lateral tail vein. The puncture site was compressed with a tissue until bleeding ceased. Mice were then returned to their cage and monitored regularly for general well-being.

2.2.11.5 Peripheral blood monitoring of leukaemia engraftment

Leukaemia engraftment was monitored weekly from ten days post-inoculation. Mice were warmed and restrained as described in Section 2.2.11.4 and approximately 50 μL (3 to 4 drops) of blood was collected from the lateral tail vein into Mini-Collect® EDTA tubes. Bleeding was stopped and mice returned to their cage.

50 μL of blood samples were transferred into flow cytometry tubes and together with 50 μL of PBS containing 1 μL of APC-conjugated anti-mouse CD45 and 1 μL PE- conjugated anti-human CD45. Excess blood from all mice were pooled and stained with equivalent amounts of isotype control antibodies to ensure accurate setting of the parameter gates for flow cytometry. All samples were incubated in the dark for 30 minutes at room temperature. Red blood cells were then lysed with 900 μL FACS lysis buffer and incubated in the dark for 30 minutes at room temperature. Approximately 3 mL of PBS was added to each tube. Samples were centrifuged, supernatant was discarded and the cells resuspended in 200 μL of PBS.

The samples were analyzed by flow cytometry using a FACSCantoTM. Compensation parameters were established for overlapped fluorophore signals. A minimum of 10,000 events was acquired for each sample. The proportion of human CD45+ (huCD45+) cells as a proportion of total human and murine CD45+ cells was calculated to determine the overall leukaemic burden.

2.2.11.6 Drug treatment of NOD/SCID mice

When the median huCD45+ in peripheral blood of all engrafted mice was approximately 1% as a proportion of total human and murine CD45+ cells, mice were randomized to receive either drug or vehicle control with treatment schedule as listed in Table 2.4. Leukaemia engraftment was monitored weekly by tail vein bleeds until huCD45+ in peripheral blood reached 25% as a proportion of total human and murine

CD45+ cells (event). Mice were humanely sacrificed by CO2 asphyxiation once they reached event.

Table 2.4: Treatment protocols for in vivo studies. Drug Vehicle Treatment schedule 10% DMSO, 30% chremophor EL, 2.5 mg/kg intraperitoneal injection on 5 Auranofin 60% saline (Huang et al., 2016) days/week for 3 weeks 25 mg/kg intraperitoneal injection on 4 Cytarabine PBS days/week for 2 weeks 0.5% methyl cellulose, 0.5% Tween- 200 mg/kg via oral administration on 5 Disulfiram 80, 99% PBS (Deng et al., 2016) days/week for 4 weeks 1.5 mg/kg via oral administration on 5 Copper PBS (Deng et al., 2016) days/week for 4 weeks

2.2.11.7 Assessment of in vivo drug efficacy

EFS was graphically represented by Kaplan-Meier analysis and survival curves were compared by log-rank test. Median EFS of control mice was subtracted from the median EFS for drug treated mice to generate a leukaemia growth delay (LGD) for each treatment.

2.2.11.8 Assessment of molecular response to drugs in vivo

Mice from the control group in each efficacy experiment were further used to examine the molecular response to drug treatment. In these mice, when huCD45+ cells were allowed to reach a level in peripheral blood of approximately 50 – 70 %, at which point 1 – 2 mice were treated with drugs as listed in Table 2.5 for for short-term treatment followed by harvest of splenocytes. Mice were humanely sacrificed by CO2 asphyxiation 4 hours after the last dose. Spleens were collected and placed into a sterile tea strainer and homogenized with RPMI media using the plunger of a 10 mL syringe. Single cell suspensions were achieved by filtering the cell mixture through a 40 μm cell strainers into 50 mL tubes. RPMI media was added to obtain a total volume of 40 mL. Ten mL of Lymphoprep solution was carefully underlaid and mononuclear cells were purified by density gradient centrifugation. Cells were cryopreserved at 30 – 50 x 106 cells/mL per cryovial in freezedown media (90% FBS with 10% DMSO). Western blotting was performed as described in Section 2.2.9.

Table 2.5: Treatment protocols for study for molecular response in vivo. Drug Treatment schedule 2.5 mg/kg intraperitoneal injection daily for 2 days Auranofin 10 mg/kg intraperitoneal injection for 1 day Cytarabine 25 mg/kg intraperitoneal injection daily for 2 days Disulfiram 200 mg/kg via oral administration daily for 3 days Copper 1.5 mg/kg via oral administration daily for 3 days

2.2.12 Statistical analyses

To assess statistical significance of the difference in IC50 values, percentage of viability or gene expression between two groups, unpaired t-test was used. For analyses involving three or more groups, one-way ANOVA with either Dunn’s or Tukey’s correction for multiple comparisons was used. Both types of analyses were performed using GraphPad Prism 7.02 (GraphPad software, San Diego, CA, USA). P-values below 0.05 were considered statistically significant.

CHAPTER 3: HIGH-THROUGHPUT SCREENING TO IDENTIFY NOVEL THERAPIES FOR HIGH-RISK LEUKAEMIA

3.1 Introduction

Acute lymphoblastic leukaemia (ALL) is the most common cancer in children, accounting for a quarter of childhood malignancies in children aged below 15 years (Pui, 2010). Advances in combination chemotherapy and treatment strategies over the past decades have increased survival rates towards 90%, however there are still groups of high-risk patients with dismal prognosis including infants with MLL gene rearrangement that have survival rates below 40% (Kotecha et al., 2014). Low survival rates and toxicities associated with currently used chemotherapeutics have pushed the development of targeted therapies such as the tyrosine kinase inhibitor, imatinib, the introduction of which improved the disease-free survival rate from 56% to 75% for Philadelphia chromosome-positive ALL paediatric patients (Biondi et al., 2012). Similar effort has also been taken for the high-risk patient group with MLL-r leukaemia with Pinometostat (EPZ-5676), a DOT1L inhibitor, entering Phase I clinical trials for adults and children with relapsed or refractory leukaemia bearing MLL gene rearrangement (NCT01684150; NCT02141828). However, so far the compound is not doing as well as was expected based on preclinical data (Stein et al., 2015).

As discussed in Chapter 1, the conventional drug discovery path involves numerous stages and processes starting from compound discovery and target identification and/or validation, followed by preclinical development requiring ADMET data, and lastly rigorous testing in humans through proof of principle and proof of concept clinical trials. Due to this time-consuming development process, it has taken more than a decade for new agents such as those mentioned above to advance from bench to bedside. As a result, an alternative and potentially more effective discovery approach based on drug repurposing has recently gained traction. Drug repurposing bypasses many development and optimization steps, shortening the timeline and decreasing the cost for drug approval.

Some efforts have been made to discover new therapeutics for MLL-r leukaemia through a drug repurposing approach. Hoeksema et al. (2011) previously performed a screen of the NCI/NIH Developmental Therapeutic Program Approved Oncology Drug Set II, a library of 89 FDA-approved anti-cancer drugs, against five paediatric MLL-r leukaemia cell lines. Out of the 89 drugs, 42 showed anti-leukaemic activity and out of these, 12 were effective against all cell lines with IC50 below 1 μM. A nucleic acid synthesis inhibitor, cladribine, a microtubule destabilizer, vinorelbine, a DNA intercalating agent and protein kinase inhibitor, valrubicin, and several other drugs also demonstrated activity in patient samples in vitro (Hoeksema et al., 2011). The study provided a basis for considering existing drugs as treatment options for MLL-r leukaemia and suggested further characterization of these drugs for future clinical trials, although no further validation of these candidates in animal models was reported. Cruickshank et al. (2017) also recently adopted the drug repurposing method to discover novel targeted therapies for MLL-r leukaemia by screening 101 FDA-approved chemotherapeutics against eight cell lines derived from infants diagnosed with MLL-r ALL. They found proteasome inhibitors, histone deacetylase (HDAC) inhibitors and cyclin dependent kinase inhibitors to be efficacious against infant ALL in vitro with most of the drugs in these three classes consistently showing low nanomolar IC50 values in all cell lines tested. One of the HDAC inhibitors, romidepsin was selected for further assessment due to its potency, as well as availability of formulation for infants and translational potential as it is FDA-approved and currently tested for hematological cancers (Cruickshank et al., 2017). Romidepsin enhanced the in vivo activity of cytarabine, a chemotherapeutic used in infant ALL therapy, as the combination of both drugs potently reduced leukaemia burden in xenografted mice. These studies indicate the potential of the drug repurposing approach as an alternative to discover new therapeutics for high-risk leukaemia.

The primary aim of this study was to use the drug repurposing approach to identify novel compounds for improved treatment of MLL-rearranged leukaemia. The approach taken was to screen a library consisting of FDA-approved drugs and pharmacologically active compounds with known targets against a MLL-r cell line, PER-485, harbouring the most common MLL translocation among paediatric patients (MLL-AF4) (Kees et al., 2003) and successfully used in previous identification of anti-MLL compounds

(Somers et al., 2016; Cruickshank et al., 2017). A parallel screen was conducted against a T-ALL cell line wild-type for MLL but isolated from a relapsed high-risk patient, CCRF-CEM (Foley et al., 1965), in order to identify those compounds selectively affecting the viability of PER-485 compared to CCRF-CEM (referred to from this point onwards as CEM). Characterization of those compounds (Strategy I; further discussed in Chapter 4) however, found the candidates unsuitable for advancement to in vivo testing due to lack of drug-like properties. Therefore, another secondary screen (Strategy II) was performed to identify more potent compounds with likely clinical potential. For this strategy, compounds that were cytotoxic to both cell lines in the primary screen were further tested at lower doses. Two potent compounds against high- risk leukaemia were identified and further characterized, as reported in Chapter 5. This chapter reports the primary screen as well as the secondary screening and further filtering associated with each of these two strategies.

3.2 Results

3.2.1 Primary screening of FDA-approved drugs and pharmacologically active compounds for activity against leukaemia cell lines

To identify novel compounds targeting MLL-r leukaemia, a primary phenotypic screen was performed with four compound libraries together containing 3707 approved drugs and/or pharmacologically active compounds. All four libraries, Prestwick (n=1200), Tocris (n=1119), LOPAC (n=1280) and Selleck (n=108), consist of a mixture of drugs approved by the US FDA as well as other agencies such as the European Medicines Agency, and biologically active compounds that covers a wide range of targets including kinases, neurotransmitter receptors and G-protein-coupled receptors. These libraries were screened against a MLL-AF4 ALL cell line derived from an infant patient sample, PER-485, and a childhood T-cell ALL cell line, CEM, as representative of a MLL-wild-type (MLL-wt) yet high-risk paediatric leukaemia. This approach required an efficient screening format and was therefore conducted in 384-well configuration.

3.2.1.1 Optimization of assay conditions for high-throughput screening

In order to conduct the screen in 384-well format, it was necessary to optimize cell seeding density and period of incubation with resazurin for the 72-hour cytotoxicity assay. Cell densities tested for each cell line were 100%, 50%, 25% and 12.5% of the optimal cell density in 96-well format. PER-485 and CEM cells were treated with 1.5 µL of dimethyl sulfoxide (DMSO), which was the volume of drug, resuspended in DMSO, to be used in the primary screen. Fluorescence was read after three-, five- and seven-hour incubation with resazurin to determine optimal incubation time, ensuring sufficient signal above background/noise was captured. Based on the readout of fluorescence (Figure 3.1), the incubation period with resazurin selected for both cell lines was seven hours as low fluorescence (<4000) was measured at other incubation periods (Figure 3.1). Optimal cell density for each cell line was determined by selecting a density at the exponential section of the seven-hour curve, which indicated exponential cell proliferation rate. Thus, cell density selected for PER-485 was 1750 cells per well and CEM, 1000 cells per well (Figure 3.1).

A P E R -4 8 5

8 0 0 0

6 0 0 0 3 h

5 h U

F 4 0 0 0 7 h R

2 0 0 0

0 0 2 0 0 0 4 0 0 0 6 0 0 0 8 0 0 0

C e ll d e n s ity /w e ll

B C E M

1 5 0 0 0

3 h

1 0 0 0 0 5 h U

F 7 h R

5 0 0 0

0 0 2 0 0 0 4 0 0 0 6 0 0 0 C e ll d e n s ity /w e ll

Figure 3.1: Optimization of cell density and resazurin incubation time for PER-485 and CEM cell lines for high-throughput screening in 384-well format. Relative fluorescence unit (RFU) measurements at 72 h post-seeding in 384 well plates after incubation with resazurin in (A) PER-485 and (B) CEM. Resazurin was added and allowed to incubate for 3, 5 or 7 hours before being detected by a plate fluorimeter. Optimal incubation with resazurin was 7 hours. Optimal cell density was selected at the exponential section of the 7-hour curves (exponential cell proliferating rate), as indicated by dotted lines: PER-485, 1750 cells/well; CEM, 1000 cells/well. The results are expressed as the mean ± SE of three independent experiments.

3.2.1.2 Results of primary screen

The optimized conditions were used to screen the compound libraries against PER-485 and CEM cell lines. Cells were treated with compounds at a single dose of 5 µM concentration for 72 hours and the viability of the cells was determined using a resazurin-based cytotoxicity assay (as described in Section 2.2.3.1). The results of this screen are displayed graphically in Figure 3.2. Among the 3707 compounds, 249 compounds were completely inactive in both cell lines. 3136 compounds killed at least 50% of CEM cells, while 3191 compounds killed 50% of PER-485 cells.

3.2.2 Secondary screen Strategy I: Identification of compounds selective against MLL-rearranged leukaemia

The primary screen data was first analyzed to identify potential MLL-selective compounds. In this first strategy (Strategy I), ‘hits’ were defined as compounds having 30% or greater growth inhibition in PER-485 compared to CEM cells (viability of CEM – viability of PER-485 ≥30%) (Figure 3.3). Positive and negative control data for the primary screen is shown in Supplementary Figure 1. 503 compounds identified to pass this criterion were further validated in both cell lines in a secondary screen at 5 µM, 1 µM and 0.25 µM concentrations in triplicates. From this validation screen, only 26 compounds were confirmed to show selective inhibition of PER-485 cells in at least two replicates at 5 μM. Among these 26, five compounds were selective at 1 μM and one continued its selectivity at 0.25 μM. To further filter the hits, 12 compounds showing 40% or greater difference in killing of PER-485 cells versus CEM cells at 5 μM were selected for a full-dose range screening in an expanded panel of cell lines (Figure 3.4).

The 12 selected compounds were tested against four cell lines including PER-485, CEM, an additional infant MLL-rearranged cell line, KOPN-8 to indicate possible activity in cells with a distinct MLL rearrangement (MLL-ENL), and a non-malignant cell line, MRC-5 using resazurin-based cytoxicity assays. Cells were treated with each compound over a dose range of 0.16 – 20 μM for 72 hours and the inhibitory concentration resulting in 50% reduction of cell survival relative to control (IC50) values were calculated. Compounds were prioritized based on their selective killing of MLL-r

180

140

100

60 Viability (%) Viability

0 337 674 1011 1348 1685 2022 2359 2696 3033 3370 3707 -20 Compound # CEM PER-485

Figure 3.2: Diagram representing the result of the primary screen of a library of approved drugs and pharmacologically active compounds. Percentage viability of MLL-rearranged cell line, PER-485 (red) and MLL-wild-type cell line, CEM (green) for 3707 compounds in a primary screen (5 μM) is displayed on the Y axis. Compounds are ranked in order of activity against the PER-485 cell line.

100 PER-485 (MLL-r) more sensitive

485) -

0 PER

- 0 337 674 1011 1348 1685 2022 2359 2696 3033 3370 3707 (CEM

% viability difference difference viability % -50

-100 Compound #

Figure 3.3: Diagram representing the result of the primary screen of a library of approved drugs and pharmacologically active compounds in terms of differential activity towards PER-485 and CEM cells. Viability difference between CEM and PER-485 in the primary screen (5 μM) as displayed on the Y axis. 503 compounds showed a higher activity towards the MLL- rearranged cell line, PER-485 compared to the MLL-wild-type cell line, CEM (i.e. CEM viability minus PER-485 viability ≥30%, above dotted line), and were therefore selected for further evaluation.

69 cells, with a requirement of a greater than 2-fold difference in IC50 between PER-485 and CEM. Three compounds, amthamine dihydrobromode, N-α-methylhistamine dihydrochloride and SB 205384, each had IC50 of above 20 μM across all cell lines. Two other compounds, iodophenpropit dihydrobromide and dihydroergotamine tartrate, did not meet the criterion of 2-fold difference in IC50 in PER-485 or KOPN-8 compared to CEM cells. These five compounds were therefore not analyzed further. Oxethezaine was confirmed to show high activity against PER-485 cells and Ro 90-7501 exhibited high potency towards KOPN-8 cells. All remaining compounds presented with at least

2-fold lower IC50 in PER-485 compared to the CEM cells, and, together with oxethazaine and Ro 90-7501 were thus selected for further evaluation (Figure 3.4).

The 12 short-listed compounds did not belong to any particular target classes, however two of the compounds, SL327 and U0126 are well-known mitogen-activated protein kinase kinase/extracellular signal–regulated kinases kinase (MEK) inhibitors and have been widely tested in leukaemia (Vrana and Grant, 2001; Ng et al., 2010) as well as other cancer types (Ma et al., 2015; Wang et al., 2017). The other selected compounds include an agonist of purinergic G protein-coupled receptor, P2Y, a Steroidogenic Factor-1 (SF-1) inhibitor, a benzodiazepine receptor agonist, a sodium channel blocker and an inhibitor of amyloid β42 fibril assembly.

Figure 3.4: Characterization of twelve candidate MLL-selective compounds in a full-dose response screen.

Cell viability was determined after 72-hour exposure to compound. Heat map of concentration with 50% cell growth reduction (IC50) according to the colour key. β-CCB: Butyl β-carboline-3-carboxylate, MEK: mitogen-activated protein kinase kinase/extracellular signal– regulated kinases kinase; P2Y: purinergic G protein-coupled receptor; SF1: Steroidogenic Factor-1; H3: histamine3 receptor; H2: histamine2 receptor; GABA-A: gamma-aminobutyric acid A.

To further investigate the selectivity of the seven short-listed compounds towards MLL- r leukaemia, the compounds were tested using 72-hour resazurin-based viability assay against an expanded leukaemia cell line panel that included additional MLL-rearranged cell lines (both ALL and AML) with different MLL gene translocations (MLL-AF4, n=4; MLL-AF9, n=2; MLL-ENL, n=1), as well as MLL-wt leukaemia cell lines (n=5). These included CALM-AF10 translocated leukaemia cell lines (n=2) that represent an aggressive leukaemia subtype that is reported to share common underlying molecular aetiological pathways with MLL-r leukaemia cells such as their dependency on the DOT1L histone-lysine methyltransferase (Chen et al., 2013).

Figure 3.5 shows the percentage viability of the panel of leukaemia cell lines treated with 10 μM of the compound and the results of the full dose response cytotoxicity assays are presented in Table 3.1 and Supplementary Figure 3. When cell lines were grouped according to genetic lesion, there was no significant difference for any compound in the mean viability at 10 µM for the MLL-r leukaemia cells compared to the MLL-wt leukaemia cells (Figure 3.6). However, two compounds, 2-chloroadenosine triphosphate and SID7969543, showed activity towards several MLL-r and related CALM-AF10 leukaemia cell lines while not affecting MLL-wt leukaemia cells, indicating potential selectivity towards these disease subgroups. For 2-chloroadenosine triphosphate (2-Cl-ATP), the IC50 ranged from 2.5 – 5.9 μM, with four out of seven

MLL-r cell lines being sensitive (defined as IC50 below 10 μM). Both CALM-AF10 cell lines were sensitive to the compound, while none of the other high-risk ALL cell lines were sensitive, with IC50 ranging from 11.5 μM to more than 20 μM (Table 3.1). For

SID7969543 (SID), three out of seven MLL-r cell lines were sensitive with IC50 ranging from 1 to 2.6 μM. As observed for 2-Cl-ATP, the two CALM-AF10 cell lines were sensitive to SID, with both having an IC50 of 1.9 μM and the other high-risk ALL cell lines were resistant, with IC50 values higher than 20 μM (Table 3.1). These compounds, 2-Cl-ATP, a P2Y purinergic receptor, agonist, and SID, an inhibitor of the Steroidogenic Factor-1 (SF-1/NR5A1), were selected for further characterization as reported in Chapter 4 (Figure 3.7). However, since those studies revealed these compounds to lack certain drug-like properties and that they were unsuitable to advance into preclinical animal studies, an alternative secondary screen was performed to identify more potent compounds for high-risk leukaemia (Section 3.2.3).

Figure 3.5: Viability of cell line panel at 10 μM concentration of each of the seven short-listed compounds. Viability was measured in MLL-rearranged (red bars), CALM-AF10 (blue bars) and MLL-wild type (green bars) leukaemia cell lines and a non-malignant cell line (black bar) in 72-hour resazurin-based assays. The results are expressed as the mean ± SE of three independent experiments. β-CCB: Butyl β-carboline-3-carboxylate. 73

Table 3.1: Cytotoxicity of seven shortlisted compounds in a high-risk leukaemia cell line panel.

IC50 (µM) Cell line Translocation Disease 2-Cl- SL β- Ro 90- SID U0126 Oxethazaine ATP 327 CCB 7501 PER-485 t(4;11) Infant ALL 2.5 1.4 7.1 2.6 1.4 3.6 8.9 RS4;11 t(4;11) Pre-B cell ALL 5.9 >20 >20 >20 >20 9 10 ALL Pre-B cell childhood SEMK2 t(4;11) 5.1 >20 >20 20 20 8.9 1.8 MLL- ALL rearranged KOPN-8 t(11;19) Infant pre-B ALL 13.4 >20 >20 >20 20 10 2.5 MV4;11 t(4;11) Childhood AML >20 >20 20 >20 >20 11.7 >20 AML MOLM-13 t(9;11) AML 5.6 1 15 15 20 3.8 10 THP-1 t(9;11) Infant AML >20 2.6 >20 >20 >20 15 >20 CEM - Childhood T-cell ALL 11.4 >20 >20 >20 >20 10.6 18.9 Relapse pre-B cell ALL REH - >20 >20 20 >20 >20 4.9 20 MLL-wild- ALL type Jurkat - Childhood T-cell ALL >20 >20 >20 >20 >20 18.9 >20 U937 CALM-AF10 AML 3.5 1.9 >20 >20 >20 9.8 >20 AML KP-MO-TS CALM-AF10 AML 4.4 1.9 10 20 8.7 5.5 >20 Normal Normal lung MRC-5 - >20 >20 >20 >20 >20 >20 >20 cells fibroblasts

IC50: inhibitory concentration resulting in 50% reduction of cell survival relative to control, IC50 values were derived from the mean of three independent experiments; ALL: acute lymphoblastic leukaemia; AML: acute myeloid leukaemia, 2-Cl-ATP: 2-chloroadenosine triphosphate, SID: SID7969543, β-CCB: Butyl β-carboline-3- carboxylate.

Figure 3.6: Viability of MLL-rearranged, CALM-AF10 and MLL-wild-type leukaemia cell lines at 10 µM compound concentration. None of the compounds show significant differences in viability between each cell group. Dots represent mean viability of three replicates. Mean viability was compared between groups by one-way ANOVA with Tukey’s correction for multiple comparisons.

Primary screen Cell lines: MLL-r PER-485 and MLL-wt CEM. Add drugs at 5 µM; Incubate for 72 hour Hit identification 503 compounds with >30% viability difference between CEM and PER-485 Hit validation Validated 503 compounds in a 3-point serial dilution – 5, 1, and 0.25 µM

Full-dose response screen Screened 12 compounds with >40% viability difference between CEM and PER-485 at 5 µM against 4 cell lines MLL-selectivity screening against leukaemia panel Screened 7 compounds showing difference in IC50 between CEM and PER-485

Selection of final compounds 2 compounds showed preferential cytotoxicty towards several MLL-rearranged cell lines

Figure 3.7: Summary of high-throughput screening of a library of approved drugs and pharmacologically active compounds for identification of candidate MLL- selective compounds (Strategy I). Schematic outline of high-throughput primary screen of Prestwick, Tocris, LOPAC and Selleck libraries containing 3707 FDA-approved drugs and pharmacologically active compounds, and further screening and filtering for compounds selectively targeting MLL-rearranged leukaemia.

3.2.3 Secondary screen Strategy II: Identification of potent novel inhibitors of high-risk leukaemia

To identify more potent inhibitors targeting MLL-r leukaemia, ‘hits’ were defined as compounds that reduced the viability of both cell lines to below 10% (i.e. >90% growth inhibition) after 72-hour treatment at 5 µM compound concentration in the primary screen. Under this criterion, 184 compounds were cytotoxic to both MLL-r cell line, PER-485 and childhood MLL-wt cell line, CEM. These hit compounds were further subjected to a secondary screen against both cell lines at a lower dose range (2, 1, 0.5 and 0.25 µM) to determine if there is a window of MLL-selectivity at lower doses that was missed at a high non-discriminate dose of 5 µM (Supplementary Table 1). Out of the 184 compounds, 28 were identified to have an estimated IC50 below 0.25 µM in both cell lines. These compounds include approved drugs currently used clinically for the therapy of paediatric ALL, such as vincristine and daunorubicin (Pui, 2012), as well as compounds previously reported to be active in vitro or in preclinical models of various types of cancers, such as apoptosis inducer, CHM-1 in hepatocellular carcinoma (Wang et al., 2008).

To further select for compounds that were selective for leukaemia and did not affect the viability of non-cancerous cells, the 28 compounds were evaluated in a full-dose response screen using 72-hour resazurin-based cytotoxicity assays against a human neuroblastoma cell line, KELLY and a non-malignant human lung fibroblast cell line, MRC-5, in addition to the cancer cell lines used in the primary screen, PER-485 and CEM. Results of the full-dose screen are shown in Figure 3.8 as a heat map showing

IC50 of each compound for the four cell lines. IC50 values of the compounds range from less than 1 nM to above 200 nM in all cell lines. In general, the PER-485 cells were actually found to be less sensitive to the selected compounds than the other cancer cell lines, which is consistent with the highly aggressive characteristic of MLL-r leukaemia and the resistant nature of the cells towards chemotherapeutics. In addition, no clear leukaemia-specificity was observed for any of the 28 compounds. The 28 compounds were further prioritized based on compound toxicity to the control cell line, MRC-5. Nine out of 28 compounds were equally cytotoxic to the MRC-5 as any of the leukaemia cells. Ten out of 28 presented with an IC50 for MRC-5 cells greater than two

77 times the IC50 for either of the leukaemia cell lines (Figure 3.8). Six of these compounds have previously been tested in paediatric leukaemia clinical trials or are currently under investigation, namely, vincristine and daunorubicin (standard paediatric ALL therapy) (Pui, 2012; Brown, 2013), mitoxantrone (Interfant-06) (Brown, 2013), clofarabine (NHL16; NCT01451515) (Burkhardt et al., 2016), bortezomib (NCT00440726) (Messinger et al., 2012; Bertaina et al., 2017), and topotecan (Hijiya et al., 2008; Inaba et al., 2010), validating the performed drug screen. Camptothecine, a topoisomerase inhibitor was in early clinical trials for several solid tumours in the 1960s and 70s but was stopped due to severe toxicity (Masuda et al., 1996). SN38 is the active metabolite of irinotecan, an analogue of camptothecine previously tested in refractory or relapsed leukaemia in children and adolescents (NCT01239485) (Saletta et al., 2014) and has been studied extensively pre-clinically in leukaemia (Cohen et al., 1999; Adams et al., 2008; Cardillo et al., 2018). Since the aim of this study was to identify novel compounds for potential therapy against high-risk leukaemia, the two remaining short- listed compounds, auranofin and disulfiram, which are FDA-approved drugs for the treatment of rheumatoid arthritis and chronic alcoholism respectively, were chosen for further study as potential anti-leukaemia agents (summarized in Figure 3.9). Detailed characterization of the effect of these drugs against high-risk leukaemia cells is the subject of Chapter 5.

Figure 3.8: Characterization of 28 potent compounds in a full dose response screen. Cell viability was determined after a 72-hour drug exposure at four concentrations (2, 1,

0.5 and 0.25 µM). Heat map of concentration with 50% cell growth reduction (IC50) shown according to the colour key. Ten compounds with lower toxicity to normal cells were identified with a majority already tested in clinical trials for paediatric leukaemia (bolded). This includes two conventional drugs that are currently used in standard therapy of paediatric ALL, vincristine and daunorubicin (marked with *). Auranofin and disulfiram were selected for further testing based on their novelty as therapeutic options for leukaemia.

Primary screen Cell lines: MLL-r PER-485 and MLL-wt CEM. Add drugs at 5 µM; Incubate for 72 hour

Hit identification 184 compounds with >90% growth inhibition at 5 µM in both cell lines

Hit validation Validated 184 compounds in a 4-point serial dilution (2, 1, 0.5, 0.25 µM) to identify MLL- selective compounds at lower doses Full-dose response screen No MLL-selective compounds identified. However, 28 compounds with potent activity in both leukaemia cell lines (IC50 < 0.25 µM), were screened against 4 cell lines Selection of final compounds 2 compounds showed novel activity against high-risk leukaemia cell lines. Neither have been clinically used in the treatment of ALL

Figure 3.9: Summary of high-throughput screening of a library of approved drugs and pharmacologically active compounds to identify potent compounds for high- risk leukaemia (Strategy II). Schematic outline of screening performed for the selection of two more potent compounds for high-risk leukaemia. Hits were identified from the same primary screen previously shown (Figure 3.2).

3.3 Discussion

Failure to increase the survival rates of patients suffering from high-risk leukaemia subgroups despite increasingly aggressive treatment regimens indicates the urgent need to develop novel, more potent and safer therapies. Drug repurposing has recently emerged as a strategy to overcome some of the challenges associated with traditional drug discovery approaches and to accelerate drug development. The aim of this study was to identify potential novel therapeutics targeting high-risk leukaemia in general, and MLL-rearranged leukaemia specifically, through a high-throughput phenotypic screen of a library composed of clinically approved and/or pharmacologically active compounds. The screen was performed using two strategies with Strategy I focusing on identifying MLL-selective compounds and Strategy II aimed at discovering more potent compounds for high-risk leukaemia. Each strategy yielded two candidate compounds for further investigation. An overall summary of screenings performed for this thesis is added into Appendix as Supplementary Figure 2.

For the first strategy, 503 compounds exhibited selective killing of a MLL-r leukaemia cell line over a MLL-wt leukaemia line in the primary screen. However, after validation in a secondary screen, only 26 compounds were confirmed as truly more selective for the MLL-r leukaemia cell line. High false positive rates have been reported in many HTS and are one of the drawbacks associated with the technology as large numbers of compounds are tested simultaneously at a single concentration (Xie, 2010; Bibette, 2012; Feng et al., 2017). It illustrates the importance of biological replicates in primary screens and the crucial confirmatory secondary screens for eliminating these errors. The major justification concerning the absence of biological replicates are usually limitations in resources and budget constraints (i.e. feasibility), as including replicates would significantly reduce the number of screening compounds. However, the gains in precision may be significant and thus outweigh the immediate cost and time considerations (Xie, 2010). Z-factor for each assay plate was calculated to determine the quality of high-throughput screening by assessing any technical, procedural or environmental factors that could cause systemic measurement error (Zhang et al., 2012). Factors above 0.5 indicate highly robust assays (Zhang et al., 1999). In the primary and both secondary screens performed, the calculated Z-factor was higher than

0.5. There was also low variability between replicates in the secondary assays demonstrating good data reproducibility.

In Strategy I, PER-485 cells generally displayed the highest sensitivity among all other lines and only upon subsequent screening that included additional MLL-r leukaemia cells, were a majority of short-listed compounds found to be non-selective of MLL-r leukaemia. Among the 26 candidate MLL-selective - compounds yielded from the screen, a handful clustered in several drug classes such as histamine receptor agonists, dopamine receptor agonists, MEK inhibitors and purinergic receptor agonists. Whilst none of the seven compounds showed complete selectivity for MLL-r cells compared to MLL-wt cell lines, a dichotomy of responses was seen amongst the MLL-r cell lines tested. The genetic diversity of the MLL-r cell lines, could be masking the MLL-r- specific vulnerability. Molecular alterations could affect in vitro behavior, adaptation and response of the cell lines to any of the lead compounds, resulting in the low numbers of compounds identified with anti-MLL activity. The status of tumour suppressor gene, TP53 is one of the genetic confounders that could influence drug response, as mutations and deletions to the gene lead to disruption in regulation to various signaling pathways such as cell proliferation, apoptosis and DNA repair (Quintas-Cardama et al., 2017), thus could allow escape from compounds that would otherwise disable key MLL-associated pathways. Although the prevalence of p53 mutation is significantly lower in haematologic malignancies compared to other tumours such as skin and lung cancers (10% vs. 50 – 100%) (Rivlin et al., 2011), TP53 alterations have been established to increase drug resistance and negatively affect survival of patients with haematologic malignancies (Wattel et al., 1994; Peller and Rotter, 2003; Stengel et al., 2017). Unfortunately in this study, apart from leukaemia subtype, lineage and MLL translocation, the genetic characteristics known about the cell lines panel is very limited. Only six cell lines have known TP53 status, three (RS4;11, MOLM-13 and REH) were wild-type and three (THP-1, U937 and CEM) were mutants. Nevertheless, based on these data, selectivity cannot be explained by the subtype, MLL translocation or the TP53 status (Table 3.1). Varied sensitivities among MLL-r leukaemia cell lines have been reported in previous screens (Hoeksema et al., 2011; Somers et al., 2016; Cruickshank et al., 2017), indicating the heterogeneity of the disease and emphasizing the need of targeted therapies for these subgroups of patients.

The diversity of responses was particularly apparent for 2-Cl-ATP and SID, which showed higher activity in additional MLL-r leukaemia cell lines compared to other compounds, and therefore were selected for further evaluation. 2-Cl-ATP, a P2Y receptor agonist, has not been widely reported in the literature, however one study found it to exert anti-tumour activities in phaeochromocytoma, neuroblastoma and glioma cell lines (D’Ambrosi et al., 2004), while SID, an inhibitor of SF-1/NR5A1, a transcription factor that plays an essential role in adrenal and gonadal development and function, has been shown to reduce proliferation of adrenocortical carcinoma cells (Doghman et al., 2010). However, due to later findings demonstrating poor drug-like qualities of both compounds (see Chapter 4) and because neither compound had been previously used in animal models or in human trials, another secondary screen was performed to identify more potent and drug-like compounds.

184 compounds toxic to both PER-485 and CEM in the primary screen were re- screened in the same cell lines at lower concentrations to identify compounds that may show a window of MLL-selectivity when used at a lower dose range. However, MLL- selectivity was not observed for any of these compounds. Nevertheless, 28 compounds showed potent cytotoxic activity against both high-risk leukaemia cell lines, with IC50 ranging from below 1 nM to more than 200 nM. Most of the compounds could be classified in three classes: nine microtubule inhibitors, five topoisomerase inhibitors or DNA intercalating agents, four sodium and potassium ATPase inhibitors or cardiotonic agents that are involved with regulation of myocardial contractility. In addition, amongst the 28 hits, two protein synthesis inhibitors, two proteasome inhibitors and two apoptosis inducers were identified. Interestingly, a recent drug screen of 101 FDA- approved chemotherapeutics also reported proteasome inhibitors, in addition to histone deacetylase (HDAC) and cyclin dependent kinase inhibitors, to have potent activities against MLL-r ALL cells in vitro (Cruickshank et al., 2017). Whilst 58 drugs from the Cruickshank library overlapped with our pharmacologically active compound library, only nine of these (doxetaxel, paclitaxel, vinblastine, vincristine, daunorubicin, mitoxantrone, topotecan, bortezomib, clofarabine) were among the 28 short-listed compounds. The other 49 drugs were filtered out during the hit validation stage (Figure

3.9) due to having estimated IC50 values above 250 nM. Among the nine, one was the proteasome inhibitor, bortezomib. The same nine drugs were also in the NCI/NIH

Developmental Therapeutic Program Approved Oncology Drug Set II screened by Hoeksema et al. (2011), whereby five of these (doxetaxel, vinblastine, daunorubicin, mitoxantrone, topotecan) were among the 12 compounds found to be active in all MLL- r cell lines tested in their screen. Three of these (daunorubicin, mitoxantrone, topotecan) were among our top ten compounds short-listed as being potent against high-risk leukaemia. The detection of these nine drugs in the studies by both Hoeksema et al. (2011) and Cruickshank et al. (2017) further validated our screening strategy in identifying novel therapies for MLL-r leukaemia as their screenings were performed against five and eight MLL-r cell lines, respectively, substantiating their results. In the current screen, the identification of two drugs currently used as paediatric ALL therapeutics and four drugs under investigation in paediatric clinical trials out of the ten hits lends further support to the strength of the drug screening design. Moreover, the screen highlighted two novel drugs worthy of further investigation as potential drugs against high-risk ALL, auranofin and disulfiram.

Auranofin and disulfiram, FDA-approved drugs for the treatment of rheumatoid arthritis and alcoholism respectively have been reported to have anti-cancer activity in several tumour types (Morrison et al., 2010; Nakaya et al., 2011; Kim et al., 2013; Kast et al., 2014). Interestingly, the anti-cancer activities of both drugs have been previously identified in drug repurposing attempts involving high-throughput screening. Auranofin was selected as a therapeutic candidate for CLL from a HTS of the National Institute of Health Chemical Genomic Centre pharmaceutical library consisting of 2816 bioactive compounds and FDA-approved drugs (Shen et al., 2013), while disulfiram was identified to have cytotoxic effects in glioma cells from a HTS of 446 small-molecule compounds that have been used in clinical trials (Lun et al., 2016). These findings not only supported our drug screen method but also reinforced the capacity of repurposing existing drugs for highly aggressive diseases and the potential of auranofin and disulfiram as therapeutic candidates for high-risk leukaemia. Although the drugs are classified as an anti-inflammatory gold salt in the case of auranofin and a proteasome inhibitor in the case of disulfiram, and have various additional targets and functions, the interesting commonality between both compounds is that they have been reported to exert their anti-cancer function through reactive oxygen species (ROS) induction in several cancers (Fiskus et al., 2014; Zou et al., 2015; Zha et al., 2014; Allensworth et

84 al., 2015). The cytotoxic effects of auranofin and disulfiram in high-risk leukaemia will be further explored in Chapter 5.

Overall, the screens have led to the identification of interesting candidates to be further investigated for the treatment of MLL-r and high-risk ALL. However, there were some limitations with the screens. The use of cell lines of different immunophenotypes, a T- ALL cell line, CEM and a pre-B ALL cell line, PER-485 might have affected the primary screen results by selecting hits with differential targets between B and T cell lineages rather than MLL rearrangement specifically. However, when the hit compounds were screened against a larger panel of leukaemia cell lines, no lineage selectivity was seen. The use of high-risk leukaemia CEM cells as a ‘control/reference’ cell line might have also decreased the number of MLL-selective candidate compounds for Strategy I as the highly resistant CEM cells would have been less affected by low and moderately active compounds. Using a pre-B ALL cell line for the screening could have resulted in a greater number of leads as there would be larger difference in cell viability between the MLL-wt and MLL-r cell lines. Nevertheless, the combination of high-risk ALL CEM and PER-485 cell lines enabled detection of highly potent compounds for the second aim of this study using Strategy II, selecting FDA-approved drugs currently used in the treatment of childhood ALL as discussed above, as well as interesting potential leads for the treatment of high-risk leukaemia.

Another limitation of this study was the potential loss of hit compounds that target other MLL translocations during the primary screen due to the HTS being performed with only one MLL-r ALL cell line, PER-485. Due to heterogeneity of MLL-r ALL (as discussed in Section 1.4.1), inclusion of an additional MLL-r ALL cell line in the primary screen, such as KOPN-8, that harbours t(11;19) translocation, the second most common gene fusion in infant MLL-r ALL (Meyer et al., 2018), might increase the chance of uncovering novel MLL-selective compounds and thus the statistical power of the screen. However as mentioned above, this would increase the cost significantly. Besides cell lines, studies have also shown the potential of using clinically relevant xenograft cells in HTS to identify novel compounds sensitizing chemo-resistant leukaemia cells (Pemovska et al., 2013; Toscan et al., 2014). In one study, a dexamethasone-resistant xenograft was screened against 40,000 compounds to select

85 dexamethasone sensitizers by firstly adding the test compounds from the drug library to the cells, followed by dexamethasone. This combination screen identified 2-(4- chlorophenoxy)-2-methyl-N-(2-(piperidin-1-yl)phenyl)propanamide, which demonstrated synergy with dexamethasone and was able to reverse glucocorticoid resistance in four out of five resistant xenografts tested (Toscan et al., 2014). This screening design could potentially identify highly relevant MLL-targeting hit compounds.

The size of the chemical library used in this study is significantly smaller than many HTS of small molecules that usually consist tens of thousands of compounds (Toscan et al., 2014; Somers et al., 2016), lowering the statistical power of the screening. However, due to the aim of the project which was to identify existing drugs or bioactive compounds that could be repurposed for MLL-r ALL or high-risk leukaemia, the number of libraries that could be screened is limited. Libraries of similar or smaller size were also reported in several repurposing screens such as Hoeksema et al. (2011), Lun et al. (2016) and Cruickshank et al. (2017) whereby the chemical libraries used had less than 500 compounds. Selection of chemical libraries for drug repurposing that consist of bioactive compounds that have not necessarily already been used in vivo should also be taken into consideration for future screenings. Although the compounds could potentially uncover novel targetable mechanisms and pathways, they could also restrict rapid progression to investigations in animal models due to further chemical and structural optimization being required. Recently, several research groups came together and created the Drug Repurposing Hub, an information resource and comprehensive drug-screening library, containing approximately 4700 compounds with details on chemical structure, mechanism of action, clinical trial status, approved indications, vendor and others (Corsello et al., 2017). The database is expected to have an enormous impact on drug repurposing pipeline and thus accelerate advancement of existing drugs into the clinic.

In summary, the two strategies taken for high-throughput screening of libraries of approved drugs and pharmacologically active compounds yielded two categories of candidates worth further study. Strategy I yielded two compounds with apparent selectivity for MLL-r leukaemia, 2-chloroadenosine triphosphate and SID7969543, but

86 each of these compounds had the drawback of having IC50 in the micromolar range among the sensitive cell lines necessitating further optimization if the compounds were to be tested in vivo. In Chapter 4, these compounds are studied in more detail in order to better characterize their activity and potentially reveal new targetable pathways in MLL-r ALL. Further investigation of compounds that were non-selective at 5 uM in Strategy II revealed a number of drugs with higher potency and potentially a faster path to the clinic. Two of these compounds will be investigated further in Chapter 5.

CHAPTER 4: CHARACTERIZATION OF NOVEL CANDIDATE MLL-SELECTIVE INHIBITORS

4.1 Introduction

In the previous chapter, two potential MLL-selective candidates were identified from a HTS of approved drugs and pharmacologically active compounds. These compounds were 2-chloroadenosine triphosphate (2-Cl-ATP) and SID7969543 (SID). 2-Cl-ATP is a P2Y receptor purinergic agonist and an anti-convulsant that commercially exists in a tetrasodium salt form (Figure 4.1). The compound has not been not been tested in vivo and not widely reported in the literature, however one study found 2-Cl-ATP to exert anti-tumour activities in phaeochromocytoma, neuroblastoma and glioma cell lines (D’Ambrosi et al., 2004). 2-Cl-ATP is also an ATP analogue and a phosphorylated derivative of 2-chloroadenosine (2-CADO), a well-known adenosine receptor agonist that has been reported to have cytotoxic effects in vitro against several cancer cell types such as astrocytoma (Ceruti et al., 2003), prostate cancer (Minelli et al., 2006) and chronic lymphocytic leukaemia (Bastin-Coyette et al., 2008). Another class of purinergic receptor is adenosine or P1 receptors. P1 and P2Y receptors are preferentially activated by the release of adenosine, and ATP respectively (King and Burnstock, 2002), however being a derivative to 2-chloroadenosine, 2-Cl-ATP could possibly affect both types of receptors. These receptors have been reported to be involved in an array of cellular responses including cell growth, differentiation and apoptosis (Burnstock and Di Virgilio, 2013). In PC12nnr5 cells, a cell line derived from a rat neuroendocrine tumour of the adrenal gland, cytotoxicity of 2-Cl-ATP was independent of P1 and P2 receptors, but relied on hydrolysis of 2-Cl-ATP into 2-CADO by serum ATPase. Once converted into 2-CADO, toxicity depended on its intracellular phosphorylation, followed by inhibition of DNA synthesis and hence cell cycle arrest and death (D’Ambrosi et al., 2004).

The second candidate MLL-selective hit compound yielded from this HTS was a Steroidogenic Factor-1 (SF-1/NR5A1) inhibitor, SID7969543. The structure of this compound is shown in Figure 4.12. SID, an isoquinolinone derivative, was first identified from ultra-high-throughput screening of approximately 65,000 compounds

88 from the National Institute of Health’s Molecular Libraries Small Molecule Repository in a cell-based transactivation assay (Madoux et al., 2008), in an effort to search for potent and selective SF-1 inhibitors. The transcription factor SF-1 regulates steroidogenic enzymes and plays key roles in adrenal and reproductive development and functions (Ferraz-de-Souza et al., 2011). Tissues in steroidogenic organs (adrenal cortex and gonads) are the major sites of SF-1 expression. In gene knockout mice, lack of sf-1 resulted in the loss of adrenal glands and gonads (Hoivik et al., 2010). Recently, the nuclear factor was reported to regulate growth of ovarian surface epithelial cancer cells (Ramayya et al., 2010). In adrenocortical carcinoma (ACC), overexpression of SF- 1 increased cell proliferation and decreased apoptosis in vitro. Transgenic mice bearing multiple copies of sf-1 rapidly develop adrenocortical hyperplasia and subsequent tumour formation, indicating its pivotal role in the development of ACC (Doghman et al., 2007; Doghman et al., 2010). In a human adrenocortical cell line expressing SF-1, H295R, SID significantly inhibited cell proliferation in the presence of doxorubicin or increased SF-1 dosage (Doghman et al., 2010). However, very little is known about the role of SF-1 if any in leukaemia cells, and like 2-Cl-ATP, the SF-1 inhibitory compound has not been tested in vivo.

In this chapter, both compounds were characterized for their potential as anti-MLL agents by determining their potency and selectivity towards MLL-r leukaemia cells through testing their effects on the viability of an expanded panel of cell lines. The activity of the compounds was further evaluated in cells derived from xenografts of patients with MLL-r ALL. Activity in peripheral blood mononuclear cells was also evaluated to establish a therapeutic window for the compounds. The compounds were additionally be tested in combination with currently used chemotherapeutics to determine if they offer any potentiating effects. Finally, in vivo stability of 2-Cl-ATP and SID was assessed through microsomal assay to reveal their ability to withstand metabolic degradation if further progressed into preclinical and clinical studies.

4.2 Results

4.2.1 Characterization of 2-chloroadenosine triphosphate as a selective inhibitor of MLL-rearranged leukaemia cells

4.2.1.1 Selectivity of 2-chloroadenosine triphosphate towards MLL-rearranged leukaemia

Initial testing described in Chapter 3 revealed the activity of 2-Cl-ATP against MLL-r and CALM-AF10 leukaemia cell lines, while not affecting MLL-wt leukaemia lines. To further characterize the selectivity of the compound, it was tested in an expanded cell line panel that included four additional cell lines derived from infants diagnosed with MLL-r ALL, seven cell lines derived from solid tumours including neuroblastoma, breast cancer and lung carcinoma, and an additional non-malignant cell line (in total: MLL-r n=11, CALM-AF10 n=2, MLL-wt n=3, solid tumours n=7, non-malignant n=2) (Table 2.1). The compound showed inhibitory activity towards each of the additional MLL-r cell lines, while not affecting solid tumours or non-malignant cells up to a concentration of 20 μM (Figure 4.2, Table 4.1). Eight out of 11 MLL-r cells had IC50 below 10 uM with three out of 11 MLL-r cells (MV4;11, THP-1 and KOPN-8) being less responsive or resistant to 2-Cl-ATP with IC50 above 10 uM (Figure 4.3A). The cell lines were grouped according to lineage and MLL status, and viability at 10 μM was compared. The mean percentage viability at 10 μM was significantly lower for MLL-r cells compared to solid tumours and non-malignant cells, although only a trend was seen when viability percentages for MLL-wt cell lines were compared to MLL-r cells, possibly due to the small number of MLL-wt cell lines and the variation in response among the MLL-r cell lines (Figure 4.3B).

Figure 4.1: Chemical structure of 2-chloroadenosine triphosphate.

1 2 5 M V 4 ;1 1 U 9 3 7

1 0 0 R S 4 ;1 1 K P M O T S )

S E M K 2 C E M

% (

7 5 y

t M O L M -1 3 R E H

l i

b 5 0 T H P -1 J U R K A T

a i

V K O P N 8 2 5 P E R -4 8 5

0 P E R -4 9 0 -6 .5 -6 .0 -5 .5 -5 .0 P E R -7 8 5 A 2 -C l-A T P c o n c e n tr a tio n [L o g M ] P E R -7 0 3 A

P E R -8 2 6 A

Figure 4.2: Cytotoxicity of 2-chloroadenosine triphosphate across a range of MLL- rearranged, CALM-AF10 and MLL-wild-type leukaemia cell lines. Dose response curves for a range of MLL-rearranged (red), CALM-AF10 (blue) and MLL-wild-type (green) leukaemia cell lines exposed to 2-chloroadenosine triphosphate (2-Cl-ATP) for 72 hours as measured by resazurin-based assay. The results are expressed as the mean ± SE of three independent experiments. Details of cell lines are found in Table 2.1 and IC50 in Table 4.1.

Table 4.1: Cytotoxicity of 2-chloroadenosine triphosphate in a panel of 25 cell lines.

Cell line Translocation Disease IC50 (µM)

PER-485 t(4;11) Infant ALL 2.5 PER-490 t(4;11) Infant ALL 4.5 PER-703A t(1;11) Infant ALL 4.3 PER-785A t(4;11) Infant ALL 3.5 ALL PER-826A Complex, t(11;19) Infant ALL 2.6 MLL-rearranged RS4;11 t(4;11) Pre-B cell ALL 5.9 SEMK2 t(4;11) Pre-B cell childhood ALL 5.1 KOPN-8 t(11;19) Infant pre-B cell ALL 13.4 MV4;11 t(4;11) Childhood AML >20 AML MOLM-13 t(9;11) AML 5.6 THP-1 t(9;11) Infant AML >20 CEM - Childhood T-cell ALL 11.4 ALL REH - Pre-B cell ALL >20 MLL-wild-type Jurkat - Childhood T-cell ALL >20 U937 CALM-AF10 AML 3.5 AML KP-MO-TS CALM-AF10 AML 4.4 KELLY - Neuroblastoma >20

BE(2)C - Childhood neuroblastoma >20

HEY - Ovarian carcinoma >20

Solid tumours 27/87 - Endometrioid ovarian cancer >20

MCF-7 - Breast adenocarcinoma >20

H460 - Lung carcinoma >20

LNCaP - Prostate carcinoma 14.9

MRC-5 - Normal lung >20 Normal cells WI-38 - Normal lung >20

Figure 4.3: Comparison of viability between cell lines treated with 10 µM 2- chloroadenosine triphosphate. (A) Viability of cell line panel at 10 µM of 2-chloroadenosine triphosphate in 72-hour resazurin-based cytotoxicity assay. (B) Comparison of cell viability at 10 µM between cell lines grouped according to type: MLL-rearranged (MLL-r), CALM-AF10, MLL- wild-type (MLL-wt), solid tumour and normal cell lines. Mean viability between groups was compared by one-way ANOVA with Dunn’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, p<0.05; **, p<0.01. 93

To determine whether 2-Cl-ATP affected the viability of the cells cells through induction of apoptosis, Annexin V staining was performed. A significant increase in the percentage of apoptotic cells was shown upon treatment of PER-485 cells with 2-Cl- ATP for 48 hours (Figure 4.4A; untreated control vs. 48 hours, p<0.0001) which increased to more than 30% of cells after 72 hours of treatment (Figure 4.4A; untreated control vs. 72 hours, p<0.0001). Thus, 2-Cl-ATP had selective cytotoxicity towards a subset of MLL-r leukaemia cell lines through apoptotic cell death.

Figure 4.4: Apoptosis in PER-485 cells over 72-hour treatment course of 2- chloroadenosine triphosphate. (A) Mean percentage increases of Annexin V-positive cells compared to vehicle control at each of the indicated time points after treatment with 2-chloroadenosine triphosphate (2-ClATP). The results are expressed as the mean ± SE of three independent experiments. Means were compared by one-way ANOVA. Asterisks represent significance levels of P-values. ****, p<0.0001; (B) Representative data for flow cytometric analysis of Annexin V/7AAD staining in PER-485 cells exposed to 2-Cl- ATP over a 72-hour treatment course.

4.2.1.2 Determination of synergy between conventional chemotherapeutics and 2- chloroadenosine triphosphate in vitro

The range of in vitro IC50 for a drug will influence the ability to achieve sufficiently high levels of active compound in animal or in patient plasma for in vivo treatment efficacy. As IC50 values for 2-Cl-ATP were in the micromolar range, combination assays were performed with drugs currently used in the treatment of paediatric ALL in an effort to identify combinations that might allow reduction in the concentration of 2- Cl-ATP required to affect MLL-r cell viability. In addition, potential synergistic or additive effects between agents could significantly reduce individual drug concentrations which in turn minimizes side effects, increases potential clinical utility and delays or prevents resistance development for targeted therapies (Sun et al., 2016).

4.2.1.2 (i) Response of leukaemia cell lines to conventional chemotherapeutics

To increase the efficacy of 2-Cl-ATP and determine the effects of the compound when used in combination with drug used in ALL patient therapy, five conventional chemotherapeutics currently used chemotherapy agents in paediatric leukaemia clinical trials were selected for subsequent testing in combination assays. Mitoxantrone, daunorubicin, etoposide and cytarabine are in use in the Interfant-06 trial (Brown, 2013), and topotecan in the NCT00187083 study (Hijiya et al., 2008). As a precursor to combination studies, it was necessary to determine the single agent in vitro activity of these drugs in the leukaemia cell line panel. Thus each drug was tested against eight MLL-rearranged leukaemia cell lines with different MLL translocations, two CALM- AF10 leukaemia cell lines and three MLL-wild-type leukaemia cell lines in a 72-hour resazurin-based cytotoxicity assay.

Dose response curves for the five chemotherapeutics are shown in Figure 4.5 and IC50 values for each cell line can be found in Table 4.2. The leukaemia cell line panel demonstrated a broad range of responsiveness to the drugs. MV4;11 is among the most resistant MLL-r cell lines, with the highest IC50 for mitoxantrone, daunorubicin and etoposide, and second highest IC50 for topotecan. THP-1 had the highest IC50 for cytarabine and second highest IC50 for etoposide. Infant cell lines, PER-485 and PER-

490 also had among the highest IC50 for etoposide and cytarabine. The KP-MO-TS cell line with CALM-AF10 translocation, had high IC50 for daunorubicin, etoposide and cytarabine and Jurkat was the most resistant MLL-wt cell line, with high IC50 for all drugs except cytarabine. The drug response profile of the leukaemia panel was plotted in order to visualize the range of responses for each drug (Figure 4.6). The least variability in responsiveness across all cell lines was observed with topotecan, whilst response to cytarabine showed the widest spectrum of sensitivity with IC50 ranging from 14 nM – 5.5 μM (Taable 4.2). Based on these drug response data, combination assays were performed as described in the next section.

Figure 4.5: Cytotoxicity of five conventional chemotherapeutics currently used in paediatric ALL therapy, against a leukaemia cell line panel. Dose response curves of leukaemia cell lines for (A) mitoxantrone (B) daunorubicin (C) etoposide (D) cytarabine and (E) topotecan in MLL-rearranged (red), CALM-AF10 (blue) and MLL-wild-type (green) leukaemia cell lines treated for 72 hours before viability was assessed through a resazurin-based assay. The results are expressed as the mean ± SE of three independent experiments.

Table 4.2: IC50 values for chemotherapeutics against a leukaemia cell line panel.

Mitoxantrone Daunorubicin Etoposide Cytarabine Topotecan Cell line Translocation (nM) (nM) (nM) (nM) (nM) PER-485 23 26 877 4424 23 PER-490 13 18 616 4235 16 MV4;11 4;11 600 220 5000 500 125 RS4;11 3 6 104 46 6 SEMK2 6 8 146 169 8 MOLM-13 2 11 49 116 10 9;11 THP-1 30 38 1092 5492 36 KOPN-8 11;19 3 7 103 79 6 U937 11 17 311 25 17 CALM-AF10 KP-MO-TS 20 92 705 699 12 CEM 12 21 313 24 21 REH Wild-type 2 5 74 14 7 Jurkat 366 139 366 156 250

IC50: inhibitory concentration resulting in 50% reduction of cell survival relative to control, IC50 values were derived from the mean of three independent experiments.

1 5

) M

n 1 0

(

g 5

o L

a M D Figure 4.6: Drug response profile of leukaemia cell line panel. Boxes indicate inter-quartile range with median (horizontal line) and whiskers indicate th th 10 and 90 percentiles. Values are calculated as log2 IC50 (nM).

4.2.1.2 (ii) Combination of conventional chemotherapeutics with 2-chloroadenosine triphosphate

To determine if 2-Cl-ATP can potentiate any of the five chemotherapy agents selected in previous section, in vitro synergy assays were performed in two cell lines, PER-485 and MOLM-13. PER-485 was selected as a representative of MLL-r leukaemia cell line characterized by high chemoresistance, while MOLM-13 was selected as a more chemoresponsive MLL-r cell line. Cells were subjected to increasing doses of 2-Cl-ATP alone, of a chemotherapy agent alone or a combination of the two simultaneously at a fixed-ratio for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay.

CalcuSyn was used to determine whether a synergistic, additive or antagonistic effect occurred between drugs in the combination studies (Chou, 2006). Combination indices (CI) were calculated at effective dose 75 (ED75) or the concentration causing 75% effect for biological effect of interest, which in this combination assay, is decrease in viability. A CI below 0.9 indicates a synergistic interaction between the tested drugs, 0.9 to 1.1 indicates an additive effect and above 1.1 an antagonistic effect (Table 2.3). The calculated indices revealed low to moderate synergy between 2-Cl-ATP with mitoxantrone, etoposide and cytarabine in both cell lines. Daunorubicin showed clear synergy with 2-Cl-ATP in MOLM-13 cells but the effect was only additive in PER-485 cells. Topotecan on the other hand did not exhibit synergism with 2-Cl-ATP in either cell line (Table 4.3, Supplementary Figure 4 and 5).

The drug reduction index (DRI) is a measure of how many fold the dose of each drug can be reduced in a synergistic combination to achieve a similar effect as if the drug were used alone (Chou, 2006). DRI was calculated to determine whether the use of 2- Cl-ATP could effectively be used to reduce the dose of each chemotherapeutic agents. Based on the results in Table 4.3, 2-Cl-ATP could reduce the dose of mitoxantrone by at least 2-fold in both cell lines, which could potentially translate to a reduction of side effects due to the use of lower concentrations of chemotherapeutics for similar effect. Similarly, the all except one chemotherapeutic, topotecan could also reduce the concentration of 2-Cl-ATP by at least 2-fold in both cell lines.

100

Table 4.3: Calculated combination index and drug reduction index for 2- chloroadenosine triphosphate in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and MOLM-13 cells. PER-485 MOLM-13 Chemodrugs DRI DRI CI CI 2-Cl-ATP Chemodrug 2-Cl-ATP Chemodrug Mitoxantrone 0.659 2.546 3.784 0.864 2.098 2.612 Daunorubicin 0.929 2.141 2.218 0.34 1.693 2.944 Etoposide 0.721 1.911 5.337 0.798 1.896 3.814 Cytarabine 0.741 1.967 3.401 0.898 2.141 2.329 Topotecan 1.139 1.870 1.680 1.000 2.312 1.770

2-Cl-ATP: 2-chloroadenosine triphosphate, CI: combination index; DRI: drug reduction index, ED75: effective dose causing 75% reduction of cell viability. Cells were subjected to increasing doses of 2-Cl- ATP alone, of a chemotherapy agent alone or a combination of the two simultaneously at a fixed-ratio for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. Drug reduction index indicates fold reduction in dose of indicated drug when 2-Cl-ATP is applied, and vice versa.

101

4.2.1.3 Effect of 2-chloroadenosine triphosphate on infant MLL-rearranged ALL patient-derived xenograft cells in vitro

Patient-derived xenografts (PDX) constitute highly relevant preclinical models in drug development. PDX are patient-derived cells that have been serially passaged in mice, in which the immunophenotypic and genotypic characteristics of these cells have been shown to remain stable and essentially unaltered throughout the in vivo expansion (Lock et al., 2002; Liem et al., 2004). To determine the activity of 2-Cl-ATP against leukaemia PDX cells in vitro, cytotoxicity assays were performed with six xenografts derived from infants diagnosed with MLL-rearranged ALL. The panel of xenografts consisted PDX with MLL-AF4, MLL-ENL and MLL-MLLT10 translocations respectively (Table 4.4) (Henderson et al., 2008; Richmond et al., 2015). Cell viability was measured after 48 hours of compound exposure in resazurin-based cytotoxicity assays. A 48-hour timepoint was selected due to the PDX cells being more susceptible to cell death in the ex vivo setting compared to cell lines, possibly due to the lack of proliferation or low proliferating rates (Liem et al., 2004). Four PDXs were highly sensitive, with IC50 within the range of 2.5 – 4.8 μM and two were more resistant

(Figure 4.7). MLL-5 cells were the most resistant to 2-Cl-ATP with an IC50 of 11.5 μM, consistent with previous data on this xenograft indicating the highly resistant nature of this PDX (Richmond et al., 2015; Khaw et al., 2016). These results show that the compound not only affects leukaemia cell lines but also highly relevant patient-derived xenografts that closely represent the original disease.

102

Table 4.4: Translocations present in panel of infant MLL-rearranged ALL PDXs. Translocation Xenograft MLL-AF4 – t(4;11) MLL-2 MLL-7 MLL-ENL – t(11;19) MLL-6 MLL-8 MLL-14 MLL-MLLT10 – t(10;11) MLL-5

See Table 2.2 for additional information.

M L L -r P D X

1 2 5

1 0 0 M L L -2 (IC 5 0 = 4 .8  M )

) M L L - 5 ( 1 1 .5  M )

% (

7 5

y M L L - 6 ( 4 .5  M )

i l i M L L - 7 ( 9 .7  M )

b 5 0 a

i M L L - 8 ( 2 .5  M ) V

2 5 M L L - 1 4 ( 2 .7  M )

0 -6 -5 -4 2 -C l-A T P c o n c e n tr a tio n [L o g M ]

Figure 4.7: Cytotoxicity of 2-chloroadenosine triphosphate in infant MLL- rearranged ALL patient-derived xenograft cells in vitro. Dose response curves for infant MLL-rearranged ALL patient derived xenografts (MLL-r PDX) upon 48-hour in vitro treatment with 2-chloroadenosine triphosphate (2- Cl-ATP) as measured by resazurin-based viability assay. The results are expressed as the mean ± SE of three independent experiments and IC50 values were derived from the mean of these experiments.

103

4.2.1.4 Markers of response to 2-chloroadenosine triphosphate in infant MLL- rearranged ALL patient-derived xenografts and patient samples

The cytotoxicity assays conducted so far have indicated a degree of variability in response in of MLL-r leukaemia cell lines and PDXs to 2-Cl-ATP. To investigate the basis of this variability, basal expression levels of potential targets of 2-Cl-ATP, five P1 and nine P2Y receptors, in the MLL-r PDX panel (n=6; available on the Gene

Expression Omnibus site (GSE52991)), were analyzed in relation to their IC50. Pearson product-moment correlation coefficient was calculated between the gene expression data of each receptor and sensitivity of the xenograft panel (IC50) to test for possible associations between the two factors. None of the P1 receptors showed any correlation between gene expression and PDX sensitivity, while expression of two P2Y receptors,

P2RY6 and P2RY14 were significantly inversely correlated with sensitivity towards the compound (Table 4.5). To determine if these receptors are differentially expressed in MLL-r leukaemia patients compared to non-MLL-r leukaemia patients, the microarray dataset of leukaemia patients from Stam et al. (2010) was analyzed. The cohort had 59 MLL-r leukaemia patients and 14 MLL-wt patients. The expression levels of most of the receptors were too low and therefore were not analyzed. However, among the ones with reasonable expression, one P1 receptor (ADORA2A) and three P2Y receptors

(P2RY8, P2RY10 and P2RY14) were expressed at a lower level in MLL-r leukaemia patients compared to non-MLL-r leukaemia patients, indicating a potentially novel MLL-r associated pathway (Table 4.6, Figure 4.8).

Besides P1 and P2Y receptors, 2-Cl-ATP has also been reported to be an inhibitor of guanylate cyclase (GC), an enzyme involved in the conversion of guanine triphosphate to cyclic guanosine monophosphate (cGMP) (Carpentieri et al., 1981). GC activity has been shown to be higher in ALL patient samples than normal donors (Carpentieri et al., 1981). An increased level of GC has also been reported in prostate cancer (Cai et al., 2007). Based on the same GSE52991 expression dataset, no correlation was found between GC activators nor GC family 1 or 2 and sensitivity of MLL-r PDXs towards 2- Cl-ATP (Table 4.7). However, in the Stam et al. (2010) patient cohort, GUCY1A3 was highly expressed by MLL-r leukaemia patients compared to non-MLL-r patients (Table

104

Table 4.5: Correlation analyses of expression of P1 and P2Y receptors against IC50 for 2-chloroadenosine triphosphate in MLL-rearranged patient-derived xenografts. Genes r value P value P1 receptors ADORA1 0.3607 0.2074 ADORA2A 0.1594 0.4329 ADORA2A-AS1 0.0003289 0.9728 ADORA2B 0.005885 0.8852 ADORA3 0.1039 0.5333 P2Y receptors

P2RY1 -0.5165 0.2941

P2RY2 0.7507 0.0855

P2RY4 -0.701 0.1207

P2RY6 0.8974 0.0153

P2RY8 -0.3577 0.4863

P2RY10 -0.5858 0.2219

P2RY12 -0.1487 0.7786

P2RY13 0.6793 0.1378

P2RY14 0.8586 0.0286

Pearson correlation coefficient (r) and P-values were generated using GraphPad Prism 7.02. Gene expression dataset obtained from Gene Expression Omnibus site (Accession no. GSE52991).

105

Table 4.6: P-values from unpaired t-tests of mRNA expression of P1 and P2Y receptors in MLL-rearranged leukaemia versus MLL-wild-type leukaemia patients based on expression microarray data of a paediatric ALL patient cohort. Genes P value P1 receptors ADORA2A 0.0041 P2Y receptors

P2RY5 0.1297

P2RY8 0.0105

P2RY10 0.0038

P2RY14 <0.0001

(Expression data obtained from Stam et al., 2010).

106

A B

C D

Figure 4.8: Expression levels of P1 and P2 receptors in MLL-rearranged leukaemia and MLL-wild-type leukaemia patients. Expression of (A) ADORA2A; (B) P2RY8; (C) P2RY10 and (D) P2RY14 in MLL- rearranged (MLL-r) and non-MLL-rearranged (MLL-wt) leukaemia patients. Each data point represents the expression level of one patient based on microarray dataset extracted from Stam et al. (2010). P-values were calculated from unpaired t-tests of mRNA expression of each receptor.

107

Table 4.7: Correlation analyses of expression of genes encoding guanylate cyclase

(GC) activators, GC family 1 and 2 against 2-chloroadenosine triphosphate IC50 for MLL-rearranged patient-derived xenografts. Genes r value P value Guanylate cyclase activators GUCA1A -0.3731 0.4663 GUCA1B -0.5617 0.2461 GUCA1C -0.08809 0.8682 GUCA2A 0.01574 0.9764 GUCA2B 0.6474 0.1646 GUCD1 0.1748 0.7405 Guanylate cyclase 1 GUCY1A2 -0.2283 0.6635 GUCY1A3 -0.6115 0.1971 GUCY1B2 0.2736 0.5998 GUCY1B3 -0.6861 0.1324 Guanylate cyclase 2 GUCY2C -0.3238 0.5312 GUCY2D 0.6779 0.139 GUCY2F -0.1256 0.8126 GUCY2GP -0.537 0.2719

Pearson correlation coefficient (r) and P-values were generated using GraphPad Prism 7.02. Gene expression dataset obtained from Gene Expression Omnibus site (Accession no. GSE52991).

108

Table 4.8: P-values from unpaired t-tests of mRNA expression of genes encoding guanylate cyclase (GC) activators, GC family 1 in MLL-rearranged leukaemia versus MLL-wild-type leukaemia patients based on expression microarray data of a paediatric ALL patient cohort.

Genes P value Guanylate cyclase 1 GUCY1A3 0.0040

(Expression data obtained from Stam et al., 2010).

1 0 p = 0 .0 0 4 0

)

o l

( 8

o i

s 6

e r

p 4

A 2

R m 0 M L L -r M L L -w t

Figure 4.9: Expression level of GUCY1A3 in MLL-rearranged leukaemia and MLL-wild-type leukaemia patients. Each data point represents the expression level of one patient based on microarray dataset extracted from Stam et al. (2010). P-values were calculated from unpaired t-tests of mRNA expression in MLL-rearranged (MLL-r) and non-MLL-rearranged (MLL-wt) leukaemia patients.

109

4.8, Figure 4.9). In all, despite these analyses, a robust biomarker could not be determined in MLL-r leukaemia cells cultured in vitro.

4.2.1.5 Effect of 2-chloroadenosine triphosphate on normal peripheral blood mononuclear cells

To obtain some insight into the therapeutic window of 2-Cl-ATP, the effect of the compound was investigated in normal peripheral blood mononuclear cells (PBMC) isolated from three healthy donors using resazurin-based viability assay. 2-Cl-ATP affected the viability of the PBMCs of all three donors with IC50 in the range of 4.2 – 5.4 μM (Figure 4.10), which is in the same range as previously reported for MLL-r ALL PDX cells, raising concerns about the presence of therapeutic window for this compound.

4.2.1.6 Comparison of activity between 2-chloroadenosine triphosphate and its parent compound, 2-chloroadenosine

As previously mentioned, 2-Cl-ATP is a derivate of adenosine receptor agonist, 2- chloroadenosine (2-CADO). It is hydrolyzed into 2-CADO by serum and could be re- phosphorylated intracellularly causing cell death (D’Ambrosi et al., 2004). To determine whether there was any similarity in the potency, activity and therapeutic window of the two compounds, 2-CADO was tested in several MLL-r and MLL-wt leukaemia cell lines that were either sensitive (n=3) or resistant (n=2) to 2-Cl-ATP. In 72-hour resazurin-based cytotoxicity assays, 2-CADO demonstrated a similar cytotoxic effect as 2-Cl-ATP (Table 4.9, Supplementary Figure 6A). To test toxicity of 2-CADO to normal cells, 72-hour resazurin-based viability assays were performed with PBMC from the same three donors previously tested for 2-Cl-ATP. 2-CADO displayed similar toxicity towards the PBMCs as 2-Cl-ATP with IC50 ranging from 3.6 – 4.5 μM (Table 4.9, Supplementary Figure 6B), suggesting that there is no difference in therapeutic window between the two compounds.

110

P B M C

1 2 5

D o n o r 1 ( IC = 5 .4  M ) 1 0 0 5 0

) D o n o r 2 ( 4 .8  M )

% (

7 5 D o n o r 3 ( 4 .2 M )

y 

l i

b 5 0

i V 2 5

0 -6 -5 -4 2 -C l-A T P c o n c e n tr a tio n [L o g M ]

Figure 4.10: Cytotoxicity of 2-chloroadenosine triphosphate in normal peripheral blood mononuclear cells. Dose response curves for 2-chloroadenosine triphosphate (2-Cl-ATP) in normal peripheral blood mononuclear cells (PBMC) from three donors in a 72-hour resazurin- based viability assay. The results are expressed as the mean ± SE of three independent experiments and IC50 values were derived from the mean of these experiments.

111

Table 4.9: Comparison of cytotoxicity of 2-chloroadenosine-triphosphate and 2-chloroadenosine.

Cell line Translocation Disease 2-Cl-ATP IC50 2-CADO IC50 (µM) (µM) MLL- ALL PER-485 t(4;11) Infant ALL 2.5 2.5 rearranged RS4;11 t(4;11) Pre-B cell ALL 5.9 7.1 AML THP-1 t(9;11) Infant AML >20 18.2 MLL- ALL CEM - Childhood T-cell ALL 11.4 6.2 wild-type REH - Pre-B cell ALL >20 >20 Donor 1 - Normal 5.4 4.5 Peripheral blood Donor 2 - Normal 4.8 3.6 mononuclear cells Donor 3 - Normal 4.2 3.6

112

4.2.1.7 In vivo stability of 2-chloroadenosine triphosphate

There are no reports on 2-Cl-ATP application in animal models and nothing is known about its stability for in vivo application. To perform a first evaluation of its metabolic stability, the compound was tested in microsomal stability assays to estimate the hepatic clearance rate (Houston, 1994; Houston and Carlile, 1997; Obach, 1999). Studies have shown that the assay could successfully predict drug clearance in animals (Chiba et al., 2009) and humans (Obach, 1999). 2-Cl-ATP exhibited high microsomal stability with 100% of the compound remaining at 90 minutes, indicating a half-life of more than 90 minutes (Figure 4.11). A stable compound enables in vivo studies and is a prerequisite for clinical testing. However it is paramount to determine its actual pharmacokinetic property as long half-life could also potentially mean difficulty in breaking down the compound and thus elimination by the kidney and liver.

113

1 2 5

)

o n

r 1 0 0

m c

e 7 5

n i

u 5 0

a o

g H a lf-life : > 9 0 m in

m 2 5

( C

0 0 2 0 4 0 6 0 8 0 1 0 0 T im e (m in )

Figure 4.11: Prediction of metabolic stability of 2-chloroadenosine triphosphate using a mouse microsomal stability assay. The compound has a half-life of >90 minutes. The results are expressed as the mean ± SE of two independent experiments.

114

4.2.2 Characterization of SID7969543 as a selective inhibitor of MLL-rearranged leukaemia cells

4.2.2.1 Selectivity of SID7969543 towards MLL-rearranged leukaemia

The second candidate MLL-selective hit compound yielded from this HTS was an SF-1 or NR5A1 inhibitor, SID7969543 (SID) (Figure 4.12). To determine the cytotoxicity profile of SID, the compound was also tested in the broad panel of 25 cell lines that includes MLL-r cell lines, MLL-wt cell lines, solid tumours and non-malignant cell lines. All cell lines sensitive towards SID were leukaemias with either MLL-r or CALM-AF10 translocation. As observed for 2-Cl-ATP, SID affected the viability of a subset of MLL-r leukaemia cells, as well as CALM-AF10 cells, while not affecting the wild-type MLL cells (Figure 4.13, Table 4.10), solid tumours or non-malignant cells. A subset of five MLL-r cell lines (PER-826A, RS4;11, SEMK2, KOPN-8 and MV4;11) with either t(4;11) or t(11;19) fusion were resistant to the compound with IC50 values above 20 μM (Figure 4.14A). Two of these lines, KOPN-8 and MV4;11 were also resistant to 2-Cl-ATP. Sensitive cell lines had IC50 values ranging between 1 – 5 μM. Whilst a significant difference in the mean percentage viability could be seen between MLL-r cell lines and solid tumours at 10 μM, only a trend was seen when comparing MLL-wt cell lines with MLL-r cells (Figure 4.14B).

To identify how SID affected the viability of the cells, Annexin V staining was performed to measure the level of apoptosis in cells over a 72-hour period of treatment. Significant induction of apoptosis was shown in approximately 35% of the cells at as early as 6 hours post-exposure to SID and persisted until 48 hours post-treatment. After 72 hours of treatment with SID, the proportion of cells undergoing apoptosis decreased slightly to approximately 20% (Figure 4.15).

115

Figure 4.12: Chemical structure of SID7969543.

1 2 5 M V 4 ;1 1 U 9 3 7

1 0 0 R S 4 ;1 1 K P M O T S )

S E M K 2 C E M

% (

7 5 y

t M O L M -1 3 R E H

l i

b 5 0 T H P -1 J U R K A T

a i

V K O P N 8 2 5 P E R -4 8 5

0 P E R -4 9 0 -6 -5 -4 P E R -7 8 5 A S ID c o n c e n tr a tio n [L o g M ] P E R -7 0 3 A

P E R -8 2 6 A

Figure 4.13: Cytotoxicity of SID7969543 across a range of MLL-rearranged, CALM-AF10 and MLL-wild-type leukaemia cell lines. Dose response curves for SID7969543 (SID) across a range of MLL-rearranged (red), CALM-AF10 (blue) and MLL-wild-type (green) leukaemia cell lines as measured in a 72-hour resazurin-based viability assay. The results are expressed as the mean ± SE of three independent experiments. Details of cell lines are found in Table 2.1 and IC50 in Table 4.10.

116

Table 4.10: Cytotoxicity of SID7969543 in a panel of 25 cell lines.

Cell line Translocation Disease IC50 (µM) PER-485 t(4;11) Infant ALL 1.4 PER-490 t(4;11) Infant ALL 1.9 PER-703A t(1;11) Infant ALL 4.9 PER-785A t(4;11) Infant ALL 5 ALL PER-826A Complex, t(11;19) Infant ALL >20 MLL-rearranged RS4;11 t(4;11) Pre-B cell ALL >20 SEMK2 t(4;11) Pre-B cell childhood ALL >20 KOPN-8 t(11;19) Infant pre-B cell ALL >20 MV4;11 t(4;11) Childhood AML >20 AML MOLM-13 t(9;11) AML 1 THP-1 t(9;11) Infant AML 2.6 CEM - Childhood T-cell ALL >20 ALL REH - Pre-B cell ALL >20 MLL-wild type Jurkat - Childhood T-cell ALL >20 U937 CALM-AF10 AML 1.9 AML KP-MO-TS CALM-AF10 AML 1.9 KELLY - Neuroblastoma >20

BE(2)C - Childhood neuroblastoma >20

HEY - Ovarian carcinoma >20

Solid tumours 27/87 - Endometrioid ovarian cancer >20

MCF-7 - Breast adenocarcinoma >20

H460 - Lung carcinoma >20

LNCaP - Prostate carcinoma >20 MRC-5 - Normal lung >20 Normal cells WI-38 - Normal lung >20

IC50: inhibitory concentration resulting in 50% reduction of cell survival relative to control, IC50 were derived from the mean of three independent experiments; ALL: acute lymphoblastic leukaemia; AML: acute myeloid leukaemia.

117

Figure 4.14: Viability of cell line panel and comparison of viability between cell types at 10 µM SID7969543. (A) Viability of cell line panel at 10 µM of SID7969543 in 72-hour resazurin-based cytotoxicity assay. (B) Comparison of cell viability at 10 µM between cell lines grouped according to type: MLL-rearranged (MLL-r), CALM-AF10, MLL-wild-type (MLL-wt), solid tumour and normal cell lines. Mean viability between groups was compared by one-way ANOVA with Dunn’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, p<0.05; **, p<0.01. 118

Figure 4.15: Apoptosis of PER-485 cells over 72-hour treatment course of SID7969543. (A) Mean percentage increases of Annexin V-positive cells compared to vehicle control at each of the indicated time points after treatment with SID7969543. The results are expressed as the mean ± SE of three independent experiments. Means were compared by one-way ANOVA. Asterisks represent significance levels of P-values. **, p<0.01, ****, p<0.0001; B) Representative data for flow cytometric analysis of Annexin V/7AAD staining in PER-485 cells exposed to SID7969543 over a 72-hour treatment course.

119

4.2.2.2 Determination of synergy between conventional chemotherapeutics with SID7969543 in vitro

SID was also tested in combination with chemotherapy agents to establish if the compound offers any synergistic effect with any the agents, and conversely to increase its efficacy and potency as a potential novel MLL-r leukaemia therapy. The in vitro synergy assays were again performed with two cell lines, PER-485 and MOLM-13 as previously described. Synergy was observed between SID and either etoposide or cytarabine in both cell lines and with mitoxantrone or topotecan in MOLM-13 with CI of 0.891 and 0.855, respectively (Table 4.11, Supplementary Figure 7 and 8). Calculated DRIs showed that SID could reduce etoposide dose more than 2-fold both cell lines. In MOLM-13, SID could reduce the dose of the other chemotherapeutics by at least 2-fold. Etoposide and cytarabine could also reduce SID concentration for similar cytotoxic effect in both cell lines.

4.2.2.3 Activity of SID7969543 in infant MLL-rearranged ALL patient-derived xenografts in vitro

To test the effect of SID in PDX cells, 48-hour resazurin-based cytotoxicity assays were performed with infant MLL-r PDXs in vitro as previously described. Remarkably, none of the PDX were affected by SID at any dose up to 20 μM (Figure 4.16).

4.2.2.4 Toxicity of SID7969543 in normal peripheral blood mononuclear cells

Using 72-hour resazurin-based viability assay, the effect of SID towards peripheral blood mononuclear cells (PBMCs) was determined. There was some effect of the compound on cells from just one donor sample, whereby approximately 50% of cells were killed at the highest dose of 20 µM (Figure 4.17).

120

Table 4.11: Calculated combination index and drug reduction index for SID7969543 in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and MOLM-13 cells. PER-485 MOLM-13 Chemodrugs DRI DRI CI CI SID Chemodrug SID Chemodrug Mitoxantrone 0.988 1.723 2.534 0.891 3.872 1.600 Daunorubicin 0.985 1.729 2.522 0.903 4.283 1.531 Etoposide 0.740 2.004 4.243 0.834 2.260 2.563 Cytarabine 0.735 1.856 6.419 0.540 4.515 3.156 Topotecan 1.312 1.868 1.355 0.855 2.954 1.948

SID: SID7969543, CI: combination index; DRI: drug reduction index, ED75: effective dose causing 75% reduction of cell viability. Cells were subjected to increasing doses of SID alone, of a chemotherapy agent alone or a combination of the two simultaneously at a fixed-ratio for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. Drug reduction index indicates fold reduction in dose of indicated drug when SID is applied, and vice versa.

121

M L L -r P D X

1 2 5

1 0 0 M L L - 2

) M L L - 5

% (

7 5

y M L L - 6

i l i M L L - 7

b 5 0 a

i M L L - 8 V 2 5 M L L - 1 4

0 -6 -5 -4 S ID c o n c e n tr a tio n [L o g M ]

Figure 4.16: Cytotoxicity of SID7969543 in infant MLL-rearranged patient derived xenograft cells in vitro Dose response curves for of SID7969543 (SID) in infant MLL-rearranged ALL patient derived xenografts (MLL-r PDX) in 48-hour resazurin-based viability assay. The results are expressed as the mean ± SE of three independent experiments.

P B M C

1 2 5

D o n o r 1 1 0 0

) D o n o r 2

% (

7 5

y D o n o r 3

l i

b 5 0

i V 2 5

0 -6 -5 -4 S ID c o n c e n tr a tio n [L o g M ]

Figure 4.17: Cytotoxicity of SID7969543 in normal peripheral blood mononuclear cells. Dose response curves for SID7969543 (SID) in normal peripheral blood mononuclear cells (PBMC) from three donors in a 72-hour resazurin-based viability assay. The results are expressed as the mean ± SE of three independent experiments.

122

4.2.2.5 In vivo stability of SID7969543

Microsomal stability assay using a mouse liver microsome, preparation was performed to predict SID metabolic stability and hepatic clearance rate. SID exhibited very low stability in this assay, with only 6% of the compound remaining at 90 minutes and calculated half-life of only 1.89 minutes (Figure 4.18).

4.2.2.6 Level of SF-1/NR5A1 in leukaemia patients

An increased level of SF-1 has been observed in childhood adrenal cortical tumours, affecting cell proliferation in vitro and triggering tumour formation in mice (Doghman et al., 2007a; 2007b; 2009). To determine if MLL-rearranged leukaemia patients differentially express SF-1 compared to leukaemia patients without the MLL rearrangement, the microarray-based gene expression dataset of leukaemia patients from Stam et al. (2010) was analyzed. This analysis indicated that whilst SF-1/NR5A1 was expressed at a reasonable levels in paediatric ALL patient samples, expression of this gene did not differ significantly between the two groups (Figure 4.19, p=0.8276).

123

1 0 0

)

o n

r 8 0

m c

e 6 0

n i

u 4 0

a o

g H a lf-life : 1 .8 9 m in

m 2 0

( C

0 0 2 0 4 0 6 0 8 0 1 0 0 T im e (m in )

Figure 4.18: Prediction of metabolic stability of SID7969543 in a mouse microsomal stability assay. SID7969543 had a half-life of 1.89 minutes. The results are expressed as the mean ± SE of two independent experiments.

p = 0 .8 2 7 6

) 1 0

o l

( 8

n o

i 6

e r

p 4

A 2

N R

m 0 M L L -r M L L -w t

Figure 4.19: Expression levels of of SF-1/NR5A1 in MLL-rearranged leukaemia and MLL-wild-type leukaemia patients. Each data point represents the expression level of one patient based on microarray dataset extracted from Stam et al. (2010). P-values were calculated from unpaired t-tests of mRNA expression in MLL-rearranged (MLL-r) and non-MLL-rearranged (MLL-wt) leukaemia patients.

124

4.3 Discussion

In this study, a high-throughput phenotypic screening of a library of clinically-approved and/or pharmacologically active compounds was performed to identify potential novel therapeutics targeting high-risk leukaemia in general, and MLL-rearranged leukaemia specifically. The candidate MLL-selective compound 2-chloroadenosine triphosphate (2-Cl-ATP), is a P2Y receptor agonist. It is also an ATP analogue and a phosphorylated derivative of 2-chloroadenosine (2-CADO), a widely used adenosine receptor agonist. Among the G-protein-coupled receptors (GPCRs), the purinergic P1 and P2Y receptors are preferentially activated by the release of either adenosine or ATP respectively (King and Burnstock, 2002). These receptors are involved in many cellular responses including cell growth, differentiation and apoptosis (Burnstock and Di Virgilio, 2013). Testing against a panel of 25 cell lines showed selectivity of 2-Cl-ATP towards a subset of MLL-rearranged and CALM-AF10 leukaemia cells with IC50 ranging between 2.5 – 5.9 μM, while not affecting wild-type MLL cells, solid tumours or normal cells. Less responsive MLL-r cell lines had IC50 ranging from 13.4 μM to above 20 μM. Based on characteristics of the resistant subset, sensitivity to 2-Cl-ATP could not be linked to a particular MLL-translocation or leukaemia lineage, indicating a distinct underlying mechanism behind the selectivity. This highlights the heterogeneity of MLL-r disease which has been widely reported and demonstrates the need of different therapeutics for specific patient subgroups (Marschalek, 2010a; Marschalek 2010b; Meyer et al., 2018).

With fairly high IC50 as a single agent in MLL-r cells, in vitro combination assays were performed and identified moderate synergy between 2-Cl-ATP and mitoxantrone, etoposide and cytarabine. These synergistic effects the indicate possibility of clinical administration with current chemotherapeutics and potentially lead to a reduction of side effects due to the use of lower concentrations of chemotherapeutics for similar effect.

Among the issues of 2-Cl-ATP as potential therapy for MLL-r leukaemia is its lack of discriminative activity against normal PBMCs. Although the in vitro assay does not completely predict in vivo toxicity to normal blood cells, it is often used as a first test to evaluate the effect of the compound on healthy haematological cells. In ex vivo viability

125 assays, the IC50 range of 4.2 – 5.4 μM was very similar to the IC50 range observed for MLL-r PDXs (2.5 – 9.7 μM). Its parental compound, 2-CADO, also demonstrated similar cytotoxicity effects in PBMCs, corresponding to an earlier report that apoptosis was induced after 48-hour compound exposure (Barbieri et al., 1998). Despite this, 2- CADO has been tested in Wistar rats at a dose range of 1 – 15 mg/kg, administered intravenously or intraperitoneally, in studies to determine the pharmacokinetic relationship of the compound with adenosine receptors (Mathot et al., 1996; Mares, 2010), indicating a possibility of in vivo testing of 2-Cl-ATP. However, a comprehensive toxicity study should be performed, to ensure an effective dose could be achieved in humans, while not causing severe or life-threatening toxicities. Additionally, actual compound half-life should be determined prior to in vivo testing. Although the half-life of >90 minutes estimated from the in vitro microsomal assays indicate high in vivo compound stability, it could also mean difficulty in breaking down the compound and thus elimination by the kidney and liver. Accumulation of 2-Cl-ATP and its byproducts may potentially cause toxic effects, and therefore should be confirmed before any testing in the animals or patients. Nevertheless, the compound has highlighted a group of purinergic receptors and GC enzymes and their activators as potentially interesting for MLL leukaemia biology and targeting.

The 2-Cl-ATP compound showed high activity in four MLL-r PDXs with IC50 between

2.5 – 4.8 μM and lower activity in two xenografts with IC50 of 9.7 and 11.5 μM. This range of response, coupled with the access to expression data from these PDX allowed some investigation of the basis of this variance in response. To investigate the possibility that variation might be due to differential expression of the target of the compound, the basal expression levels of purinergic receptors was analyzed. Differing only on one ATP group, 2-Cl-ATP is considered one of the derivatives of 2-CADO. This compound was also present in the chemical library screened, however due to lower selectivity towards PER-485, it was filtered out during subsequent testing for not meeting the 40% differential viability between CEM and PER-485. 2-CADO is an adenosine (P1) receptor agonist that has been reported to have anti-convulsant activity in vivo. It was demonstrated to suppress pentetrazol-induced seizures in immature and older rats by activating A1 adenosine receptors (Mares, 2010). The compound also affects vasodilation and heart rate, which are dependent on adenosine receptors.

126

Selective deletion of these receptors was shown to abolish 2-CADO activity (Mathot et al., 1996; Nicholls et al., 2002; Koeppen et al., 2009). Additionally, 2-CADO has also been reported to have anti-cancer activities in vitro against several cancer cell types such as astrocytoma (Ceruti et al., 2003), prostate cancer (Minelli et al., 2006) and chronic lymphocytic leukaemia (Bastin-Coyette et al., 2008) as previously mentioned.

Due to 2-Cl-ATP being a derivative of 2-CADO, basal expression level of both the P1 and P2Y classes of receptors, were analyzed across the PDX to determine if sensitivity of MLL-r leukaemia cells towards 2-Cl-ATP was correlated to these receptors. Among five P1 and nine P2Y receptors analyzed, expression of two P2Y receptors, P2RY6 and

P2RY14, inversely correlated with sensitivity towards 2-Cl-ATP, a striking result given the small sample size (n=6). In paediatric patient expression data (Stam et al., 2010), significantly decreased expression levels of P1 receptor, ADORA2A and P2Y receptors,

P2RY8, P2RY10 and P2RY14 were observed in MLL-r compared to MLL-wt groups of leukaemia patients. The association of sensitivity with low rather than high levels of receptors could potentially indicate that the compound is not working through these receptors but instead affecting an alternate pathway in the cells. In a study of PC12nnr5 cells, a cell line derived from a rat neuroendocrine tumour of the adrenal gland, cytotoxicity towards 2-Cl-ATP was independent of P1 and P2 receptors, but relied on hydrolysis of 2-Cl-ATP into 2-CADO by serum ATPase. Once converted into 2-CADO, toxicity depended on its intracellular phosphorylation, followed by inhibition of DNA synthesis and hence cell cycle arrest and death (D’Ambrosi et al., 2004). This tallies with another study on chronic lymphocytic cell line, EHEB revealing cytotoxicity of 2- CADO was caused by its intracellular phosphorylation into 2-Cl-ATP, followed by a decline in endogenous ATP causing inhibition of DNA and RNA synthesis (Bastin- Coyette et al., 2008). This could be investigated as a mechanism of 2-Cl-ATP action in MLL-r leukaemia cells in future studies by performing similar experiments involving an ATPase inhibitor such as pyridoxalphosphate-6-azophenyl-20,40-disulphonic acid (PPADS), nucleoside transport inhibitor, nitrobenzylthioinosine (NBTI), and adenosine kinase inhibitor, nitrobenzylthioinosine (ITu). These experiments would involve treatment of the cells with 2-Cl-ATP in the presence of either PPADS, NBTI or ITu. Reduced 2-Cl-ATP activity in cells incubated with PPADS would indicate that the toxicity is related to metabolism of 2-Cl-ATP into 2-CADO, while increased cell

127 viability in cells treated with 2-Cl-ATP and either NBTI or ITu would confirm the dependency of 2-Cl-ATP cytotoxicity on the phosphorylation of the metabolized 2- CADO. P1 or P2Y receptor antagonists such as 8-cyclopentyl-3-[3-[[4- (fluorosulfonyl)benzoyl]oxy]propyl]-1-propylxanth ine (FSCPX) and PIT, respectively could also be co-treated with 2-Cl-ATP to test whether toxicity of 2-Cl-ATP is mediated by the receptors. Such analyses would determine the mechanism of this purinergic receptor agonist in MLL-r leukaemia cells, which may potentially be exploited for future therapies.

As yet another potential explanation for sensitivity to 2-Cl-ATP, the expression levels of guanylate cyclase (GC) enzymes and their activators were also analyzed as 2-Cl-ATP was also reported to be an inhibitor of GC, an enzyme in the signalling cascade that catalyzes the formation of cyclic guanosine monophosphate (cGMP) (Carpentieri et al., 1981). Once formed, cGMP signalling can directly or indirectly, through phophodiesterase (PDE) activity, cross-activate cyclic adenosine monophosphate (cAMP) signalling pathways (Amarachinta et al., 2018). cAMP is thought to be involved in leukaemogenesis and leukaemic cell proliferation (Takemoto et al., 1982) and GC activity has been shown to be higher in ALL patient samples than in lymphocytes from normal donors (Carpentieri et al., 1981). In prostate cancer, the levels of the soluble GC subunit α1 were significantly elevated in tissues with advanced disease, and high levels of this protein were also sufficient to stimulate cell proliferation (Cai et al., 2007). However, the opposite was reported for breast cancer, whereby higher levels were associated with inhibition of cell survival and proliferation in vitro and in vivo (Wen et al., 2015), and elevation of the β1 subunit in patient tissues correlated with greater survival probability (Sotolongo et al., 2016). In MLL-r PDXs, basal expression levels of GC activators and GC family 1 and 2 were not significantly correlated with sensitivity towards 2-Cl-ATP. A small response variation among the PDXs with only a maximum of 5-fold change in IC50 as well as the small number of PDX (n=6) might limit this analysis, which could be expanded in future studies and the levels of these receptors could also be determined for the cell line panel. However, in a paediatric leukaemia patient cohort (Stam et al., 2010), GUCY1A3 was more highly expressed in MLL-r compared to leukaemia patients without MLL rearrangement, indicating a potential novel MLL-r associated pathway. Interestingly, one study found a

128 combination of adenosine receptor agonist (chloro-IB-MECA) and papaverine, an inhibitor of PDE, which regulates GC, to synergize and potentiate the anti-proliferative effect of dexamethasone in multiple myeloma and diffuse large B-cell lymphoma, inducing apoptosis and inhibiting proliferation by elevating cAMP (Rickles et al., 2010). Recently, HDAC inhibitor, panobinostat was shown to increase the level of soluble GC β1 in breast cancer which as previously mentioned, is associated with improved patient survival (Sotolongo et al., 2016). Future experiments using siRNA- mediated suppression could address whether GUCY1A3 or related receptors are important for sensitivity in the MLL-r cell lines and PDXs and if confirmed, some of these therapeutic strategies could be considered for testing in MLL-r leukaemia, particularly if specific subtypes that rely on the GC pathway, could be identified. These further investigations would also allow the identification of relevant biomarkers to monitor biological response for future studies.

Another MLL-selective candidate compound identified in this study was SID7969543, a well-known SF-1/NR5A1 inhibitor that has been quite extensively tested in vitro in ACC. This is the first report of the compound’s anti-cancer activity in a subgroup of MLL-r paediatric leukaemia cell lines. In this study, similar to 2-Cl-ATP, SID showed selectivity towards a subgroup of MLL-r leukaemia and CALM-AF10 cell lines while not affecting MLL-wt leukaemia, solid tumours or non-malignant cell lines. The IC50 of sensitive cell lines ranged between 1 – 5 μM. The compound affected the viability of sensitive MLL-r cells by inducing apoptosis as early as 6 hours post-exposure to SID. Apoptosis has never been reported as the mechanism of cell killing by SID in previous studies, but in a human adrenocortical cell line expressing SF-1, H295R, the compound significantly inhibited cell proliferation in the presence of Doxorubicin or increased SF- 1 dosage (Doghman et al., 2010) demonstrating its high selectivity towards SF-1. The compound did not show any activity in infant MLL-rearranged PDX panel ex vivo, possibly due to the mechanism of SID in targeting cell proliferation whereas ex vivo cultured xenograft cells show limited or no proliferation (Liem et al., 2004) as opposed to the cultured immortalized cell lines. For similar reasons, it is also difficult to conclude whether SID might affect PBMCs. However, the issue of reduced proliferation in MLL-r PDXs and PBMCs in an ex vivo setting may be solved by using a co-culture model, such as one incorporating bone marrow-derived mesenchymal stem cell (MSC)

129 to mimic in vivo microenvironment conditions. The stromal cells produce cytokines and other soluble factors that affect proliferation, maintenance and survival of leukaemia cells (Jones and Wagers, 2008). Several PDXs have demonstrated increased proliferation when co-cultured with murine bone marrow stromal cells, MS-5. Paediatric ALL PDXs, ALL-7 and ALL-11, were shown to undergo 1 – 2 cell divisions within three days (Liem et al., 2004) as measured by a fluorescent dye, carboxyfluorescein diacetate succinimidyl ester (CFSE) that covalently binds to available amines, resulting in a sequential halving of fluorescence with each cell division (Quah et al., 2007). However, there were also PDXs, ALL-17 and ALL-19, that underwent a negligible proliferation (Liem et al., 2004). Therefore, further testing of the benefit of co-culture system should be performed with several types of MSCs such as the human MSC line, HS and murine MSC lines, M2-10B4 and MS-5, to determine the most suitable line for the MLL-r ALL PDX panel. It is also important to note that the interactions existed in this co-culture system might protect the cancer cells from cytotoxic drugs, mimicking the microenvironment protection given to leukaemic cells in vivo (Mudry et al, 2000; Tesfai et al., 2012, Sison et al., 2013). Thus, a suitable in vitro drug testing model is critical when investigating drug response, particularly drugs that target cell proliferation pathways.

Combination assays revealed that SID synergizes with etoposide and cytarabine and is able to reduce the dose of etoposide by more than 2-fold. Likewise, the chemotherapeutics could reduce the dose of SID, potentially allowing sufficient levels of compound to be achieved in animal models or in patients, despite a micromolar IC50.

However, besides a high IC50, SID was predicted to have very low in vivo stability with calculated half-life of less than two minutes in a microsomal assay. This provides a possible reason for the absence of published in vivo data for SID and indicates the necessity of further molecule optimization if the compound was to be tested in vivo. The group that firstly identified the compound also reported poor compound stability with half-life of 1.24 minute in human microsomes and less than one minute in other four animal species, dog, monkey, mouse and rat (Madoux et al., 2008). They also developed two more compounds, compound 31 and 32, with improved potency and selectivity towards SF-1, however the stability of those compounds was not mentioned (Roth et al., 2008; Doghman et al., 2009). Despite possessing such unfavourable

130 properties, the identification of SID uncovered SF-1 as a possible targetable gene pathway for MLL-r leukaemia. Even though infants with MLL-r leukaemia did not show differential expression of SF-1 compared to non-MLL-rearranged leukaemia patients in a leukaemia patient dataset (Stam et al., 2010), the pathway could be further investigated in vitro such as through testing whether knockdown of SF-1 affects leukaemia cell growth or response to the compound. The basal levels of SF-1 and its levels after being subjected to SID and other available SF-1 inhibitors, including compound 31 and 32 by Roth and colleagues, could be evaluated in MLL-r cell lines and PDXs to determine if SF-1 transcriptional activity truly affects MLL-r leukaemia cell viability.

In summary, the high-throughput screening of libraries of approved drugs and pharmacologically active compounds yielded two candidates selective for MLL-r leukaemia, 2-chloroadenosine triphosphate and SID7969543. Despite their limitations in terms of potency, stability or toxicity, the compounds gave an insight into possible novel pathways that could be investigated further as potential targets for MLL- rearranged leukaemia therapy. If optimized derivatives of the candidate compounds became available, these could be tested in vivo to determine if these newly identified pathways are indeed targetable for this disease. However, as previously mentioned, the aim of this study was to identify candidates that could be repurposed for high-risk paediatric leukaemia in order to overcome the common hurdles of conventional drug development process and potentially fast-track the candidates for clinical testing. Hence, further compound development and optimization of the structural, physicochemical and biochemical properties to improve drug-like attributes were not attempted. Instead the two candidates identified as hits for high-risk leukaemia, auranofin and disulfiram, with established pharmacokinetic and pharmacodynamic data, were further investigated in the next chapter.

131

CHAPTER 5: CHARACTERIZATION OF INDUCERS OF REACTIVE OXYGEN SPECIES, AURANOFIN AND DISULFIRAM, AS POTENTIAL THERAPEUTICS FOR HIGH-RISK LEUKAEMIA

5.1 Introduction

As described in Chapter 3, due to the high IC50 of the MLL-selective candidates against MLL-r leukaemia cell lines, a subsequent secondary screen was performed to identify more potent candidates for MLL-r leukaemia. It was found that when screened at a dose of 5 μM, 184 compounds were completely toxic at to the CEM childhood T-cell ALL cell line as well as the PER-485 infant MLL-AF4 ALL cell line, and these compounds were re-screened at a lower dose range in case some MLL-selectivity would become apparent at these lower doses. Although none of the 184 compounds showed selectivity towards the MLL-r cell line over the MLL-wt cell line, further filtering was conducted in order to identify compounds that might nevertheless be useful agents for MLL-r ALL and other high-risk forms of ALL. This strategy (described in Section 3.2.3) led to the selection of auranofin and disulfiram, known reactive oxygen species (ROS) inducers and FDA-approved molecules, for further characterization based on their novelty as therapeutic options for MLL-r and other high-risk ALL.

As mentioned in Chapter 1, auranofin is a gold (I)-containing compound (Figure 5.1A) previously approved by the FDA for rheumatoid arthritis. The precise mechanism of action of auranofin is not well understood, however it has been reported that the drug decreases immune responses and suppresses inflammation, which is persistent in rheumatoid arthritis patients (Furst, 1983; Roder and Thomson, 2015). With several mild side effects, the compound was found to be safe for use in both adults and children with a plasma half-life of 15 – 25 days and elimination after 55 – 80 days (Kean et al., 2008). More recently, this drug, either alone or in combination with other therapeutics, has been reported to inhibit tumour growth in animal models for breast cancer (Liu et al., 2014), colorectal cancer (Hrabe et al., 2015), lung cancer (Dai et al., 2013; Fan et al., 2014; Li et al., 2016) and several types of leukaemias such as lymphocytic leukaemia (Simon et al., 1981), chronic myelogenous leukaemia (Chen et al., 2014) and

132

CLL (Fiskus et al., 2014). The anti-cancer activities of auranofin in CLL were discovered through a HTS of the National Institute of Health Chemical Genomic Centre pharmaceutical library against lymphocyte samples obtained from six patients (Shen et al., 2013). Further investigation found the drug to be effective against primary cells from 13 CLL patients and reported a 30-fold higher sensitivity of primary CLL cells to the compound compared to lymphocytes isolated from five normal donors, indicating a potential therapeutic window for auranofin for the treatment of CLL. The therapeutic potential of the compound was confirmed in an in vivo model of CLL using TCL-1 transgenic mice in which the drug reduced tumour cell burden and improved mouse survival (Fiskus et al., 2014). It was hypothesized that auranofin exerted its anti-cancer activity through the reduction/oxidation (redox) pathway by inhibiting enzymes such as thioredoxin reductase (TrxR) which results in increased levels of intracellular ROS causing oxidative stress followed by apoptosis (Marzano et al., 2007; Fan et al., 2014; Fiskus et al., 2014). Auranofin went into clinical trial for adult CLL patients merely two years after being identified (Weir et al., 2012), although the drug was not further pursued due to the ongoing development of more promising treatment options for CLL (Saba et al., 2013). However, as novel therapeutic options are still urgently needed for high-risk leukaemia, this compound was investigated to further characterize its in vitro and in vivo efficacy against models MLL-r and other high-risk ALL.

Another hit identified was disulfiram (Figure 5.1B), an FDA-approved drug for the treatment of alcoholism since 1951 (Franck and Jayaram-Lindstrom 2013). The drug inhibits liver enzyme aldehyde dehydrogenase thereby preventing the metabolism of the primary metabolite of alcohol, acetaldehyde which causes unpleasant effects such as facial flushing, nausea and vomiting due to accumulation of acetaldehyde in the blood following intake of alcohol (Heilig and Egli, 2006). In the body, disulfiram has a half- life of 60 to 120 hours and is rapidly reduced into its active metabolite, diethyldithiocarbamate (DDC) (Cobby et al., 1977). Due to the very low frequency of side effects, disulfiram was tested in adolescents with a dose of 250 mg (half of standard adult dose) with no reported adverse effects over the short period of treatment of up to four months. Nevertheless, liver function tests were recommended fortnightly for the first two months and then at 3 – 6 months intervals (Myers et al., 1994). Today, disulfiram is administered in combination with naltrexone, an opioid receptor

133 antagonist, and/or acamprosate, a modulator of the glutamatergic neurotransmitter system that is associated with chronic alcohol exposure and alcohol withdrawal (Fuller and Gordis, 2004; Scott et al., 2005)

Disulfiram recently gained momentum in cancer research as a potential therapy for various types of malignancies. Its anti-cancer activities have been reported with in vitro and in vivo preclinical models of breast cancer (Chen et al., 2006a; Zhang et al., 2010; Allensworth et al., 2015), prostate cancer (Iljin et al., 2009; Lin et al., 2011), glioblastoma (Li et al., 2015; Lun et al., 2016), as well as ALL (Conticello et al., 2012; Deng et al., 2016). However, it was actually in 1984 when a French group initiated a phase III clinical trial for breast cancer that the efficacy of active metabolite, DDC as an anti-cancer therapy was demonstrated. In this double-blinded trial, 64 patients with non- metastatic high-risk breast cancer who had undergone a mastectomy were treated with DDC or a placebo as an adjuvant immunotherapy for nine months. After six years of follow up, the overall survival was higher in the DDC group (81%) than in the placebo group (55%), and the DFS rates were 76% and 55% respectively (Dufour et al., 1993). Disulfiram, a known potent copper chelator, forming a copper-DDC complex (Eneanya et al., 1981; Cvek, 2012), has been shown to have greater cytotoxicity when used in combination with copper in several cancer cell lines including inflammatory breast cancer and gliomas (Allensworth et al., 2015; Li et al., 2015). Disulfiram has been reported to affect the redox pathway by depleting cellular glutathione (GSH) (Nobel et al., 1995) increasing ROS level in cells (Zha et al., 2014) and triggering depolarization of the mitochondrial membrane potential (Yang et al., 2016). These effects increase with the presence of copper (Zha et al., 2014). In recent years, disulfiram has entered clinical trials with copper alone or in combination with other drugs for several cancers including prostate cancer (NCT02963051), breast cancer (NCT03323346) and glioblastoma (NCT02678975, NCT03363659). These studies are still ongoing.

In this chapter, characterization of the activity of auranofin and disulfiram (in the presence and absence of copper) in models of MLL-r and other high-risk leukaemia is presented. Both compounds will be tested in MLL-r and MLL-wt leukaemia cell lines, as well as solid tumours and non-malignant cell lines. The compounds will additionally be tested in combination with currently used chemotherapeutics to determine if they

134 offer any chemopotentiating effect. The activity of the compounds will be evaluated with in vitro cultured high-risk leukaemia patient-derived xenografts including MLL-r ALL, but also extended to Philadephia-positive ALL, Philadelphia-like ALL and T- ALL. Finally, in vivo testing in one of the MLL-r leukaemia models will be conducted as a first step towards determining their potential as novel therapeutics for high-risk leukaemia.

135

5.2 Results

5.2.1 Auranofin and disulfiram affect viability of MLL-rearranged and other high-risk leukaemia cells in vitro

To characterize the activity of auranofin and disulfiram in MLL-r and other high-risk leukaemia, the drugs were tested against a panel of 25 cell lines which included high- risk leukaemia (n=16), solid tumour (n=7) and non-malignant cell lines (n=2) in 72- hour resazurin-based cytotoxicity assays as described (Section 2.2.4.1).

Auranofin exhibited high efficacy with sub-micromolar IC50 against all leukaemia cell lines except for Jurkat cells (Figure 5.2A). The IC50 of leukaemia cell lines ranged from

78 to 889 nM, with the exception of Jurkat cells that presented with an IC50 of 1667 nM

(Table 5.1). The IC50 values for solid tumours and normal cell lines lay within a higher range of 0.55 – 5 µM, with HEY ovarian cancer cells being the most sensitive among the solid tumour lines, with an IC50 of 548 nM (Table 5.1). The mean viability at 1 μM auranofin was significantly lower for the leukaemia cells indicating a selectivity of the compound for leukaemia compared to solid tumours (p<0.0001) or normal cell lines (p<0.001) (Figure 5.3B).

Disulfiram exhibited high potency against all leukaemia cell lines (Figure 5.2B), with

IC50 of leukaemia cell lines ranging from 45 to 81 nM, between 49 to 245 nM for solid tumours, and above 1 µM for normal cell lines (Table 5.1). Despite several solid tumour cell lines showing sensitivity to disulfiram, with IC50 below 150 nM (including the neuroblastoma and ovarian cancer cell lines), overall disulfiram had significant selectivity towards leukaemia cancer cell lines compared to solid tumour and normal cell lines (p<0.0001 in each case) (Figure 5.3D).

136

Figure 5.1: Chemical structures of auranofin and disulfiram. Chemical structures of (A) auranofin and (B) disulfiram

137

1 2 5 M V 4 ;1 1 U 9 3 7

R S 4 ;1 1 K P M O T S 1 0 0

) S E M K 2 C E M %

( M O L M -1 3

7 5 R E H

y t

i T H P - 1

l J U R K A T i

b 5 0 K O P N 8

a i

V P E R - 4 8 5

2 5 P E R - 4 9 0

P E R -7 8 5 A

0 P E R -7 0 3 A -8 -7 -6 -5 P E R -8 2 6 A A u ra n o fin c o n c e n tr a tio n [L o g M ]

M V 4 ;1 1 U 9 3 7 1 2 5 R S 4 ;1 1 K P M O T S

1 0 0 S E M K 2 C E M )

M O L M -1 3 R E H

% ( 7 5

y T H P - 1 J U R K A T

i l i K O P N 8

b 5 0 a

i P E R - 4 8 5 V 2 5 P E R - 4 9 0

P E R -7 8 5 A 0 P E R -7 0 3 A -8 .0 -7 .5 -7 .0 -6 .5 -6 .0 D is u lfira m c o n c e n tra tio n [L o g M ] P E R -8 2 6 A

Figure 5.2: Cytotoxicity of auranofin and disulfiram. Dose response curves for (A) auranofin and (B) disulfiram, across a range of high-risk leukaemia including MLL-rearranged (in red), CALM-AF10 (in blue) and MLL-wild- type (in green) leukaemia cell lines in 72-hour resazurin-based assay. The results are expressed as the mean ± SE of three independent experiments.

138

Table 5.1: Cytotoxicity of auranofin and disulfiram against a panel of leukaemia, solid tumour and non-malignant cell lines. IC (nM) Lineage Cell line Translocation Disease 50 Auranofin Disulfiram PER-485 t(4;11) Infant ALL 583 60 PER-490 t(4;11) Infant ALL 889 56 PER-703A t(1;11) Infant ALL 348 47 PER-785A t(4;11) Infant ALL 293 49 B-cell PER-826A Complex, t(11;19) Infant ALL 78 45 ALL KOPN-8 t(11;19) Infant ALL 132 45 RS4;11 t(4;11) Pre-B cell ALL 230 46 SEMK2 t(4;11) Pre-B cell childhood ALL 288 49 Leukaemia REH - Pre-B cell ALL 126 52 CEM - Childhood T-cell ALL 432 61 T-cell Jurkat - Childhood T-cell ALL 1668 58 MV4;11 t(4;11) Childhood AML 124 57

MOLM-13 t(9;11) AML 195 63

AML THP-1 t(9;11) Infant AML 667 51

U937 CALM-AF10 AML 470 65

KP-MO-TS CALM-AF10 AML 528 81 KELLY - Neuroblastoma 1498 73

BE(2)-C - Childhood neuroblastoma 2368 81

HEY - Ovarian carcinoma 548 115

Solid tumours 27/87 - Endometrioid ovarian cancer 1412 49

MCF-7 - Breast adenocarcinoma 2500 >1000

H460 - Lung carcinoma 3623 500

LNCaP - Prostate carcinoma 1585 106 MRC-5 - Normal lung 5000 >1000 Normal cells WI-38 - Normal lung 2076 >1000

139

A B

C D

Figure 5.3: Auranofin and disulfiram have potent activity against high-risk leukaemia cells in vitro. Viability of the cell line panel after a 72-hour treatment with (A) 1 µM of auranofin or (C) 100 nM of disulfiram, as measured in resazurin-based cytotoxicity assay; comparison of cell viability between leukaemia cell lines, solid tumours and normal cells at (B) 1 µM auranofin or (D) 100 nM disulfiram. Dots represent mean viability of three replicates. Mean viability between groups were compared by one-way ANOVA with Dunn’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; ***, P<0.001; ****, P<0.0001.

140

5.2.2 Auranofin and disulfiram induce apoptosis in high-risk leukaemia cells

To identify how auranofin affected the viability of cells, the level of apoptosis upon auranofin exposure was measured by Annexin V staining in two MLL-r cell lines (PER- 485, RS4;11) and two MLL-wt cell lines (REH, CEM), alongside a resistant leukaemia cell line (Jurkat, IC50>1 μM). All cells were dosed with 600 nM of auranofin for up to

48 hours. As the approximate IC50 of auranofin for PER-485, this dose of 600 nM was selected as the standard auranofin treatment dose from this point onwards unless stated otherwise. A significant increase in the percentage of apoptotic cells was seen in all sensitive cells, as early as 6 hours after treatment in PER-485, RS4;11 and REH and 12 hours in CEM cells, with the proportion of cells undergoing apoptosis increasing over time (Figure 5.4A). The increase in apoptotic cell percentage corresponded with sensitivity towards auranofin, with REH and RS4;11 having the lowest IC50 values and the highest levels of apoptosis induction compared to PER-485 and CEM. The resistant Jurkat cell line showed only 5% of cells undergoing apoptosis after being subjected to 48 hours of treatment (Figure 5.4A).

Similarly for characterization of disulfiram response, Annexin V staining was performed in the responsive PER-485, RS4;11, REH and CEM lines to monitor induction of apoptosis upon disulfiram treatment. All cells were dosed with 60 nM disulfiram (IC50 of disulfiram for PER-485) for up to 48 hours. This dose was chosen as the standard disulfiram treatment dose from this point onwards unless stated otherwise. Significant increases in the percentage of apoptotic cells were seen for all cell lines within 24 hours of treatment, with the proportion of cells undergoing apoptosis increasing over time (Figure 5.4B). Apoptosis induction was most rapid and pronounced in RS4;11 and REH cells.

141

Figure 5.4: Auranofin and disulfiram induce apoptosis in sensitive high-risk leukaemia cell lines. Flow cytometric analysis of Annexin V staining after 6 – 48 hour drug treatment with (A) auranofin at 600 nM or (B) disulfiram at 60 nM. In each case, apoptosis is expressed as percentage increase in Annexin V-positive cells compared to untreated control. The results are expressed as the mean ± SE of three independent experiments. Mean percentages of increase in Annexin V-positive cells were compared between untreated control and treatment groups by one-way ANOVA with Tukey’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

142

5.2.2.1 Impact of copper on disulfiram response

As mentioned previously, disulfiram was reported to have greater cytotoxicity against a variety of cancer cells when used in combination with copper (Cu). To test if this also applies for high-risk leukaemia, apoptosis was measured in the four cell lines after 24 hours of disulfiram treatment in the absence or presence of increasing Cu concentrations (100 nM, 200 nM, 500 nM or 1 μM). The addition of up to 1 μM Cu itself was not toxic to the cells (data not shown). A strong enhancement of apoptosis was observed in the CEM cells, which showed very modest apoptosis induction without Cu (Figure 5.4B and 5.5A), while there were very subtle or minimal changes in the other lines, which already had good apoptosis induction in the absence of Cu (Figure 5.4B and 5.5A). A significant increase in the percentage of apoptotic cells was seen in RS4;11 cells when treated with disulfiram together with 100 nM Cu compared to disulfiram alone. Although a further significant increase in cytotoxicity was not seen in PER-485 at any Cu concentration, a trend was observed for 100 nM Cu (Figure 5.5A). Since 100 nM Cu was sufficient to affect apoptosis induction by disulfiram, this dose of Cu was selected for all subsequent experiments. The level of apoptosis was also measured in the same leukaemia cell lines following treatment of disulfiram/100 nM Cu, over a time frame of 48 hours. Enhancement of apoptosis by Cu was most pronounced in CEM cells, with several-fold increase in Annexin V-positive cells compared to cells treated with disulfiram without Cu at all timepoints (Figure 5.5B). After 24 or 48 hours of treatment, addition of copper also increased the percentage of apoptotic cells significantly in PER- 485 and RS4;11, compared to disulfiram alone. Interestingly, although not significant addition of Cu reduced apoptosis in REH cells when treated with the combination for up to 24 hours (Figure 5.5B). The addition of copper also affected the IC50 of non- malignant cells, MRC-5, however there was still a significant 2-fold difference between

IC50 of MRC-5 cells and PER-485 leukaemia cells (Figure 5.6)

143

Figure 5.5: Copper increases efficacy of disulfiram in high-risk leukaemia cell lines. Flow cytometric analysis of Annexin V staining after (A) 24 hour treatment with 60 nM disulfiram (DSF), in the presence and absence of Cu at indicated concentrations or (B) 6 – 48 hour treatment with 60 nM DSF, in the presence and absence of 100 nM Cu. In each case, apoptosis is expressed as percentage increase in Annexin V-positive cells compared to untreated control. The results are expressed as the mean ± SE of three independent experiments. Mean percentages of increase in Annexin V-positive cells were compared between untreated control and treatment groups by one-way ANOVA with Tukey’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; **, P<0.01; ****, P<0.0001.

144

1 5 0 ****

)

(

0 5

C 1 0 0

m a

r 5 0

i D 0 P E R -4 8 5 M R C -5

Figure 5.6: Comparison of efficacy of disulfiram/Cu between PER-485 and non- malignant cell line MRC-5.

A significant difference in IC50 was seen between PER-485 and MRC-5 cells treated with disulfiram in combination with 100 nM copper. The results are expressed as the mean of three independent experiments. Mean IC50 between cell lines were compared by unpaired t-test. Asterisks represent significance levels of P-value. ****, P<0.0001.

145

5.2.2.2 Confirmation of apoptosis induction

As further confirmation of apoptotic cell death, the appearance of apoptotic markers were monitored by Western blotting of cellular extracts following treatment. The auranofin sensitive cell lines, PER-485, RS4;11 and REH were exposed to auranofin for 6 hours and the protein expression of Poly(ADP-ribose) polymerase (PARP), a nuclear protein involved in DNA repair and among the earliest targeted during apoptosis by proteolytic cleavage to yield cleaved PARP (Duriez and Shah, 1997; Sallmann et al., 1997), was determined. Treatment with auranofin increased the level of cleaved PARP compared to untreated cells (Figure 5.7A). Following six-hour treatment of disulfiram, increased PARP cleavage (compared to copper alone or vehicle, DMSO) was evident in PER-485 regardless of the absence or presence of copper (Figure 5.7B). However in REH cells, the level of cleaved PARP following disulfiram/Cu treatment did not reach the levels observed with disulfiram alone, which was consistent with the relative levels of apoptosis observed under these conditions in this cell line.

146

Figure 5.7: Confirmation of apoptosis induction in high-risk leukaemia cell lines. Immunoblots of cleaved PARP protein expression (represented by the lower band) in sensitive cell lines, PER-485, RS4;11 and REH following 6 hours of (A) auranofin (600 nM) or (B) disulfiram (60 nM), with and without copper (100 nM). In (A) the IC50 for auranofin is indicated alongside each cell line.

147

5.2.3 Auranofin and disulfiram kill leukaemia cells in vitro by increasing intracellular ROS levels

It has been previously reported that auranofin increases ROS levels and causes death of cancer cells including a CLL cell line (MEC-1) and CLL primary cells (Fiskus et al., 2014), gastric cancer cell lines (BGC-823 and SGC-7901) (Zou et al., 2015) and hepatocellular carcinoma cell line Hep3B (Hwang-Bo et al., 2017). Therefore, the level of ROS was measured in the panel of four sensitive cell lines (PER-485, RS4;11, REH, CEM) and one resistant cell line (Jurkat) after one hour of treatment with auranofin through flow cytometry-based detection of cell permeate 2',7'- dichlorodihydrofluorescein diacetate (DCF-DA). In the presence of ROS, DCFDA is oxidized into a highly fluorescent compound 2',7'-dichlorodihydrofluorescein (DCF), proportionally quantifying the amount of intracellular ROS. Three sensitive cell lines, PER-485, RS4;11 and REH displayed a significant increase in intracellular ROS levels after one hour of incubation with auranofin. In contrast, CEM cells did not show elevated ROS levels after a one-hour treatment with auranofin which could possibly be due to the short treatment period and the apparently longer time course for development of apoptosis in that cell line (Figure 5.4A). The level of ROS in the resistant Jurkat cells remained unchanged upon treatment with auranofin. In addition, compared to other cell lines, Jurkat cells also exhibited less ROS induction, in response to the positive control, hydrogen peroxide (Figure 5.8A).

Disulfiram has also been reported to increase ROS levels in cancer cells in vitro, including breast cancer cell lines MCF-7, MDA-MB-231 and T47D (Yip et al., 2011) and lymphoid cell line, Raji (Zha et al., 2014). To identify whether the compound also induces ROS in leukaemia cells, the level of ROS was measured with DCF-DA in the four sensitive high-risk ALL cell lines after six hours of treatment with disulfiram, in the presence and absence of copper. PER-485 cells displayed a significant increase of intracellular ROS (Figure 5.8B), while no change in ROS level was observed in the remaining cell lines compared to the control (RS4;11 data not shown due to issues with positive control) despite investigation of additional timepoints at 1, 3, 9, 12 and 24 hours (data not shown). Another fluorogenic dye, MitoSOX was then used to determine if ROS might be generated but localized to the mitochondria. This probe measures

148

B C

Figure 5.8: Auranofin and disulfiram increase intracellular ROS. (A) Intracellular ROS level as measured by flow cytometric analysis of DCFDA after one-hour auranofin (AUR) treatment (600 nM). Hydrogen peroxide (H2O2) treatment was conducted alongside as a positive control. (B), (C) Flow cytometric analysis of intracellular ROS levels after six-hour treatment with 60 nM disulfiram (DSF), in the presence and absence of 100 nM copper (Cu) as measured by (B) DCFDA, with H2O2 a positive control, and (C) MitoSOX, with menadione as a positive control. In each case, ROS level is expressed as percentage increase compared to untreated control and mean increase of ROS from three independent experiments was compared by one-way ANOVA with Tukey’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

149 mitochondrial superoxide rather than cytoplasmic hydrogen peroxides as detected by DCFDA. Significant increases in mitochondrial ROS levels were detected in RS4;11 and CEM cells at 6 hours post treatment with disulfiram, either with or without copper (Figure 5.8C). A trend for higher ROS levels upon treatment with disulfiram was observed for PER-485 and REH, although this failed to reach statistical significance. As expected, disulfiram/Cu treatment did not enhance ROS in REH cells, in concordance with previous experiments showing the absence of an additional cytotoxic effect of Cu addition (Figure 5.8C; Figure 5.5).

5.2.3.1 Treatment of auranofin or disulfiram is accompanied by induction of oxidative stress pathway

To further validate the ROS-inducing action of auranofin, the effect of the compound on the induction of several proteins involved in the oxidative stress pathway was analyzed through Western blotting of treated cells. Both nuclear factor (erythroid-derived 2)-like 2 (Nrf2), a master regulator of the antioxidant response, and haem oxygenase 1 (HMOX1), a protein encoded by the Nrf2-activated HMOX1 gene, were induced in the sensitive cells after six-hour treatment with auranofin (Figure 5.9A). In concordance with the ROS data, no increases in Nrf2 or HMOX1 were observed for treated Jurkat cells. Elevated levels of ROS are described to result in DNA damage, culminating in apoptosis (Choi and Park, 2012; Yang et al., 2017). Thus DNA damage was assessed by measuring the levels of phosphorylated H2A histone family, member X (γH2AX), a marker for DNA damage, particularly in response to formation of double-strand breaks (Huang et al., 2006). In agreement with the dynamics of response observed so far, auranofin-treated sensitive cells displayed higher levels of γH2AX, while this was not observed in resistant Jurkat cells (Figure 5.9A). To further confirm the effect auranofin has on the ROS pathway, cells were pre-treated with a ROS scavenger, N-acetyl cysteine (NAC) for one hour. The ROS scavenger prevented the increase of intracellular ROS and significantly reduced the proportion of cells undergoing apoptosis in all cases (Figure 5.9B and 5.9C). Despite the lack of detectable increase of auranofin-mediated ROS in CEM cells, NAC rescued these cells from apoptosis indicating the mechanism of cell death is mediated through ROS. Pre-treatment with NAC attenuated expression of Nrf2 and HMOX1 protein upon auranofin treatment (Figure 5.9A). These findings

150

B C

Figure 5.9: Auranofin induces higher expression of proteins in the oxidative stress pathway and pre-treatment with NAC prevents auranofin-induced ROS production and rescues cells from apoptosis. (A) Expression levels of Nrf2, HMOX1 and γH2AX upon six-hour treatment with 600 nM auranofin (AUR) were detected by Western blot and compared between sensitive cells and resistant cell line, Jurkat. Pre-incubation with 5 mM of ROS scavenger, N- acetyl cysteine (NAC) prevented the expression of these proteins in sensitive cell lines. (B) Impact of NAC on ROS induction upon treatment with auranofin as measured by DCFDA. (C) Impact of NAC on apoptosis as measured by Annexin V staining. The ROS and apoptosis results are expressed as the mean ± SE of three independent experiments. Mean percentages of increase in ROS and Annexin V-positive cells were compared between untreated control and treatment groups by one-way ANOVA with Tukey’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

151 indicate that the apoptosis induced by auranofin in leukaemia cells is largely mediated through the auranofin-induced ROS increase.

Similarly, the ROS-mediated mechanism of action of disulfiram was validated by assessing oxidative stress pathway proteins in PER-485, RS4;11 and CEM cells. Induction of Nrf2 was observed after a six-hour treatment with disulfiram and the level of expression increased in the presence of Cu (Figure 5.10A). Expression of HMOX1 was also elevated in cells treated with disulfiram, both with and without copper. The extent of DNA damage was confirmed by induction of γH2AX. PER-485 cells displayed less induction of γH2AX than other cells, which could possibly be due to the location of the oxygen species in cytoplasm resulting in less DNA damage compared to mitochondrial ROS. To further confirm the effect disulfiram and copper have on the redox pathway, cells were pre-treated with NAC for one hour prior to treatment. The ROS level in PER-485 was significantly reduced when pre-treated with NAC (Figure 5.10B). NAC also significantly reduced mitochondrial ROS in RS4;11 and CEM (Figure 5.10C). A trend for reversal of mitochondrial ROS was also observed in PER- 485 and REH cells, however this failed to reach statistical significance (Figure 5.10C). Regardless of copper addition, the level of apoptosis induced by disulfiram was significantly reduced in NAC-treated samples across all cell lines at 24 hours, with complete prevention of apoptosis by NAC evident in PER-485 cells (Figure 5.10D). Pre-treatment with NAC also attenuated expression of Nrf2 and HMOX1 protein in each line (Figure 5.10A). These data illustrated that disulfiram treatment, with or without copper, executes its function in abolishing leukaemia cells through the ROS pathway.

152

B C

153

Figure 5.10: Disulfiram induces higher expression of proteins in the oxidative stress pathway and pre-treatment with NAC prevents disulfiram-induced ROS production and rescues cells from apoptosis. (A) Expression levels of Nrf2, HMOX1 and γH2AX upon six-hour treatment with 60 nM disulfiram (DSF), in the presence and absence of 100 nM copper (Cu) were detected by Western blot in sensitive cell lines. Pre-incubation with 5 mM of ROS scavenger, N-acetyl cysteine (NAC) prevented the expression of these proteins. (B) Impact of NAC on ROS induction on PER-485 cells upon treatment with DSF ± Cu as measured by DCFDA. (C) Impact of NAC on mitochondrial ROS induction in sensitive cell lines upon treatment with DSF ± Cu as measured by MitoSOX; (D) Impact of NAC on apoptosis as measured by Annexin V staining of sensitive cell lines. The ROS and apoptosis results are expressed as the mean ± SE of three independent experiments. Mean percentages of increase in ROS and Annexin V-positive cells were compared between untreated control and treatment groups by one-way ANOVA with Tukey’s correction for multiple comparisons. Asterisks represent significance levels of P-values. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

154

5.2.4 Auranofin and disulfiram/Cu potently affect viability of high-risk leukaemia patient-derived xenografts in vitro

Next, the activities of auranofin and disulfiram/Cu were determined in a panel of paediatric high-risk ALL PDXs in vitro since these paediatric patient-derived cells are more closely representative of primary patient samples than are cultured cell lines. The panel included MLL-r infant ALL, Ph+ ALL, Ph-like ALL and T-ALL from patients either at diagnosis and relapse (Table 5.2). Cell viability was measured after 48 hours of compound exposure in resazurin-based cytotoxicity assays (Figure 5.11). Auranofin showed high potency in both MLL-r and MLL-wt xenografts in the high-risk leukaemia

PDX panel with IC50 ranging from 55 to 1597 nM. The two least responsive were of T-

ALL lineage, one derived at relapse (IC50>1 μM) and one at diagnostis (IC50 = 744 nM)

(Table 5.2, Figure 5.11A, B), while xenografts of other high-risk subtypes had IC50 below 400 nM.

Disulfiram as single agent did not affect the viability of the PDX cells (Supplementary Figure 9). However, with addition of copper (100 nM), disulfiram was able to kill the

PDX cells with similar potency to that observed for the leukaemia cell lines, with IC50 ranging from 28 to 100 nM (Table 5.2; Figure 5.11C, D). This effect could be due to a low level of available intrinsic copper in these PDXs, preventing the DDC-copper complex formation or the lack of copper transmembrane transporters which are major determinant of endogenous Cu.

Thus, both auranofin and disulfiram show potent in vitro activity against xenograft cells derived from some of the most aggressive sub-types of ALL.

155

Table 5.2: Cytotoxicity of auranofin and disulfiram/Cu in high-risk paediatric leukaemia patient-derived xenograft panel.

IC50 (nM) Subtype Xenograft Disease status at biopsy Auranofin Disulfiram/Cu MLL-2 - 377 87 MLL-5 Diagnosis 132 35 MLL-6 - 389 100 MLL-r ALL MLL-7 - 79 51 MLL-8 Diagnosis 55 81 MLL-14 Diagnosis 122 64 ALL-2 Relapse 208 74 ALL ALL-7 Diagnosis 154 35 ALL-19 Relapse 184 38 ALL-8 Relapse 1597 34 T-ALL ALL-31 Diagnosis 744 36 ALL-4 Diagnosis 119 28 Philadelphia + ALL ALL-55 Diagnosis 353 46 ALL-56 Diagnosis - 31 Philadelphia-like ALL TGT-052 Diagnosis 127 48

156

A B

C D

Figure 5.11: Auranofin and disulfiram/Cu affect viability of high-risk leukaemia PDXs in vitro. (A) Dose response curves of MLL-rearranged (MLL-r) PDXs after treatment with (A) auranofin or (C) disulfiram; dose response curves of MLL-wild-type (MLL-wt) high-risk leukaemia after treatment with (B) auranofin or (D) disulfiram, as measured in 48-hour resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments. 157

5.2.4.1 Auranofin activity in MLL-rearranged ALL patient derived xenografts

To confirm if MLL-r PDX were also affected by auranofin through the ROS pathway, two xenografts, MLL-6 and MLL-7, were treated with auranofin for six hours in vitvo, in the presence or absence of ROS scavenger, NAC. RS4;11 was also included in the assay as a positive reference sample. Two of the proteins involved in the oxidative stress pathway, Nrf2 and γH2AX, were analysed through Western blotting. The MLL-6 xenograft cells showed elevated levels of Nrf2 and γH2AX following treatment of auranofin that was rescued by NAC pre-treatment. On the other hand, despite being one of the xenografts with lowest IC50, the MLL-7 xenograft cells displayed no increase in either protein in response to treatment was observed (Figure 5.12).

158

Figure 5.12: Nrf2 and γH2AX expression in MLL-6 and MLL-7 cells treated with auranofin in vitro. Western blot of γH2AX in two PDXs, MLL-6 and MLL-7, with RS4;11 as positive control, following 6-hour in vitro treatment of 600 nM auranofin and 5 mM ROS inhibitor, NAC. An increased level of γH2AX is seen in MLL-6 xenograft cells upon treatment of auranofin and NAC prevented this increase, comparable to the effect seen in the RS4;11 leukaemia cell line treated with the compound. In contrast, MLL-7 did not respond to auranofin with an increase of γH2AX.

159

5.2.5 Determining in vivo efficacy of auranofin using patient-derived xenograft models of MLL-rearranged leukaemia

The results so far have demonstrated the potency of auranofin and disulfiram in inhibiting growth and viability of some of the most aggressive types of paediatric leukaemia cell lines and PDX cells in vitro through ROS-mediated mechanisms. The next critical step in determining their efficacies is to test each compound in preclinical animal models. The orthotopic patient-derived xenograft mouse model of ALL has become one of the gold standard measures of preclinical efficacy for drugs in development and for assessing the potential clinical utility of new or repurposed agents. In this model, a systemic disease is achieved by inoculation of patient-derived xenograft cells via the tail-vein of immunodeficient mice (Lock et al., 2002). Human leukaemic cells highly infiltrate the bone marrow and proliferate and disseminate to peripheral blood and spleen, as well as other organs, reflective of the disease course in ALL patients. The infiltration of the peripheral blood allows accurate measurement of leukaemic burden by establishing the percentage of human versus murine CD45 positive cells in the blood (Lock et al., 2002). Highly engrafted mice have their bone marrow and spleen almost entirely replaced by human leukocytes, enabling harvest of leukaemic cells, mainly from the spleen, for transplantation into subsequent mice recipients or further molecular analyses. The purified human cells from the splenocytes have been shown to maintain patient immunophenotype and genetic characteristics (Liem et al., 2004). This robust, reliable and clinically relevant murine model is currently used in the Pediatric Preclinical Testing Program (PPTP), an international collaborative initiative supported by the NCI, to test novel therapeutic agents for childhood cancers (Houghton et al., 2007).

In this section, in vivo efficacy of auranofin was tested to assess the potential of this drug for treating MLL-r leukaemia. Maximum tolerated dose (MTD) for auranofin was first determined in the non-obese diabetic/severe combined immunodeficient (NOD/SCID) ALL mouse model, followed by selection of a PDX assessment of in vivo efficacy of auranofin.

160

5.2.5.1 Determining the maximum tolerated dose of auranofin in NOD/SCID mice

Since there were no available reports on the use of auranofin in NOD/SCID mice, it was necessary to first determine the MTD for auranofin in these mice. The drug was injected intraperitoneally into mice (n=3) at increasing concentrations starting with vehicle control (0), 2.5, 5, 7.5 and 10 mg/kg for 5 consecutive days per week, over 3 weeks. The range of dose was determined from reports in the literature describing in vivo testing of auranofin whereby 10 mg/kg was the highest dose reported in mice (Fiskus et al., 2014). Mice were checked on treatment days and weighed weekly to monitor toxicity-related weight loss. No significant weight loss was observed in mice treated with any doses of auranofin (Supplementary Figure 10). One mouse in the 10 mg/kg treatment group presented with signs of toxicity such as decreased activity, ruffled coat and hunched posture after two injections. Treatment for this mouse was immediately stopped to prevent further adverse side effects. The other mice in 10 mg/kg group and all mice in the other treatment groups remained free of any external physical signs of toxicity over the course of treatment.

Three weeks after the end of treatment, mice were humanely sacrificed and post mortem investigations were conducted to determine any internal abnormalities. Mice treated with 5 to 10 mg/kg auranofin exhibited enlarged livers, while other organs appeared normal. Representative liver samples from one mouse from each of the control or drug treatment groups were processed for haemotoxylin and eosin staining and analyzed by a pathologist. Histologic examination of murine livers post administration of auranofin showed fibrotic thickening of Glisson’s capsule in treated mice. This effect appeared to occur in a dose dependent manner, given the least degree of fibrosis in the mouse treated with 2.5 mg/kg auranofin and greatest fibrosis seen following 10 mg/kg treatment (Figure 5.13B – E). Masson trichrome staining, a three-colour stain to determine hepatic toxicities, was performed for control and 5 mg/kg liver samples and showed subtle changes in the liver treated with 5 mg/kg auranofin compared to the liver from control mouse (Figure 5.14A, B). Apoptosis of hepatocytes adjacent to the hepatic vein in two large portal tracts were also observed in mice treated with 5 mg/kg auranofin (Figure 5.14D) (A. Gifford, personal communication). From this, MTD was

161 determined as 2.5 mg/kg as this dose did not induce observable adverse side effects over the course of treatment.

A B

C D

Figure 5.13: Morphologic appearance of Glisson’s capsule in livers of mice in control and treatment groups in MTD study. Histological changes of Glisson’s capsule (arrows) observed by haemotoxylin and eosin staining in livers from mice treated with vehicle or 2.5 – 10 mg/kg auranofin for three weeks. Progressive thickening of the liver capsule can be observed with increasing doses of auranofin administered. Magnification, 600x.

162

A B

C D

Figure 5.14: Morphologic appearance of Glisson’s capsule and hepatocytes in livers of mice in control and 5 mg/kg treatment groups of MTD study. Masson trichrome staining of (A) control and (B) 5 mg/kg auronofin treated livers showing marked thickening of Glisson’s capsule (stained blue) following auranofin treatment (magnification, x200). Morphologic appearance of periportal hepatocytes of (C) control liver (magnification, 400x) and (D) liver from mouse treated with 5 mg/kg auranofin (magnification, 600x), visualized by haemotoxylin and eosin staining. Drug treated liver showed periportal apoptotic hepatocytes (arrows) indicating hepatic toxicity.

163

5.2.5.2 In vivo efficacy of auranofin in a patient-derived xenograft model for high- risk MLL-rearranged leukaemia

In vivo efficacy of auranofin was assessed in the preclinical NOD/SCID ALL mouse model (Lock et al., 2002). Mice were engrafted with MLL-6 PDX cells as described (Section 2.2.11.6). The xenograft was selected after an earlier experiment showed increased expression of Nrf2 and γH2AX, two of the proteins involved in the oxidative stress pathway, in MLL-6 cells after Auranofin treatment. Despite being one of the xenografts with lowest IC50, the MLL-7 xenograft cells displayed no increase in either protein (Figure 5.12). Treatment started when the mouse cohort had a median of 1% human CD45+ cells in peripheral blood (PB). Mice were randomized into two groups (n=8 per treatment group) and given either 2.5 mg/kg of auranofin or DMSO as vehicle. The drug was delivered to the abdominal cavity via intraperitoneal (i.p) injection for five days per week, for three weeks (3 cycles: 5 days on, 2 days off). As Figure 5.15 shows, administration of auranofin did not delay leukaemia growth (p=0.4518, EFS=2.1 days). All mice in either control or treatment groups had reached event by the end of treatment.

Following this result, a short-term pharmacodynamic study was performed to determine if the treatment dose used was sufficient to have an effect on the leukaemia cells in vivo. Mice were engrafted with either MLL-6 or MLL-7, each of which showed high sensitivity in vitro, and treated with either 2.5 mg/kg auranofin or vehicle control over two days, or a single dose of 10 mg/kg auranofin. Mice were culled 4 hours after the final dose and highly infiltrated spleens were collected. The levels of Nrf2 and γH2AX were measured by Western blotting as indicators of auranofin delivery to the tissue of interest and subsequent activity. Auranofin treatment did not have an effect on Nrf2 levels at the time point of sampling. However, while an increased level of γH2AX was seen in the 10 mg/kg sample for both PDXs, no such observation was made in mice treated with 2.5 mg/kg, indicating that the treatment dose might have been too low to cause an effect on leukaemia cells in vivo (Figure 5.16).

164

Figure 5.15: In vivo efficacy testing of auranofin in MLL-rearranged ALL xenograft, MLL-6. Auranofin did not show significant activity in vivo: no significant leukaemia growth delay (LGD) was seen in MLL-6 engrafted mice compared to the control mice (p=0.4518, log-rank Mantel-Cox test). The drug was delivered to the abdominal cavity via i.p injection for 3 cycles: 5 days on, 2 days off. N=8 mice per treatment group.

Figure 5.16: Immunoblots of Nrf2 and γH2AX in MLL-6 and MLL-7 in vivo treated cells. Western blot of extracts from MLL-6 or MLL-7 cells extracted from splenocytes of MLL-6 or MLL-7 engrafted mice (one mouse per treatment group) following in vivo auranofin (AUR) treatment at the indicated doses (mg/kg). Mice were treated with either 1 dose of vehicle (0 mg/kg) or 2.5 mg/kg auranofin daily over 2 days, or 10 mg/kg for 4 hours (due to toxicity). Protein expression of Nrf2 was not distinctly different among the in vivo treatment groups. An increased expression of γH2AX was seen in spleen cells from mice treated with the 10 mg/kg dose in both PDXs but not for the 2.5 mg/kg dose.

165

5.2.5.3 Cytarabine potentiates auranofin in vitro

Drug potentiation has been shown to be achievable even in the absence of single agent efficacy. For example, Cruickshank et al. (2017) previously observed enhancement of EFS for two drugs, romidepsin and bortezomib, despite neither being efficacious as a single agent. Similarly, enhancement of auranofin activity might be possible, especially since auranofin had been reported to synergize with cytotoxic drugs and compounds in several types of cancers. Auranofin was shown to synergize with mesupron, an inhibitor of urokinase-type plasminogen activator (uPA), inducing apoptosis and inhibiting proliferation of human breast cancer cells, MCF-7 cells in vitro (Lee et al., 2017). The drug also demonstrated synergy with selenocystine, a naturally available selenoamino acid, potently killing the same breast cancer cell line through ROS-dependent apoptosis with the involvement of mitochondrial dysfunction (Liu et al., 2013). Interestingly, the same auranofin and selenocystine combination was also reported to be efficacious in non–small cell lung carcinoma (NSCLC) in vitro and in vivo (Fan et al., 2014). The drug also lethally affected NSCLC cell lines in vitro and in vivo when in combination a targeted agent, MK2206 (protein kinase B inhibitor) (Dai et al., 2013). Auranofin in combination with sirolimus/rapamycin, a mammalian target of rapamycin (mTOR), also recently entered a phase II clinical trial for patients with ovarian cancer (NCT03456700).

To investigate the possibility of a potentiating effect with auranofin, combination assays with five chemotherapeutic agents currently used in the treatment of paediatric ALL (as also selected for testing in the previous chapter) were performed against two infant MLL-r leukaemia cell lines, PER-485 and PER-490. Cells were exposed to increasing doses of auranofin or chemotherapy as single agents or a combination of the two simultaneously at fixed ratios. Cell viability was assessed using resazurin-based cytotoxity assays after 72 hour treatment and combination indices (CI) were calculated using the CalcuSyn program (Section 2.2.4.2). Auranofin showed synergy with cytarabine in both cell lines with a CI of 0.74 (Table 5.3, Supplementary Figure 11, 12). The indices revealed additive effects between auranofin and either mitoxantrone or etoposide, and antagonistic effects with either daunorubicin or topotecan (Table 5.3, Supplementary Figure 11, 12).

166

Table 5.3: Calculated combination index and drug reduction index for auranofin in combination with currently used drugs in the treatment of paediatric ALL at ED75 in PER-485 and PER-490 cells. Combination index PER-485 PER-490 Mitoxantrone 1.02 0.93 Daunorubicin 1.45 1.10 Etoposide 0.97 1.04 Cytarabine 0.74 0.74 Topotecan 1.41 1.43

Combination index below 0.9 indicates synergism, between 0.9 and 1.1 indicates additive effect, and above 1.1 signifies antagonism. ED75: effective dose causing 75% reduction of cell viability.

Figure 5.17: Immunoblots of γH2AX levels in PER-485 and REH cells treated with auranofin and cytarabine. Western blot of γH2AX phosphoprotein levels in PER-485 and REH cells treated with two concentrations of each drug alone or in combination for 48 hours. The concentrations selected for cytarabine were its IC50 of each cell line and one lower concentration. For PER-485, 2 and 4.5 μM concentrations were used, while for REH, 10 and 15 nM. Increased γH2AX phosphoprotein is evident in both cell lines when treated with combination of the two drugs.

167

ROS has been reported to induce DNA damage (Choi and Park, 2012; Yang et al., 2017), while cytarabine, a nucleoside analogue, is an established DNA-damage causing agent through its incorporation into newly synthesized DNA (Geller et al., 2001). Therefore, the potential mechanism of synergy between auranofin and cytarabine was assessed at a molecular level through detection of γH2AX, a biomarker of DNA damage, in PER-485 and REH cells. Both cell lines were treated in vitro with either auranofin alone at 300 or 600 nM, cytarabine alone at 0.5 or 1 μM, or both drugs simultaneously for 48 hours. In both cell lines, a dose-dependent increase in γH2AX was observed upon treatment with either auranofin or cytarabine as single agents (Figure 5.17). Moreover, combined treatment with both drugs strongly enhanced the level of γH2AX in PER-485 and moderately in REH, even further suggesting increased DNA damage as a possible mechanism of synergy (Figure 5.17).

5.2.5.4 Prioritizing PDX for auranofin/cytarabine in vivo combination testing

The synergy observed between cytarabine and auranofin suggested that this could be a clinically relevant combination treatment for high-risk leukaemia, such as MLL-r ALL, thus warranting further investigation in PDX models in vivo. To select a MLL-r PDX for in vivo testing, four MLL-r PDXs were investigated for their responses to the combination treatment in vitro using 48-hour resazurin-based combination assays and Western blotting of pharmacodynamic markers following a 48-hour drug treatment to determine drug effects. MLL-5, MLL-6, MLL-7 and MLL-14 were selected for these experiments.

Firstly, 48-hour resazurin-based assays were performed with a dose range of auranofin and a fixed dose of cytarabine, according to its IC50 in each xenograft. As depicted in Figure 5.18, potentiation of auranofin was evident especially at the lower dose range, in all PDX except for MLL-6. MLL-5 showed the highest fold decrease in IC50 compared to other xenografts (Table 5.4). In the second experiment, levels of γH2AX and occurrence of PARP cleavage were monitored through Western blot. None of the PDXs showed increased levels of γH2AX after treatment compared to vehicle treated cells, but MLL-5 showed an increase in PARP cleavage in response to cytarabine and in combination with auranofin and cytarabine compared to vehicle treated cells among the

168

Figure 5.18: Combination assays of auranofin and cytarabine in MLL-rearranged patient-derived xenografts. Dose response curves of MLL-5, MLL-6, MLL-7 and MLL-14 following a combined treatment of a full-dose range of auranofin and a fixed dose of cytarabine (respective

IC50 of each PDX) for 48 hours, as measured by resazurin based assays. A co-operative effect of the drugs in all PDXs except MLL-14. Dotted line on each graph represents viability of cells treated with single agent cytarabine at its IC50 concentration.

Table 5.4: Fold decrease of auranofin IC50 for MLL-rearranged patient derived xenografts in auranofin and cytarabine combination assays.

IC50 (nM) PDX Fold decrease Auranofin alone Auranofin + cytarabine MLL-5 222 103 2.2 MLL-6 445 361 1.2 MLL-7 224 164 1.4 MLL-14 112 71 1.6

IC50: inhibitory concentration resulting in 50% reduction of cell survival relative to control. IC50 of each xenograft was derived from a 48-hour resazurin-based combination assay.

169

Figure 5.19: Immunoblots following auranofin and cytarabine combination treatment in MLL-rearranged patient-derived xenografts. Western blot showing distinct PARP cleavage (represented by the lower band) in MLL-5 after 48-hour combination treatment of auranofin (AUR) and cytarabine in vitro, but not in other PDXs. No apparent increase in γH2AX levels upon treatment with auranofin or cytarabine was seen in any of the PDXs.

170 four PDXs (Figure 5.19). Based on the information obtained from these in vitro testings, MLL-5 was selected for in vivo testing.

5.2.5.5 Efficacy of combined treatment with auranofin and cytarabine in a patient- derived xenograft model for MLL-rearranged leukaemia

Auranofin was administered 5 days per week for 3 weeks, as previously determined and cytarabine was administered at 25 mg/kg for 4 consecutive days over 2 weeks (2 cycles: 4 days on, 3 days off). The cytarabine dose and treatment regimen used were based on previously optimized treatment protocols. Mice engrafted with MLL-5 were randomized into four groups (8 mice per group) when the level of human CD45 positive cells in peripheral blood reached 1%: (i) vehicle control, (ii) 2.5 mg/kg auranofin (iii) 25 mg/kg cytarabine (iv) 2.5 mg/kg auranofin plus 25 mg/kg cytarabine. The study revealed a statistically significant delay in leukaemia progression in mice treated with the combination of auranofin and cytarabine compared to either compound as single agents (auranofin, p=0.0005; cytarabine, p=0.016), although due to the aggressiveness of MLL-5 (Richmond et al., 2015; Khaw et al., 2016), the leukaemia growth delay was just three days (Figure 5.20A).

To verify the effect of drug treatment on the leukaemia cells in vivo, spleen cells were harvested from MLL-5 leukaemia-bearing mice following short-term treatment. The mice in the control group were used for this experiment, whereby once the control mice reached event and were highly engrafted (human CD45 positive cells of 50 – 70%), they were assigned randomly into one of four treatment groups and received two treatment doses, followed by culling 4 hours after the second dose. The expression levels of Nrf2 and γH2AX, were measured in all samples through Western blotting. In contrast to the observations of cultured MLL-5 cells that were treated in vitro (Figure 5.19), cytarabine clearly enhanced the level of γH2AX, although no further increase was observed in the presence of auranofin (Figure 5.20B). Additional doses of both auranofin and cytarabine might be required to establish the potential mechanism of synergy through DNA damage as observed in vitro. In this pharmacodynamics study, in vivo treatment was only possible for a period of 28 hours, while the drug combination was shown in vitro to

171

Figure 5.20: In vivo efficacy testing of auranofin and cytarabine combination in a MLL-rearranged ALL xenograft. (A) Combination of auranofin and cytarabine significantly delayed leukaemia growth in MLL-5 engrafted mice. Mice were randomized into four groups as above (8 mice per treatment group) when level of human CD45 positive cells in peripheral blood reached 1%: (i) vehicle control, (ii) 2.5 mg/kg auranofin (iii) 25 mg/kg cytarabine (iv) 2.5 mg/kg auranofin plus 25 mg/kg cytarabine. Drugs were given via intraperitoneal injection for 3 cycles (5 days on, 2 days off). Statistical analyses were performed using the log-rank Mantel-Cox test. (B) Immunoblots of spleen cell extracts prepared from mice treated with 2 doses of drug at the doses indicated in (A). Spleens were harvested 4 hours after the last dose of drugs. Protein expression of Nrf2 was not distinctly different between the treatment groups, however an increased expression of γH2AX was seen in cytarabine treated mice (alone or in combination with auranofin (AUR)).

172 take a treatment period of 48 hours to cause DNA damage. The protein levels of Nrf2 were not significantly different between the treatment groups (Figure 5.20B). Unfortunately, the level of PARP cleavage in these sample was not determined.

5.2.6 Determining in vivo efficacy of disulfiram/Cu using patient-derived xenograft models of MLL-rearranged leukaemia

The promising results of in vitro testing which demonstrated high activity of disulfiram/Cu towards leukaemia cell lines and PDXs, warrants further investigations into its ability to inhibit leukaemia growth in vivo. The outcome could potentially advance the drug into the clinic for the treatment of aggressive types of paediatric leukaemia.

5.2.6.1 Determining the maximum tolerated dose of disulfiram/Cu in NOD/SCID mice

To test the in vivo efficacy of disulfiram and copper in the NOD/SCID ALL mouse model, MTD was first determined in this mouse strain. The range of dose was determined from literature search on in vivo testing of disulfiram whereby 200 mg/kg for treatment over three weeks was the highest dose reported in mice (Deng et al., 2016). Disulfiram was given orally at increasing concentrations starting with control (0), 100, 150 and 200 mg/kg for 5 consecutive days, over 4 weeks (4 cycles: 5 days on, 2 days off). Another group of mice was also administered the same doses of disulfiram, in addition to 1.5 mg/kg copper delivered orally prior to disulfiram. To monitor toxicity- related weight loss, mice were checked on treatment days and weighed weekly. No significant weight loss was observed in mice treated with any doses of disulfiram (Supplementary Figure 13). To monitor disulfiram-mediated hepatoxicity, levels of liver enzymes, alkaline phosphatase (ALP) and alanine aminotransferase (ALT) were also tested at 1, 3 and 5 weeks post treatment in mice treated with disulfiram plus copper. Based on historical data, a level of ALP above 70 U/L and a level of ALT below 30 U/L are considered normal for this mouse strain. Neither of these analytes substantially deviated from the normal range during or after the treatment period (Table 5.5). To check for organ abnormalities, mice were humanely sacrificed three weeks after the end

173

Table 5.5: Level of alkaline phosphatase and alanine aminotransferase in leukaemia-free NOD/SCID mice during toxicity testing with disulfiram/Cu. Disulfiram dose (mg/kg) Alkaline phosphatase Alanine aminotransferase + 1.5 mg/kg Cu (U/L) (U/L) Day 7 post treatment 0 107 31 100 107 35 150 90 25 200 81 27 Day 21 post treatment 0 107 25 100 94 36 150 75 25 200 77 31 Day 35 post treatment 0 90 24 100 78 27 150 72 24 200 70 30

Normal range: alkaline phosphatase >70 U/L, alanine aminotransferase <30 U/L.

174 of treatment for post mortem investigations. None of the mice showed any abnormalities. From this study, the highest dose of disulfiram (200 mg/kg) was considered safe as it did not induce observable adverse side effects over the course of treatment either in the presence or absence of 1.5 mg/kg copper and therefore was selected for efficacy studies.

5.2.6.2 In vivo efficacy testing of disulfiram/Cu in a patient-derived xenograft model for MLL-rearranged leukaemia

To determine the efficacy of disulfiram/Cu in vivo, three of the most sensitive MLL-r PDXs in vitro, MLL-5, MLL-7 and MLL-14 (Section 5.4.2), were selected for engraftment in NOD/SCID mice. The treatment groups were (i) vehicle control (ii) 200 mg/kg disulfiram (iii) 1.5 mg/kg Cu and (iv) 200 mg/kg disulfiram plus 1.5 mg/kg copper. To gather information about drug efficacy in as many PDX as possible yet using a minimal number of mice, each treatment group consisted of just two mice each, with a control group of four mice. Although the small number of mice makes it difficult to achieve statistical significance, this efficient experimental design allows for the heterogeneity of the disease in terms of response without substantially increasing the experimental cost. A recent study of historical data demonstrated high reproducibility of the NOD/SCID mouse PDX model. It showed that even when one mouse was selected randomly from the treatment group, it gave prediction accuracy of 95% (Murphy et al, 2016). Disappointingly, the administration of disulfiram or disulfiram/Cu did not appear to delay the rate of leukaemia growth for any of the PDX (Figure 5.21A). To investigate whether disulfiram/Cu treatment was able to exert its action on the leukaemia cells in vivo at the dose used, the induction of ROS pathway protein Nrf2 and γH2AX levels were analyzed by Western blotting of spleen cells harvested from treated mice. In this study, once control mice reached event and were highly engrafted, they were randomized into four groups and given three doses of either disulfiram alone, disulfiram/Cu, copper alone or vehicle before harvesting spleen cells for preparation of protein extracts. Western blotting revealed that the levels of γH2AX increased in MLL- 5 disulfiram/Cu-treated samples indicating a greater extent of DNA damage compared to disulfiram treatment alone (Figure 5.21B). Disulfiram did not increase the level of γH2AX in MLL-7 and MLL-14, regardless of the presence of copper (Figure 5.21B).

175

Figure 5.21: In vivo efficacy testing of disulfiram/Cu in MLL-rearranged ALL patient xenograft models. Mice received vehicle control, 1.5 mg/kg copper (Cu), 200 mg/kg disulfiram (DSF) or 200 mg/kg disulfiram plus 1.5 mg/kg copper. Copper and disulfiram were given via oral administration for 4 cycles (5 days on, 2 days off). (A) Disulfiram in the presence or absence of copper did not delay leukaemia growth. (B) Immunoblots of spleen cell extracts prepared from mice engrafted with the indicated PDX and treated with three doses of drug. The level of γH2AX increased in MLL-5 disulfiram/Cu-treated cells indicating the treatment caused DNA damage. 176

5.2.6.3 Potentiation of disulfiram/Cu in high-risk leukaemia

Following the inability to demonstrate of disulfiram efficacy in vivo, combination assays with five currently used chemotherapeutic agents were performed in an infant MLL cell lines, PER-485 to determine synergy between disulfiram and the established drugs. Cells were subjected to disulfiram alone, chemotherapy agent alone or a combination of the two simultaneously at fixed ratios. Cell viability was assessed using resazurin-based cytotoxity assay after 72 hours. Combination indices were calculated using the CalcuSyn program as previously described. Disulfiram did not show synergy with any of the chemotherapeutics (Table 5.6, Figure 5.22A-E).

In the absence of copper, disulfiram has been reported to synergize with auranofin in hepatoma cancer cell lines (Huang et al., 2016) and in the presence of copper these agents synergized in a breast cancer cell line (Papaioannou et al., 2014). Therefore combination assays were performed with disulfiram and auranofin to assess whether these two drugs may have synergistic anti-leukaemia effects in PER-485 cells. Using the Calcusyn algorithm, the calculated CI indicated the drugs were antagonistic towards each other (CI=1.12). However, from the graphical plot it was clear that there was some co-operation between the two drugs (Figure 5.22F).

Another method of determining the likelihood of synergy is the Bliss Independence model which was recently recommended by the NIH in determining drug synergy, which calculates a drug combination effect using non-linear regression under the assumption that the drugs act independently and there is no effect from drug-drug interactions, but that each drug effect contributes to the final result (Zhao et al., 2014; Foucquier and Guedj, 2015). This differs from the Chou-Talalay CI method that takes into account drug-drug interactions and linearizes drug effects by logarithmic transformation (linear regression), which could exaggerate experimental error and reduce accuracy of determining drug effects (Zhao et al., 2004; Chou, 2006). The Bliss Independence method calculates the predicted response based on additive effect of both drugs, represented by a dotted curve and this analysis was conducted for the chemotherapeutic combinations as well as for the disulfiram-auranofin combination (Figure 5.22A-F). If the observed combination effect is below this curve, the

177 combination is declared synergistic, and if above, antagonistic (Zhao et al., 2014; Foucquier and Guedj, 2015). The Excess over Bliss (EOB) statistic was also calculated (Table 5.6). Values of 0 indicate additive effect, whereas positive or negative values demonstrate synergistic or antagonistic behaviour, respectively (Bansal et al., 2014; Goswami et al., 2015). Combination of disulfiram and auranofin gave a high positive value that was higher than the values achieved with any of the chemotherapeutic drug combinations. Based on this analysis, the Bliss results appears to more closely describe the dose response combination curves in which disulfiram appeared to strongly enhance the activity of auranofin in PER-485 cells (Table 5.6, Figure 5.22F), suggesting that the Chou-Talalay method may have been confounded by the transformation error due to its linear logarithmic transformation method.

178

Figure 5.22: Combination assays of disulfiram and other drugs in vitro and analysis using Bliss Independence methodology. Cells were subjected to increasing doses of disulfiram alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine, (E) topotecan or (F) auranofin alone or a combination of the two simultaneously at a fixed-ratio for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments. Synergy was observed between disulfiram and auranofin in PER-485 when analyzed according to the Bliss method. The dotted curve in each graph represents the predicted response based on additive effect of both drugs.

179

Table 5.6: Calculated combination index (CI) and excess over Bliss (EOB) values for disulfiram in combination with currently used drugs in the treatment of paediatric ALL and auranofin at ED75 in PER-485 cells. CI EOB

Mitoxantrone 0.97 0.120 Daunorubicin 1.23 0.081 Etoposide 1.17 0.096 Cytarabine 0.97 0.126 Topotecan 1.59 -0.004 Auranofin 1.12 0.289

Synergy occurs when CI<0.9 or positive EOB values, additive effect 0.91.1 or negative EOB values. In the EOB estimation, values of 0 indicate additive effect, and positive and negative values indicate synergy and antagonistic effects, respectively. ED75: effective dose causing 75% reduction of cell viability.

180

5.3 Discussion

In this chapter, the activities of two potent agents identified from the high-throughput screening of libraries containing approved drugs and pharmacologically active compounds were further characterized in high-risk ALL. The candidates were auranofin, an FDA-approved drug previously used for the treatment of rheumatoid arthritis, and disulfiram, an FDA-approved drug used for the treatment of alcoholism, both of which have been previously reported to induce ROS in vitro in cancer cells including haematological cancers. ROS, include oxygen radicals such as superoxide anion and hydroxyl radicals, as well as non-radical oxidizing agents such as hydrogen peroxide that could be easily converted into radicals (Jang and Surh, 2003; Bayr, 2005), In healthy cells, ROS are essential in maintaining normal cellular physiological functions as they act as signalling molecules to control the growth, proliferation, and differentiation of cells (Chen et al., 2017). In mammalian cells, the species bind non- covalently to specific receptors or affect signal transduction molecules involved in promoting cell growth. A variety of enzymatic and non-enzymatic cellular metabolism processes can generate ROS, especially mitochondrial-catalyzed electron transport reactions, as well peroxisome activity and activities of several types of leukocytes during inflammation (Bayr, 2005; Udensi and Tchounwou, 2014). On the other hand, an excess of ROS can induce DNA damage and oxidative stress that could disrupt cellular function and cause cell death, indicating the critical need to maintain ROS homeostasis.

Cancer cells have been shown to have higher levels of ROS production compared to normal cells due to aberrant processes induced by abnormalities such as mutations (Szatrowski and Nathan, 1991). ROS are involved in multiple roles in cancer cells including survival, proliferation and metastasis (Wang and Yi, 2008). Increased ROS levels has been shown to drive the migratory and invasive activity of metastatic bladder tumour cells by activating a key signalling protein for metastasis, p130Cas (Hempel et al., 2013). ROS were also demonstrated to induce hypermethylation of the E-cadherin promoter that encodes a cell adhesion molecule, down-regulating the expression of E- cadherin (Lim et al., 2008). Decreased level of E-cadherin is correlated with poor prognosis in hepatocellular carcinoma patients (Endo et al., 2000).

181

In this study, a particular sensitivity of MLL-r and other high-risk leukaemias to two ROS-inducing agents, auranofin and disulfiram, was observed. ROS over-production due to redox dysregulation has been reported in several types of leukaemia, affecting proliferation, survival and differentiation of leukaemia cells (Lewandowski et al., 2010). A study performed in 80 ALL patients found higher levels of protein carbonylation (oxidation by ROS) and enzyme superoxide dismutase (SOD) activities, in addition to decreased antioxidants compared to healthy controls, supporting the hypothesis that there is a persistence of oxidative stress in ALL (Battisti et al., 2008). In a cohort of 49 AML patients, extracellular ROS production was demonstrated to be highly elevated in the primary AML blast samples with approximately 10-fold higher compared to normal human CD34+ hematopoietic cells (Hole et al., 2013). The high ROS generated was associated with reduced antioxidant expression and strongly promoted the proliferation of AML cells. The blasts were also found to suppress the stress signalling that would normally limit proliferative response, giving an advantage to the leukaemic clone (Hole et al., 2013). Higher levels of oxidative stress-related parameters such as GSH and SOD levels were also evident in AML patients who went on to relapse compared to those who did not, indicating the role played by ROS in the advancement of the disease (Zhou et al., 2010). Increased ROS levels have also been reported in CLL (Carew et al., 2004; Trachootham et al., 2008), CML (Singh et al., 2009; Singh et al., 2012) and APL (Dong et al., 2004) either in vitro or in clinical samples.

Intrinsically elevated levels of ROS in leukaemia cells mean they are more vulnerable to further stress, and several chemotherapeutic agents have been used to exploit this condition by tipping the intracellular ROS balance, causing cellular and DNA damage and subsequently cell death. Some of the ROS inducers include anthracyclines such as daunorubicin and its hydroxylated form, doxorubicin, which are among the standard induction therapy drugs for paediatric leukaemia. Besides their established mechanisms of action of intercalating into DNA, and preventing DNA replication and transcription by inhibiting topoisomerase II (Forrest et al., 2012), anthracyclines are also known to affect the redox pathway through free radical formation and lipid peroxidationthat could induce DNA damage (Gewirtz, 1999). ROS levels in leukaemia cells have also been exploited to overcome drug resistance. Triptolide, a compound used in traditional

182

Chinese medicine, had been shown to reverse cytarabine and doxorubicin resistance in an ALL cell line and primary ALL patient samples through the production of ROS, increasing ROS-induced mitochondrial injuries and DNA damage in vitro and significantly prolonging survival of mice engrafted with the resistant ALL cells (Zhao et al., 2016), indicating potential clinical benefits of ROS-inducing agents for the treatment of chemoresistant leukaemia. Additionally, basal ROS levels have been reported to affect leukaemia cell sensitivity towards chemotherapeutics (Yi et al., 2002; Zhou et al., 2003). Together these studies illustrate the influence of endogenous ROS production in responsiveness towards drugs and the potential of utilizing this condition in advancing the treatment of leukaemia patients.

In the current study, auranofin was identified as a candidate drug for MLL-r and other high-risk leukaemia. In recent years, anti-cancer activities of this drug were reported in several types of malignancies including leukaemias such as lymphocytic leukaemia (Simon et al., 1981), chronic myelogenous leukaemia (Chen et al., 2014) and CLL (Fiskus et al., 2014). Here, auranofin showed preferential anti-leukaemic activity when it was tested against a panel of 16 leukaemia cell lines with IC50 ranging from 78 to 889 nM for 15 out of the 16 lines. In contrast, IC50 values for solid tumour and normal cell lines fell within a higher range of 1.4 – 5 µM, with only the exception of HEY cells (0.548 μM). Thus, leukaemia cells were selectively killed compared to solid tumours (p<0.0001) and normal cell lines (p=0.0001) and this occurred through ROS-induced apoptosis. In most sensitive cell lines, PER-485, RS4;11 and REH, ROS-induced expression of Nrf2, HMOX1 and DNA damage biomarker, γH2AX were evident after 6 hours of treatment, and attenuated by ROS scavenger, NAC. Pre-treatment with NAC also prevented increase in ROS levels and rescued cells from apoptosis, indicating a ROS-mediated mechanism of killing. Although not tested in this study, auranofin has been shown to inhibit another anti-oxidant TrxR, in CLL cells (Fiskus et al., 2014) and ovarian cancer cell lines, expression of which was inversely related to auranofin sensitivity (Marzano et al., 2007). In a multidrug-resistant CML cell line, K562/ADM, auranofin was shown to affect the redox balance by inhibiting TrxR activity, resulting in decreased cell viability and induction of apoptosis. This demonstrated the potency of auranofin in sensitizing cells with acquired resistance, potentially becoming a candidate to overcome drug resistance in CML patients (Liu et al., 2011). In another study,

183 auranofin also exhibited a ROS-mediated mechanism in depleting a cancer side- population (SP), a subpopulation of stem-like cancer cells of human lung cancer cell line, A549, which could be partially neutralized by pre-incubation with NAC (Hou et al., 2018). These studies indicate anti-cancer effects of auranofin through the ROS pathway. In the ex vivo testing, the compound also demonstrated high potency in MLL- r, Ph+, Ph–like and B-ALL PDXs with IC50 values between 55 and 389 nM. T-ALL xenografts showed higher IC50 of 744 nM and 1597 nM, possibly indicating lineage- specific attributes that result in the cells relying on the ROS pathway to a lesser extent than the B-ALL xenografts. Further investigations involving ROS scavenger NAC should be performed across the PDX panel to determine if the sensitivity towards auranofin is related to the extent of dependency to the ROS pathway. Biomarkers of the oxidative stress pathway used in this study, Nrf2, HMOX1 and γH2AX, should also be probed across the PDX panel to confirm the results.

Auranofin was also tested in vivo, however an adverse effect on the liver was observed in the murine model that prevented the use of doses above 2.5 mg/kg. Some of the previously reported common adverse effects associated with auranofin are gastrointestinal discomforts including loose stool, which occurs in 40% of patients, and watery diarrhoea, developing in up to 5% of patients. Other effects including skin irritations occur in approximately 20% of patients within the first year of treatment, and up to 12% of users develop mouth ulcerations and stomatitis (Furst, 1983; Kean et al., 1997; Kean et al., 2008). There are reports showing rare hepatic toxicity in 4% patients receiving auranofin as a treatment of rheumatoid arthritis, indicated through abnormal liver enzymes, however they are not clinically apparent liver injuries (Weisman and Hannifin, 1979; Kean et al., 1997). The adverse effect in this study could be an observation that is limited to the NOD/SCID mouse strain as there are published studies where doses 3 – 10 mg/kg of the drug were used in nude BALB/c and DBA2 strains, as well as TCL-1 transgenic mice (Simon et al., 1981; Huang et al., 2015; Fiskus et al., 2014). Nevertheless, the dose of 2.5 mg/kg used in this study is equivalent to a human dose of 12.2 mg, which is similar to that administered to patients in the CLL clinical trial without any adverse effects (12 mg daily) (Saba et al., 2013).

184

Auranofin was shown to potentiate the effect of cytarabine in vitro, potentially by increasing DNA damage. When this combination was tested in vivo, a significant delay of leukaemia growth was observed, despite being only three days. Whilst MLL-5 was selected for this study due to apparent response in vitro as measured by PARP cleavage, historically this PDX is one of the most aggressive in the panel (Richmond et al., 2015; Khaw et al., 2016; and unpublished communication) and therefore testing in additional MLL-r PDX is warranted. This discrepancy also highlights the difficulties often encountered when using in vitro models to predict in vivo responses and raises the issue of the need for more representative in vitro models. As discussed previously (Section 4.3), co-culturing leukaemia cells with bone marrow MSC can support cell proliferation (Yang et al., 2013). Utilization of this co-culture model was able to predict in vivo sensitivity of a BH3 mimetic, ABT-263, in paediatric leukaemia xenografts. First, the level of leukaemia cell viability in this model was defined using a training subset of 17 xenografts to categorize responders and non-responders. The established predictive- model was then used to blindly re-test the xenografts which demonstrated high sensitivity (50%) and specificity (100%) (Suryani et al., 2014). Going forward, a similar co-culture assay could be used to prioritize xenografts for animal testing and predict the efficacy of auranofin in vivo. Besides poor correlation between in vitro and in vivo response, the lack of response of ROS biomarker, Nrf2 and DNA damage biomarker, γH2AX, could possibly be due to pharmacodynamic study sampling timepoint. Nrf2 protein expression has been reported to increase upon exposure to ROS (Fiskus et al., 2014), as well as decreased after a long period of high ROS levels in cells (Zha et al., 2014). Therefore, a suitable sampling timepoint is critical in determining the role of ROS as the mechanism of action. Further investigation on the in vivo sampling timepoint could be performed in MLL-5 and other PDXs to properly evaluate the role of ROS caused by the auranofin and cytarabine drug combination. As for γH2AX, the protein expression did not increase in mice treated with the drug combination. This again could be due to the sampling timepoint as in the pharmacodynamics study, the mice were only treated with two doses of drugs. While the combination was partially effective in in vivo study after 3 weeks of drug treatment. Similar to Nrf2, optimization of pharmacokinetic study timepoint could address this issue of γH2AX response and verify the capacity of ROS as the mechanism of toxicity of the drug combination. Nevertheless, the combination of auranofin with cytarabine, a currently used drug for

185 the treatment of high-risk paediatric ALL, should be confirmed in other PDXs as their synergy could be potentially clinically relevant.

Disulfiram was also identified as a potential drug candidate for high-risk leukaemia. Although its activity was recently reported in adult B-ALL, this is the first time its anti- cancer properties were demonstrated in a range of MLL-r and other high-risk paediatric leukaemia PDXs. Testing of disulfiram efficacy against a panel of cell lines demonstrated IC50 values for leukaemia cells from 45 to 81 nM, solid tumours between 49 to 245 nM, and above 1 µM for normal cell lines. Whilst disulfiram was not especially selective towards MLL-r cells, it did exhibit significant selectivity towards leukaemia cells compared to solid tumours and normal cell lines (p<0.0001). The increased cytotoxic effect observed when disulfiram was coupled with copper suggests the combination eradicates cells through ROS production. Copper has long been known to catalyse oxidation-reduction reactions, leading to ROS generation and apoptosis in cells (Osredkar and Sustar, 2011; Kaplan and Mayron, 2016; Corce et al., 2016; Kalinowski et al., 2016; Mustafa and AlSharif, 2018). Due to the critical role of copper as a cofactor for key cellular enzymes, its transport is strictly controlled. Plasma proteins such as albumin deliver copper to its transmembrane transporter, CTR1 (SLC31A1) and once inside the cell, it is bound and trafficked by cytosolic metallochaperones such as ATOX1 to specific destinations (Baffoe et al., 2015; Denoyer et al., 2015). The ability of disulfiram to chelate copper and form disulfiram/Cu complex improves the transport of copper into cancer cells (Baffoe et al., 2015), resulting in oxidative stress. In a comprehensive review of the literature by Gupte and Mumper (2008), cancer patients were found to have significantly higher copper levels compared to non-cancer patients. These include solid tumours such as breast, cervical, ovarian, lung, colorectal and prostate cancers, as well as haematological cancers such as Non-Hodgkin’s lymphoma and several types of leukaemia. In leukaemia, patients have been shown to have increased levels of copper in serum and in leukaemic lymphocytes indicating a relationship between concentrations of this metal and immune response (Carpentieri et al., 1986; Dayer et al., 2015) and response to treatment (Hrgovcic et al., 1968), with patients with relapsed or progressive disease having elevated copper level compared to patients in remission or with stable disease (Kaiafa et al., 2012). With leukaemia patients having elevated levels of serum copper,

186 treatment with disulfiram could be very effective in targeting malignant cells. Besides disulfiram, other drugs such as clioquinol (CQ), an FDA-approved topical antifungal agent, repurposed in combination with copper has been found to increase cell death by inhibition of proteasome in prostate cancer cell lines in vitro compared to CQ alone. When tested in vivo without copper, the drug showed significant tumour suppression and strong inhibition of proteasomal chymotrypsin-like activity associated with CQ/Cu complex inhibition, supporting the hypothesis that the organic ligand CQ could interact with tumour cellular copper (Chen et al., 2007). Increased activity of CQ/Cu was also demonstrated in leukaemia cell lines in vitro, however the study also reported loss of leukaemia selectivity compared to normal haematopoietic cells hence cautioned potential clinical tests of co-treatment of CQ with copper (Mao et al., 2009). In the current study, the therapeutic window of disulfiram between PER-485 and non- malignant cell line, MRC-5 also shifted upon copper addition, however no apparent toxicity was seen in mice treated with disulfiram/Cu indicating the copper dose administered was tolerable.

In the current study, disulfiram was indeed found to act through ROS induction MLL-r and other high-risk leukaemia cells. The levels of Nrf2 and HMOX1 protein in disulfiram-treated cells confirmed induction of an oxidative stress response, which was further supported by ROS-mediated DNA damage as demonstrated by the expression of γH2AX and the ability of a ROS inhibitor, NAC to reverse these effects. In this study, ROS level increased significantly in PER-485, as detected using DCFDA which measures intracellular ROS species, and in RS4;11 and CEM, as detected using MitoSOX which specifically measures mitochondrial superoxides. The requirement to use these different methods in order to detect ROS in the different cell lines suggests the localization of ROS is largely restricted to either cytoplasm or mitochondria. In turn, this could also reflect the type of oxidative species generated and hence the activity of SOD in the mitochondria of each cell line. SOD catalyses the dismutation of superoxide into molecular oxygen and to a less reactive species, hydrogen peroxide, which could diffuse freely across membrane (Fisher, 2009; Udensi and Tchounwou, 2014).

Besides affecting the redox pathway, disulfiram has also been reported to have several other functions including its well-characterized activity as an aldehyde dehydrogenase

187

(ALDH) inhibitor for the treatment of alcoholism (Petersen, 1992; Robinson, 2015), and its ability to inhibit the activity of O6-methylguanine-DNA methyltransferase (MGMT) (Lun et al., 2016), an anti-mutagenic DNA repair protein that plays a crucial role in the defence against alkylating agents (Paranjpe et al., 2014). ALDH has been shown to be highly expressed in cancer stem cells (CSC) including leukaemia, and is used as a marker of stemness (Gerber et al., 2012; Kim et al., 2013b; Li et al., 2014; Hoang et al., 2015). There are extensive published data on heightened activity of ALDH in leukaemia stem cells (Gasparetto et al., 2012; Venton et al., 2016) and association between ALDH activity and clinical outcome (Cheung et al., 2007; Ran et al., 2009). Inhibition of ALDH could also contribute to higher oxidative stress as it prevents metabolism of endogenous and exogenous aldehydes generated during ROS-mediated lipid peroxidation (Singh et al., 2013). Disulfiram/Cu has been shown effective in inhibiting recurrence in ALDH-positive non-small cell lung cancer-stem-like cells (Liu et al., 2016), and glioblastoma multiforme (GBM) and GBM-stem-like cells (Liu et al., 2012) by inhibiting the ALDH pathway. Further investigations could determine if this is also the case for high-risk leukaemia by blotting for ALDHA1A, the main target of disulfiram (Moore et al., 1998; Koppaka et al., 2012) and MGMT in MLL-r cell lines and xenograft panel. Additionally, activity of ALDH1A1 and other family members could also be established to determine the ALDH status of these leukaemia cells compared to solid tumours.

When disulfiram was tested against MLL-r infant and other high-risk ALL PDXs in vitro, its activity was found to depend on the addition of copper. Possibly, this effect was due to low levels of available copper in these PDXs in vitro, preventing DDC- copper complex formation or the lack of copper transmembrane transporters such as CTR1/SLC31A1, CTR2/SLC31A2 and DMT1 (Denoyer et al., 2015). Using the Gene Expression Omnibus database previously mentioned, the PDXs displayed a moderate basal expression levels of these targets (data not shown). However, further investigation has to be performed with normal lymphocytes and other ALL PDXs that are sensitive to disulfiram alone in vitro, to determine if the copper transporters play any role in disulfiram sensitivity.

188

In vivo efficacy of disulfiram was determined in three MLL-r most sensitive PDX models using a modified experimental design with only two mice per treatment group. However, neither disulfiram nor disulfiram/Cu caused leukaemia growth delay. As previously discussed, an improved in vitro model that takes into account microenvironmental factors may better predict in vivo responses. Such a model could then be used to prioritize PDX for in vivo testing. Based on this study using the 2-mouse efficacy model, the design is of great value for screening new compounds as it only highlights highly efficacious in vivo activities (long leukaemia growth delay). This screening model could deliver high quality in vivo information with minimal number of mice and hence minimal cost. However, its actual strength and potential could only be determined after screening a large number of xenografts and examining the results with historical data from experiments performed in a standard in vivo experimental design as previously described by Murphy et al. (2016). The advantage of the 2-mouse model compared to the 1-mouse model recommended by Murphy et al. (2016), is having an additional mouse in each treatment group as a backup in the event of unfortunate circumstances of non-drug-related effect. During disulfiram in vivo testing, a mouse from each copper and disulfiram treatment group had to be culled due to spontaneous thymic lymphoma (thymoma), which was shown to be common in NOD/SCID mouse due to the absence of immune system (Prochazka et al., 1992). The extra mouse enabled experiment to continue and produce valid results.

Combination assays were performed with five chemotherapeutics currently used in paediatric ALL therapy but calculated CI values revealed lack of synergy between disulfiram and these agents. However, due to recent findings showing synergistic effects between auranofin and disulfiram (Huang et al., 2016; Papaioannou et al., 2014), the combination was investigated in MLL-r cell line, PER-485. Despite an apparent synergistic effect being indicated graphically, the CI calculated using the Chou-Talalay method indicated antagonism between the two drugs. As an alternative method for determining synergy, the Bliss Independence model was used to confirm the effect. This method uses non-linear regression modeling with the assumption of the absence of effect from drug-drug interaction, in contrast with the Chou-Talalay method which takes into account these interactions and uses linear regression modeling to calculate drug combination effect (Chou, 2004; Zhao et al., 2014; Foucquier and Guedj,

189

2015). Although the latter method is highly popular for evaluating combination therapy, it may sometimes be less suitable since logarithmic data transformation using the linear regression model distorts experimental error and reduces accuracy of drug effect compared to non-linear regression (Zhao et al., 2004). Based on the Bliss Independence model which was recently recommended by the NIH in determining drug synergy, disulfiram and auranofin were indeed found to synergize in the PER-485 MLL-r cell line as demonstrated in the dose response curve (Figure 5.22F), similar to previously published data of disulfiram alone in hepatoma cancer cell lines (Huang et al., 2016) and in the presence of copper in a breast cancer cell line (Papaioannou et al., 2014). Both studies reported proteotoxicity through the inhibition of proteasome-associated deubiquitinases resulting in accumulation of misfolded proteins and protein aggregates which induce heat shock protein activation, and increase of ROS levels by both drugs simultaneously limiting the ability of cancer cells to cope with the assaults. This combination therefore warrants further investigation MLL-r and other in high-risk ALL.

In summary, two potent FDA-approved drugs with ROS inducing properties, auranofin and disulfiram, were identified from the high-throughput screening of approved drugs and biologically active compounds. Both drugs showed selective in vitro potency towards high-risk leukaemia cells regardless of MLL status and induced generation of ROS in sensitive cells, leading to apoptosis. Auranofin displayed synergy with the paediatric ALL drug cytarabine, both in vitro and in the NOD/SCID ALL mouse model through ROS-mediated DNA damage. Disulfiram potency was shown to increase when co-treated with copper, which could be used as an advantage to target cancer cells, given their elevated copper levels. The drug unfortunately did not delay leukaemia growth in vivo, and was antagonistic with most chemotherapeutics tested. However, it showed strong synergy with auranofin, which could further assault leukaemia cells and suppress their development. The reasons underlying the particular selectivity of these drugs towards high-risk leukaemia and the mechanisms behind their synergistic activity are important questions for further research, investigation of which could ultimately result in repurposing auranofin and disulfiram for the treatment of high-risk leukaemia.

190

CHAPTER 6: CONCLUSIONS AND FUTURE DIRECTIONS

Acute lymphoblastic leukaemia (ALL) is the most common cancer in children accounting for a quarter of paediatric malignancies. Continuous advancements in risk- directed combination chemotherapy over the last five decades have improved the clinical outcome of this once fatal disease such that patient survival rates are close to 90% in recent years (Hunger et al., 2012). Despite this, patients with high-risk subtypes including MLL-rearranged leukaemia, Philadelphia-positive ALL, Philadelphia-like ALL and early T-cell precursor ALL still have a dismal prognosis. Despite extensive research being performed towards the development of new drugs and targeted therapies for these high-risk patient groups, the number of drug approvals remains disappointingly low with only 3 – 8% tested compounds entering clinical trials finally approved for clinical use in the oncology field due to failure in late stage, large scale clinical trials (Williams, 2011). The high failure rate, coupled with the escalating costs of research and development have forced a change in the drug discovery pipeline in the form of drug repurposing, to take advantage of the comprehensive amount of ADMET data already available on a large number of drugs, potentially bypassing multiple phases of testing.

The aim of this study was to identify novel compounds targeting high-risk leukaemia with a special focus on MLL-rearranged ALL, by performing a cell-based high- throughput screen on a library of approved drugs and pharmacologically active compounds. The work in this thesis has successfully revealed two lead compounds with selectivity towards subsets of MLL-rearranged and CALM-AF10 leukaemia cell lines; 2-chloroadenosine triphosphate, a purinergic receptor (P2Y) agonist and guanylate cyclase inhibitor, and SID7969543 which targets Steroidogenic Factor-1 (SF- 1/NR5A1). Since these compounds have known targets, their identification has revealed potentially interesting pathways for further study in the the context of MLL-r ALL.

Further characterization of the candidate MLL-selective compounds indicated the in vitro efficacy of 2-Cl-ATP against MLL-r ALL patient-derived xenografts. Curiously, sensitivity to this compound among the PDX was associated with decreased expression of several P2Y receptors including P2RY14, which was additionally found to be

191 expressed at low levels in MLL-r patients compared to patients without MLL gene rearrangement in a cohort of paediatric ALL patients. Despite the fact that this specific compound has not previously been tested in models for ALL, the purigenic signaling pathways in general have been investigated as therapeutic targets in several cancers. The targeting of these pathways has shown preclinical efficacy in promyelocytic leukaemia whereby an agonist caused apoptosis in leukaemic cells (Gessi et al., 2011), and in triple negative breast cancer, whereby mice treated with a combination of doxorubicin and an A2A receptor antagonist, SCH58261 significantly delayed tumour progression (Loi et al., 2013). The current findings, whereby low levels of the receptors were associated with sensitivity to 2-Cl-ATP are completely consistent with these observations and therefore the significance of these receptors as the relevant target of 2- Cl-ATP or as markers of sensitivity still needs further investigation. An alternate hypothesis is that cytotoxicity of 2-Cl-ATP is independent of the receptors but instead relies on metabolism of the compound into 2-chloroadenosine (2-CADO) by ATPase and subsequent intracellular phosphorylation of 2-CADO (D’Ambrosi et al., 2004; Bastin-Coyette et al., 2008). This could be further pursued in MLL-r leukaemia cell lines and PDXs by monitoring the effects of inhibitors of ATPase and adenosine kinase on ATPase mediated cytotoxicity. MLL-r patients in the Stam et al. (2010) cohort also demonstrated significantly higher basal expression compared to patients without MLL gene rearrangement, of guanylate cyclase (GC) GUCY1A3, which encodes an α subunit of GC, sGCα1, shown to be a therapeutic target for several cancers including breast cancer (Sotolongo et al., 2016) and head and neck squamous cell carcinoma (Tuttle et al., 2016). Even though a biomarker could not be clearly determined from current data, further investigations are definitely worthwhile for future studies. Such studies could involve direct genetic suppression of target receptors to assess their role in MLL ALL biology and survival, as well as the response to 2-Cl-ATP. Unfortunately, 2-Cl-ATP also similarly affected peripheral blood mononuclear cells from healthy donors in vitro, with similar range of IC50 to the MLL-r ALL PDX panel, questioning the existence of a therapeutic window. Nevertheless, further studies should still be performed to investigate the true potential of 2-Cl-ATP in MLL-r leukaemia. Even though the compound has its limitations for in vivo application, future studies in could potentially still be carried out after establishing possible in vivo toxicities associated with the compound. If unacceptable levels of adverse effects are observed, subsequent work

192 should be directed at examining its derivatives such as 2-CADO which has been tested in animal models (Mathot et al., 1996; Mares, 2010). Guanylate cyclase inhibitors such as ODQ and NS 2028 could also be tested against MLL-r cell lines and an expanded panel of MLL-r PDXs to determine if sensitivity towards 2-Cl-ATP is associated with the levels of guanylate cyclase. Both of these inhibitors have been tested in vivo (Tseng et al., 2011; Morbidelli et al., 2010), which therefore opens up the possibility for preclinical examinations if initial results were promising.

The second MLL-selective candidate SID7969543, showed activity against a subset of MLL-r and CALM-AF10 leukaemia cell lines and synergy with etoposide and cytarabine. However, this cytotoxic effect could not be validated in vivo because no activity was observed in the MLL-r ALL PDX panel in vitro. One explanation could be that the compound targets processes related to cell proliferation, whereas the ALL PDX do not proliferate in vitro. To investigate whether this is the case, future studies can be carried out with the compound in an MSC-PDX co-culture model, which has proven to support proliferation of leukaemia cells (Jones and Wagers, 2008). Considering the poor stability of SID with an estimated half-life of less than two minutes in microsomal stability assay, another commercially available SF-1 inhibitor (CAS: 868224-75-3) can be utilized to further investigate the potential of SF-1 inhibition in influencing MLL-r leukaemia cell survival. Another critical issue to be addressed is the levels of SF-1 expression in MLL-r leukaemia cells. Based on the Human Protein Atlas database, expression of SF-1 is recorded in only two leukaemia cell lines, HEP (erythroid leukemia, a rare form of AML (MacKinnon et al., 2013)) and K-562 (CML), and not in the cell lines also in our panel, THP-1, U937 and REH. It is therefore paramount that the basal levels of SF-1 expression and its levels after being subjected to SID and other SF-1 inhibitors be evaluated in a panel of MLL-r leukaemia cell lines and PDXs to determine if SF-1 transcriptional activity truly contributes to MLL-r ALL progression or whether the effects seen so far are due to off target effects of SID. In sum, both compounds identified through Strategy I revealed potential new targetable pathways in MLL-r leukaemia that could be further explored and possibly exploited therapeutically.

The studies undertaken for Chapter 5 revealed the potential of two ROS inducing compounds, auranofin and disulfiram as inhibitors of high-risk leukaemia cell lines and

193 xenografts in vitro. These two FDA-approved drugs were further advanced into MLL-r ALL patient-derived xenograft mouse models to determine in vivo efficacy in highly relevant preclinical models of this disease. Whilst strong ROS-mediated cell killing of high-risk leukaemia cells by auranofin was displayed in vitro, auranofin in combination with cytarabine, a drug currently used in paediatric ALL therapy, only demonstrated a modest leukaemia growth delay of three days in an aggressive MLL-r ALL PDX mouse model. Likewise, disulfiram showed a highly potent anti-leukaemia activity in vitro, but failed to prolong survival of mice treated with the drug alone or with the addition of copper, which previously demonstrated significant efficacy in in vivo studies including models of B-ALL (Deng et al., 2016) and breast cancer (Allensworth et al., 2015). As significant in vivo effects were lacking with either auranofin or disulfiram as single agents, the dose and treatment scheme of each drug could be further increased and optimized to enhance their effectiveness. Then, potential for in vivo efficacy should be established in an extended panel of MLL-r PDXs, as well as other high-risk leukaemia PDX models including Ph+ and Ph-like ALL to determine the true potential of auranofin and disulfiram as high-risk leukaemia therapies.

In vivo efficacy of the interesting combination of auranofin and disulfiram for which synergy was observed in vitro, should be further explored to determine if it offers increased efficacy. The auranofin/disulfiram combination could also be assessed in combination with the current treatment backbone of ALL or with other targeted therapies. Despite the clinical successes of several targeted therapies such as imatinib and venetoclax, unfortunately, many of the recently developed, molecular-guided targeted therapies are found not to be as potent as postulated based on preclinical data. It is thought that they often target molecules or signaling pathways that are not crucial drivers to leukaemogenesis, and therefore produce transient effects (Pui, 2010). Emergence of therapy resistance has also been a challenge in sustaining drug efficacy (Dharmaraja, 2014). Limited responses and disease progression were recently reported in the adult clinical trial of DOT1L inhibitor, EPZ5676 (Pinometostat) (Stein and Tallman, 2015). However, combination of ROS induction through auranofin or disulfiram along with MLL-specific therapy, simultaneously targeting MLL-r leukaemogenesis and tipping the balance of the high ROS levels in the leukaemia, could potentially increase treatment efficacy, and result in better clinical outcome. Compared

194 to other ROS inducers such as fenretinide (Ruvolo et al., 2010; Zhang et al., 2013), menadione (Baran et al., 2014; Park et al., 2017) and quercetin (Maso et al., 2014; Calgarotto et al., 2018) that are currently under investigations for several subtypes of leukaemia, agents such as auranofin and disulfiram could potentially advance more quickly into the clinics due to their known toxicity profiles as both have been used by patients for several decades. Further work in optimizing and enhancing efficacy, as well as in identifying biomarker could potentially put forward these drugs as less toxic alternatives to existing drugs associated with ROS-mediated cell killing such as doxorubicin and mitoxantrone (Dharmaraja, 2014), which have been reported to cause long-term cardiotoxicity as previously discussed (section 1.3.3). In this thesis, a robust biomarker for in vivo studies was unable to be determined due to a seemingly poor correlation between in vitro and in vivo response. The protein expression of Nrf2 and γH2AX did not reflect the leukaemia growth delay observed in vivo. As discussed in Chapter 5, this could be due to the short-term drug treatment for the pharmacodynamics studies, and a longer treatment protocol might resolve this issue. Ultimately, if confirmed as the mechanism, future clinical trials of auranofin and disulfiram could also incorporate weekly measurement of ROS levels in patients, as a biomarker to predict patient response, through peripheral blood samples using DCFDA or a similar fluorescent probe, dihydroethidium (DHE) which has been previously used for ROS monitoring in CLL patients treated with auranofin (Saba et al., 2013).

A major challenge faced in this and other studies aimed at identifying novel active drug compounds starting with in vitro cell-based screenings, is the absence of a straightforward in vitro test to predict in vivo sensitivity and often the absence of a clear correlation between in vitro and in vivo drug sensitivity. Factors that could cause this include the lack of microenvironment characteristics that are present in vivo such as interaction with mesenchymal cells that influence drug response (Mudry et al., 2000; Aljitawi et al., 2014) and the absence of a competent immune system in the currently used animal models which does not mimic the human biological systems (Walsh et al., 2017).

In this study, both auranofin and disulfiram demonstrated potent activity against all MLL-r ALL PDXs in vitro, however when tested in vivo, significant leukaemia growth

195 delays were not achieved in those PDXs. Possible explanations for this discrepancy are the fact that the in vitro culture model used does not support proliferation of xenograft cells and lacks tumour microenvironment components such as cytokines and other soluble factors, as well as surrounding mesenchymal cells in vivo. The lack of direct correlations between in vitro and in vivo sensitivity has stimulated research into the development of in vitro models to better mimic the in vivo setting. The development of co-culture systems has been shown to protect leukaemia cells from cytotoxic drugs (Mudry et al, 2000). A study by Sison et al. showed that bone marrow stromal cells can protect MLL-r infant ALL primary cells from spontaneous apoptosis and lestaurtinib- mediated cytotoxicity, and inhibition of bone marrow microenvironment C-X-C chemokine receptor type 4 (CXCR-4) signaling, enhanced the efficacy of lestaurtinib in vivo (Sison et al., 2013). Expression levels of CXCR-4 and other genes involved in tumour microenviroment such as bone morphogenetic protein 2 (BMP-2) were reported to be elevated in precursor B-ALL cells compared to non-malignant CD34+ specimens. This expression signature was also shown to be similar in three patient cohorts (Tesfai et al., 2012), indicating the importance of cellular interactions within the bone marrow microenvironment in conferring drug resistance and suggesting that the suppression of these pathways would mobilize the leukaemic cells, resensitizing them to chemotherapeutic agents.

Recently, 3-dimensional (3D) stromal-based models for leukaemia were established to further mimic the 3D microenvironment by seeding human bone marrow mesenchymal stem cells (MSC) into synthetic scaffolds, followed by leukaemia cells of interest. These models not only provide the 3D mechanical structures that enable further cell- extracellular matrix interactions, but also allow leukaemia cells to establish niches within the MSC scaffolds, further supporting their survival and proliferation (Aljitawi et al., 2014). A comparison of AML cell line, HL-60 cultured in suspension, in 2D MSC monolayer format and in 3D scaffold format showed decreased sensitivity towards doxorubicin or cytarabine in cells cultured in the 3D system. Higher proliferation was also observed in the cells in the 3D system even when exposed to chemotherapy, indicating the protective niches provided by the MSCs. In the clinical setting, these cells could represent highly resistant residual cells that may ultimately result in disease relapse (Aljitawi et al., 2014). In another 3D model, 3-dimensional microfluidic tri-

196 culture system, bone marrow stem cells and human osteoblasts are embedded in collagen I, and injected into microchannels. Once the collagen I has been allowed to gellify, leukaemia cells and culture medium are pumped through to mimic the in vivo bone marrow microenvironment (Bruce et al., 2015). Leukaemia cells cultured on this dynamic platform showed reduced sensitivity towards cytarabine, thus enhanced cell viability compared to 2D and 3D static models, indicating more microenvironment protection conferred to the leukaemic cells (Bruce et al., 2015). These studies further demonstrate the importance of developing in vitro drug testing models that incorporate the tumour microenvironment aspect to mimic the in vivo conditions and responsiveness for future investigations of drug efficacy, which might allow better translatability to the in vivo setting.

In the cancer drug discovery pipeline, murine models have been a powerful preclinical tool not only in testing therapeutic efficacy, but also mechanism discovery and understanding origin and evolution of a disease. However, even with ample in vivo data, still the effects seen in patients can be insufficient. A reason for this might be the lack of immune system in the mice which result in animal models that do not fully recapitulate human biological systems (Walsh et al., 2017). Over the past 20 years, mouse models of human leukaemia have constantly evolved to generate an ideal model that represents and mimics the human disease. Genetically modified mouse models (GEMMs) is one class of models developed to better understand cancer by introducing mouse oncogenes such as the BCR-ABL1 for Ph+ ALL or chromosomal translocations such as TEL-AML1 (ETV-RUNX1) that is present in almost a quarter of childhood ALL patients (Mullighan, 2012b). For MLL-r leukaemia, transgenic MLL-AF9, MLL-ENL, MLL-AF10 and MLL-ELL mice develop MLL-r AML (Cook and Pardee, 2013; Jacoby et al., 2014). One translocation highly associated with MLL-r ALL, MLL-AF10, has been modeled with a knock-in mouse using gene targeting to fuse human AF4 to exon 7 of mouse MLL, developing lymphoid and myeloid hyperplasia and hematologic malignancies, frequently B-cell lymphomas, after prolonged latency (Chen et al., 2006b; Jacoby et al., 2014). Another model uses a recombination system and invertor technology with an inverted AF4 allele targeted to an intron in the MLL gene, generating MLL-AF4 oncogene in mice which develop B-cell neoplasias of mature phenotype (Hauer et al., 2014; Jacoby et al., 2014). These transgenic mouse models are unable to mimic MLL-r

197

ALL and thus are not ideal in studying the highly heterogeneous disease. Therefore patient-derived xenograft models using immunocompromised mice are highly utilized for MLL-r ALL testing.

In this study, the immune-deficient NOS/SCID mouse model was used for testing in vivo efficacy of auranofin and disulfiram. This xenograft model gives systemic disease that mimics the course of disease in patients and has been shown to respond to drugs in a manner that significantly correlated to clinical outcome (Lock et al., 2002; Liem et al., 2004). The mouse strain was developed from the backcrossing of the CB17-scid strain, the first severe immunodeficient mice mouse strain with mutation in Prkdcscid (protein kinase DNA activated also known as the Scid mutation) (Belizario, 2009), with the non- obese diabetic (NOD/Lt) strain. The resulting NOD/SCID strain lacks immune cells such as T-, B- and natural killer cells, with minimal cytokine production and stromal support, as well as having reduced macrophages and dendritic cell functions (Ito et al., 2012). The reduced level and activity of innate immunity increased the engraftment rate of human leukaemia cells, either patient-derived or cell lines, which advanced investigations into tumour biology, proof-of concept from in vitro studies and efficacy of therapeutics on human tumour cells (Kohnken et al., 2017). However the lack of immune system in these xenograft models might also affect the response to drug treatments due to the loss of interactions of immune cells with chemotherapeutics and other compounds which in turn can affect the anti-tumour immune response.

Recently developed immunocompetent humanized mouse models such as the Hu-PBL- SCID (human peripheral blood lymphocytes (PBLs)) and the BLT (bone marrow-liver- thymus) models, may address the issue of immune system in previous xenograft mouse models. The Hu-PBL-SCID mouse model is a model which human PBLs are injected intraperitoneally or intravenously into immunodeficient mice to engraft human immune system by primarily engrafting T-cells, while the BLT mouse model is a model whereby mice are firstly transplanted with fragments of human fetal liver and thymus, followed by intravenous injection of human CD34+ haematopoietic stem cells obtained from the fetal liver to establish a functional human immune system, before engrafting the mice with tumour cells (Brehm et al., 2014). These models, however, have their own limitations. The mice of the Hu-PBL-SCID model develop lethal graft-versus-host-

198 disease (GVHD) within 4 – 8 weeks after engraftment, limiting the experimental window (Brehm, 2014; Walsh et al., 2017), while in the BLT model, the cytokines produced are mostly by the mouse stromal cells, which direct the subsequent development of the immune cells. Due to the inefficiency of cross-reaction of mouse cytokines with the human receptors, the resulting microenvironment does not fully support human immune cells such as myeloid and natural killer cells (Spits, 2014). Two mouse strains, MITRG and MISTRG were developed to overcome this issue by knocking in four human genes encoding cytokines for innate human immune cell development, replacing their mouse counterparts. The absence of mouse cytokine genes reconstitutes a patient-derived immune system, developing functional myeloid cells including monocytes and natural killer cells in the peripheral blood and lymphoid tissues (Rongvaux et al., 2014). Drawbacks still exist with this model such as suboptimal development of mouse and human red blood cells which can cause anemia and limit the use of mice for long-term experiments (Rongvaux et al., 2014; Spits, 2014). Nevertheless, these strains facilitate the interactions between immune and cancer cells, and integrate the tumour microenvironment factor that was lacking in previous models. Future development of immunocompetent mouse models from patient HSCs or PBLs that could better recapitulate heterogeneous diseases such as MLL-r leukaemia would advance drug testing and provide superior insight into leukaemogenesis.

In summary, the data presented in this study provide insight into the drug repurposing process and highlight the potential of this approach in identifying novel therapeutics for high-risk leukaemia. The candidate MLL-selective compounds identified revealed potential novel pathways that could be therapeutically targeted in subgroups of MLL-r leukaemia. The two FDA-approved drugs identified, which showed strong ROS- mediated anti-leukaemic activities, provide further support for the therapeutic potential of ROS-targeting therapeutics in high-risk leukaemia. With well-known toxicity profiles, both drugs could be put forward into the clinic as non-toxic alternatives of ROS-inducing compounds instead of further investigating potentially more toxic ROS inducers. The work outlined within this thesis has formed the basis of potential testing of repurposed drugs in combination with current chemotherapy regimens, which may lead to improved treatment of this childhood malignancy.

199

APPENDIX

Supplementary Figure 1: Absorbance of CEM and PER-485 cell lines treated with either positive or negative control in the primary screen. Absorbance (relative fluoresnce unit, RFU) of CEM and PER-485 cells when treated with positive control (10 µM cytarabine) or negative control (5 µM DMSO) in the primary drug screen

200

Supplementary Figure 2: Primary and secondary high-throughput screenings. Overall summary of high-throughput screenings performed.

201

Supplementary Figure 3: Cytotoxicity of seven shortlisted compounds across a cell line panel. Dose response curves of each of the seven compounds short-listed from the secondary screen (Strategy I) across a range of MLL-rearranged and MLL-wild-type leukaemia cell lines and a non-malignant cell line. Two compounds, 2-chloroadenosine triphosphate and SID7969543 showed a trend of selectivity towards MLL-rearranged cells. The results are expressed as the mean ± SE of three independent experiments.

202

Supplementary Figure 4: Combination assays of 2-chloroadenosine triphosphate with conventional agents in PER-485. Cells were subjected to increasing doses of 2-chloroadenosine triphosphate (2-Cl-ATP) alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine or (E) topotecan alone, or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

203

Supplementary Figure 5: Combination assays of 2-chloroadenosine triphosphate with conventional agents in MOLM-13. Cells were subjected to increasing doses of 2-chloroadenosine triphosphate (2-Cl-ATP) alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine or (E) topotecan alone, or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

204

A 2 -C A D O (c e ll lin e s )

1 2 5

1 0 0

P E R -4 8 5 (IC 5 0 = 2 .5  M )

) %

( R S 4 ;1 1 (7 .1  M )

7 5

y t

i T H P -1 (1 8 .2  M )

l i

b 5 0

a C E M (6 .2  M )

i V 2 5 R E H (> 2 0  M )

0 -6 -5 -4 2 -C A D O c o n c e n tr a tio n [L o g M ]

B 2 -C A D O (P B M C )

l o

r 1 2 5

o c

D o n o r 1 (IC = 4 .5  M )

t 1 0 0 5 0

s n

i D o n o r 2 (3 .6  M )

a 7 5

g D o n o r 3 (3 .6  M )

l a

v 5 0

r u

s 2 5

e c

% -6 -5 -4 2 -C A D O c o n c e n tr a tio n [L o g M ]

Supplementary Figure 6: Cytotoxicity of 2-chloroadenosine. Dose response curves of 2-chloroadenosine (2-CADO) in (A) leukaemia cell lines and (B) normal peripheral blood mononuclear cells (PBMC) in 72-hour resazurin-based viability assay. The results are expressed as the mean ± SE of three independent experiments and IC50 values were derived from the mean of these experiments.

205

Supplementary Figure 7: Combination assays of SID7969543 with conventional agents in PER-485. Cells were subjected to increasing doses of SID7969543 (SID) alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine or (E) topotecan alone, or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

206

Supplementary Figure 8: Combination assays of SID7969543 with conventional agents in MOLM-13. Cells were subjected to increasing doses of SID7969543 (SID) alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine or (E) topotecan alone, or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

207

A D is u lfira m - M L L -r P D X

1 2 5 M L L - 2

1 0 0 M L L - 5

) M L L - 6

% (

7 5 M L L - 7

y t

i M L L - 8

l i

b 5 0 M L L - 1 4

i V 2 5

0 -8 .0 -7 .5 -7 .0 -6 .5 -6 .0 D is u lfira m c o n c e n tra tio n [L o g M ]

B D is u lfira m - M L L -w t h ig h -ris k le u k a e m ia P D X

1 5 0 A L L - 2 1 2 5 A L L - 4

) A L L - 7

1 0 0

% (

A L L - 8

y t

i 7 5

l A L L - 1 9

i b

a A L L - 3 1

i 5 0 V A L L - 5 5 2 5 A L L - 5 6

0 T G T - 5 2 -8 .0 -7 .5 -7 .0 -6 .5 -6 .0 D is u lfira m c o n c e n tra tio n [L o g M ]

Supplementary Figure 9: Disulfiram as single agent does not affect viability of high-risk leukaemia PDXs in vitro. (A) Dose response curves of MLL-rearranged (MLL-r) ALL PDXs and (B) MLL-wild- type (MLL-wt) high-risk ALL PDX after 48 hours of treatment as measured by resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

208

Supplementary Figure 10: Weight of mice during the auranofin MTD study. Mice weight were monitored over the 3-week treatment period with 4 doses of auranofin and 3 weeks after treatment ended.

209

Supplementary Figure 11: Combination assays of auranofin with conventional agents in PER-485. Cells were subjected to increasing doses of auranofin alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine or (E) topotecan alone, or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

210

Supplementary Figure 12: Combination assays of auranofin with conventional agents in PER-490. Cells were subjected to increasing doses of auranofin alone, (A) mitoxantrone, (B) daunorubicin, (C) etoposide, (D) cytarabine (E) topotecan alone or a combination of the two simultaneously at fixed-ratios for 72 hours, and viability was assessed using resazurin-based cytotoxicity assay. The results are expressed as the mean ± SE of three independent experiments.

211

Supplementary Figure 13: Weight of mice during the disulfiram MTD study. Mice weight were monitored over the 4-week treatment period with 4 doses of disulfiram with the presence and absence of copper and 3 weeks after treatment ended.

212

Supplementary Table 1: IC50 of 184 potent compounds in a secondary screen against CEM and PER-485 cell lines.

CEM PER-485 Compound Name IC50 (µM) IC50 (µM) (S)-(+)-Camptothecin 0.70 0.65 2-methoxyestradiol 0.28 1.68 2-Methoxyestradiol 0.62 1.36 2-Methoxyestradiol 0.49 2.02 7-Chloro-4-hydroxy-2-phenyl-1,8-naphthyridine 1.24 >2 A23187, free acid 0.35 0.52 A-77636 hydrochloride 0.82 >2 Actinomycin D <0.25 0.61 Adriamycin <0.25 0.32 Albendazole <0.25 0.66 Alexidine dihydrochloride 1.53 >2 AM 92016 hydrochloride >2 >2 Ammonium pyrrolidinedithiocarbamate 0.32 0.46 Amsacrine hydrochloride <0.25 0.38 Anisomycin <0.25 <0.25 ARP 101 >2 >2 Auranofin <0.25 <0.25 Auranofin 0.21 <0.25 Bax channel blocker >2 >2 Benzethonium chloride >2 >2 beta-Lapachone >2 >2 BIX 01294 trihydrochloride hydrate 1.77 >2 BNTX maleate 1.34 >2 BNTX maleate salt hydrate 0.52 >2 Bortezomib <0.25 <0.25 Brefeldin A <0.25 <0.25 Brefeldin A from Penicillium brefeldianum <0.25 <0.25 BVT 948 0.93 >2 Calcimycin 0.31 0.42 Calmidazolium chloride >2 >2 Calmidazolium chloride 1.05 1.70 Camptothecin <0.25 <0.25 Camptothecine (S,+) <0.25 <0.25 Cantharidic Acid >2 >2 Cantharidin >2 >2 Cantharidin >2 1.72 CCMQ >2 >2 CD 437 <0.25 0.34

213

CGP-74514A hydrochloride 1.15 >2 Chelerythrine chloride >2 >2 Chlorhexidine 1.84 >2 Chloroxine >2 >2 CHM-1 hydrate <0.25 <0.25 Chrysene-1,4-quinone 1.90 1.53 Ciclopirox ethanolamine 1.72 >2 Cladribine <0.25 1.20 Cladribine <0.25 1.11 Clofarabine <0.25 <0.25 Clofarabine <0.25 <0.25 Colchicine <0.25 <0.25 Colchicine <0.25 <0.25 Colchicine <0.25 <0.25 CPT 11 >2 >2 Cycloheximide <0.25 0.29 Cycloheximide <0.25 0.26 Cyclosporin A 0.31 1.08 Cytarabine <0.25 >2 Cytarabine(Cytosar-U) <0.25 >2 Cytosine-1-beta-D-arabinofuranoside hydrochloride <0.25 >2 D-64131 <0.25 <0.25 Daunorubicin hydrochloride <0.25 <0.25 Daunorubicin hydrochloride <0.25 <0.25 Dequalinium chloride hydrate 0.55 >2 Dequalinium dichloride 1.60 >2 Digitoxigenin <0.25 0.55 Digoxigenin <0.25 0.78 Digoxin <0.25 <0.25 Dihydroouabain 0.54 1.30 Diphenyleneiodonium chloride <0.25 1.97 Diphenyleneiodonium chloride <0.25 1.54 Disulfiram <0.25 <0.25 Docetaxel <0.25 <0.25 Docetaxel <0.25 <0.25 Doxorubicin hydrochloride <0.25 0.34 Doxorubicin hydrochloride <0.25 0.35 Ellipticine 0.36 >2 Emetine dihydrochloride hydrate <0.25 <0.25 ER 27319 maleate 0.86 >2 Etoposide <0.25 1.01 Etoposide(Etopophos) <0.25 0.63 FCCP 1.19 >2 Fenbendazole 0.66 >2

214

Flubendazol 1.54 >2 Gemcitabine <0.25 0.86 Gemcitabine hydrochloride <0.25 1.12 Gemcitabine Hydrochloride <0.25 1.11 Genistein >2 >2 GW 843682X 1.54 >2 Haloprogin >2 0.86 Homoharringtonine <0.25 <0.25 IC 261 0.63 1.11 IKK 16 1.83 2.34 IMD 0354 0.47 >2 IMS2186 1.31 >2 INCA-6 >2 >2 Irinotecan 0.30 0.49 Irinotecan Hydrochloride 0.55 0.82 JTC 801 1.97 >2 KF 38789 0.59 0.91 Lanatoside C <0.25 <0.25 LY 2183240 1.05 1.17 Malonoben 0.32 0.83 Maprotiline hydrochloride >2 >2 Mebendazole <0.25 0.44 Methiazole 0.28 5.74 Methyl benzethonium chloride 1.42 >2 MG 132 0.80 1.94 MG 624 >2 >2 Mitoxantrone <0.25 <0.25 Mitoxantrone <0.25 <0.25 Mitoxantrone dihydrochloride <0.25 <0.25 Monensin sodium salt 0.47 0.49 Mycophenolic acid 0.67 >2 Niclosamide 0.35 1.48 Niclosamide 0.31 1.39 NNC 26-9100 1.63 >2 Nocodazole <0.25 <0.25 Nocodazole <0.25 <0.25 Nocodazole <0.25 <0.25 N-Oleoyldopamine 0.57 1.26 NS8593 hydrochloride >2 >2 NSC 3852 0.25 1.01 NSC 632839 hydrochloride >2 >2 NSC 663284 >2 >2 NSC 95397 1.04 >2 NSC 95397 1.16 >2

215

OLDA 0.72 1.41 Ouabain <0.25 <0.25 Ouabain <0.25 <0.25 Oxibendazol 1.05 1.60 Oxyphenbutazone >2 >2 Paclitaxel <0.25 <0.25 Paclitaxel(Taxol) <0.25 <0.25 PALDA 0.64 1.32 Parbendazole <0.25 <0.25 Parthenolide >2 >2 Parthenolide 1.80 >2 Parthenolide 1.07 1.47 PD 407824 0.55 0.99 PD-166285 hydrate 0.99 0.77 PD173952 ~0.25 1.48 PD-407824 0.53 0.92 PMPA (NAALADase inhibitor) >2 >2 Podophyllotoxin <0.25 <0.25 Podophyllotoxin <0.25 <0.25 Proscillaridin A <0.25 <0.25 Pyrrolidinedithiocarbamate ammonium 0.71 0.60 Pyrvinium pamoate <0.25 0.43 Quinacrine dihydrochloride >2 >2 Quinacrine dihydrochloride dihydrate >2 >2 Ro 106-9920 1.12 1.30 Ro 31-8220 mesylate 1.03 0.95 Rotenone 0.37 0.96 Ryuvidine 0.25 >2 Sanguinarine chloride 1.33 >2 SB 225002 0.48 1.19 SCH 202676 hydrobromide >2 >2 SCH 79797 dihydrochloride 0.40 0.79 Scriptaid 1.00 0.93 SCS 0.66 >2 SN 38 <0.25 <0.25 Stattic 0.90 1.59 Taxol <0.25 <0.25 Tetraethylthiuram disulfide <0.25 <0.25 Thapsigargin <0.25 <0.25 Thimerosal 0.57 0.52 Thonzonium bromide 0.74 1.44 Topotecan <0.25 <0.25 Topotecan Hydrochloride <0.25 <0.25 Topotecan hydrochloride hydrate <0.25 <0.25

216

Tyrphostin A9 0.25 1.24 Vinblastine sulfate <0.25 <0.25 Vinblastine sulfate salt <0.25 <0.25 Vincristine sulfate <0.25 <0.25 Vincristine sulfate <0.25 <0.25 Vincristine Sulfate <0.25 <0.25 Vorinostat 0.50 0.91 Y 29794 oxalate <0.25 1.13 YM 298198, Desmethyl- >2 >2 Zalcitabine >2 >2 ZM 39923 hydrochloride >2 >2 ZM 39923 hydrochloride >2 >2 ZM 447439 0.34 2.07 ZM 449829 >2 >2

217

REFERENCES

Adams, D. J., Sandvold, M. L., Myhren, F., Jacobsen, T. F., Giles, F., & Rizzieri, D. A. (2008). Anti proliferative activity of ELACYT™ (CP-4055) in combination with cloretazine (VNP40101M), idarubicin, gemcitabine, irinotecan and topotecan in human leukemia and lymphoma cells. Leukemia & Lymphoma, 49(4), 786-797.

Adams, M. J., & Lipshultz, S. E. (2005). Pathophysiology of anthracycline- and radiation-associated cardiomyopathies: implications for screening and prevention. Pediatric Blood & Cancer, 44(7), 600-606.

Advani, A. S., & Pendergast, A. M. (2002). Bcr–Abl variants: biological and clinical aspects. Leukemia Research, 26, 713-720.

Aljitawi, O. S., Li, D., Xiao, Y., Zhang, D., Ramachandran, K., Stehno-Bittel, L., . . . Garimella, R. (2014). A novel three-dimensional stromal-based model for in vitro chemotherapy sensitivity testing of leukemia cells. Leukemia & Lymphoma, 55(2), 378- 391.

Allensworth, J. L., Evans, M. K., Bertucci, F., Aldrich, A. J., Festa, R. A., Finetti, P., . . . Devi, G. R. (2015). Disulfiram (DSF) acts as a copper ionophore to induce copper- dependent oxidative stress and mediate anti-tumor efficacy in inflammatory breast cancer. Molecular Oncology, 9(6), 1155-1168.

Amarachintha, S., Harmel-Laws, E., & Steinbrecher, K. A. (2018). Guanylate cyclase C reduces invasion of intestinal epithelial cells by bacterial pathogens. Scientific Reports, 8, 1521.

Ansari, K. I., & Mandal, S. S. (2010). Mixed lineage leukemia: roles in gene expression, hormone signaling and mRNA processing. The FEBS Journal, 277, 1790- 1804.

Aricò, M., Valsecchi, M. G., Camitta, B., Schrappe, M., Chessells, J., Baruchel, A., . . . Masera, G. (2000). Outcome of treatment in children with Philadelphia chromosome-

218 positive acute lymphoblastic leukemia. The New England Journal of Medicine, 342(14), 998-1006.

Armstrong, S. A., Staunton, J. E., Silverman, L. B., Pieters, R., Boer, M. L., Minden, M. D., . . . Korsmeyer, S. J. (2002). MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia. Nature Genetics, 30(1), 41-47.

Ashburn, T. T., & Thor, K. B. (2004). Drug repositioning: identifying and developing new uses for existing drugs. Nature Reviews Drug Discovery, 3(8), 673-683.

Assouline, S., Culjkovic, B., Cocolakis, E., Rousseau, C., Beslu, N., Amri, A., . . . Borden, K. L. (2009). Molecular targeting of the oncogene eIF4E in acute myeloid leukemia (AML): a proof-of-principle clinical trial with ribavirin. Blood, 114, 257-260.

Assouline, S., Culjkovic-Kraljacic, B., Bergeron, J., Caplan, S., Cocolakis, E., Lambert, C., . . . Borden, K. L. (2015). A phase I trial of ribavirin and low-dose cytarabine for the treatment of relapsed and refractory acute myeloid leukemia with elevated eIF4E. Haematologica, 100(1), e7-e9.

Baffoe, C. S., Nguyen, N., Boyd, P., Wang, W., Morris, M., & McConville, C. (2015). Disulfiram-loaded immediate and extended release vaginal tablets for the localised treatment of cervical cancer. Journal of Pharmacy and Pharmacology, 67(2), 189-198.

Bailey, L. C., Lange, B. J., Rheingold, S. R., & Bunin, N. J. (2008). Bone-marrow relapse in paediatric acute lymphoblastic leukaemia. The Lancet Oncology, 9(9), 873– 883.

Balgobind, B., Zwaan, C., Pieters, R., & Heuvel-Eibrink, M. V. (2011). The heterogeneity of pediatric MLL-rearranged acute myeloid leukemia. Leukemia, 25, 1239-1248.

Bansal, M., Yang, J., Karan, C., Menden, M. P., Costello, J. C., Tang, H., . . . Community, N.-D. (2014). A community computational challenge to predict the activity of pairs of compounds. Nature Biotechnology, 32(12), 1213-1222.

219

Baran, I., Ionescu, D., Filippi, A., Mocanu, M. M., Iftime, A., Babes, R., . . . Ganea, C. (2014). Novel insights into the antiproliferative effects and synergism of quercetin and menadione in human leukemia Jurkat T cells. Leukemia Research, 38(7), 836-849.

Barbieri, D., Abbracchio, M. P., Salvioli, S., Monti, D., Cossarizza, A., Ceruti, S., . . . Franceschi, C. (1998). Apoptosis by 2-chloro-2'-deoxy-adenosine and 2-chloroadenosine in human peripheral blood mononuclear cells. Neurochemistry International, 32(5-6), 493-504.

Basso, G., Veltroni, M., Valsecchi, M. G., Dworzak, M. N., Ratei, R., Silvestri, D., . . . Gaipa, G. (2009). Risk of relapse of childhood acute lymphoblastic leukemia is predicted by flow cytometric measurement of residual disease on day 15 bone marrow. Journal of Clinical Oncology, 27, 5168-5174.

Bastin-Coyette, L., Smal, C., Cardoen, S., Saussoy, P., Van Den Neste, E., & Bontemps, F. (2008). Mechanisms of cell death induced by 2-chloroadenosine in leukemic B-cells. Biochemical Pharmacology, 75(7), 1451-1460.

Battisti, V., Maders, L. D., Bagatini, M. D., Santos, K. F., Spanevello, R. M., Maldonado, P. A., . . . Morsch, V. M. (2008). Measurement of oxidative stress and antioxidant status in acute lymphoblastic leukemia patients. Clinical Biochemistry, 41(7-8), 511-518.

Bayir, H. (2005). Reactive oxygen species. Critical Care Medicine, 33(12 Suppl), S498-501.

Behm, F. G. (2012). Immunophenotyping. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 72-112). Cambridge: Cambridge University Press.

Belizário, J. E. (2009). Immunodeficient Mouse Models: An Overview. The Open Immunology Journal, 2, 79-85.

Bernt, K. M., & Hunger, S. P. (2014). Current concepts in pediatric Philadelphia chromosome-positive acute lymphoblastic leukemia. Frontiers in Oncology, 4(54), 1- 21.

220

Bernt, K. M., Zhu, N., Sinha, A. U., Vempati, S., Faber, J., Krivtsov, A. V., . . . Armstrong, S. A. (2011). MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. Cancer Cell, 20(1), 66-78.

Bertaina, A., Vinti, L., Strocchio, L., Gaspari, S., Caruso, R., Algeri, M., . . . Locatelli, F. (2017). The combination of bortezomib with chemotherapy to treat relapsed/refractory acute lymphoblastic leukaemia of childhood. British Journal of Haematology, 176, 629-636.

Bibette, J. (2012). Gaining confidence in high-throughput screening. Proceedings of the National Academy of Sciences, 109(3), 649-650.

Biondi, A., Schrappe, M., Lorenzo, P. D., Castor, A., Lucchini, G., Gandemer, V., . . . Valsecchi, M. G. (2012). Imatinib after induction for treatment of children and adolescents with Philadelphia-chromosome-positive acute lymphoblastic leukaemia (EsPhALL): a randomised, open-label, intergroup study. The Lancet Oncology, 13(9), 936-945.

Bisgin, H., Liu, Z., Kelly, R., Fang, H., Xu, X., & Tong, W. (2012). Investigating drug repositioning opportunities in FDA drug labels through topic modeling. BMC Bioinformatics, 13 Suppl 15, S6.

Blatt, J., & Corey, S. J. (2013). Drug repurposing in pediatrics and pediatric hematology oncology. Drug Discovery Today, 18(1-2), 4-10.

Blatt, J., Farag, S., Corey, S. J., Sarrimanolis, Z., Muratov, E., Fourches, D., . . . Janzen, W. P. (2014). Expanding the scope of drug repurposing in pediatrics: The Children’s Pharmacy Collaborative. Drug Discovery Today, 19(11), 1696-1698.

Boer, J. M., & den Boer, M. L. (2017). BCR-ABL1-like acute lymphoblastic leukaemia: From bench to bedside. European Journal of Cancer, 82, 203-218.

Boolell, M., Allen, M. J., Ballard, S. A., Gepi-Attee, S., Muirhead, G. J., Naylor, A. M., . . . Gingell, C. (1996). Sildenafil: an orally active type 5 cyclic GMP-specific phosphodiesterase inhibitor for the treatment of penile erectile dysfunction. International Journal of Impotence Research, 8(2), 47-52.

221

Borden, K. L., & Culjkovic-Kraljacic, B. (2010). Ribavirin as an anti-cancer therapy: Acute Myeloid Leukemia and beyond? Leukemia & Lymphoma, 51(10), 1805-1815.

Borowitz, M. J., Devidas, M., Hunger, S. P., Bowman, W. P., Carroll, A. J., Carroll, W. L., . . . for the Children's Oncology Group. (2008). Clinical significance of minimal residual disease in childhood acute lymphoblastic leukemia and its relationship to other prognostic factors: a children's oncology group study. Blood, 111, 5477-5485.

Bradley, D. (2005). Why big pharma needs to learn the three 'R's. Nature Reviews Drug Discovery, 4(6), 446.

Brehm, M. A., Wiles, M. V., Greiner, D. L., & Shultz, L. D. (2014). Generation of improved humanized mouse models for human infectious diseases. Journal of Immunological Methods, 410, 3-17.

Brown, P. (2013). Treatment of infant leukemias: challenge and promise. Hematology Am Soc Hematol Educ Program, 2013, 596-600.

Bruce, A., Evans, R., Mezan, R., Shi, L., Moses, B. S., Martin, K. H., . . . Yang, Y. (2015). Three-Dimensional Microfluidic Tri-Culture Model of the Bone Marrow Microenvironment for Study of Acute Lymphoblastic Leukemia. PLoS One, 10(10), e0140506.

Bug, G., Burchert, A., Wagner, E.-M., Kröger, N., Berg, T., Güller, S., . . . Ottmann, O. G. (2017). Phase I/II study of the deacetylase inhibitor panobinostat after allogeneic stem cell transplantation in patients with high-risk MDS or AML (PANOBEST trial). Leukemia, 31, 2523–2525.

Bulman, C. A., Bidlow, C. M., Lustigman, S., Cho-Ngwa, F., Williams, D., Rascón, A. A., . . . Sakanari, J. A. (2015). Repurposing Auranofin as a Lead Candidate for Treatment of Lymphatic Filariasis and Onchocerciasis. PLOS Neglected Tropical Diseases, 9(2), e0003534.

Burbury, K. L., Bishton, M. J., Johnstone, R. W., Dickinson, M. J., Szer, J., & Prince, H. M. (2011). MLL-aberrant leukemia: complete cytogenetic remission following

222 treatment with a histone deacetylase inhibitor (HDACi). Annals of Hematology, 90, 847-849.

Burkhardt, B., Mueller, S., Khanam, T., & Perkins, S. L. (2016). Current status and future directions of T-lymphoblastic lymphoma in children and adolescents. British Journal of Haematology, 173, 545–559.

Burnstock, G., & Di Virgilio, F. (2013). Purinergic signalling and cancer. Purinergic Signalling, 9, 491-540.

Cai, C., Chen, S.-Y., Zheng, Z., Omwancha, J., Lin, M.-F., Balk, S. P., & Shemshedini, L. (2007). Androgen regulation of soluble guanylyl cyclasea1 mediates prostate cancer cell proliferation. Oncogene, 26, 1606-1615.

Calgarotto, A. K., Maso, V., Junior, G. C., Nowill, A. E., Filho, P. L., Vassallo, J., & Saad, S. T. (2018). Antitumor activities of Quercetin and Green Tea in xenografts of human leukemia HL60 cells. Scientific Reports, 8, 3459.

Camos, M., & Colomer, D. (2006). Molecular biology in acute leukemia. Clinical and Translational Oncology, 8, 550-559.

Campana, D. (2010). Minimal residual disease in acute lymphoblastic leukemia. Hematology, 1, 7-12.

Campbell, C. T., Haladyna, J. N., Drubin, D. A., Thomson, T. M., Maria, M. J., Yamauchi, T., . . . Daigle, S. R. (2017). Mechanisms of Pinometostat (EPZ-5676) Treatment Emergent Resistance in MLL Rearranged Leukemia. Molecular Cancer Therapeutics, 16(8), 1669-1679.

Capparelli, E. V., Bricker-Ford, R., Rogers, M. J., McKerrow, J. H., & Reed, S. L. (2016). Phase I Clinical Trial Results of Auranofin, a Novel Antiparasitic Agent. Antimicrobial Agents and Chemotherapy, 61(1), e01947-16.

Cardillo, T. M., Govindan, S. V., Zalath, M. B., Rossi, D. L., Wang, Y., Chang, C.-H., & Goldenberg, D. M. (2018). IMMU-140, a Novel SN-38 Antibody-Drug Conjugate Targeting HLA-DR, Mediates Dual Cytotoxic Effects in Hematologic Cancers and Malignant Melanoma. Molecular Cancer Therapeutics, 17(1), 150-160.

223

Carew, J. S., Nawrocki, S. T., Xu, R. H., Dunner, K., McConkey, D. J., Wierda, W. G., . . . Huang, P. (2004). Increased mitochondrial biogenesis in primary leukemia cells: the role of endogenous nitric oxide and impact on sensitivity to fludarabine. Leukemia, 18(12), 1934-1940.

Carpentieri, U., Minguell, J. J., & Gardner, F. H. (1981). Adenylate cyclase and guanylate cyclase activity in normal and leukemic human lymphocytes. Blood, 57(5), 975-978.

Carpentieri, U., Myers, J., Thorpe, L., Daeschner III, C. W., & Haggard, M. E. (1986). Copper, Zinc, and Iron in Normal and Leukemic Lymphocytes from Children. Cancer Research, 46, 981-964.

Caudell, D., & Aplan, P. D. (2008). The role of CALM–AF10 gene fusion in acute leukemia. Leukemia, 22, 678-685.

Caveney, E., Caveney, B. J., Turner, J. R., & Gourgiotis, L. (2011). Pharmaceutical interventions for obesity: a public health perspective. Diabetes, Obesity and Metabolism, 13(6), 490-497.

Ceruti, S., Beltrami, E., Matarrese, P., Mazzola, A., Cattabeni, F., Malorni, W., & Abbracchio, M. P. (2003). A key role for caspase-2 and caspase-3 in the apoptosis induced by 2-chloro-2'-deoxy-adenosine (cladribine) and 2-chloro-adenosine in human astrocytoma cells. Molecular Pharmacology, 63(6), 1437-1447.

Chandra, J., Tracy, J., Loegering, D., Flatten, K., Verstovsek, S., Beran, M., . . . Kaufmann, S. H. (2006). Adaphostin-induced oxidative stress overcomes BCR/ABL mutation-dependent and -independent imatinib resistance. Blood, 107(6), 2501-2506.

Chen, D., Cui, Q. C., Yang, H., & Dou, Q. P. (2006a). Disulfiram, a clinically used anti- alcoholism drug and copper-binding agent, induces apoptotic cell death in breast cancer cultures and xenografts via inhibition of the proteasome activity. Cancer Research, 66(21), 10425-33.

Chen, D., Cui, Q. C., Yang, H., Barrea, R. A., Sarkar, F. H., Sheng, S., . . . Dou, Q. P. (2007). Clioquinol, a therapeutic agent for Alzheimer's disease, has proteasome-

224 inhibitory, androgen receptor-suppressing, apoptosis-inducing, and antitumor activities in human prostate cancer cells and xenografts. Cancer Research, 67(4), 1636-1644.

Chen, L., Deshpande, A. J., Banka, D., Bernt, K. M., Dias, S., Buske, C., . . . Armstrong, S. A. (2013). Abrogation of MLL-AF10 and CALM-AF10-mediated transformation through genetic inactivation or pharmacological inhibition of the H3K79 methyltransferase Dot1l. Leukemia, 27(4), 813-822.

Chen, W., Li, Q., Hudson, W. A., Kumar, A., Kirchhof, N., & Kersey, J. H. (2006b). Amurine Mll-AF4 knock-in model results in lymphoid and myeloid deregulation and hematologic malignancy. Blood, 108(2), 669-677.

Chen, X., Shi, X., Zhao, C., Li, X., Lan, X., Liu, S., . . . Liu, J. (2014). Anti-rheumatic agent auranofin induced apoptosis in chronic myeloid leukemia cells resistant to imatinib through both Bcr/Abl-dependent and -independent mechanisms. Oncotarget, 5(19), 9118-9132.

Chen, Y.-F., Liu, H., Luo, X.-J., Zhao, Z., Zou, Z.-Y., Li, J., . . . Liang, Y. (2017). The roles of reactive oxygen species (ROS) and autophagy in the survival and death of leukemia cells. Critical Reviews in Oncology/Hematology, 112, 21-30.

Chena, C.-W., & Armstrong, S. A. (2015). Targeting DOT1L and HOX gene expression in MLL-rearranged leukemia and beyond. Experimental Hematology, 43, 673–684.

Cheung, A. M., Wan, T. S., Leung, J. C., Chan, L. Y., Huang, H., Kwong, Y. L., . . . Leung, A. Y. (2007). Aldehyde dehydrogenase activity in leukemic blasts defines a subgroup of acute myeloid leukemia with adverse prognosis and superior NOD/SCID engrafting potential. Leukemia, 21(7), 1423-30.

Chiba, M., Ishii, Y., & Sugiyama, Y. (2009). Prediction of hepatic clearance in human from in vitro data for successful drug development. The AAPS Journal, 11(2), 262-276.

Childhood ALL Collaborative Group. (1996). Duration and intensity of maintenance chemotherapy in acute lymphoblastic leukaemia: overview of 42 trials involving 12 000 randomised children. The Lancet, 347, 1783-1788.

225

Choi, Y. H., & Park, H. S. (2012). Apoptosis induction of U937 human leukemia cells by diallyl trisulfide induces through generation of reactive oxygen species. Journal of Biomedical Science, 19, 50.

Chong, C. R., & Sullivan, D. J. (2007). New uses for old drugs. Nature, 448(7154), 645-646.

Chou, T. C. (2006). Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacological Reviews, 58(3), 621-681.

Chowdhury, T., & Brady, H. J. (2008). Insights from clinical studies into the role of the MLL gene in infant and childhood leukemia. Blood Cells, Molecules, and Diseases, 40, 192–199.

Cobby, J., Mayersohn, M., & Selliah, S. (1977). The rapid reduction of disulfiram in blood and plasma. Journal of Pharmacology and Experimental Therapeutics, 202(3), 724-731.

Cohen, D. P., Adams, D. J., Flowers, J. L., Wall, M. E., Wani, M. C., Manikumar, G., . . . Silber, R. (1999). Pre-clinical evaluation of SN-38 and novel camptothecin analogs against human chronic B-cell lymphocytic leukemia lymphocytes. Leukemia Research, 23(11), 1061-1070.

Conticello, C., Martinetti, D., Adamo, L., Buccheri, S., Giuffrida, R., Parrinello, N., . . . Di Raimondo, F. (2012). Disulfiram, an old drug with new potential therapeutic uses for human hematological malignancies. International Journal of Cancer, 131(9), 2197- 2203.

Cook, G. J., & Pardee, T. S. (2013). Animal Models of Leukemia: Any closer to the real thing? Cancer Metastasis Reviews, 32, 63-76.

Cooper, S. L., & Brown, P. A. (2015). Treatment of Pediatric Acute Lymphoblastic Leukemia. Pediatric Clinics of North America, 62(1), 61-73.

226

Corcé, V., Gouin, S. G., Renaud, S., Gaboriau, F., & Deniaud, D. (2016). Recent advances in cancer treatment by Iron Chelators. Bioorganic and Medicinal Chemistry Letters, 26(2), 251-256.

Corsello, S. M., Bittker, J. A., Liu, Z., Gould, J., McCarren, P., Hirschman, J. E., . . . Golub, T. R. (2017). The Drug Repurposing Hub: a next-generation drug library and information resource. Nature Medicine, 23(4), 405-408.

Coustan-Smith, E., Mullighan, C. G., Onciu, M., Behm, F. G., Raimondi, S. C., Pei, D., . . . Campana, D. (2009). Early T-cell precursor leukaemia: a subtype of very high-risk acute lymphoblastic leukaemia. The Lancet Oncology, 10, 147-156.

Cruickshank, M. N., Ford, J., Cheung, L. C., Heng, J., Singh, S., Wells, J., . . . Kees, U. R. (2017). Systematic chemical and molecular profiling of MLL-rearranged infant acute lymphoblastic leukemia reveals efficacy of romidepsin. Leukemia, 31, 40-50.

Cvek, B. (2012). Nonprofit drugs as the salvation of the world's healthcare systems: the case of Antabuse (disulfiram). Drug Discovery Today, 17(9-10), 409-412.

D’Ambrosi, N., Costanzi, S., Angelini, D. F., Volpini, R., Sancesario, G., Cristalli, G., & Volonte, C. (2004). 2-ClATP exerts anti-tumoural actions not mediated by P2 receptors in neuronal and glial cell lines. Biochemical Pharmacology, 67(4), 621-630.

Dai, B., Yoo, S. Y., Bartholomeusz, G., Graham, R. A., Majidi, M., Yan, S., . . . Roth, J. A. (2013). KEAP1-dependent synthetic lethality induced by AKT and TXNRD1 inhibitors in lung cancer. Cancer Research, 73(17), 5532-5543.

Daigle, S. R., Olhava, E. J., Therkelsen, C. A., Basavapathruni, A., Jin, L., Boriack- Sjodin, P. A., . . . Pollock, R. (2013). Potent inhibition of DOT1L as treatment of MLL- fusion leukemia. Blood, 122, 1017-1025.

227

Daigle, S. R., Olhava, E. J., Therkelsen, C. A., Majer, C. R., Sneeringer, C. J., Song, J., . . . Pollock, R. M. (2011). Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer Cell, 20(1), 53-65.

Dastugue, N., Suciu, S., Plat, G., Speleman, F., Cavé, H., Girard, S., . . . Benoit, Y. (2013). Hyperdiploidy with 58-66 chromosomes in childhood B-acute lymphoblastic leukemia is highly curable: 58951 CLG-EORTC results. Blood, 121(13), 2415-2423.

Dayer, D., Asadi, Z. T., Samie, M., Reisiyan, N., Kalantarian, G., & Vakhshiteh, F. (2015). Evaluation of Serum Copper Levels in Patients with Leukemia and Lymphoma. Trends in Life sciences, 4(3), 266-270. den Boer, M. L., Slegtenhorst, M. v., Menezes, R. X., Cheok, M. H., Buijs-Gladdines, J. G., Peters, S. T., . . . Pieters, R. (2009). A subtype of childhood acute lymphoblastic leukaemia with poor treatment outcome: a genome-wide classifi cation study. The Lancet Oncology, 10, 125-134.

Deng, M., Jiang, Z., Li, Y., Zhou, Y., Li, J., Wang, X., . . . Xu, B. (2016). Effective elimination of adult B-lineage acute lymphoblastic leukemia by disulfiram/copper complex in vitro and in vivo in patient-derived xenograft models. Oncotarget, 7(50), 82200-82212.

Denoyer, D., Masaldan, S., La Fontaine, S., & Cater, M. A. (2015). Targeting copper in cancer therapy: 'Copper That Cancer'. Metallomics, 7(11), 1459-1476.

Devi, G. S., Prasad, M. H., Saraswathi, I., Raghu, D., Rao, D. N., & Reddy, P. P. (2000). Free radicals antioxidant enzymes and lipid peroxidation in different types of leukemias. Clinica Chimica Acta, 293(1-2), 53-62.

Dharmaraja, A. T. (2017). Role of Reactive Oxygen Species (ROS) in Therapeutics and Drug Resistance in Cancer and Bacteria. Journal of Medicinal Chemistry, 60(8), 3221- 3240.

Dickson, M., & Gagnon, J. P. (2004). Key factors in the rising cost of new drug discovery and development. Nature Reviews Drug Discovery, 3, 417-429.

228

Doan, T., & Massarotti, E. (2005). Rheumatoid arthritis: an overview of new and emerging therapies. The Journal of Clinical Pharmacology, 45(7), 751-762.

Doghman, M., Cazareth, J., Douguet, D., Madoux, F., Hodder, P., & Lalli, E. (2009). Inhibition of Adrenocortical Carcinoma Cell Proliferation by Steroidogenic Factor-1 Inverse Agonists. The Journal of Clinical Endocrinology & Metabolism, 94, 2178-2183.

Doghman, M., Karpova, T., Rodrigues, G. A., Arhatte, M., De Moura, J., Cavalli, L. R., . . . Lalli, E. (2007). Increased steroidogenic factor-1 dosage triggers adrenocortical cell proliferation and cancer. Molecular Endocrinology, 21(12), 2968-2987.

Doghman, M., Madoux, F., Hodder, P., & Lalli, E. (2010). Identification and characterization of steroidogenic factor-1 inverse agonists. Methods in Enzymology, 485, 3-23.

Dong, J.-m., Zhao, S.-g., Huang, G.-y., & Liu, Q. (2004). NADPH Oxidase-mediated Generation of Reactive Oxygen Species is Critically Required for Survival of Undifferentiated Human Promyelocytic Leukemia Cell Line HL-60. Free Radical Research, 38(6), 629-637.

Doogue, M. P., & Polasek, T. M. (2013). The ABCD of clinical pharmacokinetics. Therapeutic Advances in Drug Safety, 4(1), 5-7.

Druker, B. (2004). Imatinib as a paradigm of targeted therapies. Advances in Cancer Research, 91, 1-30.

Duffner, U., Abdel-Mageed, A., Younge, J., Tornga, C., Scott, K., Staddon, J., . . . Kidd, P. (2016). The possible perils of targeted therapy. Leukemia, 30(7), 1619-21.

Dufour, P., Lang, J. M., Giron, C., Duclos, B., Haehnel, P., Jaeck, D., . . . Oberling, F. (1993). Sodium dithiocarb as adjuvant immunotherapy for high risk breast cancer: a randomized study. Biotherapy, 6(1), 9-12.

Duriez, P. J., & Shah, G. M. (1997). Cleavage of poly(ADP-ribose) polymerase: a sensitive parameter to study cell death. Biochemistry and Cell Biology, 75(4), 337-349.

229

Ekins, S., Williams, A. J., Krasowski, M. D., & Freundlich, J. S. (2011). In silico repositioning of approved drugs for rare and neglected diseases. Drug Discovery Today, 16(7-8), 298-310.

Endo, K., Ueda, T., Ueyama, J., Ohta, T., & Terada, T. (2000). Immunoreactive e- cadherin, alpha-catenin,beta-catenin, and gamma-catenin proteins in hepatocellular carcinoma: Relationships with tumor grade, clinicopathologic parameters, and patients' survival. Human Pathology, 31(5), 558-565.

Eneanya, D. I., Bianchine, J. R., Duran, D. O., & Andresen, B. D. (1981). The Actions and Metabolic Fate of Disulfiram. Annual Review of Pharmacology and Toxicology, 21, 575-596.

Enghusen Poulsen, H., Loft, S., Andersen, J. R., & Andersen, M. (1992). Disulfiram therapy--adverse drug reactions and interactions. Acta Psychiatrica Scandinavica, 369, 65-66.

Ensor, H. M., Schwab, C., Russell, L. J., Richards, S. M., Morrison, H., Masic, D., . . . Moorman, A. V. (2011). Demographic, clinical, and outcome features of children with acute lymphoblastic leukemia and CRLF2 deregulation: results from the MRCALL97 clinical trial. Blood, 117(7), 2129-2136.

Fan, C., Zheng, W., Fu, X., Li, X., Wong, Y.-S., & Chen, T. (2014). Enhancement of auranofin-induced lung cancer cell apoptosis by selenocystine, a natural inhibitor of TrxR1 in vitro and in vivo. Cell Death & Disease, 5(4), e1191.

Feng, T., Basu, P., Sun, W., Ku, H. T., & Mack, W. J. (2017). Optimal design for high- throughput screening via false discovery rate control. arXiv:1707.03462 [stat.AP], 1-24.

Ferraz-de-Souza, B., Lin, L., & Achermann, J. C. (2011). Steroidogenic factor-1 (SF-1, NR5A1) and human disease. Molecular and Cellular Endocrinology, 336, 198-205.

Field-Smith, A., Morgan, G. J., & Davies, F. E. (2006). Bortezomib (Velcade™) in the Treatment of Multiple Myeloma. Therapeutics and Clinical Risk Management, 2(3), 271-279.

230

Fisher, A. B. (2009). Redox Signaling Across Cell Membranes. Antioxidants & Redox Signaling, 11(6), 1349-56.

Fiskus, W., Saba, N., Shen, M., Ghias, M., Liu, J., Gupta, S. D., . . . Bhalla, K. N. (2014). Auranofin induces lethal oxidative and endoplasmic reticulum stress and exerts potent preclinical activity against chronic lymphocytic leukemia. Cancer Research, 74(9), 2520-2532.

Foley, G. E., Lazarus, H., Farber, S., Uzman, B. G., Boone, B. A., & McCarthy, R. E. (1965). Continuous culture of human lymphoblasts from peripheral blood of a child with acute leukemia. Cancer, 18, 522-529.

Forghieri, F., Luppi, M., & Potenza, L. (2015). Philadelphia chromosome-positive Acute Lymphoblastic Leukemia. Hematology, 20(10), 618-619.

Forrest, R. A., Swift, L. P., Rephaeli, A., Nudelman, A., Kimura, K., Phillips, D. R., & Cutts, S. M. (2012). Activation of DNA damage response pathways as a consequence of anthracycline-DNA adduct formation. Biochemical Pharmacology, 83(12), 1602-1612.

Foucquier, J., & Guedj, M. (2015). Analysis of drug combinations: current methodological landscape. Pharmacology Research & Perspectives, 3(3), e00149.

Franck, J., & Jayaram-Lindstrom, N. (2013). Pharmacotherapy for alcohol dependence: status of current treatments. Current Opinion in Neurobiology, 23, 692-699.

Fuller, R. K., & Gordis, E. (2004). Does disulfiram have a role in alcoholism treatment today? Addiction, 99(1), 21-24.

Furst, D. E. (1983). Mechanism of action, pharmacology, clinical efficacy and side effects of auranofin. An orally administered organic gold compound for the treatment of rheumatoid arthritis. Pharmacotherapy, 3(5), 284-98.

Gan, F., Cao, B., Wu, D., Chen, Z., Hou, T., & Mao, X. (2011). Exploring Old Drugs for the Treatment of Hematological Malignancies. Current Medicinal Chemistry, 18, 1509-1514.

231

Gasparetto, M., Sekulovic, S., Brocker, C., Tang, P., Zakaryan, A., Xiang, P., . . . Smith, C. (2012). Aldehyde dehydrogenases are regulators of hematopoietic stem cell numbers and B-cell development. Experimental Hematology, 40(4), 318–329.

Geller, H. M., Cheng, K. Y., Goldsmith, N. K., Romero, A. A., Zhang, A. L., Morris, E. J., & Grandison, L. (2001). Oxidative stress mediates neuronal DNA damage and apoptosis in response to cytosine arabinoside. Journal of Neurochemistry, 78(2), 265- 275.

Gerber, J. M., Smith, B. D., Ngwang, B., Zhang, H., Vala, M. S., Morsberger, L., . . . Jones, R. J. (2012). A clinically relevant population of leukemic CD34(+)CD38(-) cells in acute myeloid leukemia. Blood, 119(15), 3571-3577.

Gessi, S., Merighi, S., Sacchetto, V., Simioni, C., & Borea, P. A. (2011). Adenosine receptors and cancer. Biochimica et Biophysica Acta, 1808, 1400-1412.

Gewirtz, D. A. (1999). A critical evaluation of the mechanisms of action proposed for the antitumor effects of the anthracycline antibiotics adriamycin and daunorubicin. Biochemical Pharmacology, 57(7), 727-741.

Goswami, C. P., Cheng, L., Alexander, P. S., Singal, A., & Li, L. (2015). A New Drug Combinatory Effect Prediction Algorithm on the Cancer Cell Based on Gene Expression and Dose-Response Curve. Pharmacometrics & Systems Pharmacology, 4(2), e9.

Green, D. M., Kawashima, T., Stovall, M., Leisenring, W., Sklar, C. A., Mertens, A. C., . . . Robison, L. L. (2009). Fertility of female survivors of childhood cancer: a report from the childhood cancer survivor study. Journal of Clinical Oncology, 27(16), 2677- 85.

Green, D. M., Kawashima, T., Stovall, M., Leisenring, W., Sklar, C. A., Mertens, A. C., . . . Robison, L. L. (2010). Fertility of male survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. Journal of Clinical Oncology, 28(2), 332-339.

232

Greif, P. A., Tizazu, B., Krause, A., Kremmer, E., & Bohlander, S. K. (2008). The leukemogenic CALM/AF10 fusion protein alters the subcellular localization of the lymphoid regulator Ikaros. Oncogene, 27(20), 2886-2896.

Grembecka, J., He, S., Shi, A., Purohit, T., Muntean, A. G., Sorenson, R. J., . . . Cierpicki, T. (2012). Menin-MLL Inhibitors Reverse Oncogenic Activity of MLL Fusion Proteins in Leukemia. Nature Chemical Biology, 8(3), 277-284.

Gupta, S. C., Sung, B., Prasad, S., Webb, L. J., & Aggarwal, B. B. (2013). Cancer drug discovery by repurposing: teaching new tricks to old dogs. Trends in Pharmacological Sciences, 34(9), 508-517.

Gupte, A., & Mumper, R. J. (2009). Elevated copper and oxidative stress in cancer cells as a target for cancer treatment. Cancer Treatment Reviews, 35, 32-46.

Haddox, C. L., Mangaonkar, A. A., Chen, D., Shi, M., He, R., Oliveira, J. L., . . . Elliott, M. A. (2017). Blinatumomab-induced lineage switch of B-ALL with t(4:11)(q21;q23) KMT2A/AFF1 into an aggressive AML: pre- and post-switch phenotypic, cytogenetic and molecular analysis. Blood Cancer Journal, 7(9), e607.

Hamre, H., Zeller, B., Kanellopoulos, A., Kiserud, C. E., Aakhus, S., Lund, M. B., . . . Ruud, E. (2013). High Prevalence of Chronic Fatigue in Adult Long-Term Survivors of Acute Lymphoblastic Leukemia and Lymphoma During Childhood and Adolescence. Journal of Adolescent and Young Adult Oncology, 2(1).

Hantschel, O. (2012). Structure, Regulation, Signaling, and Targeting of Abl Kinases in Cancer. Genes & Cancer, 3(5-6), 436-446.

Harned, T. M., & Gaynon, P. S. (2008). Relapsed acute lymphoblastic leukemia: current status and future opportunities. Current Oncology Reports, 10, 453-458.

Harrison, C. J. (2000). The genetics of childhood acute lymphoblastic leukaemia. Best Practice & Research Clinical Haematology, 3(3), 427 - 439.

Hauer, J., Borkhardt, A., Sánchez-García, I., & Cobaleda, C. (2014). Genetically engineered mouse models of human B-cell precursor leukemias. Cell Cycle, 13(18), 2836-46.

233

Haydu, J. E., & Ferrando, A. A. (2013). Early T-cell Precursor Acute Lymphoblastic Leukemia (ETP TALL). Current Opinion in Hematology, 20(4), 369-373.

Heilig, M., & Egli, M. (2006). Pharmacological treatment of alcohol dependence: Target symptoms and target mechanisms. Pharmacology & Therapeutics, 111(3), 855– 876.

Henderson, M. J., Choi, S., Beesley, A. H., Baker, D. L., Wright, D., Papa, R. A., . . . Kees, U. R. (2008). A xenograft model of infant leukaemia reveals a complex MLL translocation. British Journal of Haematology, 140(6), 716-719.

Hijiya, N., Stewart, C. F., Zhou, Y., Campana, D., Coustan-Smith, E., Rivera, G. K., . . . Gajjar, A. (2008). Phase II study of topotecan in combination with dexamethasone, asparaginase, and vincristine in pediatric patients with acute lymphoblastic leukemia in first relapse. Cancer, 112(9), 1983-1991.

Hilden, J. M., Dinndorf, P. A., Meerbaum, S. O., Sather, H., Villaluna, D., Heerema, N. A., . . . Reaman, G. H. (2006). Analysis of prognostic factors of acute lymphoblastic leukemia in infants: report on CCG 1953 from the Children’s Oncology Group. Blood, 108, 441-451.

Hill, J. M., Kornblith, A. B., Jones, D., Freeman, A., Holland, J. F., Glicksman, A. S., . . . Holland, J. C. (1998). A comparative study of the long term psychosocial functioning of childhood acute lymphoblastic leukemia survivors treated by intrathecal methotrexate with or without cranial radiation. Cancer, 82(1), 208-218.

Hoang, V. T., Buss, E. C., Wang, W., Hoffmann, I., Raffel, S., Zepeda-Moreno, A., . . . Lutz, C. (2015). The rarity of ALDH(+) cells is the key to separation of normal versus leukemia stem cells by ALDH activity in AML patients. The International Journal of Cancer, 137(3), 525-536.

Hoeksema, K. A., Jayanthan, A., Cooper, T., Gore, L., Trippett, T., Boklan, J., . . . Narendran, A. (2011). Systematic in-vitro evaluation of the NCI/NIH Developmental Therapeutics Program Approved Oncology Drug Set for the identification of a candidate drug repertoire for MLL-rearranged leukemia. OncoTargets and Therapy, 4, 149-168.

234

Hoivik, E. A., Lewis, A. E., Aumo, L., & Bakke, M. (2010). Molecular aspects of steroidogenic factor 1 (SF-1). Molecular and Cellular Endocrinology, 315, 27-39.

Horstmann, E., McCabe, M. S., Grochow, L., Yamamoto, S., Rubinstein, L., Budd, T., . . . Grady, C. (2005). Risks and Benefits of Phase 1 Oncology Trials, 1991 through 2002. The New England Journal of Medicine, 352, 895-904.

Horton, T. M., Perentesis, J. P., Gamis, A. S., Alonzo, T. A., Gerbing, R. B., Ballard, J., . . . Moscow, J. A. (2014). A Phase 2 study of bortezomib combined with either idarubicin/cytarabine or cytarabine/etoposide in children with relapsed, refractory or secondary acute myeloid leukemia: a report from the Children's Oncology Group. Pediatric Blood & Cancer, 61, 1754-1760.

Houghton, P. J., Morton, C. L., Tucker, C., Payne, D., Favours, E., Cole, C., . . . Smith, M. A. (2007). The pediatric preclinical testing program: description of models and early testing results. Pediatric Blood & Cancer, 49(7), 928-940.

Houston, J. B. (1994). Utility of in vitro drug metabolism data in predicting in vivo metabolic clearance. Biochemical Pharmacology, 47(9), 1469-1479.

Houston, J. B., & Carlile, D. J. (1997). Prediction of Hepatic Clearance from Microsomes, Hepatocytes, and Liver Slices. Drug Metabolism Reviews, 29(4), 891-922.

Hrabe, J. E., O’Leary, B. R., Fath, M. A., Rodman, S. N., Button, A. M., Domann, F. E., . . . Mezhir, J. J. (2015). Disruption of thioredoxin metabolism enhances the toxicity of transforming growth factor β-activated kinase 1 (TAK1) inhibition in KRAS-mutated colon cancer cells. Redox Biology, 5, 319-327.

Hrgovcic, M., Tessmer, C. F., Minckler, T. M., Mosier, B., & Taylor, G. H. (1968). Serum copper levels in lymphoma and leukemia. Special reference to Hodgkin's disease. Cancer, 21(4), 743-755.

Huang, H., Liao, Y., Liu, N., Hua, X., Cai, J., Yang, C., . . . Liu, J. (2016). Two clinical drugs deubiquitinase inhibitor auranofin and aldehyde dehydrogenase inhibitor disulfiram trigger synergistic anti-tumor effects in vitro and in vivo. Oncotarget, 7(3), 2796-808.

235

Huang, X., Tanaka, T., Kurose, A., Traganos, F., & Darzynkiewicz, Z. (2006). Constitutive histone H2AX phosphorylation on Ser-139 in cells untreated by genotoxic agents is cell-cycle phase specific and attenuated by scavenging reactive oxygen species. International Journal of Oncology, 29(2), 495-501.

Hudson, M. M. (2010). Reproductive outcomes for survivors of childhood cancer. Obstetrics & Gynecology, 116(5), 1171-1183.

Hughes, J. P., Rees, S., Kalindjian, S. B., & Philpott, K. L. (2011). Principles of early drug discovery. British Journal of Pharmacology, 162(6), 1239-1249.

Hunger, S. P., & Mullighan, C. G. (2015). Acute Lymphoblastic Leukemia in Children. The New England Journal of Medicine, 373, 1541-1552.

Hunger, S. P., Lu, X., Devidas, M., Camitta, B. M., Gaynon, P. S., Winick, N. J., . . . Carroll, W. L. (2012). Improved Survival for Children and Adolescents With Acute Lymphoblastic Leukemia Between 1990 and 2005: A Report From the Children's Oncology Group. Journal of Clinical Oncology, 30(14), 1663-1669.

Hwang-Bo, H., Jeong, J. W., Han, M. H., Park, C., Hong, S. H., Kim, G. Y., . . . Choi, Y. H. (2017). Auranofin, an inhibitor of thioredoxin reductase, induces apoptosis in hepatocellular carcinoma Hep3B cells by generation of reactive oxygen species. General Physiology and Biophysics, 36(2), 117-128.

Iljin, K., Ketola, K., Vainio, P., Halonen, P., Kohonen, P., Fey, V., . . . Kallioniemi, O. (2009). High-throughput cell-based screening of 4910 known drugs and drug-like small molecules identifies disulfiram as an inhibitor of prostate cancer cell growth. Clinical Cancer Research, 15(19), 6070-6078.

Imamura, T., Morimoto, A., Ikushima, S., Kakazu, N., Hada, S., Tabata, Y., . . . Imashuku, S. (2002). A novel infant acute lymphoblastic leukemia cell line with MLL- AF5q31 fusion transcript. Leukemia, 2302-2308.

Inaba, H., Azzato, E. M., & Mullighan, C. G. (2017). Integration of Next-Generation Sequencing to Treat Acute Lymphoblastic Leukemia with Targetable Lesions: The St. Jude Children’s Research Hospital Approach. Frontiers in Pediatrics, 5, 258.

236

Inaba, H., Stewart, C. F., Crews, K. R., Yang, S., Pounds, S., Pui, C. H., . . . Ribeiro, R. C. (2010). Combination of cladribine plus topotecan for recurrent or refractory pediatric acute myeloid leukemia. Cancer, 116(1), 98-105.

Iorio, F., Rittman, T., Ge, H., Menden, M., & Saez-Rodriguez, J. (2013). Transcriptional data: a new gateway to drug repositioning? Drug Discovery Today, 18(7-8), 350-357.

Ito, R., Takahashi, T., Katano, I., & Ito, M. (2012). Current advances in humanized mouse models. Cellular & Molecular Immunology, 9, 208-214.

Iyer, N. S., Balsamo, L. M., Bracken, M. B., & Kadan-Lottick, N. S. (2015). Chemotherapy-only treatment effects on long-term neurocognitive functioning in childhood ALL survivors: a review and meta-analysis. Blood, 126(3), 346-353.

Jacoby, E., Chien, C. D., & Fry, T. J. (2014). Murine Models of Acute Leukemia: Important Tools in Current Pediatric Leukemia Research. Frontiers in Oncology, 4, 95.

Jain, N., Lamb, A. V., O’Brien, S., Ravandi, F., Konopleva, M., Jabbour, E., . . . Khoury, J. (2016). Early T-cell precursor acute lymphoblastic leukemia/lymphoma (ETP-ALL/LBL) in adolescents and adults: a high-risk subtype. Blood, 127(15), 1863- 1869.

Jang, J.-H., & Surh, Y.-J. (2003). Potentiation of cellular antioxidant capacity by Bcl-2: implications for its antiapoptotic function. Biochemical Pharmacology, 66, 1371–1379.

Jeha, S. (2012). Development therapeutics. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 616-631). Cambridge: Cambridge University Press.

Jin, G., & Wong, S. T. (2014). Toward better drug repositioning: prioritizing and integrating existing methods into efficient pipelines. Drug Discovery Today, 19(5), 637- 644.

Jones, D. L., & Wagers, A. J. (2008). No place like home: anatomy and function of the stem cell niche. Nature Reviews Molecular Cell Biology, 9(1), 11-21.

237

Kaiafa, G. D., Saouli, Z., Diamantidis, M. D., Kontoninas, Z., Voulgaridou, V., Raptaki, M., . . . Perifanis, V. (2012). Copper levels in patients with hematological malignancies. European Journal of Internal Medicine, 23(8), 738–741.

Kalinowski, D. S., Stefani, C., Toyokuni, S., Ganz, T., Anderson, G. J., Subramaniam, N. V., . . . Richardson, D. R. (2016). Redox cycling metals: Pedaling their roles in metabolism and their use in the development of novel therapeutics. Biochimica et Biophysica Acta, 1863(4), 727-748.

Kantarjian, H. M., Shah, N. P., Cortes, J. E., Baccarani, M., Agarwal, M. B., Undurraga, M. S., . . . Hochhaus, A. (2012). Dasatinib or imatinib in newly diagnosed chronic- phase chronic myeloid leukemia: 2-year follow-up from a randomized phase 3 trial (DASISION). Blood, 119(5), 1123-1129.

Kantarjian, H., Stein, A., Gökbuget, N., Fielding, A. K., Schuh, A. C., Ribera, J.-M., . . . Topp, M. S. (2017). Blinatumomab versus Chemotherapy for Advanced Acute Lymphoblastic Leukemia. The New England Journal of Medicine, 376, 836-847.

Kaplan, J. H., & Maryon, E. B. (2016). How Mammalian Cells Acquire Copper: An Essential but Potentially Toxic Metal. The Biophysical Journal, 110(1), 7-13.

Kast, R. E., Karpel-Massler, G., & Halatsch, M. E. (2014). CUSP9* treatment protocol for recurrent glioblastoma: aprepitant, artesunate, auranofin, captopril, celecoxib, disulfiram, itraconazole, ritonavir, sertraline augmenting continuous low dose temozolomide. Oncotarget, 5(18), 8052-8082.

Kean, W. F., & Kean, I. R. (2008). Clinical pharmacology of gold. Inflammopharmacology, 16(3), 112-125.

Kean, W. F., Hart, L., & Buchanan, W. W. (1997). Auranofin. British Journal of Rheumatology, 36(5), 560-572.

Kees, U. R., Ford, J., Watson, M., Murch, A., Ringńer, M., Walker, R. L., & Meltzer, P. (2003). Gene expression profiles in a panel of childhood leukemia cell lines mirror critical features of the disease. Molecular Cancer Therapeutics, 2(7), 671-677.

238

Kentsis, A., Topisirovic, I., Culjkovic, B., Shao, L., & Borden, K. L. (2004). Ribavirin suppresses eIF4E-mediated oncogenic transformation by physical mimicry of the 7- methyl guanosine mRNA cap. Proceedings of the National Academy of Sciences, 101(52), 18105-11810.

Khaw, S. L., Suryani, S., Evans, K., Richmond, J., Robbins, A., Kurmasheva, R. T., . . . Lock, R. B. (2016). Venetoclax responses of pediatric ALL xenografts reveal sensitivity of MLL-rearranged leukemia. Blood, 128(10), 1382-1395.

Kim, D.-Y., Joo, Y.-D., Lim, S.-N., Kim, S.-D., Lee, J.-H., Lee, J.-H., . . . Lee, K.-H. W. (2015). Nilotinib combined with multiagent chemotherapy for newly diagnosed Philadelphia-positive acute lymphoblastic leukemia. Blood, 126(6), 746-756.

Kim, N.-H., Park, H. J., Oh, M.-K., & Kim, I.-S. (2013). Antiproliferative effect of gold(I) compound auranofin through inhibition of STAT3 and telomerase activity in MDA-MB 231 human breast cancer cells. BMB Reports, 46(1), 59-64.

Kim, S. K., Kim, H., Lee, D. H., Kim, T. S., Kim, T., Chung, C., . . . Lim, D. S. (2013). Reversing the intractable nature of pancreatic cancer by selectively targeting ALDH- high, therapy-resistant cancer cells. PLoS One, 8(10), e78130.

King, B. F., & Burnstock, G. (2002). Purinergic receptors. In M. Pangalos, & C. Davies, Understanding G protein-coupled receptors and their role in the CNS (3 ed., pp. 422- 438). Oxford: Oxford University Press.

Koeppen, M., Eckle, T., & Eltzschig, H. K. (2009). Selective deletion of the A1 adenosine receptor abolishes heart-rate slowing effects of intravascular adenosine in vivo. PLoS One, 4(8), e6784.

Kohnken, R., Porcu, P., & Mishra, A. (2017). Overview of the Use of Murine Models in Leukemia and Lymphoma Research. Frontiers in Oncology, 7, 22.

Koo, H. H. (2011). Philadelphia Chromosome-Positive Acute Lymphoblastic Leukemia in Childhood. Korean Journal of Pediatrics, 54(3), 106-110.

Koppaka, V., Thompson, D. C., Chen, Y., Ellermann, M., Nicolaou, K. C., Juvonen, R. O., . . . Vasiliou, V. (2012). Aldehyde dehydrogenase inhibitors: a comprehensive

239 review of the pharmacology, mechanism of action, substrate specificity, and clinical application. Pharmacological Reviews, 64(3), 520-539.

Kotecha, R. S., Gottardo, N. G., Kees, U. R., & Cole, C. H. (2014). The evolution of clinical trials for infant acute lymphoblastic leukemia. Blood Cancer Journal, 4(4), e200.

Kouroukis, T. C., Baldassarre, F. G., Haynes, A. E., Imrie, K., Reece, D. E., & Cheung, M. C. (2014). Bortezomib in multiple myeloma: systematic review and clinical considerations. Current Oncology, 21(4), e573-e603.

Koyama, D., Kikuchi, J., Hiraoka, N., Wada, T., Kurosawa, H., Chiba, S., & Furukawa, Y. (2014). Proteasome inhibitors exert cytotoxicity and increase chemosensitivity via transcriptional repression of Notch1 in T-cell acute lymphoblastic leukemia. Leukemia, 28, 1216-1226.

Kwong, Y. L., & Todd, D. (1997). Delicious poison: arsenic trioxide for the treatment of leukemia. Blood, 89(9), 3487-3488.

Lander, E. S. (1999). Array of hope. Nature Genetics, 21(1 Suppl), 3-4.

Lanzkowsky, P. (2010). Manual of Pediatric Hematology and Oncology (5 ed.). San Diego: Academic Press.

Lee, J. E., Kwon, Y. J., Baek, H. S., Ye, D. J., Cho, E., Choi, H. K., . . . Chun, Y. J. (2017). Synergistic induction of apoptosis by combination treatment with mesupron and auranofin in human breast cancer cells. Archives of Pharmacal Research, 40(6), 746- 759.

Lemmers, B., Arnoulet, C., Fossat, C., Chambost, H., Sainty, D., Gabert, J., & Schiff, C. (2000). Fine characterization of childhood and adult acute lymphoblastic leukemia (ALL) by a proB and preB surrogate light chain-specific mAb and a proposal for a new B cell ALL classification. Leukemia, 14, 2103-2111.

Leung, W., & Pulsipher, M. A. (2012). Hematopoietic stem cell and natural killer cell transplantation. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 527-548). Cambridge: Cambridge University Press.

240

Lewandowski, D., Barroca, V., Ducongé, F., Bayer, J., Van Nhieu, J. T., Pestourie, C., . . . Roméo, P. H. (2010). In vivo cellular imaging pinpoints the role of reactive oxygen species in the early steps of adult hematopoietic reconstitution. Blood, 115(3), 443-452.

Li, H., Hu, J., Wu, S., Wang, L., Cao, X., Zhang, X., . . . Fang, B. (2016). Auranofin- mediated inhibition of PI3K/AKT/mTOR axis and anticancer activity in non-small cell lung cancer cells. Oncotarget, 7(3), 3548-58.

Li, Y. Y., & Jones, S. J. (2012). Drug repositioning for personalized medicine. Genome Medicine, 4(3), 27.

Li, Y., Fu, S. Y., Wang, L. H., Wang, F. Y., Wang, N. N., Cao, Q., . . . Wu, C. F. (2015). Copper improves the anti-angiogenic activity of disulfiram through the EGFR/Src/VEGF pathway in gliomas. Cancer Letters, 369(1), 86-96.

Li, Z., Xiang, Y., Xiang, L., Xiao, Y., Li, F., & Hao, P. (2014). ALDH Maintains the Stemness of Lung Adenoma Stem Cells by Suppressing the Notch/CDK2/CCNE Pathway. PLoS ONE, 9(3), e92669.

Liem, N. L., Papa, R. A., Milross, C. G., Schmid, M. A., Tajbakhsh, M., Choi, S., . . . Lock, R. B. (2004). Characterization of childhood acute lymphoblastic leukemia xenograft models for the preclinical evaluation of new therapies. Blood, 103(10), 3905- 3914.

Lim, S. O., Gu, J. M., Kim, M. S., Kim, H. S., Park, Y. N., Park, C. K., . . . Jung, G. (2008). Epigenetic changes induced by reactive oxygen species in hepatocellular carcinoma: methylation of the E-cadherin promoter. Gastroenterology, 135(6), 2128- 2140.

Lin, J., Haffner, M. C., Zhang, Y., Lee, B. H., Brennen, W. N., Britton, J., . . . Carducci, M. A. (2011). Disulfiram is a DNA demethylating agent and inhibits prostate cancer cell growth. The Prostate, 71(4), 333-343.

Liu, C., Liu, Z., Li, M., Li, X., Wong, Y. S., Ngai, S. M., . . . Chen, T. (2013). Enhancement of auranofin-induced apoptosis in MCF-7 human breast cells by selenocystine, a synergistic inhibitor of thioredoxin reductase. PLoS One, 8(1), e53945.

241

Liu, N., Li, X., Huang, H., Zhao, C., Liao, S., Yang, C., . . . Liu, J. (2014). Clinically used antirheumatic agent auranofin is a proteasomal deubiquitinase inhibitor and inhibits tumor growth. Oncotarget, 5(14), 5453-5471.

Liu, P., Brown, S., Goktug, T., Channathodiyil, P., Kannappan, V., Hugnot, J. P., . . . Darling, J. L. (2012). Cytotoxic effect of disulfiram/copper on human glioblastoma cell lines and ALDH-positive cancer-stem-like cells. British Journal of Cancer, 107(9), 1488–1497.

Liu, X., Wang, L., Cui, W., Yuan, X., Lin, L., Cao, Q., . . . Yang, J. (2016). Targeting ALDH1A1 by disulfiram/copper complex inhibits non-small cell lung cancer recurrence driven by ALDH-positive cancer stem cells. Oncotarget, 7(36), 58516-58530.

Locatelli, F., Schrappe, M., Bernardo, M. E., & Rutella, S. (2012). How I treat relapsed childhood acute lymphoblastic leukemia. Blood, 120(14), 2807-2816.

Lock, R. B., Liem, N., Farnsworth, M. L., Milross, C. G., Xue, C., Tajbakhsh, M., . . . Marshall, G. M. (2002). The nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model of childhood acute lymphoblastic leukemia reveals intrinsic differences in biologic characteristics at diagnosis and relapse. Blood, 99(11), 4100- 4108.

Loi, S., Pommey, S., Haibe-Kains, B., Beavis, P. A., Darcy, P. K., Smyth, M. J., & Stagg, J. (2013). CD73 promotes anthracycline resistance and poor prognosis in triple negative breast cancer. PNAS, 110(27), 11091-11096.

Lun, X., Wells, J. C., Grinshtein, N., King, J. C., Hao, X., Dang, N. H., . . . Senger, D. L. (2016). Disulfiram when Combined with Copper Enhances the Therapeutic Effects of Temozolomide for the Treatment of Glioblastoma. Clinical Cancer Research, 22(15), 3860-3875.

Ma, M., Wang, X., Tang, J., Xue, H., Chen, J., Pan, C., . . . Shen, S. (2012). Early T-cell precursor leukemia: a subtype of high risk childhood acute lymphoblastic leukemia. Frontiers in Medicine, 6(4), 416-420.

242

Ma, X., Wang, Q., Liu, Y., Chen, Y., Zhang, L., Jiang, M., . . . Han, J. (2015). Inhibition of tumor growth by U0126 is associated with induction of interferon-γ production. International Journal of Cancer, 136(4), 771-783.

Mackinnon, R. N., Wall, M., Zordan, A., Nutalapati, S., Mercer, B., Peverall, J., & Campbell, L. J. (2013). Genome organization and the role of centromeres in evolution of the erythroleukaemia cell line HEL. Evolution, Medicine, and Public Health, 2013(1), 225-240.

Madeira, J. M., Renschler, C. J., Mueller, B., Hashioka, S., Gibson, D. L., & Klegeris, A. (2013). Novel protective properties of auranofin: inhibition of human astrocyte cytotoxic secretions and direct neuroprotection. Life Sciences, 92(22), 1072-1080.

Madoux, F., Li, X., Chase, P., Zastrow, G., Cameron, M. D., Conkright, J. J., . . . Hodder, P. (2008). Potent, Selective and Cell Penetrant Inhibitors of SF-1 by Functional Ultra-High-Throughput Screening. Molecular Pharmacology, 73(6), 1776-1784.

Manda, G., Isvoranu, G., Comanescu, M. V., Manea, A., Debelec Butuner, B., & Korkmaz, K. S. (2015). The redox biology network in cancer pathophysiology and therapeutics. Redox Biology, 5, 347-357.

Mann, G., Attarbaschi, A., Schrappe, M., Lorenzo, P. D., Peters, C., Hann, I., . . . Pieters, R.-9. (2010). Improved outcome with hematopoietic stem cell transplantation in a poor prognostic subgroup of infants with mixed-lineage-leukemia (MLL)–rearranged acute lymphoblastic leukemia: results from the Interfant-99 Study. Blood, 116(15), 2644-2650.

Mao, X., Li, X., Sprangers, R., Wang, X., Venugopal, A., Wood, T., . . . Schimmer, A. D. (2009). Clioquinol inhibits the proteasome and displays preclinical activity in leukemia and myeloma. Leukemia, 23(3), 585-590.

Mareš, P. (2010). Anticonvulsant action of 2-chloroadenosine against pentetrazol- induced seizures in immature rats is due to activation of A1 adenosine receptors. The Journal of Neural Transmission, 117(11), 1269-1277.

243

Margolin, J. F., Rabin, K. R., Steuber, C. P., & Poplack, D. G. (2011). Acute Lymphoblastic Leukemia. In P. A. Pizzo, & D. G. Poplack, Principles and Practice of Pediatric Oncology (6 ed.). Philadelphia, United States: Lippincott Williams & Wilkins.

Marschalek, R. (2010b). Mechanisms of leukemogenesis by MLL fusion proteins. British Journal of Haematology, 152, 141-154.

Marschalek, R. (2010a). Mixed lineage leukemia: roles in human malignancies and potential therapy. The FEBS Journal, 277, 1822-1831.

Marschalek, R. (2016). Systematic Classification of Mixed-Lineage Leukemia Fusion Partners Predicts Additional Cancer Pathways. Annals of Laboratory Medicine, 36(2), 85-100.

Martinelli, G., Iacobucci, I., Storlazzi, C. T., Vignetti, M., Paoloni, F., Cilloni, D., . . . Foa`, R. (2009). IKZF1 (Ikaros) Deletions in BCR-ABL1–Positive Acute Lymphoblastic Leukemia Are Associated With Short Disease-Free Survival and High Rate of Cumulative Incidence of Relapse: A GIMEMA AL WP Report. Journal of Clinical Oncology, 27(31), 5202-5207.

Marzano, C., Gandin, V., Folda, A., Scutari, G., Bindoli, A., & Rigobello, M. P. (2007). Inhibition of thioredoxin reductase by auranofin induces apoptosis in cisplatin-resistant human ovarian cancer cells. Free Radical Biology & Medicine, 42(6), 872-881.

Maso, V., Calgarotto, A. K., Franchi, G. C., Nowill, A. E., Filho, P. L., Vassallo, J., & Saad, S. T. (2014). Multitarget effects of quercetin in leukemia. Cancer Prevention Research, 7(12), 1240-1250.

Masuda, N., Kudoh, S., & Fukuoka, M. (1996). Irinotecan (CPT-11): pharmacology and clinical applications. Critical Reviews in Oncology/Hematology, 24(1), 3-26.

Mathews, V., Chendamarai, E., George, B., Viswabandya, A., & Srivastava, A. (2011). Treatment of Acute Promyelocytic Leukemia with Single-Agent Arsenic Trioxide. Mediterranean Journal of Hematology and Infectious Diseases, 3(1), e2011056.

244

Mathoôt, R. A., Soudijn, W., Breimer, D. D., Ijzerman, A. P., & Danhof, M. (1996). Pharmacokinetic-haemodynamic relationships of 2-chloroadenosine at adenosine A1 and A2a receptors in vivo. British Journal of Pharmacology, 118(2), 369-377.

Maude, S. L., Dolai, S., Delgado-Martin, C., Vincent, T., Robbins, A., Selvanathan, A., . . . Teachey, D. (2015). Efficacy of JAK/STAT pathway inhibition in murine xenograft models of early T-cell precursor (ETP) acute lymphoblastic leukemia. Blood, 125(11), 1759-1767.

Mayr, L. M., & Bojanic, D. (2009). Novel trends in high-throughput screening. Current Opinion in Pharmacology, 9, 580-588.

McBride, W. G. (1977). Thalidomide embryopathy. Teratology, 16(1), 79-82.

McCabe, B., Liberante, F., & Mills, K. I. (2015). Repurposing medicinal compounds for blood cancer treatment. Annals of Hematology, 94, 1267-1276.

Medina-Sanson, A. (2016). Introduction: Childhood Leukemia. In J. M. Mejía- Aranguré, Etiology of Acute Leukemias in Children (pp. 1-48). Switzerland: Springer International Publishing.

Meshinchi, S., Woods, W. G., Stirewalt, D. L., Sweetser, D. A., Buckley, J. D., Tjoa, T. K., . . . Radich, J. P. (2001). Prevalence and prognostic significance of Flt3 internal tandem duplication in pediatric acute myeloid leukemia. Blood, 97, 89-94.

Messinger, Y. H., Gaynon, P. S., Sposto, R., van der Giessen, J., Eckroth, E., Malvar, J., & Bostrom, B. C. (2012). Bortezomib with chemotherapy is highly active in advanced B-precursor acute lymphoblastic leukemia: Therapeutic Advances in Childhood Leukemia & Lymphoma (TACL) Study. Blood, 120(2), 285-290.

Meyer, C., Burmeister, T., Gröger, D., Tsaur, G., Fechina, L., Renneville, A., . . . Marschalek, R. (2018). The MLL recombinome of acute leukemias in 2017. Leukemia, 32, 273-284.

Minelli, A., Bellezza, I., Agostini, M., Bracarda, S., & Culig, Z. (2006). Mechanism of 2-chloroadenosine toxicity to PC3 cell line. The Prostate, 66(13), 1425-1436.

245

Mirabell, C. K., Johnson, R. K., Hill, D. T., Faucette, L. F., Girard, G. R., Kuo, G. Y., . . . Crooke, S. T. (1986). Correlation of the in vitro cytotoxic and in vivo antitumor activities of gold(I) coordination complexes. Journal of Medicinal Chemistry, 29(2), 218-223.

Mishra, S., Zhang, B., Cunnick, J. M., Heisterkamp, N., & Groffen, J. (2006). Resistance to Imatinib of Bcr/Abl P190 Lymphoblastic Leukemia Cells. Cancer Research, 66(10), 5387-5393.

Mody, R., Li, S., Dover, D. C., Sallan, S., Leisenring, W., Oeffinger, K. C., . . . Neglia, J. P. (2008). Twenty-five-year follow-up among survivors of childhood acute lymphoblastic leukemia: a report from the Childhood Cancer Survivor Study. Blood, 111(12), 5515-5523.

Moore, S. A., Baker, H. M., Blythe, T. J., Kitson, K. E., Kitson, T. M., & Baker, E. N. (1998). Sheep liver cytosolic aldehyde dehydrogenase: the structure reveals the basis for the retinal specificity of class 1 aldehyde dehydrogenases. Structure, 6(12), 1541-1551.

Moorman, A. V. (2012). The clinical relevance of chromosomal and genomic abnormalities in B-cell precursor acute lymphoblastic leukaemia. Blood Reviews, 26, 123-135.

Moorman, A. V., Ensor, H. M., Richards, S. M., Chilton, L., Schwab, C., Kinsey, S. E., . . . Harrison, C. J. (2010). Prognostic effect of chromosomal abnormalities in childhood B-cell precursor acute lymphoblastic leukaemia: results from the UK Medical Research Council ALL97/99 randomised trial. The Lancet Oncology, 11(5), 429-438.

Morbidelli, L., Pyriochou, A., Filippi, S., Vasileiadis, I., Roussos, C., Zhou, Z., . . . Papapetropoulos, A. (2010). The soluble guanylyl cyclase inhibitor NS-2028 reduces vascular endothelial growth factor-induced angiogenesis and permeability. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 298(3), R824-832.

Morrison, B. W., Doudican, N. A., Patel, K. R., & Orlow, S. J. (2010). Disulfiram induces copper-dependent stimulation of reactive oxygen species and activation of the extrinsic apoptotic pathway in melanoma. Melanoma Research, 20(1), 11-20.

246

Moyle, G. J., Youle, M., Higgs, C., Monaghan, J., Prince, W., Chapman, S., . . . Nelson, M. R. (1998). Safety, Pharmacokinetics, and Antiretroviral Activity of the Potent, Specific Human Immunodeficiency Virus Protease Inhibitor Nelfinavir: Results of a Phase I/II Trial and Extended Follow-up in Patients Infected with Human Immunodeficiency Virus. The Journal of Clinical Pharmacology, 38(8), 736-743.

Mudry, R. E., Fortney, J. E., York, T., Hall, B. M., & Gibson, L. F. (2000). Stromal cells regulate survival of B-lineage leukemic cells during chemotherapy. Blood, 96, 1926-1932.

Muffly, L., & Larson, R. A. (2012). Improving outcomes in childhood T-cell acute lymphoblastic leukemia: promising results from the Children’s Oncology Group incorporating nelarabine into front-line therapy. Translational Pediatrics, 1(2), 120- 122.

Mullighan, C. G. (2012b). Molecular genetics of acute lymphoblastic leukemia. In C.- H. Pui, Childhood Leukemias (3 ed., pp. 168-203). Cambridge: Cambridge University Press.

Mullighan, C. G. (2012a). The molecular genetic makeup of acute lymphoblastic leukemia. Hematology, 2012(1), 389-396.

Mullighan, C. G., Phillips, L. A., Su, X., Ma, J., Miller, C. B., Shurtleff, S. A., & Downing, J. R. (2008). Genomic Analysis of the Clonal Origins of Relapsed Acute Lymphoblastic Leukemia. Science, 322(5906), 1377-1380.

Mullighan, C. G., Su, X., Zhang, J., Radtke, I., Phillips, L. A., Miller, C. B., . . . Children's Oncology Group. (2009). Deletion of IKZF1 and Prognosis in Acute Lymphoblastic Leukemia. The New England Journal of Medicine, 360, 470-480.

Murphy, B., Yin, H., Maris, J. M., Kolb, E. A., Gorlick, R., Reynolds, C. P., . . . Houghton, P. J. (2016). Evaluation of Alternative In Vivo Drug Screening Methodology: A Single Mouse Analysis. Cancer Research, 76(19), 5798-5809.

247

Mustafa, S. K., & AlSharif, M. A. (2018). Copper (Cu) an Essential Redox-Active Transition Metal in Living System—A Review Article. American Journal of Analytical Chemistry, 9(1), 15-26.

Myers, W. C., Donahue, J. E., & Goldstein, M. R. (1994). Disulfiram for alcohol use disorders in adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 33(4), 484-489.

Nakaya, A., Sagawa, M., Muto, A., Uchida, H., Ikeda, Y., & Kizaki, M. (2011). The gold compound auranofin induces apoptosis of human multiple myeloma cells through both down-regulation of STAT3 and inhibition of NF-κB activity. Leukemia Research, 35, 243–249.

Ness, K. K., Armenian, S. H., Kadan-Lottick, N., & Gurney, J. G. (2011). Adverse effects of treatment in childhood acute lymphoblastic leukemia: general overview and implications for long-term cardiac health. Expert Review of Hematology, 4(2), 185-197.

Neumann, M., Coskun, E., Fransecky, L., Mochmann, L. H., Bartram, I., Heesch, S., . . . Baldus, C. D. (2013). FLT3 mutations in early T-cell precursor ALL characterize a stem cell like leukemia and imply the clinical use of tyrosine kinase inhibitors. PLoS One, 8(1), e53190.

Neumann, M., Heesch, S., Gokbuget, N., Schwartz, S., Schlee, C., Benlasfer, O., . . . Baldus, C. D. (2012). Clinical and molecular characterization of early T-cell precursor leukemia: a high-risk subgroup in adult T-ALL with a high frequency of FLT3 mutations. Blood Cancer Journal, 2(1), e55.

Ng, M.-H. J., Ng, R. K., Kong, C.-T., Jin, D.-Y., & Chan, L.-C. (2010). Activation of Ras-dependent Elk-1 activity by MLL-AF4 family fusion oncoproteins. Experimental Hematology, 38(6), 481-488.

Nguyen, K., Devidas, M., Cheng, S.-C., La, M., Raetz, E., Carroll, W., . . . ML Loh, f. (2008). Factors influencing survival after relapse from acute lymphoblastic leukemia: a Children's Oncology Group study. Leukemia, 22, 2142-2150.

248

Nicholls, J., Hourani, S. M., & Hall, J. M. (2002). Characterization of adenosine receptors mediating the vasodilator effects of adenosine receptor agonists in the microvasculature of the hamster cheek pouch in vivo. Autonomic and Autacoid Pharmacology, 22(4), 209-214.

Nobel, C. I., Kimland, M., Lind, B., Orrenius, S., & Slater, A. F. (1995). Dithiocarbamates induce apoptosis in thymocytes by raising the intracellular level of redox-active copper. The Journal of Biological Chemistry, 270(44), 26202-8.

Novac, N. (2013). Challenges and opportunities of drug repositioning. Trends in Pharmacological Sciences, 34(5), 267-272.

Obach, R. S. (1999). Prediction of human clearance of twenty-nine drugs from hepatic microsomal intrinsic clearance data: An examination of in vitro half-life approach and nonspecific binding to microsomes. Drug Metabolism & Disposition, 27(11), 1350- 1359.

Onciu, M., & Pui, C.-H. (2012). Diagnosis and classification. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 21-48). Cambridge: Cambridge University Press.

Osredkar, J., & Sustar, N. (2011). Copper and Zinc, Biological Role and Significance of Copper/Zinc Imbalance. Journal of Clinical Toxicology, S3, 1.

Palumbo, A., & Anderson, K. (2011). Multiple Myeloma. The New England Journal of Medicine, 364, 1046-1060.

Pammolli, F., Magazzini, L., & Riccaboni, M. (2011). The productivity crisis in pharmaceutical R&D. Nature Reviews Drug Discovery, 10, 428-438.

Papaioannou, M., Mylonas, I., Kast, R. E., & Brüning, A. (2013). Disulfiram/copper causes redox-related proteotoxicity and concomitant heat shock response in ovarian cancer cells that is augmented by auranofin-mediated thioredoxin inhibition. Oncoscience, 1(1), 21-29.

Paranjpe, A., Zhang, R., Ali-Osman, F., Bobustuc, G. C., & Srivenugopal, K. S. (2014). Disulfiram is a direct and potent inhibitor of human O6-methylguanine-DNA

249 methyltransferase (MGMT) in brain tumor cells and mouse brain and markedly increases the alkylating DNA damage. Carcinogenesis, 35(3), 692-702.

Park, C. S., Shen, Y., Lewis, A. H., Suppipat, K., Puppi, M., Tomolonis, J., . . . Lacorazza, H. D. (2017). Vitamin K3 (menadione) Induces Cytotoxicity in Chronic Myeloid Leukemia Stem Cells By Upregulating DYRK2 through the Inhibition of SIAH2 Ubiquitin Ligase. Blood, 130, 4171.

Parker, C., Waters, R., Leighton, C., Hancock, J., Sutton, R., Moorman, A. V., . . . Saha, V. (2010). Effect of mitoxantrone on outcome of children with first relapse of acute lymphoblastic leukaemia (ALL R3): an open-label randomised trial. The Lancet, 376(9757), 2009-2017.

Paul, M. K., & Mukhopadhyay, A. K. (2004). Tyrosine kinase – Role and significance in Cancer. International Journal of Medical Sciences, 1(2), 101-115.

Paul, S. M., Mytelka, D. S., Dunwiddie, C. T., Persinger, C. C., Munos, B. H., Lindborg, S. R., & Schacht, A. L. (2010). How to improve R&D productivity: the pharmaceutical industry’s grand challenge. Nature Reviews Drug Discovery, 9(3), 203- 214.

Paulsson, K., Forestier, E., Lilljebjörn, H., Heldrup, J., Behrendtz, M., Young, B. D., & Johansson, B. (2010). Genetic landscape of high hyperdiploid childhood acute lymphoblastic leukemia. Proceedings of the National Academy of Sciences, 107(50), 21719-21724.

Peller, S. & Rotter, V. (2003). Human Mutation. 21, 277-284.

Pemovska, T., Kontro, M., Yadav, B., Edgren, H., Eldfors, S., Szwajda, A., . . . Wennerberg, K. (2013). Individualized systems medicine strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discovery, 3(12), 1416-1429.

Petersen, E. N. (1992). The pharmacology and toxicology of disulfiram and its metabolites. Acta Psychiatrica Scandinavica. Supplementum, 369, 7-13.

250

Pieters, R., Schrappe, M., Lorenzo, P. D., Hann, I., Rossi, G. D., Felice, M., . . . Valsecchi, M. (2007). A treatment protocol for infants younger than 1 year with acute lymphoblastic leukaemia (Interfant-99): an observational study and a multicentre randomised trial. Lancet, 370, 240-250.

Pizzo, P. A., & Poplack, D. G. (2002). Principles and Practice of Pediatric Oncology (4 ed.). Philadelphia: Lippincott Williams & Williams.

Plodkowski, R. A., Nguyen, Q., Sundaram, U., Nguyen, L., Chau, D. L., & St Jeor, S. (2009). Bupropion and naltrexone: a review of their use individually and in combination for the treatment of obesity. Expert Opinion on Pharmacotherapy, 10(6), 1069-1081.

Polley, E., Kunkel, M., Evans, D., Silvers, T., Delosh, R., Laudeman, J., . . . Teicher, B. A. (2016). Small Cell Lung Cancer Screen of Oncology Drugs, Investigational Agents, and Gene and microRNA Expression. The Journal of the National Cancer Institute, 108(10).

Prochazka, M., Gaskins, H. R., Shultz, L. D., & Leiter, E. H. (1992). The nonobese diabetic scid mouse: model for spontaneous thymomagenesis associated with immunodeficiency. Proceedings of the National Academy of Sciences, 89(8), 3290- 3294.

Pui, C.-H. (1995). Childhood leukemias. The New England Journal of Medicine, 332, 1618-1630.

Pui, C.-H. (2009). Acute lymphoblastic leukemia: introduction. Seminars in Hematology, 46, 1-2.

Pui, C.-H. (2010). Recent Research Advances in Childhood Acute Lymphoblastic Leukemia. Journal of the Formosan Medical Association, 109(11), 777–787.

Pui, C.-H. (2012). Acute lymphoblastic leukemia. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 332-366). Cambridge: Cambridge University Press.

Pui, C.-H., & Evans, W. E. (1998). Acute Lymphoblastic Leukemia. The New England Journal of Medicine, 339(9), 605-615.

251

Pui, C.-H., & Evans, W. E. (2006). Treatment of acute lymphoblastic leukemia. The New England Journal of Medicine, 354, 166-178.

Pui, C.-H., Campana, D., Pei, D., Bowman, W. P., Sandlund, J. T., Kaste, S. C., . . . Relling, M. (2009). Treating Childhood Acute Lymphoblastic Leukemia without Cranial Irradiation. The New England Journal of Medicine, 360(26), 2730-2741.

Pui, C.-H., Chessells, J. M., Camitta, B., Baruchel, A., Biondi, A., Boyett, J. M., . . . Schrappe, M. (2003). Clinical heterogeneity in childhood acute lymphoblastic leukemia with 11q23 rearrangements. Leukemia, 17, 700-706.

Pui, C.-H., Gaynon, P. S., Boyett, J. M., Chessells, J. M., Baruchel, A., Kamps, W., . . . Camitta, B. (2002). Outcome of treatment in childhood acute lymphoblastic leukaemia with rearrangements of the 11q23 chromosomal region. The Lancet, 359(9321), 1909- 1915.

Pui, C.-H., Robison, L. L., & Look, A. T. (2008). Acute lymphoblastic leukaemia. The Lancet, 371, 1030-1043.

Pui, C.-H., Yang, J. J., Hunger, S. P., Pieters, R., Schrappe, M., Biondi, A., . . . Mullighan, C. (2015). Childhood Acute Lymphoblastic Leukemia: Progress Through Collaboration. Journal Of Clinical Oncology, 33(27), 2938-2948.

Quah, B. J., Warren, H. S., & Parish, C. R. (2007). Monitoring lymphocyte proliferation in vitro and in vivo with the intracellular fluorescent dye carboxyfluorescein diacetate succinimidyl ester. Nature Protocols, 2(9), 2049-2056.

Quintas-Cardama, A., Hu, C., Qutub, A., Qiu, Y. H., Zhang, X., Post, S. M., Zhang, N., Coombes, K. & Kornblau, S. M. (2017). p53 pathway dysfunction is highly prevalent in acute myeloid leukemia independent of TP53 mutational status. Leukemia. 31, 1296- 1305.

Ramayya, M. S., Shenga, M., Moroz, K., Hill, S. M., & Rowan, B. G. (2010). Human steroidogenic factor-1 (hSF-1) regulates progesterone biosynthesis and growth of ovarian surface epithelial cancer cells. Journal of Steroid Biochemistry & Molecular Biology, 119, 14-25.

252

Ran, D., Schubert, M., Pietsch, L., Taubert, I., Wuchter, P., Eckstein, V., . . . Ho, A. D. (2009). Aldehyde dehydrogenase activity among primary leukemia cells is associated with stem cell features and correlates with adverse clinical outcomes. Experimental Hematology, 37(12), 1423–1434.

Ravandi, F., O’Brien, S. M., Cortes, J. E., Thomas, D. M., Garris, R., Faderl, S., . . . Kantarjian, H. M. (2015). Long-Term Follow-Up of a Phase 2 Study of Chemotherapy Plus Dasatinib for the Initial Treatment of Patients With Philadelphia Chromosome– Positive Acute Lymphoblastic Leukemia. Cancer, 121, 4158-4164.

Ravandi, F., O’Brien, S., Thomas, D., Faderl, S., Jones, D., Garris, R., . . . Kantarjian, H. (2010). First report of phase 2 study of dasatinib with hyper-CVAD for the frontline treatment of patients with Philadelphia chromosome–positive (Ph+) acute lymphoblastic leukemia. Blood, 116(12), 2070-2077.

Rayes, A., McMasters, R. L., & O’Brien, M. M. (2016). Lineage Switch in MLL- Rearranged Infant Leukemia Following CD19-Directed Therapy. Pediatric Blood & Cancer, 63(6), 1113-1115.

Redaelli, A., Laskin, B. L., Stephens, J., Botteman, M., & Pashos, C. (2005). A systematic literature review of the clinical and epidemiological burden of acute lymphoblastic leukaemia (ALL). European Journal of Cancer Care, 14(1), 53-62.

Richardson, P. G., Sonneveld, P., Schuster, M. W., Irwin, D., Stadtmauer, E. A., Facon, T., . . . Anderson, K. C. (2005). Bortezomib or High-Dose Dexamethasone for Relapsed Multiple Myeloma. The New England Journal of Medicine, 352, 2487-2498.

Rickles, R. J., Pierce, L. T., Giordano III, T. P., Tam, W. F., McMillin, D. W., Delmore, J., . . . Lee, M. S. (2010). Adenosine A2A receptor agonists and PDE inhibitors: a synergistic multitarget mechanism discovered through systematic combination screening in B-cell malignancies. Blood, 116(4), 593-602.

Rivlin, N., Brosh, R., Oren, M. & Rotter, V. (2011). Mutations in the p53 Tumor Suppressor Gene: Important Milestones at the Various Steps of Tumorigenesis. Genes & Cancer. 2(4), 466-474.

253

Roberts, K. G., & Mullighan, C. G. (2015). Genomics in acute lymphoblastic leukaemia: insights and treatment implications. Nature Reviews Clinical Oncology, 12, 344-357.

Roberts, K. G., Li, Y., Payne-Turner, D., Harvey, R. C., Cheng, J., Zhang, J., . . . Mullighan, C. (2013). Genomic Characterization and Experimental Modeling Of BCR- ABL1-Like Acute Lymphoblastic Leukemia. Blood, 122(21), 232.

Roberts, K. G., Li, Y., Payne-Turner, D., Harvey, R. C., Yang, Y.-L., Pei, D., . . . Mullighan, C. G. (2014). Targetable Kinase-Activating Lesions in Ph-like Acute Lymphoblastic Leukemia. The New England Journal of Medicine, 371, 1005-1015.

Robinson, S. S. (2015). The Role of Disulfiram in Alcohol Metabolism and the Treatment of Alcoholism. In R. Barth, & M. A. Benvenuto, Ethanol and Education: Alcohol as a Theme for Teaching Chemistry (pp. 191–206). Washington, D.C., United States: American Chemical Society.

Robison, L. L. (2011). Late effects of acute lymphoblastic leukemia therapy in patients diagnosed at 0-20 years of age. Hematology American Society of Hematology Education Program, 2011, 238-242.

Roder, C., & Thomson, M. J. (2015). Auranofin: Repurposing an Old Drug for a Golden New Age. Drugs in R&D, 15, 13-20.

Rongvaux, A., Willinger, T., Martinek, J., Strowig, T., Gearty, S. V., Teichmann, L. L., . . . Flavell, R. A. (2014). Development and function of human innate immune cells in a humanized mouse model. Nature Biotechnology, 32(4), 364-372.

Roth, J., Madoux, F., Hodder, P., & Roush, W. R. (2008). Synthesis of small molecule inhibitors of the orphan nuclear receptor steroidogenic factor-1 (NR5A1) based on isoquinolinone scaffolds. Bioorganic & Medicinal Chemistry Letters, 18, 2628-2632.

Rubnitz, E, J., Crews, K. R., Pounds, S., Yang, S., Campana, D., . . . Ribeiro, R. C. (2009). Combination of cladribine and cytarabine is effective for childhood acute myeloid leukemia: results of the St Jude AML97 trial. Leukemia, 23(8), 1410-1416.

254

Rubnitz, J. E., & Pui, C.-H. (1997). Childhood Acute Lymphoblastic Leukemia. The Oncologist, 2(6), 374-380.

Rubnitz, J. E., Inaba, H., & Ribeiro, R. C. (2012). Acute myeloid leukemia. In C.-H. Pui, Childhood Leukemias (3 ed., pp. 395-420). Cambridge: Cambridge University Press.

Ruggiero, A., Triarico, S., Trombatore, G., Battista, A., Dell'acqua, F., Rizzari, C., & Riccardi, R. (2013). Incidence, clinical features and management of hypersensitivity reactions to chemotherapeutic drugs in children with cancer. European Journal of Clinical Pharmacology, 69(10), 1739-1746.

Ruvolo, V. R., Karanjeet, K. B., Schuster, T. F., Brown, R., Deng, Y., Hinchcliffe, E., & Ruvolo, P. P. (2010). Role for PKC δ in Fenretinide-Mediated Apoptosis in Lymphoid Leukemia Cells. Journal of Signal Transduction, 2010, 584657.

Saba, N. S., Ghias, M., Manepalli, R., Schorno, K., Weir, S., Austin, C., . . . Wiestner, A. (2013). Auranofin Induces a Reversible In-Vivo Stress Response That Correlates With a Transient Clinical Effect In Patients With Chronic Lymphocytic Leukemia. Blood, 122, 3819.

Saletta, F., Wadham, C., Ziegler, D. S., Marshall, G. M., Haber, M., McCowage, G., . . . Byrne, J. A. (2014). Molecular profiling of childhood cancer: Biomarkers and novel therapies. BBA Clinical, 1, 59-77.

Sallmann, F. R., Bourassa, S., Saint-Cyr, J., & Poirier, G. G. (1997). Characterization of antibodies specific for the caspase cleavage site on poly(ADP-ribose) polymerase: specific detection of apoptotic fragments and mapping of the necrotic fragments of poly(ADP-ribose) polymerase. Biochemistry and Cell Biology, 75(4), 451-456.

Santana, V. M., Hurwitz, C. A., Blakley, R. L., Crom, W. R., Luo, X., Roberts, W. M., . . . Krance, R. A. (1994). Complete hematologic remissions induced by 2- chlorodeoxyadenosine in children with newly diagnosed acute myeloid leukemia. Blood, 84(4), 1237-1242.

255

Santana, V. M., Mirro, J. J., Kearns, C., Schell, M. J., Crom, W., & Blakley, R. L. (1992). 2-Chlorodeoxyadenosine produces a high rate of complete hematologic remission in relapsed acute myeloid leukemia. Journal of Clinical Oncology, 10(3), 364-370.

Savage, N. M., Kota, V., Manaloor, E. J., Kulharya, A. S., Pierini, V., Mecucci, C., & Ustun, C. (2010). Acute leukemia with PICALM-MLLT10 fusion gene: diagnostic and treatment struggle. Cancer Genetics and Cytogenetics, 202(2), 129-132.

Schrappe, M., Beier, R., & Burger, B. (2003). New teatment strategies in childhood acute lymphoblastic leukemia. Best Practice & Research Clinical Haematology, 15(4), 729-740.

Schrappe, M., Hunger, S. P., Pui, C.-H., Saha, V., Gaynon, P. S., Baruchel, A., . . . Zimmermann, M. (2012). Outcomes after Induction Failure in Childhood Acute Lymphoblastic Leukemia. The New England Journal of Medicine, 366(15), 1371-1381.

Schuessler, B. (2006). What do we know about duloxetine’s mode of action? Evidence from animals to humans. BJOG: An International Journal of Obstetrics & Gynaecology, 113 Suppl 1, 5-9.

Schultz, K. R., Bowman, W. P., Aledo, A., Slayton, W. B., Sather, H., Devidas, M., . . . Camitta, B. (2009). Improved Early Event-Free Survival With Imatinib in Philadelphia Chromosome–Positive Acute Lymphoblastic Leukemia: A Children’s Oncology Group Study. The Journal of Clinical Oncology, 27(31), 5175-5181.

Schultz, K. R., Carroll, A., Heerema, N. A., Bowman, W. P., Aledo, A., Slayton, W. B., . . . Camitta, B. f. (2014). Long-term follow-up of imatinib in pediatric Philadelphia chromosome-positive acute lymphoblastic leukemia: Children’s Oncology Group Study AALL0031. Leukemia, 28, 1467-1471.

Schwartz, C. L. (1999). Long-term survivors of childhood cancer: the late effects of therapy. The oncologist, 4(1), 45-54.

256

Scott, L. J., Figgitt, D. P., Keam, S. J., & Waugh, J. (2005). Acamprosate: a review of its use in the maintenance of abstinence in patients with alcohol dependence. CNS Drugs, 19(5), 445-464.

Shen, M., Zhang, Y., Saba, N., Austin, C. P., Wiestner, A., & Auld, D. S. (2013). Identification of Therapeutic Candidates for Chronic Lymphocytic Leukemia from a Library of Approved Drugs. PLoS One, 8(9), e75252.

Shen, Z., Chen, G., Ni, J., Li, X., Xiong, S., Qiu, Q., . . . Wang, Z. (1997). Use of arsenic trioxide (As2O3) in the treatment of acute promyelocytic leukemia (APL): II. Clinical efficacy and pharmacokinetics in relapsed patients. Blood, 89, 3354-3360.

Shepherd, G. M. (2003). Hypersensitivity reactions to chemotherapeutic drugs. Clinical Reviews in Allergy & Immunology, 24(3), 253-262.

Shim, J. S., & Liu, J. O. (2014). Recent Advances in Drug Repositioning for the Discovery of New Anticancer Drugs. International Journal of Biological Sciences, 10(7), 654-663.

Simon, T. M., Kunishima, D. H., Vibert, G. J., & Lorber, A. (1981). Screening trial with the coordinated gold compound auranofin using mouse lymphocyte leukemia P388. Cancer Research, 41(1), 94-97.

Singh, M. M., Irwin, M. E., Gao, Y., Ban, K., Shi, P., Arlinghaus, R. B., . . . Chandra, J. (2012). Inhibition of the NADPH oxidase regulates heme oxygenase 1 expression in chronic myeloid leukemia. Cancer, 118(13), 3433-3445.

Singh, R. K., Tripathi, A. K., Tripathi, P., Singh, S., Singh, R., & Ahmad, R. (2009). Studies on biomarkers for oxidative stress in patients with chronic myeloid leukemia. Hematology/Oncology and Stem Cell Therapy, 2(1), 285-288.

Singh, S., Brocker, C., Koppaka, V., Chen, Y., Jackson, B. C., Matsumoto, A., . . . Vasiliou, V. (2013). Aldehyde dehydrogenases in cellular responses to oxidative/electrophilic stress. Free Radical Biology & Medicine, 56, 89–101.

Sison, E. A., Rau, R. E., McIntyre, E., Li, L., Small, D., & Brown, P. (2013). MLL- rearranged acute lymphoblastic leukaemia stem cell interactions with bone marrow

257 stroma promote survival and therapeutic resistance that can be overcome with CXCR4 antagonism. British Journal of Haematology, 160(6), 785-797.

Slany, R. K. (2009). The molecular biology of mixed lineage leukemia. Haematologica, 94(7), 984-993.

Smith, M., Arthur, D., Camitta, B., Carroll, A. J., Crist, W., Gaynon, P., . . . Ungerleider, R. (1996). Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. Journal of Clinical Oncology, 14, 18-24.

Somers, K., Chudakova, D. A., Middlemiss, S. M., Wen, V. W., Clifton, M., Kwek, A., . . . Henderson, M. J. (2016). CCI-007, a novel small molecule with cytotoxic activity against infant leukemia with MLL rearrangements. Oncotarget, 7(29), 46067-46087.

Sotolongo, A., Mónica, F. Z., Kots, A., Xiao, H., Liu, J., Seto, E., . . . Murad, F. (2016). Epigenetic regulation of soluble guanylate cyclase (sGC) β1 in breast cancer cells. The FASEB Journal, 30(9), 3171-3180.

Spits, H. (2014). New models of human immunity. Nature Biotechnology, 32(4), 335– 336.

Stam, R. W., Schneider, P., Hagelstein, J. A., van der Linden, M. H., Stumpel, D. J., de Menezes, R. X., . . . Pieters, R. (2010). Gene expression profiling-based dissection of MLL translocated and MLL germline acute lymphoblastic leukemia in infants. Blood, 115(14), 2835-44.

Stanulla, M., & Schrappe, M. (2009). Treatment of childhood acute lymphoblastic leukemia. Seminars in Hematology, 46, 52-63.

Stegmaier, K., Corsello, S. M., Ross, K. N., Wong, J. S., DeAngelo, D. J., & Golub, T. R. (2005). Gefitinib induces myeloid differentiation of acute myeloid leukemia. Blood, 106(9), 2841-2848.

Stegmaier, K., Ross, K. N., Colavito, S. A., O’Malley, S., Stockwell, B. R., & Golub, T. R. (2004). Gene expression–based high-throughput screening (GE-HTS) and application to leukemia differentiation. Nature Genetics, 36(3), 257-263.

258

Stein, E. M., & Tallman, M. S. (2015). Mixed lineage rearranged leukaemia: pathogenesis and targeting DOT1L. Current Opinion in Hematology, 22(2), 92-96.

Steinherz, P. G., Gaynon, P. S., Breneman, J. C., Cherlow, J. M., Grossman, N. J., Kersey, J. H., . . . Bleyer, W. A. (1996). Cytoreduction and prognosis in acute lymphoblastic leukemia – the importance of early marrow response: report from the Children’s Cancer Group. Journal of Clinical Oncology, 14, 389-398.

Stengel, A., Kern, W., Haferlach, T., Meggendorfer, M., Fasan, A. & Haferlach, C. (2017). The impact of TP53 mutations and TP53 deletions on survival varies between AML, ALL, MDS and CLL: an analysis of 3307 cases. Leukemia. 31, 705-711.

Strittmatter, S. M. (2014). Old drugs learn new tricks. Nature Medicine, 20(6), 590-591.

Stubbs, M. C., Kim, W., Bariteau, M., Davis, T., Vempati, S., Minehart, J., . . . Armstrong, S. A. (2015). Selective Inhibition of HDAC1 and HDAC2 as a Potential Therapeutic Option for B-ALL. Clinical Cancer Research, 21(10), 2348-2358.

Stumpel, D. J., Schneider, P., Seslija, L., Osaki, H., Williams, O., Pieters, R., & Stam, R. W. (2012). Connectivity mapping identifies HDAC inhibitors for the treatment of t(4;11)-positive infant acute lymphoblastic leukemia. Leukemia, 26(4), 682-692.

Sukhai, M. A., Spagnuolo, P. A., Weir, S., Kasper, J., Patton, L., & Schimmer, A. D. (2011). New sources of drugs for hematologic malignancies. Blood, 117(25), 6747- 6755.

Sun, W., Sanderson, P., & Zheng, W. (2016). Drug combination therapy increases successful drug repositioning. Drug Discovery Today, 21(7), 1189-1195.

Sundberg, S. A. (2000). High-throughput and ultra-high-throughput screening: solution- and cell-based approaches. Current Opinion in Biotechnology, 11(1), 47-53.

Suryani, S., Carol, H., Chonghaile, T. N., Frismantas, V., Sarmah, C., High, L., . . . Lock, R. B. (2014). Cell and molecular determinants of in vivo efficacy of the BH3 mimetic ABT-263 against pediatric acute lymphoblastic leukemia xenografts. Clinical Cancer Research, 20(17), 4520-31.

259

Szatrowski, T. P., & Nathan, C. F. (1991). Production of Large Amounts of Hydrogen Peroxide by Human Tumor Cells. Cancer Research, 51, 794-798.

Szczepański, T., Harrison, C. J., & Dongen, J. J. (2010). Genetic aberrations in paediatric acute leukaemias and implications for management of patients. The Lancet Oncology, 11, 880-889.

Takahashi, K., Inukai, T., Imamura, T., Yano, M., Tomoyasu, C., Lucas, D. M., . . . Koshi. (2017). Anti-leukemic activity of bortezomib and carfilzomib on B-cell precursor ALL cell lines. PLoS ONE, 12(12), e0188680.

Takemoto, D. J., Dunford, C., Vaughn, D., Kramer, K. J., Smith, A., & Powell, R. G. (1982). Guanylate cyclase activity in human leukemic and normal lymphocytes. Enzyme inhibition and cytotoxicity of plant extracts. Enzyme, 27(3), 179-188.

Tan, D., Phipps, C., Hwang, W. Y., Tan, S. Y., Yeap, C. H., Chan, Y. H., . . . Investigators, S. (2015). Panobinostat in combination with bortezomib in patients with relapsed or refractory peripheral T-cell lymphoma: an open-label, multicentre phase 2 trial. The Lancet Haematology, 2(8), e326-e333.

Tasian, S. K., Loh, M. L., & Hunger, S. P. (2017). Philadelphia chromosome–like acute lymphoblastic leukemia. Blood, 130(19), 2064-2072.

Te, H. S., Randall, G., & Jensen, D. M. (2007). Mechanism of Action of Ribavirin in the Treatment of Chronic Hepatitis C. Gastroenterology & Hepatology, 3(3), 218-225.

Teitell, M. A., & Pandolfi, P. P. (2009). Molecular Genetics of Acute Lymphoblastic Leukemia. The Annual Review of Pathology: Mechanisms of Disease, 4, 175-198.

Tesfai, Y., Ford, J., Carter, K. W., Firth, M. J., O'Leary, R. A., Gottardo, N. G., . . . Kees, U. R. (2012). Interactions between acute lymphoblastic leukemia and bone marrow stromal cells influence response to therapy. Leukemia Research, 36(3), 299- 306.

Thomas-Teinturier, C., Allodji, R. S., Svetlova, E., Frey, M. A., Oberlin, O., Millischer, A. E., . . . De Vathaire, F. (2015). Ovarian reserve after treatment with alkylating agents during childhood. Human reproduction, 30(6), 1437-1446.

260

Topisirovic, I., Guzman, M. L., McConnell, M. J., Licht, J. D., Culjkovic, B., Neering, S. J., . . . Borden, K. L. (2003). Aberrant eukaryotic translation initiation factor 4E- dependent mRNA transport impedes hematopoietic differentiation and contributes to leukemogenesis. Molecular and Cellular Biology, 23(24), 8992-9002.

Toscan, C. E., Failes, T., Arndt, G. M., & Lock, R. B. (2014). High-throughput screening of human leukemia xenografts to identify dexamethasone sensitizers. Journal of Biomolecular Screening, 19(10), 1391-1401.

Trachootham, D., Zhang, H., Zhang, W., Feng, L., Du, M., Zhou, Y., . . . Huang, P. (2008). Effective elimination of fludarabine-resistant CLL cells by PEITC through a redox-mediated mechanism. Blood, 112(5), 1912-1922.

Tseng, K. Y., Caballero, A., Dec, A., Cass, D. K., Simak, N., Sunu, E., . . . West, A. R. (2011). Inhibition of striatal soluble guanylyl cyclase-cGMP signaling reverses basal ganglia dysfunction and akinesia in experimental parkinsonism. PLoS One, 6(11), e27187.

Tuttle, T. R., Mierzwa, M. L., Wells, S. I., Fox, S. R., & Ben-Jonathan, N. (2016). The cyclic GMP/protein kinase G pathway as a therapeutic target in head and neck squamous cell carcinoma. Cancer Letters, 370, 279–285.

Udensi, U. K., & Tchounwou, P. B. (2014). Dual effect of oxidative stress on leukemia cancer induction and treatment. Journal of Experimental & Clinical Cancer Research, 33, 106.

Venton, G., Pérez-Alea, M., Baier, C., Fournet, G., Quash, G., Labiad, Y., . . . Costello, R. T. (2016). Aldehyde dehydrogenases inhibition eradicates leukemia stem cells while sparing normal progenitors. Blood Cancer Journal, 6(9), e469.

Vitale, C., Griggio, V., Todaro, M., Salvetti, C., Boccadoro, M., & Coscia, M. (2017). Magic pills: new oral drugs to treat chronic lymphocytic leukemia. Expert Opinion on Pharmacotherapy, 18(4), 411-425. von Stackelberg, A., Locatelli, F., Zugmaier, G., Handgretinger, R., M, T. T., Rizzari, C., . . . Gore, L. (2016). Phase I/Phase II Study of Blinatumomab in Pediatric Patients

261

With Relapsed/Refractory Acute Lymphoblastic Leukemia. Journal of Clinical Oncology, 34(36), 4381-4389.

Vrana, J. A., & Grant, S. (2001). Synergistic induction of apoptosis in human leukemia cells (U937) exposed to bryostatin 1 and the proteasome inhibitor lactacystin involves dysregulation of the PKC/MAPK cascade. Blood, 97(7), 2105-2114.

Vrooman, L. M., & Silverman, L. B. (2016). Treatment of Childhood Acute Lymphoblastic Leukemia: Prognostic Factors and Clinical Advances. Current Hematologic Malignancy Reports, 11, 385-394.

Walsh, N., Kenney, L., Jangalwe, S., Aryee, K.-E., Greiner, D. L., Brehm, M. A., & Shultz, L. D. (2017). Humanized mouse models of clinical disease. Annual Review of Pathology, 12, 187-215.

Wang, J., & Yi, J. (2008). Cancer cell killing via ROS: To increase or decrease, that is the question. Cancer Biology & Therapy, 7(12), 1875-1884.

Wang, W., Zhou, J., Zhao, L., & Chen, S. (2017). Combination of SL327 and Sunitinib Malate leads to an additive anti-cancer effect in doxorubicin resistant thyroid carcinoma cells. Biomedicine & Pharmacotherapy, 88, 985-990.

Wang, X., Chen, C. W., & Armstrong, S. A. (2016). The role of DOT1L in the maintenance of leukemia gene expression. Current Opinion in Genetics & Development, 36, 68-72.

Wartha, K., Herting, F., & Hasmann, M. (2014). Fit-for purpose use of mouse models to improve predictivity of cancer therapeutics evaluation. Pharmacology & Therapeutics, 142, 351-361.

Wattel, E., Preudhomme, C., Hecquet, B., Vanrumbeke, M., Quesnel, B., Dervite, I., Morel, P. & Fenaux, P. (1994). p53 Mutations Are Associated With Resistance to Chemotherapy and Short Survival in Hematologic Malignancies. Blood. 84(9), 3148- 3157.

262

Weir, S. J., DeGennaro, L. J., & Austin, C. P. (2012). Repurposing approved and abandoned drugs for the treatment and prevention of cancer through public-private partnership. Cancer Research, 72(5), 1055-1058.

Weisman, M. H., & Hannifin, D. M. (1979). Management of rheumatoid arthritis with oral gold. Arthritis & Rheumatology, 22(8), 922-925.

Wells, J., Jain, N., & Konopleva, M. (2017). Philadelphia Chromosome–Like Acute Lymphoblastic Leukemia: Progress in a New Cancer Subtype. Clinical Advances in Hematology & Oncology, 15(7), 554-561.

Wen, H. C., Chuu, C. P., Chen, C. Y., Shiah, S. G., Kung, H. J., King, K. L., . . . Chang, C. H. (2015). Elevation of soluble guanylate cyclase suppresses proliferation and survival of human breast cancer cells. PLoS One, 10(4), e0125518.

Williams, R. (2011). Discontinued drugs in 2010: oncology drugs. Expert Opinion on Investigational Drugs, 20(11), 1479-1496.

Winters, A. C., & Bernt, K. M. (2017). MLL-Rearranged Leukemias—An Update on Science and Clinical Approaches. Frontiers in Pediatrics, 5, 1-21.

Woo, J. S., Alberti, M. O., & Tirado, C. A. (2014). Childhood B-acute lymphoblastic leukemia: a genetic update. Experimental Hematology & Oncology, 3(16).

Wu, Z., Wang, Y., & Chen, L. (2013). Network-based drug repositioning. Molecular BioSystems, 9, 1268.

Xie, X.-J. (2010). Statistical considerations for high throughput screening data. Frontiers in Biology, 5(4), 354-360.

Yang, J., Zhao, X., Tang, M., Li, L., Lei, Y., Cheng, P., . . . Yuan, Z. (2017). The role of ROS and subsequent DNA-damage response in PUMA-induced apoptosis of ovarian cancer cells. Oncotarget, 8(14), 23492-23506.

Yang, Y., Mallampati, S., Sun, B., Zhang, J., Kim, S.-B., Lee, J.-S., . . . Sun, X. (2013). Wnt pathway contributes to the protection by bone marrow stromal cells of acute

263 lymphoblastic leukemia cells and is a potential therapeutic target. Cancer Letters, 333, 9-17.

Yang, Y., Zhang, K., Wang, Y., Li, M., Sun, X., Liang, Z., . . . Zhu, L. (2016). Disulfiram chelated with copper promotes apoptosis in human breast cancer cells by impairing the mitochondria functions. Scanning, 38(6), 825-836.

Yi, J., Gao, F., Shi, G., Li, H., Wang, Z., Shi, X., & Tang, X. (2002). The inherent cellular level of reactive oxygen species: one of the mechanisms determining apoptotic susceptibility of leukemic cells to arsenic trioxide. Apoptosis, 7(3), 209-215.

Yip, N. C., Fombon, I. S., Liu, P., Brown, S., Kannappan, V., Armesilla, A. L., . . . Wang, W. (2010). Disulfiram modulated ROS-MAPK and NFκB pathways and targeted breast cancer cells with cancer stem cell-like properties. British Journal of Cancer, 104(10), 1564-74.

Young, D. W. (2017). Considerations Related to Small-molecule Screening Collections. In J. A. Bittker, & N. T. Ross, High Throughput Screening Methods: Evolution and Refinement (pp. 16-36). Cambridge: The Royal Society of Chemistry.

Zang, R., Li, D., Tang, I.-C., Wang, J., & Yang, S.-T. (2012). Cell-Based Assays in High-Throughput Screening for Drug Discovery. International Journal of Biotechnology for Wellness Industries, 1(1), 31-51.

Zha, J., Chen, F., Dong, H., Shi, P., Yao, Y., Zhang, Y., . . . Xu, B. (2014). Disulfiram targeting lymphoid malignant cell lines via ROS-JNK activation as well as Nrf2 and NF-kB pathway inhibition. Journal of Translational Medicine, 12, 163.

Zhang, H., Chen, D., Ringler, J., Chen, W., Cui, Q. C., Ethier, S. P., . . . Wu, G. (2010). Disulfiram treatment facilitates phosphoinositide 3-kinase inhibition in human breast cancer cells in vitro and in vivo. Cancer Research, 70(10), 3996-4004.

Zhang, H., Mi, J. Q., Fang, H., Wang, Z., Wang, C., Wu, L., . . . Wang, K. K. (2013). Preferential eradication of acute myelogenous leukemia stem cells by fenretinide. Proceedings of the National Academy of Sciences, 110(14), 5606-5611.

264

Zhang, J., Chung, T. D. Y. & Oldenburg, K. R. (1999). A simple statistical parameter for use in evaluation and validation of high throughput screening assays. Journal of Biomolecular Screening, 4(2), 67-73.

Zhang, T., Chen, G., Wang, Z., Wang, Z., Chen, S., & Chen, Z. (2001). Arsenic trioxide, a therapeutic agent for APL. Oncogene, 20, 7146-7153.

Zhang, Y., Chen, A., Yan, X.-M., & Huang, G. (2012). Disordered epigenetic regulation in MLL-related leukemia. International Journal of Hematology, 96(4), 428- 437.

Zhao, H., Shi, P., Deng, M., Jiang, Z., Li, Y., Kannappan, V., . . . Xu, B. (2016). Low dose triptolide reverses chemoresistance in adult acute lymphoblastic leukemia cells via reactive oxygen species generation and DNA damage response disruption. Oncotarget, 7(51), 85515-85528.

Zhao, J., Wang, C., Song, Y., Liu, Y., & Fang, B. (2015). Treatment of refractory/relapsed adult acute lymphoblastic leukemia with bortezomib-based chemotherapy. International Journal of General Medicine, 8, 211-214.

Zhao, L., Wientjes, M. G., & Au, J. L.-S. (2004). Evaluation of Combination Chemotherapy: Integration of Nonlinear Regression, Curve Shift, Isobologram, and Combination Index Analyses. Clinical Cancer Research, 10, 7994-8004.

Zhao, W., Sachsenmeier, K., Zhang, L., Sult, E., Hollingsworth, R. E., & Yang, H. (2014). A New Bliss Independence Model to Analyze Drug Combination Data. Journal of Biomolecular Screening, 19(5), 817-821.

Zhou, F. L., Zhang, W. G., Wei, Y. C., Meng, S., Bai, G. G., Wang, B. Y., . . . Chen, S. P. (2010). Involvement of oxidative stress in the relapse of acute myeloid leukemia. The Journal of Biological Chemistry, 285(20), 15010-15015.

Zhou, G. B., Zhang, J., Wang, Z. Y., Chen, S. J., & Chen, Z. (2007). Treatment of acute promyelocytic leukaemia with all-trans retinoic acid and arsenic trioxide: a paradigm of synergistic molecular targeting therapy. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 362(1482), 959-971.

265

Zhou, Y., Hileman, E. O., Plunkett, W., Keating, M. J., & Huang, P. (2003). Free radical stress in chronic lymphocytic leukemia cells and its role in cellular sensitivity to ROS-generating anticancer agents. Blood, 101(10), 4098-4104.

Zhu, H.-H., Wu, D.-P., Jin, J., Li, J.-Y., Ma, J., Wang, J.-X., . . . Huang, X.-J. (2016). Long-term survival of acute promyelocytic leukaemia patients treated with arsenic and retinoic acid. British Journal of Haematology, 174(5), 820–822.

Zhu, J., Chen, Z., Lallemand-Breitenbach, V., & de The, H. (2002). How acute promyelocytic leukaemia revived arsenic. Nature Reviews Cancer, 2, 705-713.

Zou, P., Chen, M., Ji, J., Chen, W., Chen, X., Ying, S., . . . Liang, G. (2015). Auranofin induces apoptosis by ROS-mediated ER stress and mitochondrial dysfunction and displayed synergistic lethality with piperlongumine in gastric cancer. Oncotarget, 6(34), 36505–36521.

Zwaan, M., Söderhäll, S., Brethon, B., Luciani, M., Rizzari, C., Sternberg, D., . . . Pieters, R. (2015). A Phase 1/2, Open-Label, Dose-Escalation Study of Midostaurin in Pediatric Patients (Pts) with Relapsed or Refractory (R/R) Acute Leukemia: Final Results of Study ITCC-024 (CPKC412A2114). Blood, 126, 2564.

266