Immunomodulation and Vaccination with RHDV VLP

Braeden Donaldson

A thesis submitted for the degree of

Doctor of Philosophy

University of Otago, Dunedin, New Zealand

November 2016

Abstract

Colorectal cancer (CRC) is an insidious disease, responsible for almost 700,000 deaths annually worldwide. The incidence of CRC in New Zealand is more than double the world average, at 48.1 and 36.7 cases per 100,000 annually for men and women respectively, with the highest rates of distant metastases on presentation around the world at 24% of cases. The predominance of this late stage presentation may be attributed to poor public awareness, exacerbated by the current absence of the nationwide screening program that is due to begin in 2017. Combined with limited long-term treatment options, the prognosis for patients diagnosed with CRC in New Zealand is relatively poor. The availability of alternative treatment options, such as immunotherapy, may help to improve patient outcomes, promoting remission and recovery while limiting the development of adverse side effects. Immunotherapy involves the engagement and manipulation of the patients own immune system, re-educating the effector cells of the immune system to recognise, respond and eliminate tumour cells. One of the most promising types of immunotherapy is vaccination, capable of reinvigorating the immune system to recognise and target molecules associated with CRC.

This study used virus-like particles (VLP) derived from Rabbit hemorrhagic disease virus (RHDV) as a vaccine construct for the treatment of CRC. VLP are a form of subunit vaccine composed of components derived from a virus, but without the ability to infect or replicate. These inert shells can be modified to incorporate heterologous antigens, such as those derived from tumour cells, as well as additional molecules that can manipulate and modify the immune responses induced. In previous work undertaken with RHDV VLP, these molecules have included adjuvants such as α-galactosylceramide and deoxyribonucleic acid (DNA) oligonucleotides containing unmethylated CpG islands (CpGs), as well as compounds that promote uptake, such as mannose and surface conjugated CpGs. During this study, it was found that RHDV VLP have a natural affinity for CpGs, facilitating the co-delivery of this adjuvant into antigen-presenting cells (APCs) that uptake VLP. This property was identified based on the protection of CpG oligonucleotides from DNase digestion in the presence of RHDV VLP, and the further retention of CpGs following dialysis. The functionality of CpGs associated with RHDV VLP was investigated using the RAW-Blue reporter cell line, and confirmed in assays evaluating the activation of murine bone marrow-derived dendritic cells (BMDCs). CpGs delivered in association with RHDV VLP were found to induce enhanced activation of BMDCs in comparison to an equivalent quantity of CpGs administered alone.

This enhancement could not be explained by the direct effects of RHDV VLP on these cells, as VLP alone was found to be slightly suppressive relative to untreated controls.

RHDV VLP were further found to have a natural affinity for other short nucleic acids, such as short interfering RNA (siRNA). This affinity was identified following dialysis of siRNA laced RHDV VLP. The efficacy of siRNA delivery was investigated using signal transducer and activator of transduction (STAT) 3 siRNA, based upon its ability to knockdown STAT3 protein levels in a model cell line following administration in combination with RHDV VLP. The functionality of STAT3 siRNA delivery by RHDV VLP was further investigated based on the hypothesised ability to interfere with interleukin-6 (IL-6) secretion in activated murine macrophages. Although RHDV VLP was found to be capable of delivering STAT3 siRNA to murine macrophages, inducing a significant knockdown in STAT3 protein levels, this effect did not functionally translate into significant alterations in IL-6 secretion.

RHDV VLP-based vaccines developed for the treatment of cancer have previously included the incorporation of model antigens, limiting direct relevance of these vaccines to targeting true tumour-associated antigens. More recently, a study involving the incorporation of the melanoma-associated antigen gp100 proved that RHDV VLP-based vaccines can induce responses against tumour-associated antigens. However, this study also showed that targeting gp100 alone was insufficient for inducing responses that could induce complete remission of murine melanoma tumours, indicating that escape mechanisms remained intact. This may be overcome by inducing simultaneous immune responses against multiple tumour-associated antigens, potentially limited the ability of tumours to adapt and escape.

In order to test this hypothesis, RHDV VLP were produced that contained an epitope derived from murine topoisomerase IIα (T.VP60), an epitope derived from murine survivin (S.VP60) or both in combination (TS.VP60). The identity of these chimaeric VLP was confirmed through sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE), electron microscopy and mass spectrometry. Their ability to induce targeted immune responses against their respective epitopes was assessed with in vivo cytotoxicity assays. The efficacy of these chimaeric VLP constructs to combat subcutaneously engrafted MC38-OVA murine CRC tumours in combination with CpGs as an adjuvant was evaluated in therapeutic tumour trials, with RHDV VLP containing the model chicken ovalbumin epitope SIINFEKL (SIIN.VP60) used as a model control. Both T.VP60 and S.VP60 delayed the growth of MC38-OVA tumours with comparable efficacy to SIIN.VP60, attributable to 60% overall

2 survival amongst mice combined across all trials. TS.VP60 exhibited further delays in the growth of MC38-OVA tumours, attributable to 67% overall survival amongst mice combined across all trials. Although this trend for enhanced efficacy with vaccination targeting multiple tumour-epitopes was repeatable between trials, the difference between mono- and multi-target therapies was not statistically significant. All mice that survived following vaccination were rechallenged with a subcutaneous MC38-OVA tumour on the opposing flank at day 100, and by day 200 none of these mice had grown tumours. Although these results are promising, combinations of greater than two target antigens may be required to provide significant benefits with a multi-target vaccine.

Finally, RHDV VLP was assessed for its ability to be expressed by the oncolytic virus Vesicular stomatitis virus (VSV). The purpose of this investigation was to facilitate the development of a combination vaccine capable of utilising the targeted tumour cell-specific lytic infection of an oncolytic virus, and the therapeutic vaccination properties of RHDV VLP. To this end, recombinant VSV expressing RHDV VP60 (VSV-VP60), SIIN.VP60 (VSV-SIIN.VP60) and TS.VP60 (VSV-TS.VP60) were produced. The viability of recombinant VSV was assessed by plaque assay, the incorporation of each VLP construct was confirmed through sequencing, and the production of each VLP in baby hamster kidney (BHK) cells was identified by Western blot and electron microscopy. In MC38-OVA tumour trials, none of the VLP expressing VSV constructs provided significant benefits in either growth rate or survival. The absence of the usual oncolytic properties of VSV may be due to impaired or altered activity in the MC38-OVA murine CRC model system, or through impairment in combination with RHDV VLP expression. The apparent inefficacy of the VLP vaccines expressed by each VSV construct may be due to inadequate dosage, with in vivo expression from the administered quantity of recombinant VSV possibly incapable of recapitulating levels known to be required for successful RHDV VLP vaccination. These results indicate that although VSV is able to express RHDV VLP, further work is required to facilitate the development of this potential combination therapy.

Acknowledgments

Although I rarely find myself at a deficit in the expression of my thoughts and opinions, there are few words that I feel can adequately convey the appreciation I have for each person that has aided me along this path. It has been a long journey, and this thesis represents the culmination of my work under the guidance of so many people.

To my loving wife Catherine, thank you for your support throughout my undergraduate, postgraduate and PhD studies. You have been my rock throughout this process, providing a stable platform on which to keep my feet grounded. Thank you for your patience, especially during the six months I spent overseas. I thought of you every waking moment while I was away, and counted down the days until our reunion. To my family, especially my parents and sister, thank you so much for your support throughout my studies. It was always comforting to know that there was a warm, welcoming home available should I need it. I appreciate everything that you have done for me, even throughout the trials of illness and infirmity you have each endured.

Throughout my PhD studies, I was lucky to receive primary supervision from Sarah Young and Vernon Ward. I couldn’t have wished for a better combination, as together they form a dream team for both research and support. Sarah, thank you for your patience throughout this process, your positivity, optimism and support were greatly appreciated. You have done so much for me throughout my studies and beyond, everything that I have academically I owe to you. I can only hope that someday I may be able to repay you through my continued research and teaching. Vernon, thank you for your dedication, guidance and for providing the critical eye that shaped this work. Without you, I would have been lost amongst the multiple facets of this project. You, along with Sarah, have helped me to appreciate the bigger picture, and to prioritize the important components of my research. Thank you both for helping me through this process, through better and worse, and out the other side.

Thank you to the Department of Pathology, its staff and students, and especially to the Young group past and present. Sara, Helen, Melanie, Francesco, Estelle, Katrin, Yasmin, Nick, Silke, thank you for your support and friendship throughout my PhD. Simon, Kunyu, and Katie, thank you for helping with the mouse handling throughout my project. Without your help, I would need to add many more sleepless nights to my list. Together you all created a fantastic working environment, and I am so glad that I had the opportunity to work with such a great group of people.

Thank you also to the Department of Microbiology and Immunology, its staff and students, and in particular the Ward group past and present. Jo, Emily, Colin, Sarah S., Tim, Holly, Kiri, Prasanth, Alice, thank you for helping me throughout my PhD, and for all the laughs we’ve shared in the office together. Zabeen, thank you for your help with mouse handling during my PhD, and for providing a daily dose of just the right proportions of enthusiasm and madness. Farah, thank you for always being there when I needed advice and support. You brightened each day, and your positivity has been sorely missed since you left. Vivienne, you are the heart and soul of the Ward group. Thank you so much for helping with, quite literally, everything I did during my time in Vernon’s laboratory. We would all be lost without you, thank you so much for everything you do. Together, you all formed a warm, inviting environment. It was a pleasure to work alongside each of you.

The University of Otago is one of the best places in the world to study and work, with so many wonderful people spread across its campus. I would like to specifically thank Margaret Baird for her advisory role. I could not have asked for a more widely respected, knowledgeable, and kind advisor, and I regret that her passing came before the completion of this work. I would also like to thank Richard Easingwood for assisting me with the electron microscopy undertaken during this PhD, Greg Walker for his advice and for providing me access to the Department of Pharmacy, and Sigurd Wilbanks for helping me analyse and evaluate some of the biochemical components of my project. Thank you all for your support, and to all the staff and students past and present throughout the University that have guided or assisted me along the way.

During my time at the Mayo Clinic, I was brought under the tutelage of another great supervisor, Richard Vile. Although I was initially unsure in this new, foreign environment, Richard made me feel welcome in his laboratory. A perfect British gentleman, Richard was infallibly kind and caring, and provided a nurturing environment that encouraged both independent and collaborative research. Richard, thank you for providing me with the tools and support I needed to become a better researcher. My time in your laboratory is something that I will always cherish and remember fondly.

Thank you to each of the members of the Vile group for your help and support during my time at the Mayo Clinic. Memy, thank you for organising the animal ethics approval for me, and for inviting me into your home. Thanks giving was a fantastic evening, and I was so glad to spend it with you along with others from our lab and neighbouring labs. Tim, thank you

5 for tolerating my intrusion into the laboratory, and for everything that you did for me during my time there. Your unique blend of professionalism, military-grade precision and experience was both refreshing and inspiring. Jill, thank you very much for your help with mouse handling throughout my time at the Mayo Clinic, and for continuing the final trial after I had left. I always enjoyed our conversations while we administered vaccines, and I hope that your family are all doing well. Karishma, Christian, Kevin, Julia, Crystal, thank you for your friendship and support throughout my time at the Mayo Clinic. I will always remember our movie nights together, and the atrocious movies we watched. Shane, thank you so much for everything. Our ‘gentleman’s club’ meetings at the cafeteria each day were the highlight of my time in Rochester. I wish you all the very best, and I hope that I get to see each of you again someday.

Thank you to the Freemasons of Otago/Southland for awarding me an Oncology Fellowship Scholarship to undertake my PhD. John Dennison, John Steele, Les Brenssell and Alistair Cowan, thank you for all your support. Thank you to the Health Research Council, the University of Otago and the Department of Pathology for providing funding that contributed to the work undertaken during my PhD. Thank you to the Otago School of Medical Sciences, the Dunedin School of Medicine, the Maurice and Phyllis Paykel Trust, the Australasian Society of Immunology and the Centre for Translational Cancer Research for funding my travel for attendance at conferences to present my work. I would also like to thank Fulbright New Zealand, the Mayo Clinic and its staff and students for supporting my exchange during my PhD. This experience proved invaluable, and I thank you all for giving me the opportunity to represent New Zealand abroad.

Table of Contents Abstract ...... 0 Acknowledgments ...... 4 Table of Contents ...... 7 List of Tables ...... 12 List of Figures ...... 13 List of Abbreviations ...... 15 1 Introduction...... 22 1.1 Cancer...... 22 1.1.1 Neoplasia ...... 23 1.1.2 Tumourigenesis ...... 28 1.2 The Immune Response to Tumourigenesis ...... 34 1.2.1 Innate Immune Response ...... 35 1.2.3 Adaptive Immune Response ...... 43 1.3 Colorectal Cancer ...... 54 1.3.1 Familial Colorectal Cancer ...... 54 1.3.2 Spontaneous Mutagenesis ...... 55 1.3.3 Additional Risk Factors ...... 58 1.3.4 Angiogenesis in Colorectal Cancer ...... 60 1.3.5 Colorectal Cancer Staging ...... 61 1.3.6 Approved Treatment ...... 63 1.3.7 Monoclonal Antibodies in Clinical Trials for Colorectal Cancer ...... 65 1.4 Therapeutic Vaccination ...... 69 1.4.1 Virus-like Particles ...... 70 1.5 Project Objectives ...... 76 1.5.1 Hypotheses ...... 77 2 Materials and Methods ...... 78 2.1 Development of Chimaeric RHDV VLP ...... 78 2.1.1 Permits and Approvals ...... 78 2.1.2 Construction of Expression Plasmids ...... 79 2.1.3 Production of Recombinant Baculovirus ...... 86 2.1.4 Production and Purification of RHDV VLP ...... 88 2.1.5 Analysis of RHDV VLP ...... 89 2.2 Association of Nucleic Acids with RHDV VLP...... 91

2.2.1 Detection of Endogenous Nucleic Acids in RHDV VLP ...... 91 2.2.2 Disassembly and Reassembly of RHDV VLP ...... 91 2.2.3 Loading/Association of CpGs with RHDV VLP ...... 93 2.2.4 RAW-Blue SEAP/NF-κB Reporter Cell Line ...... 94 2.2.5 Murine Bone Marrow-derived Dendritic Cells ...... 95 2.2.6 Association of siRNA with RHDV VLP ...... 98 2.3 In Vivo Assays and Trials ...... 99 2.3.1 Ethical Approvals ...... 99 2.3.2 In Vivo Cytotoxicity Assay ...... 100 2.3.3 In Vitro (VITAL) Cytotoxicity Assay ...... 102 2.3.4 MC38-OVA...... 104 2.4 Recombinant Vesicular Stomatitis Virus ...... 107 2.4.1 Construction of Expression Plasmids ...... 107 2.4.2 Production of Recombinant VSV ...... 110 2.5 Statistical Analyses ...... 113 3 Nucleic Acids Associated with RHDV VLP for Immunomodulation...... 114 3.1 Introduction ...... 114 3.1.1 Nucleic Acids and VLP ...... 114 3.1.2 Immunomodulation with VLP Vaccines ...... 116 3.2 Objectives...... 118 3.2.1 Aims...... 118 3.3 Results ...... 119 3.3.1 Production of Chimaeric RHDV VLP Designed for Nucleic Acid Association ...... 119 3.3.2 Endogenous Nucleic Acids in RHDV VLP ...... 120 3.3.3 Disassembly and Reassembly of RHDV VLP ...... 122 3.3.4 Qualification and Detection of RHDV VLP Disassembly ...... 123 3.3.5 Encapsulation of CpGs into RHDV VLP ...... 127 3.3.5 CpGs Associated with RHDV VLP Enhances NF-κB Activation ...... 129 3.3.6 CpGs Associated with RHDV VLP Enhances BMDC Activation ...... 130 3.3.7 Association of siRNA with RHDV VLP ...... 132 3.3.8 Knockdown of Stat3 with siRNA associated with RHDV VLP ...... 133 3.4 Discussion ...... 135 3.4.1 Nucleic Acids Associated with RHDV VLP ...... 135 3.4.2 CpGs associated with RHDV VLP ...... 136

3.4.3 siRNA associated with RHDV VLP ...... 137 3.4.3 Conclusions ...... 138 4 Development of a Virus-based Vaccine for Colorectal Cancer...... 139 4.1 Introduction ...... 139 4.1.1 Incorporating Multiple Colorectal Cancer Epitopes into RHDV VLP ...... 139 4.1.2 Prime-Boost and Primary-Recurrent Vaccination ...... 141 4.1.3 Colorectal Cancer-associated Epitopes ...... 142 4.1.4 Spacer and Linker Sequences used in Chimaeric RHDV VLP ...... 144 4.2 Objectives ...... 146 4.2.1 Aims ...... 146 4.3 Results ...... 147 4.3.1 Chimaeric RHDV VLP containing Murine CRC Epitopes ...... 147 4.3.2 Production of Chimaeric RHDV VLP ...... 149 4.3.3 Simultaneous N- and C-terminus Insertion into RHDV VP60 ...... 152 4.3.4 Vaccination with chimaeric RHDV VLP ...... 155 4.3.5 Vaccination with Chimaeric RHDV VLP using a Multi-dosage Regimen ...... 157 4.3.6 Kinetics of MC38-OVA Tumours in B57BL/6 Mice ...... 160 4.3.7 Chimaeric RHDV VLP as an Immunotherapy for CRC ...... 163 4.3.8 Rechallenge of Tumour-free Mice...... 167 4.3.9 Effect of Pre-existing Immunity to RHDV VLP ...... 168 4.3.10 Expression of RHDV VLP by Vesicular Stomatitis Virus ...... 170 4.3.11 VSV Expressing RHDV VLP as an Immunotherapy for CRC ...... 173 4.4 Discussion ...... 176 4.4.1 Chimaeric RHDV VLP Vaccine for Colorectal Cancer ...... 176 4.4.2 Targeting Multi-epitopes with a Chimaeric VLP Vaccine ...... 177 4.4.3 Combination Vaccine Therapies ...... 178 4.4.4 Conclusions ...... 179 5 Future Directions and Conclusion ...... 181 5.1 Delivery of Molecules Associated with RHDV VLP ...... 181 5.1.1 Adjuvant Delivery ...... 181 5.1.2 Manipulation of Gene Expression ...... 182 5.2 RHDV VLP Vaccines for Cancer ...... 188 5.2.1 Chimaeric RHDV VLP ...... 188 5.2.2 Conjugated RHDV VLP ...... 190

5.2.3 Personalised RHDV VLP Vaccines for Cancer ...... 191 5.2.4 Combination Vaccination ...... 193 5.3 Conclusion ...... 194 6 References...... 195 7 Appendix I: Supplementary Figures ...... 239 8 Appendix I: Antibody Compendium ...... 241 9 Appendix II: Recipes ...... 242 9.1 Assorted PBS Solutions ...... 242 9.1.1 10X PBS ...... 242 9.1.3 iPBS...... 242 9.1.4 cPBS ...... 242 9.2 Transformation Reagents...... 242 9.2.1 LB Broth and Agar ...... 242 9.2.2 Transformation Buffer I ...... 243 9.2.3 Transformation Buffer II ...... 243 9.3 Electrophoresis Reagents ...... 243 9.3.1 50X TAE Buffer ...... 243 9.3.2 Agarose Gels...... 243 9.3.3 2X Sample Buffer ...... 243 9.3.4 SDS-PAGE Stacking Gel Buffer ...... 244 9.3.5 SDS-PAGE Resolving Gel Buffer ...... 244 9.3.6 SDS-PAGE Gel Production ...... 244 9.3.7 10X SDS-PAGE Electrophoresis Buffer ...... 244 9.3.8 Coomassie Blue Stain ...... 245 9.3.9 Destain Solution ...... 245 9.4 Western Blot Reagents ...... 245 9.4.1 Anode Buffer I ...... 245 9.4.2 Anode Buffer II ...... 245 9.4.3 Cathode Buffer ...... 245 9.4.4 PBS-T Wash Buffer ...... 245 9.5 Cell Culture Reagents ...... 246 9.5.1 Trypan Blue ...... 246 9.5.2 SeaPlaque Agarose ...... 246 9.5.3 cIMDM ...... 246

9.6 VLP-specific Reagents ...... 246 9.6.1 CsCl Solutions ...... 246 9.6.2 Disassembly Buffer ...... 246 9.7 Flow Cytometry Reagents ...... 247 9.7.1 Red Cell Lysis Buffer ...... 247 9.7.2 FACS Buffer ...... 247 9.7.3 AutoMACS Buffer ...... 247 9.7.4 AutoMACS Running Buffer ...... 247 10 Appendix III: Presentations and Publications ...... 248 10.1 Presentations ...... 248 10.2 Publications ...... 249

List of Tables Table 2.1 Summary of Chimaeric RHDV VLP Constructs ...... 79 Table 2.2 Pre-synthesised Recombinant VP60 N-terminus Sequences ...... 80 Table 2.3 Expression Plasmid Sequencing Primers ...... 86 Table 4.1 Linker Comparison for TopIIα Processing ...... 148 Table 4.2 Linker Comparison for Survivin Processing ...... 148

List of Figures Figure 1.1 Development of Colorectal Carcinoma through Accumulation of Mutations ...... 25 Figure 1.2 Neoplastic Transformation of Cervical Squamous Epithelium ...... 27 Figure 1.3 Comparisons of Hypothesised Mechanisms of Tumourigenesis ...... 33 Figure 1.4 The Immunoediting Hypothesis ...... 35 Figure 1.5 Cross-presentation Pathways ...... 45 Figure 1.6 Subsets of CD4+ T Cells ...... 48 Figure 1.7 Signalling Pathways Commonly Mutated in Colorectal Cancer ...... 57 Figure 1.8 The Immunoscore for Staging of Colorectal Cancer ...... 63 Figure 1.9 Structural conformation repertoire of VLP ...... 72 Figure 2.1 Development of Recombinant VP60 Expression Plasmids ...... 85 Figure 2.2 Flow Cytometry Gating Strategy ...... 97 Figure 2.3 In Vivo Cytotoxicity Assay ...... 102 Figure 2.4 In Vitro Cytotoxicity Assay ...... 104 Figure 2.5 Stucture of the pUC57-Simple.VSV-VLP3X Plasmid ...... 108 Figure 3.1 Production of VP60, R8.VP60 and DBS.VP60 ...... 120 Figure 3.2 Endogenous Nucleic Acids in VP60 and R8.VP60 VLP ...... 121 Figure 3.3 Disassembly and Reassembly of RHDV VLP ...... 123 Figure 3.4 Quantification of RHDV VLP Disassembly ...... 124 Figure 3.5 Detection of RHDV VLP Disassembly and Reassembly ...... 126 Figure 3.6 Encapsulation of CpGs into RHDV VLP ...... 128 Figure 3.7 Quantification of CpG Associated with RHDV VLP ...... 129 Figure 3.8 NF-κB Activation in RAW-blue Cells ...... 130 Figure 3.9 BMDC Activation with CpGs Associated with RHDV VLP ...... 131 Figure 3.10 Association of siRNA with RHDV VLP ...... 133 Figure 3.11 Knockdown of STAT3 by siRNA associated with RHDV VLP ...... 134 Figure 3.12 Association Between BKV VLP and a Linearised Plasmid ...... 136 Figure 4.1 CRC Epitope RHDV VP60 Constructs ...... 147 Figure 4.2 Production of T.VP60, S.VP60 and TS.VP60 Expression Plasmids ...... 150 Figure 4.3 Production and Purification of T.VP60, S.VP60 and TS.VP60 VLP ...... 151 Figure 4.4 Mass Spectrometric Analysis of T.VP60, S.VP60 and TS.VP60 ...... 152 Figure 4.5 Production of S.VP60.T and TS.VP60.T Expression Plasmids ...... 153 Figure 4.6 Production and Purification of S.VP60.T and TS.VP60.T VLP ...... 154 Figure 4.7 In vivo Cytotoxicity Assay ...... 156 Figure 4.8 Multi-dosage Regimens for Enhancing Epitope-specific Cytotoxicity ...... 159 Figure 4.9 MC38-OVA Engraftment and Growth Rate Kinetics...... 161 Figure 4.10 Confirmation TopIIα, Survivin and Ovalbumin Expression in MC38-OVA ...... 163 Figure 4.11 Efficacy of VLP constructs against engrafted MC38-OVA tumours ...... 165 Figure 4.12 Individual tumour growth curves ...... 166 Figure 4.13 Rechallenge with MC38-OVA ...... 167 Figure 4.14 Prime-boost Vaccination with RHDV VLP ...... 169 Figure 4.15 Structure of Vescicular Stomatitis Virus (VSV) ...... 171 Figure 4.16 Adaptation of RHDV VP60 Constructs for the VSV Expression system ...... 172 Figure 4.17 Detection of RHDV VP60 VLP Produced by Recombinant VSV ...... 173 Figure 4.18 Cytopathic effect of recombinant VSV ...... 174

Figure 4.19 Screening VSV constructs for efficacy against MC38-OVA tumours ...... 175 Figure 5.1 Modelling Predictions of VP60, R8.VP60 and DBS.VP60...... 184 Figure 5.2 Charge Maps of Proposed Recombinant VP60 for Enhanced siRNA Association ...... 185 Figure 5.3 Predicted 3D Models of PRM.VP60 and DPS.VP60 ...... 186 Figure 5.4 Proposed VP60 Construct for Long Peptide Insertion ...... 189 Figure 5.5 Proposed Peptide Conjugation Constructs ...... 190 Figure 5.6 Establishment of a Tumour-associated Peptide Library ...... 192 Figure 7.1 Chimaeric VLP Sequencing Data ...... 239 Figure 7.2 Mass Spectrometry Data...... 240

List of Abbreviations

5-FU – 5-Fluorouracil BRAF – Serine/Threonine-protein Kinase A2aR – Adenosine A2A Receptor B-Raf AAY – Alanine Alanine Tyrosine BRCA – Breast Cancer

AcMNPV – Autographa californica Breg – B Regulatory Cell Multicapsid Nucleopolyhedrovirus Btag – Bluetongue Virus Capsid Protein ADAM17 – A Distintegrin and VP7 Metalloproteinase Domain 17 BTLA – B- and T-Lymphocyte ADCC – Antibody-Dependent Cellular Attenuator Cytotoxicity BTN - Butyrophilin

AEC – Animal Ethics Committee CaCl2 – Calcium Chloride AIM2 – Absent in Melanoma 2 Cas – CRISPR-associated Protein AJCC – American Joint Committee on CCL – Chemokine (C-C Motif) Ligand Cancer CCMV – Cowpea Chlorotic Mottle Virus ALL – Alanine Leucine Leucine CD – Cluster of Differentiation Alum – Hydrated Aluminium Sulfate CDC – Complement-dependent APC – Adenomatous Polyposis Coli Cytotoxicity APCs – Antigen Presenting Cells CDCA1 – Cadmium-specific Carbonic APS – Ammonium Persulphate Anhydrase 1 ATCC – American Type Culture cDCs – Conventional Dendritic Cells Collection CEA – Carcinoembryonic Antigen ATP – Adenosine Triphosphate CFSE – Carboxyfluorescein AXL – Tyrosine-protein Kinase Receptor Succinimidyl Ester UFO Precursor cIMDM – Complete Iscove’s Modified B16-OVA – B16 Murine Melanoma Dulbecco’s Media expressing Ovalbumin CIMP – CpG Island Methylator BCR – B Cell Receptor Phenotype BER – Base Excision Repair CIN – Cervical Intraepithelial Neoplasia BHK – Baby Hamster Kidney CLIP – Class II-associated Invariant BKV – BK Virus Chain Protein BMDCs – Bone Marrow-derived CML – Chronic Myelogenous Leukaemia Dendritic Cells COX2 – Cyclooxygenase 2 BOK – Book of Knowledge

15 cPBS – Coupling Phosphate Buffered DTT - Dithiothreitol Saline EBNA – Epstein-Barr Virus Nuclear CPE – Cytopathic Effect Antigen CpG – Unmethylated CpG Island EBV – Epstein-Barr Virus CRC – Colorectal Cancer EGFR – Epithelial Growth Factor CRISPR – Clustered Regulatory Receptor Interspaced Short Palindomic Repeats EGTA – Ethylene Glycol Tetraacetic CsCl – Caesium Chloride Acid CSF-1 – Colony Stimulating Factor 1 ELISA – Enzyme-linked Immunosorbent CT – Core Tumour Assay CTLA – Cytotoxic T-Lymphocyte- EMT – Epithelial-to-Mesenchymal associated Protein Transition CXC – Cysteine-X-Cysteine ER – Endoplasmic Reticulum DAMPs – Damage-associated Molecular ERK – Extracellular Receptor Kinase Patterns ERMA – Environmental Risk DBS.VP60 – N-terminus HPV-16 DNA- Management Authority binding Site Recombinant VP60 ESCC – Esophageal Squamous Cell DCC – Deleted in Colorectal Cancer Carcinoma DC-SIGN – Dendritic Cell-specific ESO – Esophageal Cancer Antigen Intercellular Adhesion Molecule-3- FACS – Fluorescence-activated Cell grabbing Non-Integrin Sorting ddH2O – Double Distilled Water FAP – Familial Adenomatous Polyposis DFS – Disease-free Survival FAS – Fas Cell Surface Death Receptor dH2O – Distilled Water FBS – Fetal Bovine Serum DLS – Dynamic Light Scattering Fc – Fragment Crystallisable DMEM – Dulbecco’s Modified Eagle FCS – Fetal Calf Serum Media FDA – Food and Drug Administration DMSO - Dimethylsulfoxide FHV – Feline Herpesvirus DNA – Deoxyribonucleic Acid FOLFOX – 5-FU, Leucovorin and DPBS – Dulbecco’s Phosphate Buffered Oxaliplatin Saline FOX – Forkhead Box DPS – DNA-binding Proteins from GATA – GATA-binding Factor Starved Cells GBM – Glioblastoma Multiforme DSS – Disease-specific Survival

GD3 – Ganglioside Precursor IFN - Interferon Disialohematoside 3 Ig - Immunoglobulin GGS – Glycine Glycine Serine IGF – Insulin-like Growth Factor GITR – Glucocorticoid-induced TNFR- IGF1R – IGF1 Receptor related Protein IgSF – Immunoglobulin Superfamily GM-CSF – Granulocyte Macrophage IL - Interleukin Colony Stimulating Factor IM – Invasive Margin Gr1 – GALEX Release 1 IMP-3 – IGF-II mRNA Binding Protein 3 GTP – Guanine-5’-Triphosphate iNKT – Type I Natural Killer T Cell HBV – Hepatitis B Virus iNOS – Inducible Nitric Oxide Synthase HCAs – Heterocyclic Amines IPCS – International Program on HCC – Human Hepatocellular Carcinoma Chemical Safety HER2 – Human Epithelial Growth Factor IR – Infra-red Receptor 2 IRS1 – Insulin Receptor Substrate 1

HIF1α – Hypoxia-inducible Factor 1α iTreg – Induced T Regulatory HLA – Human Leukocyte Antigen JAK – Janus Kinase HMGB1 – High Mobility Group Box-1 JCV – John Cunningham Virus hMLH – Human MutL Homolog JECFA – Joint FAO/WHO Expert HNPCC – Hereditary Non-polyposis Committee on Food Additives

Colon Cancer KCH3COO – Potassium Acetate HNSCC – Head and Neck Squamous KCl – Potassium Chloride

Cell Cancer KH2PO4 – Potassium Phosphate

HPV – Human Papillomavirus KHCO3 – Potassium Carbonate HSPs – Heat Shock Proteins KIRs – Killer-cell Immunoglobulin-like HSV – Herpes Simplex Virus Receptors HuNoV – Human Norovirus KOH – Potassium Hydroxide IBDV – Infectious Bursal Disease Virus KRAS – Kirsten Rat Sarcoma Viral IBSC – Institutional Biological Safety Oncogene Homolog Committee KRT – Keratin Cytoskeletal ICAM – Intercellular Adhesion Molecule L/D – Live/Dead ID – Inhibitor of DNA Binding LAIR – Leukocyte-associated IDO – Indoleamine 2,3-Dioxygenase Immunoglobulin-like Receptor IEDB – Immune Epitope Database LB – Luria Broth IFA – Incomplete Freund’s Adjuvant

LCMV – Lymphocytic M-MDSCs – Monocytic Myeloid-derived Chorionmeningitis Virus Suppressor Cells LCR – Non-coding Long Control Region MMP – Matrix Metalloproteinase LFA – Lymphocyte Function-associated MMR – Mismatch Repear Antigen MMTV – Wingless-type Mouse LGP2 – Laboratory of Genetics and Mammary Tumour Virus Physiology 2 MOI – Multiplicity of Infection LIRs – Leukocyte Inhibitory Receptors MoMLV – Moloney Murine Leukemia LL-LCMV – Lewis’ Lung Carcinoma Virus LMP1 – Epstein-Barr Virus Latent MOPS – 3-(N- Membrane Protein 1 Morpholino)propanesulfonic Acid LPS - Lipopolysaccharide MPL-A – Monophosphoryl Lipid A LY6K – Lymphocyte Antigen 6 Complex MR1 – MHC-related Protein 1 Locus K mRNA – Messenger Ribonucleic Acid mAb – Monoclonal Antibody MS2 – Bacteriophage MS2 Coat Protein MAGE – Melanoma Antigen MUC1 – Mucin 1 MAIT – Mucosal Associated Invariant T MVA-T7 – Replication Defective Cells Vaccinia (T7 RNA Polymerase) MAPK – Mitogen-activated Protein MVD – Microvessel Density

Kinase Na2HPO4 – Disodium Hydrogen MDA5 – Melanoma Differentiation- Phosphate associated Gene 5 NaCl – Sodium Chloride

MDSCs – Myeloid-derived Suppressor NaH2PO4.2H2O – Sodium Dihydogen Cells Phosphate Dehydrate MET – Hepatocyte Growth Factor NAT – N-Acetyltransferase Receptor Precursor NCBI – National Center for MFI – Median Fluorescence Intensity Biotechnology

MgCl2 – Magnesium Chloride NFAT – Nuclear Factor of Activated T MHC – Major Histocompatibility Cells Complex NF-κB – Nuclear Factor-kappa-B miRNA – Micro Ribonucleic Acid NH4Cl – Ammonium Chloride MITF – Microphthalmia-associated NK – Natural Killer Transcription Factor NKT – Natural Killer T NLRs – Nod-like Receptors

NOS2 – Nitric Oxide Synthase 2 PTEN – Phosphatase and Tensin NRP - Neuropilin Homologue NSCLC – Non-Small Cell Lung Cancer Qβ – Bacteriophage Qβ Capsid Protein nTreg – Natural T Regulatory R8.VP60 – N-terminus Octaarginine OD – Optical Density Recombinant VP60 ORFs – Open Reading Frames RAG2 – Recombination Activating Gene OS – Overall Survival 2 OTI – Ovalbumin SIINFEKL-specific RAGs – Recombination Activating Genes Transgenic Mice RbCl – Rubidium Chloride PAGE – Polyacrylamide Gel RB-DMEM – RAW-Blue Dulbecco’s Electrophoresis Modified Eagle Media PAHs – Polycyclic Aromatic RBPJ – Recombinant Binding Protein Hydrocarbons Suppressor of Hairless PAMPs – Pattern-associated Molecular RHDV – Rabbit Hemorrhagic Disease Patterns Virus PAP – Prostatic Acid Phosphatase RIG-1 – Retinoic Acid-inducible Gene 1 PBS – Phosphate Buffered Saline RLRs – RIG-1-like Receptors PCPS – Proteasome Cleavage Prediction RNA – Ribonucleic Acid Server ROR – RAR-related Orphan Receptor PD – Programmed Cell Death Protein ROSs – Reactive Oxygen Species pDCs – Plasmacytoid Dendritic Cells S.VP60 – N-terminus Survivin PDCs – Poorly Differentiated Clusters Recombinant VP60 PDGF – Platelet-derived Growth Factor S.VP60.T – N-terminus Survivin, C- PDL – Programmed Cell Death Ligand terminus Topoisomerase IIα Recombinant PFA - Paraformaldehyde VP60 PI3K – Phosphoinositide 3-Kinase SCF – Scientific Committee on Food PLAU – Plasminogen Activator SCFAs – Short Chain Fatty Acids Urokinase SDS – Sodium Dodecyl Sulfate PMN - Polymorphonuclear SEAP – Secreted Alkaline Phosphate PMN-MDSCs – Polymorphonuclear Sf21 – Spodoptera frugiperda 21 Myeloid Derived Suppressor Cells Sf900-III – Spodoptera frugiperda 900- PP2A – Protein Phosphatase 2 III PRRs – Pattern Recognition Receptors

SHP2 – Src Homology Region 2- TEMED - Tetramethylethylenediamine containing Protein Tyrosine Phosphatase TEX – T Exhausted Cell 2 TGF – Transforming Growth Factor shRNA – Short Hairpin Ribonucleic Acid TH – T Helper Cell SIIN.VP60 – N-terminus SIINFEKL TILs – Tumour Infiltrating Lymphocytes Recombinant VP60 TIM – T Cell Apoptosis through Binding siRNA – Small Interfering Ribonucleic to Mucin Domain-containing Protein Acid TK – Thymidine Kinase SOC – Super Optimal Broth with TKK – TKK Protein Kinase Catabolite Repression TLRs – Toll-like Receptors SOCS – Suppressor of Cytokine TNF – Tumour Necrosis Factor Signalling TNFRSF – Tumour Necrosis Factor SP – Specific Binding Protein Receptor Superfamily Spl – Splenocyte TNM – Tumor Node Metastasis SSL – Serine Serine Leucine TopIIα – Topoisomerase IIα STAT – Signal Transducer and Activator TRAIL – TNF-related Apoptosis of Transcription Inducing Ligand STNV – Satellite Tobacco Necrosis Virus TRAILR – TRAIL Receptor

SV – Sample Volume Treg – T Regulatory Cell

T.VP60 – N-terminus Topoisomerase IIα TRM – T Tissue-resident Memory Recombinant VP60 tRNA – Transfer Ribonucleic Acid TAAs – Tumour-associated Antigens TS.VP60 – N-terminus Topoisomerase TAE – Tris Acetate EDTA IIα and Survivin Recombinant VP60 TAMs – Tissue-associated Macrophages TS.VP60.T – N-terminus Topoisomerase TAP – Transporter Associated with IIα and Survivin and C-terminus Antigen processing Topoisomerase IIα Recombinant VP60

TC – Cytotoxic T cell TSAs – Tumour-specific Antigens TCF – T Cell Factor UICC – Union for International Cancer

TCM – T Central Memory Cell Control TCR – T Cell Receptor US – United States

Teff – T Effector Cell UT - Untreated

TEM – T Effector Memory Cell VEGF – Vascular Endothelial Growth TEM – Transmission Electron Factor Microscopy VEGFR – VEGF Receptor

VITAL – In Vitro Cytotoxicity Assay VSV-TS.VP60 – TS.VP60 Recombinant VLP – Virus-like Particle VSV VP60.T – C-terminus Recombinant VP60 VSV-VP60 – VP60 Recombinant VSV VPD – Violet Proliferation Dye WHO – World Health Organisation VSV – Vesicular Stomatitis Virus Wnt – Wingless-type MMTV Integration VSV-SIIN.VP60 – SIIN.VP60 Site Recombinant VSV αFP – α-Fetoprotein

1 Introduction

1.1 Cancer Cancer is a disease caused by the growth of a malignant neoplasm or tumour amongst normal tissue in any of a variety of organ systems, distinguishable from benign tumours primarily by its ability to invade tissues throughout the body. Amongst western countries, almost half of the population will be diagnosed with a cancerous tumour at some stage during their lives [1], representing a significant disease burden. In New Zealand, around 340 new cases of cancer per 100,000 people are registered annually, and cancer is responsible for around 120 annual deaths per 100,000 people [2]. The most common cancers amongst New Zealand men are prostate cancer, colorectal cancer (CRC), melanoma and lung cancer, and amongst New Zealand women are breast cancer, CRC, melanoma and lung cancer. This shared propensity for melanoma is likely due to a combination of geographic and ethnic factors, with a significantly higher incidence rate amongst residents of European descent. Trends in both incidence rate and mortality appear to correlate with increasing ambient UV levels towards southerly latitudes, but differences between genders also suggest contributions from confounding factors such as lifestyle and recreational UV exposure [3]. The propensity for developing CRC is more difficult to explain, with New Zealand having amongst the highest incidence rates in the world at 48.1 and 36.7 cases per 100,000 annually for men and women respectively [4, 5]. Besides familial and other genetic factors, potential risk factors for developing CRC include living a sedentary lifestyle [6, 7], obesity [8] and diabetes mellitus [9, 10]. Additional site-specific risk factors also include cigarette smoking [11] and prior cholecystectomy [12, 13] in proximal or right-handed colon cancer, and alcohol consumption [14] and dietary habits [15] in distal or left-handed colon cancer. New Zealand is also still in the process of initiating a national screening programme for CRC beginning in 2017, and the absence of such screening may be at least partially responsible for these elevated rates.

As with benign tumours, cancer arises from the neoplastic transformation of an individual cell, triggering a cascade of events that must fail to recognise and remedy the emergence of this aberrant cell if tumourigenesis is to be the result. It is important to emphasise that cancer is not a singular disease, but rather a term broadly used to describe multiple conditions with similarities in pathology. For each type of cancer, there is also a spectrum of disease, with individual subtypes distinguishable between patients. For example, distinctions can be made amongst a cohort of patients with CRC based on anatomical location, such as between

22 proximal (right-handed) from the caecum through to the splenic flexure or distal (left-handed) from descending colon to rectum [16]. Alternatively, they may be grouped based on broad genetic origins, such as between microsatellite stable and microsatellite instable cancer [17, 18]. These subtypes present with different incidence rates, are associated with different risk factors and necessitate different treatment. This becomes further complicated at the patient level, with differences in genetic background rendering each cancer within the same subtype individually unique. Therefore, each case presents as its own personalised disease, and although it might fall within a broader sub-classification, there may be no guarantee that treatment recommended for that type will be effective for an individual patient. Personalisation of treatment is a complicated concept, and is heavily dependent upon our understanding behind some of the fundamental concepts in the pathology of cancer: how it originates, how it develops in the body, and how the body responds to it.

1.1.1 Neoplasia Discernible from metaplasia and dysplasia, an alteration in mature cell state and phenotype respectively, neoplasia is the abnormal proliferation of a cell and may be preceded by either of these precursor aberrant changes. An individual neoplastic cell may arise through a multitude of modalities, but inherently possesses a generally shared phenotypic profile, even amongst differing source tissues. Dysregulated proliferation, abnormal maturation, and extended longevity, these hallmarks of neoplastic cells are a consequence of the principal origin of neoplasia, alterations in intracellular signalling or regulatory pathways commonly induced by genetic mutation. From the moment of conception, our cells begin their lifelong accumulation of somatic mutations, acquired companions of their inheritable germ-line counterparts. The rate of acquisition in the average human cell is estimated at around 1.06x10- 6 mutations per cell division [19]. Although this may appear relatively insignificant, when considering that we undergo an estimated total number of cell divisions in the order of 1x1016 [20], we may each experience around 1x1010 mutations in our lifetime. Age is not in our favour, an increasing life expectancy worldwide increases the average number of cell divisions, and hence the number of somatic mutations we experience. These statistics might suggest the inevitability of tumourigenesis, but we also possess mechanisms that can recognise and respond to these early transformational events.

Besides the accumulation of spontaneous mutations during cell division, the genomic material in a cell can also be directly mutated by external environmental factors. Some key examples include the formation of pyrimidine dimers induced by exposure to light in the ultraviolet spectrum, DNA lesions or breaks induced by ionising radiation, or the various damaging effects of exposure to different chemical compounds or molecules. While exposure to some mutagenic agents is relatively fixed, such as background radiation from both natural and artificial sources estimated at around 3.01 mSv per year worldwide [21], many will vary depending on nation of residence, employment and lifestyle. Cigarette smoking, consumption of processed foods, or use of tanning beds are all commonly recognised sources of increased exposure to mutagenic agents, but some lesser known sources include the formation polycyclic aromatic hydrocarbons in chargrilled meats [22, 23], nitrate or nitrite contaminated drinking water [24], or the formation of trihalomethanes in disinfectant treated water [25, 26]. Occupational exposure is another important consideration, such as to chemicals prior to identification as carcinogens. Factory workers who worked with various chemicals such as dioxin and chlorophenols have been found to have an elevated risk of developing cancer [27, 28], possibly due to the absence of appropriate protective clothing. Some exposure is inevitable, and while it may be prudent to minimise it is impossible to negate. Exposure to carcinogens and the subsequent induction of mutations does not guarantee tumourigenesis, and this is inherently due to the nature of the mutations that result. Only mutations associated with specific types of genes tend to facilitate neoplasia. These genes may be referred to as either tumour suppressor genes or oncogenes. Tumour suppressor genes include many genes involved in regulating cell division, apoptosis or DNA repair. Deleterious or dysfunctionalising mutations in these genes can induce dysplasia, immortalisation or the accumulation of further mutations. Oncogenes are involved in similar pathways but from the opposing perspective, providing positive drive often under tight regulation. Overexpression or amplification of oncogenes boosts their activity above normal physiological levels, promoting neoplastic transformation.

The concept of a single mutation inducing guaranteed spontaneous neoplastic transformation of a mutated cell is erroneous. The validity of this claim is probably best demonstrated with inheritable forms of cancer, such as breast cancer arising in a patient with either a breast cancer (BRCA) 1 or 2 mutation. BRCA1 and BRCA2 are tumour suppressor genes involved in DNA repair, and mutations in these have been linked with development of breast cancer and ovarian cancer [29, 30]. Inheritance of either of these mutations in women is associated

24 with a lifetime risk of breast cancer of between 60 to 85 percent, and a lifetime risk of ovarian cancer of between 15 to 40 percent [31, 32]. Although these risk levels are significantly elevated above normal, they do not provide a guarantee of cancer development in susceptible individuals. Subsequent accumulation of further mutations in alternative tumour suppressor genes, or activation of oncogenes drives the completion of mutation-induced neoplasia. This process can be viewed through the progressive stages of precancerous intestinal polyp formation through to the development of CRC especially under the context of microsatellite instability, although microsatellite instable CRC accounts for only around 15% of cases [33]. Mutations in the tumour suppressor gene adenomatous polyposis coli (APC), a key regulator of proliferation in the intestinal epithelium, are associated with both familial and spontaneous CRC [34]. Compounded with mutations that activate oncogenes such as Kirsten rat sarcoma viral oncogene homolog (KRAS), or disrupt additional tumour suppressor genes such as deleted in CRC (DCC), the dysregulated, hyperplastic epithelium can give rise to the formation of multiple benign polyps throughout the colon. Although initially benign, these precancerous polyps can accumulate further mutations during their dysregulated growth, such as in the tumour suppressor gene p53, can induce transformation into carcinoma [35]. Further mutations may alter the resulting tumour, enhancing its growth or allowing it to metastasise to distant tissues (Figure 1.1).

Figure 1.1 Development of Colorectal Carcinoma through Accumulation of Mutations Combinations of mutations drive the progression of neoplastic transformation, beginning with the formation of precancerous lesions such as intestinal polyps, through to the development of carcinoma in situ and metastasis. Primary mutations capable of driving neoplastic transformation of cells include those that promote the expression of oncogenes (e.g. KRAS), or impair the expression of tumour suppressor genes (e.g. APC, DCC and P53) [35].

The primary concern of each individual cell is self-preservation, activating a series of response mechanisms in response to significant DNA damage. Some of the first mechanisms engaged are regulators to cell cycle checkpoints such as 53BP1, BRCA1 and MDC1, through a signal transduction pathway mediated by ATM and ATR that leads to cell cycle arrest [36]. Cessation of cell division provides the opportunity to engage repair mechanisms before attempting genomic replication, which would result in subsequent inheritance of the mutation in progeny cells. If the damage induced is sufficiently severe, these signal transduction pathways can also activate mediators of apoptosis, such as p53, and other early prevention mechanisms of neoplastic transformation and tumourigenesis. DNA repair mechanisms are vast and varied in eukaryotic cells, ranging from methylguanine DNA methyltransferases for removal of the O6-methyl group from O6-methylguanine in DNA, individual base pair excision by DNA glycosylases, oligomer excision through RPA, TFIIH, XPA, XPC, XPF- ERCC1 and XPG [37-39], and double-strand break ligation through homologous recombination or non-homologous end-joining mechanisms [40-44]. Intuitively, mutations or defects in many components of these pathways have been linked with dysfunctional DNA repair, mutation accumulation and neoplasia. The ultimate fate of the cell is defined by the efficacy of these mechanisms, with a failure to recover resulting in persistent senescence, apoptosis, or neoplastic transformation.

An alternative important source of induced neoplasia is infection with specific pathogens, such as oncogenic strains of human papillomavirus (HPV) and Epstein-Barr virus (EBV). These viruses are noteworthy both for their oncogenic potential, but also for their ubiquity in infection amongst humans throughout the world, rendering their unique host interactions worthy of mention. The HPV genome contains eight open reading frames (ORFs) in addition to a non-coding long control region (LCR), and includes the two oncogenes, E6 and E7, which bind to two major tumour suppressor proteins, p53 and pRb respectively [45-48]. Proteasomal degradation of p53 through the ubiquitin ligase activity of E6 essentially immortalises the cell [49-51], while competitive binding of E7 for pRb allows the transcription factor E2F to drive progression through the cell cycle, inducing dysregulated cell proliferation [52]. Infection of cervical squamous epithelial cells induces transformation into koilocytes, with a hyperplastic and hyperchromic nuclei, and a perinuclear halo. The progression of HPV-induced neoplasia from carcinoma in situ through to cervical cancer is a model example of progressive neoplastic transformation induced by an oncogenic virus, and can be visualised in histological sections of cervical biopsies (Figure 1.2).

Figure 1.2 Neoplastic Transformation of Cervical Squamous Epithelium Distinguishable stages of neoplastic transformation can often be visualised by histological staining of sections from cervical cancer biopsies. Cervical cancer is caused by infection of squamous epithelial cells with human papillomavirus, and it predominantly arises at the cervical transformation zone. The transformation zone contains metaplastic cells transitioning between squamous and columnar epithelium, rendering them particularly susceptible to virus-induced neoplastic transformation. Early stages of cervical carcinoma in situ, such as cervical intraepithelial neoplasia (CIN) 1-2 are characterised by the presence of HPV-infected koilocytes, with nuclear enlargement, hyperchromasia and a perinuclear halo. Further progression is marked by a significant increase in the nuclear to cytoplasmic ratio in CIN3, followed by invasion of the basement membrane with complete transition into invasive cervical cancer [53].

Lymphoblastoidal transformation of EBV infected latent B cells is a complex process induced by a combination of viral proteins, although the essential proteins for transformation are Epstein-Barr virus nuclear antigen (EBNA) 2, EBNA3C and Epstein-Barr virus latent membrane protein 1 (LMP1) [54]. EBNA2 is a transactivator that targets promoters through interaction with recombining binding protein suppressor of hairless (RBPJ) DNA-binding protein, a Notch signalling pathway component [55], and is essential for latency III transformation of the host cell [56]. EBNA3C is also required for transformation, and although it is known to disrupt the G1 cell cycle checkpoint in some cell lines [57], and induce aneuploidy by disrupting mitotic spindle checkpoints [57, 58], its role in transformation may actually involve suppression of p16INK4A and p14ARF [59]. Finally, LMP1 induces constitutive (NF-κB) activation by ligating to molecules involved in signal transduction from TNF superfamily receptors [60, 61]. Chronic inflammation induced by persistent pathogenic infection, such as Helicobacter pylori infection in the stomach, can also induce changes in cells that can lead to neoplastic transformation [62, 63]. Further

27 information regarding the association between chronic inflammation and neoplasia will be discussed in a later section on the immune response to tumourigenesis.

1.1.2 Tumourigenesis Upon neoplastic transformation, a cell will usually undergo dysregulated cell division to form a mass known as a neoplasm, or tumour. This process of tumour formation, or tumourigenesis, is dependent upon the source of the original neoplastic cell, as some malignancies such as those of haematological origin may not initially form a recognisable mass. The physiological process of tumourigenesis involves not just the proliferation of a neoplastic cell mass, but also the establishment of supportive vasculature through angiogenesis, and connective tissue through fibrogenesis. Growth in excess of nutrient supply may result in pockets of hypoxia, potentially leading to cell death and necrosis. The anatomical structure of a solid neoplasm differs depending on the cell type of origin, but is generally accepted to consist of heterogeneous populations of tumour cells, rather than a single clonal mass. Considering the mutagenic origin of neoplasia, this would suggest that tumour cells continue to evolve as they proliferate, accumulating further mutations and changing in phenotypic presentation. Early elucidations of this process focused on a linear progression model, in which the neoplastic progenitor cell seeds progeny that are increasingly aberrant as they acquire further mutations [64-67]. Although logical in principle, this model fumbles if the primary tumour breaches its anatomical confinement, often a basement membrane, and becomes metastatic. Tumour cells liberated from the tumour mass must have the ability to adapt and survive in transit within the lymph or blood, and have the capacity to invaginate into distal tissues to seed subsequent growth. These mechanisms would need to occur over timescales difficult to explain with the linear progression model alone. Current theories, such as the cancer stem cell and phenotype-switching hypotheses, attempt to account for and explain this discrepancy.

1.1.2.1 Cancer Stem Cells Normal adult stem cell populations can be found throughout the body, providing a stable regenerative source for tissue growth and repair. Any individual stem cell population may be defined by possessing a combination of distinct characteristics, including self-regenerative potential, the ability to proliferate, and to produce multiple cellular lineages amongst progeny

[68]. The cancer stem cell hypothesis proposes that a small population of cells within a tumour possesses these characteristics, acting as a self-sustaining reservoir that drives and maintains tumour growth [69]. These cancer stem cells are thought to originate from the neoplastic transformation of normal stem cells, or through the acquisition of stem-like cell properties in a mutagenic progenitor cell [70]. The exact definition of a cancer stem cell varies between publications, an inconsistency that impairs the formation of a unified theory on the subject. Sometimes the cancer stem cell is considered the original source of a primary tumour, giving rise to multiple cell lineages to form the heterogeneous tumour mass. Alternatively, cancer stem cells are considered a subpopulation of tumour cells with the capacity to initiate tumour formation, or seed distant metastases. Finally, when viewed in the context of treatment efficacy, cancer stem cells may be defined as a reservoir of drug-resistant cells responsible for tumour regeneration, seeding recurrence following regression [71]. The erratic application of these definitions is the primary source of criticism for this theory, deviating significantly from direct equivalency with normal stem cells.

The cancer stem cell hypothesis began to unravel due to compounding theoretical insight, flawed supporting experimental evidence, and the accumulation of opposing experimental evidence. References to equivalency between tumour initiating cells and cancer stem cells was made despite the absence of experimentation demonstrating self-regenerative potential [72]. This flaw was further exacerbated by the observation that isolated tumour initiating cells were unable to seed xenograft tumours with corresponding heterogeneity identical to that of the original tumour [73]. Another major assumption was that the accumulation of mutations occurred only in the cancer stem cell population, while the heterogeneous tumour bulk of progeny cells were generally static. This concept was challenged by the observation that differentiated progeny tumour cells have the potential to acquire stem-like cell properties through acquisition of mutations [73]. The primary benefit heralded by this theory was the proposed ability to selectively target the cancer stem cell population, but the ability of progeny cells to acquire this phenotype and potentially replace the target cancer stem cell population presents as a significant flaw in this rational. Development of therapeutic resistance was also inconsistent with this theory, as cancer stem cells, like normal stem cells, were theorised to be inherently resistant to the cytotoxic drugs used for chemotherapy that target rapidly dividing cells. The susceptible progeny that make up the bulk of the tumour would be eliminated, while the resistant cancer stem cells remained; however, this theory suggests that these cancer stem cells would seed a recurrent tumour similarly consisting of

29 susceptible progeny, but observations indicate that the entire recurrent tumour bulk acquires drug resistance [72]. This finding instead suggests that the administration of a therapeutic agent results in the selective proliferation of an inherently resistant subpopulation that passes its resistance on to its progeny.

The assumed equivalency with normal stem cells resulted in the most criticism for the theory, with the suggestion that the term ‘cancer stem cell’ may have been hastily applied to a tumour subpopulation possessing some stem-like cell properties [74, 75]. Despite the controversies surrounding this theory, research continues to attempt to formulate a model of tumourigenesis involving cancer stem cells, incorporating the ability of differentiated tumour cells to acquire self-regenerative abilities through de-differentiation [76, 77]. The existence of cancer stem cells was recently independently verified by three groups, which included confirmation of the fluidity in the phenotype of this population of cells [78-80]. This fluid cancer stem cell model incorporates tumour heterogeneity and plasticity of progenitor cells, supported by work in breast cancer, most notably the observation of transition of breast cancer cells into a cancer stem cell-like state dependent on ZEB1, a epithelial-mesenchymal transition regulator [81-83]. Interestingly, this shift has seen this model become increasingly compatible with one of its major competitor theories, the phenotype-switching hypothesis.

1.1.2.2 Phenotype-Switching The phenotype-switching hypothesis suggests that the heterogeneous population of cells within a tumour represents a phenotypic spectrum polarised by either proliferative or invasive properties, regulated by micro-environmental stimuli [75]. This hypothesis was formulated in studies on melanoma that identified distinct tumour cell populations with polar opposite phenotype based on migratory and proliferative potential, with concordant differences in transcriptional signature [84, 85]. Isolated melanoma cells with a proliferative phenotype were found to seed tumours that grew significantly quicker than those seeded with invasive phenotype cells, but the resulting excised tumour from either displayed an identical heterogeneous phenotype with immunohistochemical staining [85]. Strikingly, almost all individual melanoma cells appear to have this capacity to initiate tumour growth, with matching heterogeneity [86, 87]. The proposed trigger for phenotype instability was hypersensitivity of tumour cells to fluctuating micro-environmental conditions, promoting transition to a suitable phenotype by inducing activation of various signal transduction pathways [75]. Potential micro-environmental triggers include hypoxia and acidosis, known 30 factors associated with proliferative growth in excess of vascular supply. Alternative selective stresses may also include inflammation, or the administration of therapeutic regimens that target rapidly dividing cells, such as DNA-damaging agents and deactylase inhibitors [88].

One of the primary strengths of the phenotype-switching hypothesis is its ability to explain the seeding and growth of distant metastases. This theory suggests that as the primary tumour proliferates it experiences crises where its mass exceeds the capacity of supporting tissues, resulting in pockets of degeneration in local micro-environmental conditions, and stimulating the conversion of subpopulations of tumour cells into an invasive phenotype [75]. Acquiring an invasive phenotype allows these cells to seek more preferential conditions permissive to growth, migrating through the circulatory or lymphatic systems until a sufficiently supportive microenvironment is encountered, before converting back to a proliferative state. This mechanism may explain why metastatic tumours of different tumour types tend to form only in specific tissues and organs. A similar effect can be observed in normal melanoblast biology, which display inherent potential for activation of a migratory state to promote distribution from the neural crest along the dorso-lateral and ventral during embryogenesis, followed by deactivation upon reaching the epidermal basal layer and other specific sites around the body [89, 90]. A reversible invasive phenotype has also been observed amongst normal melanocytes in the presence of extracellular matrix modifying molecules secreted by metastatic melanoma cells [91].

The polarised distinction of phenotypic states in melanoma has been verified, and appears with be primarily defined by expression of either melanocyte differentiation markers such as microphthalmia-associated transcription factor (MITF) in a proliferative phenotype, or negative regulators of the wingless-type mouse mammary tumour virus (MMTV) integration site (Wnt)/β-catenin pathway such as Wnt5a in an invasive phenotype. High expression of Wnt5a has been previously implicated in melanoma metastasis [85, 92], and correlates with induced resistance to administration of serine/threonine-protein kinase B-Raf (BRAF) inhibitors in melanoma patients [93]. Proliferative melanoma cells have also been observed to adopt an invasive phenotype in response to mitogen-activated protein kinase (MAPK) inhibitors [94], indicating that targeted therapy may promote phenotype-switching and drive metastasis. This theory is corroborated by the observation that BRAF inhibition can enhance metastasis in drug resistant tumours [95]. A similar mechanism has also been observed amongst non-small cell lung cancer (NSCLC) cell lines, based on an epithelial-to-

31 mesenchymal transition (EMT) in response to administration of an EGFR inhibitor [96, 97]. Resistant NSCLC cell lines express high ZEB1 and vimentin, known markers of mesenchymal cells, and had increased migratory potential [98]. Drug-induced EMT also correlated with an increase in tyrosine-protein kinase receptor UFO precursor (AXL) expression, an invasive cell marker in melanoma that is also upregulated in NSCLC patients with acquired epithelial growth factor receptor (EGFR) inhibitor resistance [99].

The observed similarities between the phenotype-switching hypothesis and EMT suggest that these theories may be analogous, merely different means of explaining the same observed phenomena. This association also establishes a tentative link to the cancer stem cell hypothesis, as immortalized mammary epithelial cells have been demonstrated to acquire stem cell-like properties by undergoing EMT [100]. This link has been further corroborated by the observation of two distinct phenotypic states polarised in proliferative or migratory potential amongst cancer stem cells in epithelial tumours, with the ability to actively interconvert between states [101, 102]. Anatomically, populations of quiescent, migratory, mesenchymal-like cancer stem cells have been identified to predominantly colonise the invasive boundaries of breast cancer, while proliferative, epithelial-like cancer stem cells congregated within the tumour centre [101]. This distribution suggests that, similar to the theorised regulation of phenotype-switching in melanoma, that microenvironmental stimuli may regulate polarisation of cancer stem cell populations into these distinct phenotypic groups. Such similarities may be a manifestation of a shared EMT-like mechanism, that all these theories when compared side by side may be explaining the same phenomenon with subtle differences between tumour types (Figure 1.3). In particular, both the phenotype- switching mechanism and EMT transformation process can be identified in the updated fluid model of the cancer stem cell hypothesis, demonstrating the overlap developing between these theories.

Figure 1.3 Comparisons of Hypothesised Mechanisms of Tumourigenesis The cancer stem cell hypothesis has changed significantly over the years based on new experimental findings. It is notable that the most recent fluid model contains both the phenotype-switching hypothesis mechanism independent of cell division, and the well-defined phenomenon of epithelial to mesenchymal transformation. The similarities between these models may indicate that these hypotheses are describing the same mechanisms in distinct but related ways [75, 103].

1.2 The Immune Response to Tumourigenesis Most theories that postulate the process of tumourigenesis appear to brush over what is probably the most important contributor towards shaping tumour formation, the immune system. Every cell in our body is subject to surveillance, intermittently scanned by sentinel immune cells. This concept provided the foundation for the cancer immunosurveillance hypothesis conceived by Burnet and Thomas in the 1950s [104, 105], which outlines the ability of specific immune populations to detect indicators of neoplastic transformation on the surface of cells. Although inherently logical based on the pathological knowledge of the time, subsequent experiments by Stutman involving tumour implantation in either athymic or nude mice called this relationship into question [106, 107], particularly due to evidence that marked immunodeficiency appeared to have no effect on cancer susceptibility. Although this resulted in the abandonment of the cancer immunosurveillance hypothesis, debate continued over the subject, and particularly on the ability of emerging tumour cells to provide necessary danger signals for stimulation of the immune system [108, 109]. It wasn’t until quite some time later that technology advanced sufficiently to provide meaningful insight. In particular was the identification of the role of interferon (IFN)-γ in promoting immune- mediated rejection of implanted tumours [110], and the susceptibility of (RAG2)-/- mice lacking T cells, B cells and natural killer T (NKT) cells to spontaneous or carcinogen-induced tumours [111, 112]. As similar reports began to emerge from other laboratories, a major revision of the immunosurveillance hypothesis was undertaken. The result was the immunoediting hypothesis, which attempted to consolidate recent experimental findings by describing three discernible phases of interaction, and how each of these phases influenced both tumourigenesis and tumour composition [113-115].

The initial elimination phase is a modified translation of the original immunosurveillance hypothesis, involving the detection of molecular indicators of neoplastic transformation secreted or expressed on the surface of each cell. Collaboration between specific components of the innate and adaptive immune systems promotes the establishment of a response specific for these aberrant markers, and the targeted elimination of susceptible tumour cells. Eradication brings the process to completion, the desired homeostatic outcome; however, tumour cells may evade detection through a variety of mechanisms, persistence establishing a dynamic equilibrium. Under the influence of selective pressure from the immune system, this equilibrium phase drives genetic and phenotype evolution of the tumour by eliminating susceptible progeny. The surviving population may become poorly immunogenic, or even

34 induce a tolerogenic phenotype amongst infiltrating leukocytes [116]. Proliferation of these immunoevasive tumour cells induces the final phase, escape, resulting in the formation of a clinically recognisable neoplasm (Figure 1.4). This mechanism brings to light several sobering truths regarding tumourigenesis, that subclinical neoplastic transformation may be a relatively common phenomenon controlled by immunosurveillance, and that a clinically relevant tumour represents an inherent or induced failure in immune function.

Figure 1.4 The Immunoediting Hypothesis The immunoediting hypothesis encompasses three distinct phases during which the immune system responds to neoplastic transformation of cells, and the formation of a solid tumour mass. These phases include the elimination phase (A), in which detection of aberrant markers triggers detection of tumour cells by initially the innate, and then the adaptive immune systems. During this phase, susceptible tumour cells are eliminated, driving divergent evolution of the tumour. Failure to thrive can result in complete elimination of the tumour at this stage, and prevention of tumour persistence. However, if some of the tumour cells can adapt and survive, this interaction progresses to the equilibrium phase (B). During this phase, susceptible tumour cells continue to be eliminated, but resistant tumour cells compete through defensive mechanisms such as masking immunogenic molecules, or producing anti-inflammatory cytokines. Once the anti-tumour immune response subdues, the tumour reaches the escape phase (C), in which it proliferates to form a continuously growing tumour mass [116].

1.2.1 Innate Immune Response The innate immune system encompasses all immune cells and molecules with the capacity to detect and respond to an acute infection from a novel source. This response is generally non-specific, inflammatory, and rapid, conferring immediate protection to the host, but without generating memory. The innate immune system is not unique to vertebrates, and includes the restriction modification system in bacteria, the prophenoloxidase system in some invertebrates, and antimicrobial peptides produced by various organisms. In humans, the 35 primary components of the innate immune system that are involved in tumour immunoediting are sentinel mononuclear leukocytes such as macrophages and dendritic cells, and natural killer (NK) cells; however, complement proteins and polymorphonuclear (PMN) leukocytes can play a role in tumourigenesis. Normal cells in the body have inherent self-protection mechanisms against aberrant activation and recognition from the innate immune system. Unfortunately these mechanisms may be retained or even amplified in tumour cells, such as overexpression of membrane-bound complement regulatory proteins including CD46, CD55 and CD59 [117, 118]. This added protection is beneficial, as the tumour microenvironment may be skewed towards complement activation, as demonstrated with immunofluorescently labelled C3 cleavage products bound along the tumour vasculature in a TC-1 syngeneic murine model of cervical cancer [119]. These complement cleavage products were found to originate from the classical pathway of activation, due to concordance in C1q and C3 distribution in engrafted tumours, possibly due to localisation of complement activating immunoglobulins, or direct C1q deposition. Counterintuitively, complement activation has been suggested to promote tumourigenesis, based on impaired tumour growth in C3 deficient mice. Administration of a peptidic C5a receptor antagonist generated a similar effect, suspected to impair the formation of an immunosuppressive tumour microenvironment, which is usually mediated by C5a-based recruitment of myeloid-derived suppressor cells (MDSCs) [119].

1.2.1.1 Danger Signals The distribution of sentinel mononuclear leukocytes differs between tissues, but they each universally express a repertoire of non-specific receptors for detecting various groups of molecules commonly associated with specific types of infection. The Toll-like receptors (TLRs) are a group of pattern recognition receptors (PRRs) that recognise pathogen- associated molecular patterns (PAMPs). They are evolutionarily conserved amongst vertebrates and invertebrates, and derivative forms can even be identified in bacteria and plants. TLRs compose an assortment of related receptors categorized into groups that recognise molecules associated with potential infectious agents, such as double-stranded RNA (TLR3) [120] and flagellin peptide (TLR5) [121]. Some TLRs can also be stimulated by components of degenerative tumour cells, such as DNA, RNA and nucleic acid degradation products [122], heat shock proteins [123, 124], uric acid [125] and high mobility group box-1 (HMGB1) nuclear protein [126]. HMGB1 detection is dependent upon TLR4,

36 with abnormal TLR4 expression accelerating metastatic progression in patients with anthracycline-treated breast cancer [127]. Stimulation of TLRs induces the release of inflammatory cytokines such as IL-12 and TNF-α, in addition to up-regulating expression of chemokine receptors such as CCR2, CCR5 and CCR7, and co-stimulatory molecules such as CD40, CD80 and CD86 [128].

The inflammasome complex consists of members of another PRR family, Nod-like receptors (NLRs), specifically NLRP1, NLRP3 and NLRC4 [129]. Inflammasome complexes formed from each of these proteins have been implicated in host defence against specific pathogens, including responses to Bacillus anthracis mediated by the NLRP1 inflammasome [130, 131], responses to Gram-negative bacteria mediated by the NLRC4 inflammasome [132-134], and responses to several species of bacteria, viruses and fungi mediated by the NLRP3 inflammasome [135-137]. The NLRP3 inflammasome has also been implicated in some roles associated with cancer. Extracellular (ATP), such as that released by necrotic tumour cells, can be detected by dendritic cells through the P2X7 receptor, resulting in activation of the NLRP3 inflammasome complex [138]. This complex regulates IL-1B secretion from dendritic cells, which is associated with priming of naïve CD8+ T cells to secrete IFN-γ [139, 140]. Interference with any step in this pathway has been demonstrated to detrimentally influence the immune response to oxaliplatin-treated tumour cells, highlighting the importance of the role of the NLRP3 inflammasome in this process [139]. The NLRP3 inflammasome has also been implicated in dampening the inflammatory processes involved in colitis and colitis-associated cancer, acting as a negative regulator of tumourigenesis [141]. The final class of PRRs are the RIG-1-like receptors (RLRs), including laboratory of genetics and physiology 2 (LGP2), melanoma differentiation-associated gene 5 (MDA5) and retinoic acid-inducible gene (RIG-1). RLRs are RNA helicases, screening the cytoplasm for viral RNA, and inducing a type-1 interferon-mediated anti-viral response [142, 143].

Through the pathological process associated with tumourigenesis, additional receptors present on the surface of innate immune cells can be stimulated. Damage-associated molecular patterns (DAMPs) released amidst the necrotic core of some tumours can trigger inflammatory processes and amplified recruitment of immune cell populations. DAMPs are important inducers of haematopoietic stem cell mobilization, and include mitochondrial DNA, formylpeptides, and ATP [144, 145]. Tumours themselves may also originate in a site of chronic inflammation induced by an infectious pathogen, or induced by alternative intrinsic factors. The inflammatory environment of each tumour differs depending on the

37 release of proinflammatory mediators by tumour or supporting stromal cells, and the presence of tissue damage, pathogens or commensal organisms. A proinflammatory wound-healing microenvironment predominated by M2 macrophages and T helper (TH) 2 cells promotes tumourigenesis and angiogenesis, and decreases expression of intracellular adhesion molecules, which can facilitate invasion of surrounding tissues [146, 147]. Local tumour- associated inflammation is also associated with systemic comorbidities of cancer, such as anorexia, cachexia and immunosuppression [148]. Chronic inflammation can promote tumourigenesis by inducing genomic instability and DNA damage by increasing levels of reactive oxygen species, and may increase the rate of genetic and epigenetic mutations [149]. Some tumours can even directly influence haematopoiesis to skew immune cell populations by producing regulatory signalling molecules such as GM-CSF [147, 150].

1.2.1.2 Tumour-associated Macrophages Tumour-associated macrophages (TAMs) are a negative prognostic marker for multiple cancer types, including bladder cancer [151, 152], breast cancer [153, 154], CRC [155, 156], kidney cancer [157] and melanoma [158, 159]. When activated, macrophages may diverge towards either of two bipolar phenotypes with generally opposing functions, the M1 and M2 phenotypes. Although these phenotypes appear starkly contrasted, recent evidence suggests that they may actually represent the polar ends of a broad phenotypic spectrum, with respecialisation of macrophages possible, and multiple intermediate phenotypes. M1 macrophages are characterised by the expression of nitric oxide synthase 2 (NOS2) and proinflammatory cytokines such as IL1β, IL6, IL12, IL23 and TNF-α. M1 macrophages are bactericidal, and promote anti-tumour immune function. M2 macrophages have completely opposing activity, promoting tissue repair and tumour progression. Although macrophages involved in the innate immune response to initial tumour formation may be proinflammatory [160], resident macrophages in established tumours are converted to promote tumourigenesis [161, 162].

The tumour microenvironment is predominantly biased towards TH2 cells, characterised by the production of transforming growth factor (TGF) β1, arginase 1, and infiltration with CD4+ T cells [163]. Within this environment TAMs become polarised towards an M2 phenotype through production of molecules such as IL-4 from TH2 cells [164, 165], and growth factors such as colony stimulating factor 1 (CSF-1) and GM-CSF produced by tumour cells [166,

167]. TAMs secrete (CCL) 20 to recruit natural T regulatory (nTreg) cells in CRC [168], and 38 express both IL-10 and TGF-β, stimulating upregulation of forkhead box (FOX) P3 + expression in CD4 T cells and promoting an induced T regulatory (iTreg) phenotype [162,

169-171]. The role of iTreg cells in CRC is heavily debated, as they appear to dampen both CRC-promoting inflammation, while also inhibiting anti-tumour immunity. Local production of IL-10 and TGF-β by TAMs can also impair cytotoxic T lymphocyte function, and inhibit

TH1 and TH2 cell activity [172, 173]. TAMs can also deplete L-arginine in their local microenvironment by secreting arginase 1, producing urea and L-ornithine [174, 175]. Depletion of L-arginine inhibits re-expression of the CD3 ζ chain following internalisation from TCR signalling, impairing T cell function [176, 177].

1.2.1.3 Dendritic Cells There is a broad variety of DC populations throughout the body, some of which are specialised for canonical DC roles such antigen processing and presentation. Although most DCs originate from monocytes in humans [178], alternative progenitors are a more common source of DCs in mice [179, 180]. There are two major subsets of DCs, conventional DCs (cDCs) and plasmacytoid DCs (pDCs). While both subsets share common progenitors, their differentiation, morphology, markers and functions are distinct. In mice, cDCs can be sub- classified as migratory DCs such as Langerhan cells, CD103+ dermal DCs and CD11b+ dermal DCs, and lymphoid resident DCs such as CD4+ DCs, CD8α+ DCs and CD8α-/CD4- DCs [181]. In humans, dermal DCs can also be classified into subsets that include CD1a+ DCs, CD14+ DCs, and CD141+ DCs [182]. pDCs share morphological characteristics with their namesake, plasma cells, express TLR7 and TLR9, and produce IFN-α following activation [183]. Abnormal myelopoiesis in cancer patients results in reduced production of functionally competent DCs, increased production of immature myeloid cells, and increased tumour infiltration with immature DCs [184]. The tumour microenvironment also impairs DC function due to hypoxia, and accumulation of extracellular adenosine and lactate [185, 186]. Upregulation of hypoxia-inducible factor 1α (HIF1α) in DCs in this hypoxia tumour microenvironment promotes TH2 cell activation [187], while the presence of adenosine increases expression of vascular endothelial growth factor (VEGF), IL-6, IL-8, IL-10, cyclooxygenase (COX)-2, TGF-β and indoleamine 2,3-dioxygenase (IDO) [188]. Major histocompatibility complex (MHC)-II+, CD11b+, CD11c+ tumour-infiltrating DCs can actively suppress CD8+ T cells by producing arginase 1, a mechanism usually associated with TAMs and MDSCs [189]. IDO-expressing DCs can accumulate in some cancers [190,

191], depleting L-tryptophan, promoting T cell apoptosis, and enhancing the suppressive abilities of FOXP3-expressing Treg cells [192].

1.2.1.4 Granulocytes Granulocytes are characterised by the presence of cytoplasmic granules, the most common granulocyte being PMN leukocytes, also known as neutrophils. Mature granulocytes can infiltrate human tumours, and may promote tumour angiogenesis and metastasis [193-196]. Tumour cells and associated stromal cells can attract neutrophils by producing cysteine-X- cysteine (CXC) chemokines and prokineticin 2 [197, 198]. Granulocytes may promote angiogenesis by expressing matrix metalloproteinase (MMP)-9, which induces VEGF expression in tumour cells [199]. Release of elastase can also degrade insulin receptor substrate 1 (IRS1), facilitating interactions between phosphoinositide 3-kinase (PI3K) and platelet-derived growth factor (PDGF) to promote proliferation of tumour cells [200]. Like macrophages, neutrophils can also shift between two polarised pro-tumour and anti-tumour phenotypes. TGF-β induces neutrophils to assume a pro-tumour N2 phenotype, expressing arginase 1, and reducing expression of TNF, CCL3 and intercellular adhesion molecule (ICAM) 1. Depletion of these N2 neutrophils can increase CD8+ T cell activity in tumour- bearing animals [201].

1.2.1.5 Myeloid-derived Suppressor Cells In mice, MDSCs are defined as CD11b+ and GALEX release 1 (Gr1)+, markers of monocytic and granulocytic origin [202]. These cells can suppress T cell responses, and block DC maturation. The majority of MDSCs have a granulocytic or polymorphonuclear phenotype (PMN-MDSCs) characterised by expression Ly6G, while the remaining Ly6C+, Ly6G- MDSCs are termed monocytic MDSCs (M-MDSCs). M-MDSCs are significantly more immunosuppressive than their polymorphonuclear counterparts, and are associated with suppression of T cell activation through production of arginine 1 and inducible nitric oxide synthase (iNOS) [203-206]. M-MDSCs also express varying levels of classical monocytic markers, including F4/80, CD115, Ly6B and CCR2 [207]. Although M-MDSCs are similar to inflammatory monocytes in phenotype and morphology, these populations are functionally distinct. Similarly, PMN-MDSCs are functionally distinct from neutrophils, predominantly immunosuppressive with reduced phagocytic and chemotactic activity, increased expression

40 of arginine 1 and myeloperoxidase, higher expression of CD115 and CD244 and reduced expression of CXCR1 and CXCR2 [208, 209].

MDSC immunosuppressive activity is facilitated by a range of mechanisms. L-arginine and L-cysteine depletion induces downregulation of the CD3 ζ chain, inducing proliferative arrest of antigen-stimulated T cells [177, 210]. Production of reactive oxygen species (ROSs) and reactive nitrogen species such as hydrogen peroxide and peroxynitrite enable MDSCs to interfere with IL-2 receptor signalling [211], and can induce desensitisation of the T cell receptor [212]. Expression of a disintegrin and metalloproteinase domain 17 (ADAM17) on the surface of MDSCs decreases CD62L expression on naïve CD4+ and CD8+ T cells, limiting recirculation of T cells back to the lymph nodes [213]. Peroxynitrite-induced modification of CCL2 can also impair infiltration of CD8+ T cells into the tumour core [214]. Surface expression of galectin 9 on MDSCs can directly induce T cell apoptosis through binding to mucin domain-containing protein (TIM) 3. MDSCs also promote the clonal expansion of nTreg cells, and can convert naïve CD4+ T cells into iTreg cells [215]. While suppression by MDSCs in peripheral lymphoid organs is predominantly antigen-specifc, requiring antigen presentation by MDSCs and direct contact with T cells, within the tumour microenvironment MDSCs can function antigen-nonspecifically [212, 216, 217]. MDSCs can also hinder macrophage and dendritic cell function by producing IL-10, and impairing production of IL-12 [218, 219].

1.2.1.6 Natural Killer Cells Similar to other lymphocyte populations, NK cells express a complex assortment of both stimulatory and inhibitory cell surface receptors that activate competitive counteractive signalling mechanisms to regulate activation in response to stimuli. Other than their aptly named natural cytotoxicity receptors, NK cells also express the variant human leukocyte antigen (HLA)-E detecting receptor CD94, and the Ig receptor CD16. The primary inhibitory signalling receptor are the killer-cell immunoglobulin-like receptors (KIRs) with immunoglobulin-like extracellular domains for interacting with classical MHC-I (HLA-A, - B and -C) and nonclassical HLA-G. Additional inhibitory signals can also be mediated by leukocyte inhibitory receptors (LIRs) and inhibitory isoforms of Ly49. The interactions between these receptors and their signalling pathways is orchestrated to enable NK cells to respond to both elevated expression of stimulatory ligands, and downregulated expression of the MHC-I receptor [220, 221]. Overexpression of stimulatory ligands can be induced by 41 neoplastic transformation and stress-induced DNA damage, while downregulation of MHC- I expression is a mechanism of immunoevasion shared between both tumour cells and some viral infections [222]. Either the loss of inhibitory stimuli through detection of MHC-I, or the presence of overwhelming stimulatory signals can induce NK cell activation [223, 224]. Once activated, NK cell-mediated apoptosis of target cells is induced through the same mechanisms as in other cytotoxic lymphocyte populations, including secretion of granzyme and perforin, stimulation of death receptors such as TNF-related apoptosis-inducing ligand (TRAIL) or Fas, or through release of IFN-γ [225].

1.2.1.7 Natural Killer T Cells NKT cells are a population of T cells that share both the properties of T cells and NK cells. NKT cells are thought to contribute to immune surveillance for tumours in a similar fashion as NK cells [226, 227]. There are two types of NKT cells, type I NKT (iNKT) cells, which express the invariant Vα24Jα18 TCR in humans and Vα14Jα18 TCR in mice [228], and type II NKT cells with more variant TCRs. iNKT cells recognise lipids and glycolipids presented on the non-classical MHC protein CD1d. They can recognise and respond to exogenous glycolipids such as α-galactosylceramide, endogenous glycolipids such as isoglobotriosylceramide, and tumour-associated gangliosides such as ganglioside precursor disialohematoside 3 (GD3) [229]. When activated, iNKT cells upregulate expression of CD25, the IL-2 receptor α-chain, and CD69, as well as the cytokines IL-2, IL-4 and IFN-γ [230]. While iNKT cells are an important component of an anti-tumour immune response, especially through co-ordination of NK cells and CD8+ T cells through secretion of IFN-γ [231, 232], and cross-licensing of CD8+ DCs [233], type II NKT cells have been implicated in an immunosuppressive role [234, 235].

1.2.1.8 Mucosal Associated Invariant T Cells Mucosal associated invariant T (MAIT) cells are a form of T cell with innate-like properties, expressing a semi-invariant TCR. Tissue-infiltrating and circulating MAIT cells can be characterised based on the expression of CD161 or IL-18Rα on their surface, along with the Vα 7.2 TCR chain [236, 237]. The semi-variant TCR on MAIT cells can recognise microbial vitamin B metabolites presented on MHC-related protein 1 (MR1) [238, 239]. When activated, MAIT cells produce IFN-γ, TNF-α and IL-17A [240]. MAIT cells are

42 predominantly found in the lamina propria of the intestines, and are thought to provide innate protection against infection and inflammation. MAIT cells have been identified to provide protection against inflammatory bowel disease in humans [241], and confer protection against a range of infectious pathogens [242, 243]. Recent studies have found that MAIT cells infiltrate into the tumours of CRC patients, with a corresponding reduction in the circulating population. Based on this finding, MAIT cells have been proposed to have a role in surveillance of mucosal-associated cancer [244, 245]. Further work is required in order to fully elucidate the role of MAIT cells in CRC.

1.2.3 Adaptive Immune Response The adaptive immune system is a shared feature amongst jawed vertebrates, and its activity is mediated through specific populations of lymphocytes. These lymphocyte populations possess recombination-activating genes (RAGs), responsible for rearranging their corresponding antigen receptor such as those encoded by the immunoglobulin (Ig) and T cell receptor (TCR) genes. Rearrangement of the sequences encoding the variable domains of these proteins generates sufficient combinations to enable the adaptive immune system to theoretically respond to any antigenic sequence in nature; however, each individual combination is only capable of binding a single antigen. This combination of broad coverage and high specificity allows the adaptive immune system to respond to any potential pathogen, but may be incapable of facilitating immediate protection against infection from a different strain of the same pathogen. An adaptive immune response induced against tumour cells follows identical rules, facilitated by detection of neo-antigens or tumour-associated antigens (TAAs) processed and presented on the surface of the tumour cells themselves, or APCs.

Neo-antigens are tumour-specific, and can include mutagenic proteins or oncogenic viral proteins [246, 247]. TAAs on the other hand are otherwise normal ‘self’ molecules that are abnormally expressed by tumours, such as markers of differentiation (e.g. tyrosinase), cancer-testis antigens (e.g. oesophageal cancer antigen (ESO), melanoma antigen (MAGE)), oncofetal antigens (e.g. carcinoembryonic antigen (CEA), α-fetoprotein (αFP)) and various overexpressed proteins (e.g. p53, human epithelial growth factor receptor 2 (HER2)) [248- 251]. As TAAs represent inherently normal self-antigens, recognition by the adaptive immune system is usually suppressed through central deletion of reactive T cells in the thymus, or peripheral tolerisation [252]. Antigen recognition can also be hampered by

43 variability between HLA haplotypes [249, 253] and antigenic strength [254, 255], with relatively weak, cryptic antigens requiring prolonged exposure at an elevated concentration to induce a potent immune response [256].

1.2.3.1 Antigen Presentation APCs such as macrophages, dendritic cells and B cells actively sample their local microenvironment for potential antigenic molecules through non-specific internalisation mechanisms such as phagocytosis and macropinocytosis. Specific molecules can also be internalised through receptor-mediated endocytosis by binding to receptors present on the surface of APCs, such as the Fc receptor for binding the fragment crystallisable (Fc) region of antibodies, C-type lectin receptors, and various scavenging receptors. Upon formation of a phagosome or endosome, ingested materials are exposed to degradative enzymes from fusion with a lysosome. Proteases activated by acidification of the resulting phagolysosome cleave proteins, liberating short peptide chains. These enzymes also cleave the invariant chain complexed with MHC-II along the internal surface of the phagolysosome, producing a short peptide called class II-associated invariant chain protein (CLIP) that binds to the peptide- binding domain of MHC-II. HLA-DM subsequently facilitates the excision of CLIP from the groove, replacing it with peptides processed from endocytosed material [257, 258]. Fusion of the phagolysosome with the cell membrane results in presentation of the processed exogenous peptides loaded in MHC-II on the surface of APCs, recognisable by antigen- specific CD4+ T cells.

While predominantly only dedicated APCs are responsible for presenting exogenous antigens on MHC-II, all nucleated cells have the capacity to monitor their own intracellular microenvironment for endogenous antigens. Endogenous antigens are routinely degraded by the proteasome in the cytoplasm, marked for degradation through tagging processes such as ubiquitination. These can consist of any proteins translated by the cell, including viral proteins, tumour-associated antigens and normal self-antigens. Degraded endogenous antigens are transported to the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP). MHC-I molecules can be found along the inner surface of the ER, complexed with β2-microglobulin. Peptides derived from endogenous antigens are loaded onto MHC-I by a peptide loading complex, which includes the chaparones calreticulin and tapasin, the thiol-oxidoreductase ERp57, and TAP [259]. Loaded MHC-I molecules are packed into vesicles, and sent to the cell membrane through the Golgi. Fusion of the vesicle 44 with the cell membrane and exocytosis results in the display of peptide loaded MHC-I on the surface of the cell, recognisable by antigen-specific CD8+ T cells.

1.2.3.2 Cross-presentation Cross-presentation is the process of presenting exogenous antigens on MHC-I rather than MHC-II. This mechanism has primarily been studied in dendritic cells, but has also been identified in macrophages [260], B cells [261] and sinusoidal endothelial cells [262]. Cross- presentation is important for providing protection against intracellular infections, and cellular abnormalities such as neoplastic transformation. Known mechanisms of cross-presentation include the endosome-to-cytosol pathway, ER-endosome fusion, the CD74-dependant pathway, and the receptor-recycling pathway (Figure 1.5) [263].

Figure 1.5 Cross-presentation Pathways APCs cross-present epitopes derived from exogenous antigen onto MHC-I through multiple pathways including: (A) Gap Junctions; (B) Endosome-to-cytosol Pathway; (C) ER-Endosome Fusion; (D) CD74- dependent Pathway; (E) Receptor Recycling; and (F) Exosomes. Adapted from Groothuis et al 2005 [264].

Internalised exogenous antigens can enter the cytoplasm through the endosome-to-cytosol pathway, undergoing proteasomal degradation and entering the MHC-I loading pathway through TAP [265, 266]. ER-endosome fusion bypasses both the proteasome and TAP, enabling direct loading of degraded proteins from the endosome onto MHC-I in the ER. CD74 is an invariant chain chaperone that escorts MHC-I molecules from the ER to

45 endolysosomes through intervening vesicles, and even facilitates the loading of MHC-I with exogenous antigen [263]. Completely independent of ER localised MHC-I, the receptor recycling pathway of cross-presentation can retrieve MHC-I molecules from the cell membrane, reload them with new peptides and return the MHC-I exogenous antigen complex to the cell surface. Additionally, mechanisms of sharing antigenic material between cells, such as gap junctions and formation of exosomes, can also indirectly facilitate cross- presentation of antigen.

1.2.3.3 Antigen Cross-Dressing In addition to processing exogenous and endogenous antigen for presentation through both the standard antigen presentation and cross-presentation pathways, there is a third mechanism through which APCs may present antigen. This process has been termed antigen cross- dressing, as it involves APCs acquiring the preloaded MHC molecules from the surface of other cells, and displaying them on their own surface. It is important to note that this process is not unique to professional APCs, as the original study identifying intercellular transfer of MHC molecules observed the transfer of MHC-II from B cells to T cells [267]. The transfer of MHC between cells can be mediated by a process known as trogocytosis, the intercellular transfer of plasma membrane and anchored proteins between cells that come into close contact [268], such as during the formation of an immunological synapse. Alternatively, antigen cross-dressing can occur due to the phagocytosis of exosomes containing preloaded MHC molecules [269], or it can be mediated through the exchange of surface membrane proteins by tunnelling nanotubes [270]. The exchange of MHC molecules between cells has been observed between APCs and T cells, APCs and other APCs, epithelial cells and APCs, and even between tumour cells and APCs [269, 271, 272]. This last interaction is of particular interest, as it highlights an inherent alternative mechanism through which APCs may directly acquire tumour-associated antigens from tumour cells.

1.2.3.4 Activation of CD4+ T Helper Cells In concert with the recognition of a specific antigen, cells of the adaptive immune system also require additional stimuli to indicate underlying abnormality, such as active infection or inflammation. CD4+ T cells are dependent upon the ability of APCs to recognise danger signals such as PAMPs and DAMPs, and upregulate the expression of co-stimulatory

46 molecules. APCs harbouring tumour antigens and activated by PAMPs/DAMPs drain from the tumour site through the lymph, entering the nearest draining lymph node through the subcapsular sinus perculates at the cortex. Interactions between APCs and T cells occur primarily within the lymph node paracortex [273], and are initially stabilised by cell-adhesion molecules, such as CD2, lymphocyte function-associated antigen (LFA)-1 and ICAM-3 on the surface of T cells, which bind to CD58, ICAM-1/2 and DC-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN) on the surface of APCs respectively. Priming of CD4+ T cells is a biphasic process, beginning with the detection of antigen loaded in MHC- II on the surface of an APC through the CD4+ T cell’s TCR, and completed by a cytokine- driven polarisation phase that ultimately determines the phenotype of the active T cell. There + are currently seven discernible subsets of CD4 T cells, T helper (TH) 1, TH2, TH9, TH17, + TH22, T follicular helper (TFH) cells and T regulatory (Treg) cells. Each CD4 T cell subset has its own characteristic gene expression and cytokine secretion profile, as illustrated in

Figure 1.6. TH1 cells are predominantly involved in the cell-mediated arm of the adaptive immune system, while TH2 and TFH cells favour the humoral arm. TH9 cells are a proinflammatory subset of CD4+ T cells [274, 275], with roles in tumour and parasite immunity identified in mice [276-278], and a potential penchant for fungal infection and dermatitis in humans [279, 280]. TH17 cells have roles in inflammatory disease processes [281-283], and fungal infections [284], while TH22 cells are a recently described variant that do not produce

IL-17 [285-287]. Treg cells suppress effector T cell activity, including both the iTreg and nTreg subtypes.

Figure 1.6 Subsets of CD4+ T Cells Upon activation following recognition of antigen/MHC-II complexes presented by APCs, naïve CD4+ T cells can differentiate into any of five distinct subsets of T helper cells. Which subset the CD4+ T cell will differentiate into is determined the factors involved in their activation, in addition to stimulation of particular activation pathways induced by specific cytokines present in the local microenvironment. These subsets include the classical TH1 and TH2 cells, induced through expression of either T-bet or GATA-3 respectively.

Alternatively, naïve CD4+ T cells can differentiate into TH17 cells through induced expression of RORγt, iTreg cells through induction of FoxP3 expression, or TFH cells through induction of Bcl-6 expression. Adapted from Yamane et al 2013 [288-291].

Polarisation of CD4+ T cells begins with the strength of TCR signalling, with weak signalling favouring TH2 differentiation, while strong signalling promotes a TH1 phenotype [292]. Transient activation of extracellular receptor kinase (ERK) with weak TCR signalling induces expression of GATA-binding factor (GATA)-3 and production of IL-2, which in turn activates STAT5. These weakly stimulated T cells differentiate into TH2 cells, producing IL- 4 and retaining STAT5 activation [293]. Strong TCR signalling prevents this early GATA-3 expression and suppresses STAT5 activation, with persistent ERK activation and endogenous + production of IL-12 facilitating differentiation into TH1 cells. Differentiation of naïve CD4

T cells into TH17 cells is a multistep process, initiated by the presence of TGF-β and IL-6 48 during TCR stimulation. This combination induces expression of IL-23R and RAR-related orphan receptor (ROR)γt, and secretion of IL-17A, IL-17F and IL-21 [294-297]. IL-21 synergises with TGF-γ to continue IL-17 production, dependent on STAT3. Naïve CD4+ T cells under similar conditions but with weak TCR stimulation express FoxP3 rather than IL-

23R and RORγt, promoting differentiation into iTreg cells [298-300]. TFH cells are essential for the formation of germinal centres [301-303], and support B cells during differentiation into memory or plasma cells [304].

In addition to signalling through the TCR, naïve CD4+ T cells require additional signals to activate completely, such as from co-signalling molecules and cytokines. Co-signalling molecules are predominantly members of either the immunoglobulin superfamily (IgSF) or the tumour necrosis factor receptor superfamily (TNFRSF), and can serve as either co- stimulatory or co-inhibitory molecules. The IgSF group includes various receptor subfamilies such as CD28, B7, CD226, TIM, CD2, butyrophilin (BTN) and leukocyte-associated immunoglobulin-like receptor (LAIR), each containing multiple co-signalling molecules of various functions, including notable examples including CD28, cytotoxic T-lymphocyte- associated protein (CTLA) 4, programmed cell death protein (PD) 1, PD ligand (PDL) 1 and CD80 [305]. Amongst the TNFRSF are subfamilies such as Type-V, Type-L and Type-S, with notable examples including CD137 (4-1BB), CD134 (OX40), CD27, glucocorticoid- induced TNFR-related protein CD357 (GITR), Fas cell surface death receptor (FAS), CD40 and TRAIL receptor (TRAILR) 1-4. The most well described co-signalling molecule interaction is that of CD28 with the B7 family of receptors. Binding between CD28 expressed on the surface of naïve CD4+ T cells and CD80/86 on the surface of an APC provides stimulatory signals that promote the activation of both cells [306]. This signalling interaction can be disrupted by expression of CTLA4 on the surface of the CD4+ T cell, which competes with CD28 for CD80/86 binding, and produces signals that inhibit activation. The net ratio of stimulatory to inhibitory signals amongst all co-signalling molecules involved determines whether the CD4+ T cell will be activated. The modality of stimulation, combined with costimulation and cytokine signalling determines the type of T helper cell generated, or whether the cell becomes a memory CD4+ T cell. Memory T cells will be further discussed in the next section.

1.2.3.5 Activation of CD8+ T Cells The activation of CD8+ T cells follows the same essential steps as has been described for CD4+ T cells, including the recognition of antigen, and signalling through co-signalling molecules and cytokines. In this instance, the TCR of the CD8+ T cells recognises antigen loaded onto MHC-I, forming an interaction stabilised by CD8. The CD8+ T cell requires further stimulation in the form of co-stimulatory molecule interactions such as between CD28 and CD80/86, and stimulation with cytokines such as IL-2, TNF-α and IFN-γ. These cytokines can be supplied by activated TH1 cells, a means of ‘help’ provided to induce + + proliferation and differentiation of the CD8 T cell into a cytotoxic T cell (TC). CD8 T cells that do not receive appropriate co-stimulation, particularly from IL-2, will become anergic and non-responsive [307]. Similar to T helper cells, TC cells can be separated into three subsets as defined by their surface marker expression, cytokine profile and effector function.

These subsets are TC1, TC2 and TC17 [308, 309], each of which are comparable with their T helper cell counterpart in regards to their primary differentiation regulator gene, cytokine profile, and function. Activated effector TC cells can induce apoptosis in target cells through either of two mechanisms, the calcium-independent Fas/Fas-L-mediated pathway, and the calcium-dependent perforin/granzyme-mediated pathway [310]. Both of these pathways are activated through engagement of the TC cells TCR with its specific epitope loaded MHC-I on the surface of target cells. Upon engagement, Fas-L molecules on the surface of the TC cell bind to Fas molecules on the target cell, inducing caspase-dependent apoptosis. This pathway is also utilised in the maintenance of lymphocyte homeostasis, and in the regulation of lymphocyte proliferation. The TCR-epitope-MHC-I interaction also stimulates the exocytosis of lytic granules from the TC cell, which contain cytotoxic effector molecules including serglycin, perforin and granzyme. Serglycin is a proteoglycan that acts as a scaffold for the cytotoxic perforin-granzyme complex. Perforin forms pores in the cell membrane of the target cells, facilitating entrance of granzyme. Granzyme is a serine protease capable of inducing apoptosis in the target cell. TC cell effector function can also be mediated through the release of cytokines, including IFN-γ, TNF-α and TNF-β [311].

+ In addition to becoming effector TC cells upon activation, CD8 T cells also have the capacity to become memory T cells. There are two primary described subsets of memory T cells, applicable to both CD4+ T cells and CD8+ T cells. These two subsets are T central memory

(TCM) and T effector memory (TEM) cells, distinguishable based on differences in cell surface molecule expressions and effector function [312]. TCM cells express CCR7 and CD62L on

50 their surface and can localise to the lymph nodes, while TEM cells are devoid of these expression markers and localise to non-lymphoid tissues. Both subsets can be found in the spleen and circulating in the blood [313, 314]. While TCM cells are capable of expansive proliferation and produce IL-2, TEM cells are generally less proliferative and produce IFN-γ. A third subset of memory T cells that permanently reside in peripheral tissues has recently been described, and has been termed T tissue-resident memory (TRM) cells [315, 316]. TRM cells share the absence of CD62L and CCR7 surface expression with TEM cells, but also notably express CD69 and CD103. Memory T cells respond more rapidly when exposed to their specific antigen epitope loaded on MHC-I [317, 318].

1.2.3.6 Suppression of Effector T Cells

The efficacy of T effector (Teff) cells upon reaching the tumour is largely dependent on the immune microenvironment. Infiltration with suppressive immune cell populations such as

IL-10 secreting CD4+/CD25+/FoxP3+ T regulatory (Treg) cells, CD1d-restricted NKT cells, immature dendritic cells, pDCs and MDSCs can significantly impair Teff function in the local tumour microenvironment [212, 319, 320]. Further suppression is induced through expression of various inhibitory molecules such as PD-L1, B7x, CD276 (B7-H3), HLA-E and HLA-G [321-323], or by release of suppressive factors such as TGF-β, VEGF, IL-10 and gangliosides [324-326]. Acquired tolerance can be induced in Teff cells through expression of IDO by tumour cells or IDO-competent APCs. IDO-competent dendritic cells in particular can induce anergy, apoptosis or conversion of Teff cells into iTreg cells [191, 327, 328]. Treg cells secrete suppressive cytokines such as TGF-β and IL-35, and can directly suppress Teff cell function through contact-dependent mechanisms, potentially mediated through CTLA-4 expressed on Teff cells [329]. MDSCs employ a complex suppressive mechanism involving inducible nitric oxide synthase or arginase-1, inducing apoptosis, inhibiting proliferation, or converting Teff cells into iTreg cells [211, 330, 331]. Although these suppressive mechanisms are crucial for preventing rampant autoimmunity, they present a major complication in both the initial adaptive immune response to tumourigenesis, and subsequent treatment with immunotherapy.

1.2.3.7 T cell Exhaustion Exhaustion of T cells was originally observed in CD8+ T cells during chronic lymphocytic chorionmeningitis virus (LCMV) infection in mice [332, 333]. The exhaustion phenotype is predominantly characterised by increased expression of inhibitory co-signalling molecules such as PD-1, CTLA-4, leukocyte activation protein (LAG)-3 and TIM-3 [334]. T cell exhaustion is usually observed in association with chronic infection or inflammation, induced by the persistence of antigen and inflammatory molecules. As a state of exhaustion develops, CD8+ T cells gradually lose effector function. IL-2 production and ex vivo cytolytic activity are the first functions lost, followed by loss of TNF-α and IFN-γ production, and the ability to degranulate. Terminally exhausted T (TEX) cells are deleted entirely, possibly due to overstimulation-induced death [334, 335]. TEX cells can no longer undergo antigen- independent proliferation, and lose sensitivity to IL-7 and IL-15 [336-338]. PD-1+ tumour- infiltrating lymphocytes (TILs) have been identified in breast [339], lung [340], ovarian [341], prostate [342] and melanoma [343, 344]. These dysfunctional TILs have reduced effector function, and may encompass T cells in various stages of exhaustion. Reinvigoration of TEX cells is possible, primarily induced using monoclonal antibodies to block inhibitory co-signalling molecules, particularly PD-1/PD-L1 [345-347]. PD-1 expression on T cells is induced upon activation, regulated by the transcription factors nuclear factor of activated T cells (NFAT), T-bet, B lymphocyte-induced maturation protein (Blimp)-1 and FoxO1 [348-

351]. TEX reinvigoration increases proliferation, cytokine production and restores cytolytic activity, returning towards normal T effector cell function [345].

1.2.3.8 Activation of B cells B cells are the primary mediator of the humoral arm of the adaptive immune system. Each B cell possesses a uniquely constructed receptor on their surface, known as the B cell receptor (BCR). The BCR binds to antigen in its native unprocessed form, allowing it to bind to a wide range of molecules such as proteins, glycoproteins and polysaccharides, as well as binding to the surface components of intact pathogens such as viruses and bacteria [310]. Binding of the BCR to its complementary antigen produces stimulatory signals, pre- activating the B cell. The B cell internalises the BCR-antigen complex and processes the antigen through the exogenous pathway, loading epitopes onto MHC-II and presenting them on the surface. Activated TH cells capable of recognising these MHC-II-epitope complexes

52 on the surface of the B cell provide secondary activation signals through interactions between co-stimulatory molecules, and release of cytokines.

Activated B cells, also known as plasma cells, secrete antibodies, immunoglobulins with an identical variable binding domain to their original BCR. This allows the antibody to bind to its complementary antigen in its native form, just as the BCR once did. Antibodies have three primary functional roles; neutralisation of toxins and viruses, opsonisation and agglutination of extracellular pathogens, and activation of the complement system. All plasma cells initially produce class M immunoglobulins (IgM), a pentameric form particularly effective at complement activation and extracellular pathogen agglutination. Dependent upon the cytokine signals it receives, a plasma cell can switch the class of its antibody to class G immunoglobulins (IgG), a monomeric class that predominates in the serum. Further stimulation with select cytokines can induce class-switching to enable the production of class A immunoglobulins (IgA), a monomeric class that can combine into a dimeric form for secretion along mucosal surfaces, or class E immunoglobulins (IgE), a monomeric class that bind to the surface of mast cells, capable of triggering mast cell degranulation and histamine release during inflammatory processes [310]. There also exists a fifth class of antibody, class D immunoglobulins (IgD), but its role is not well understood. It has been suggested that IgD antibodies may have a role in basophil activation [352].

B cells do not appear to play a major role in the primary immune response to cancer, probably limited to their function as an APC. However, antibodies capable of binding specifically to receptors on the surface of tumour cells, or that interfere with interactions between inhibitory receptors on the surface of immune cells have been extensively studied as a form of monoclonal antibody therapy. This type of anti-cancer therapy will be discussed in a later section. It is worth noting that there is a population of B cells that has been identified in mice that can promote tumour growth by suppressing the immune system [353, 354]. These regulatory B (Breg) cells secrete IL-10, and promote the generation of Treg cells [355]. They also play a role in suppressing inflammation through restoring balance between TH1 and TH2 cell responses [354, 356]. Matching populations of Breg cells were also been identified in humans [357], and have recently been implicated in promoting immune privilege of cancer through PD-1-PD-L1 interactions [358].

1.3 Colorectal Cancer CRC refers to the development of a malignant tumour anywhere throughout the colon or rectum, arising from the transformation of normal colonic epithelium into colon adenocarcinoma. CRC is the third most common cancer amongst men, and the second most common amongst women worldwide, with an annual incidence estimated at over 1.36 million in 2012, and is accountable for almost 700,000 deaths annually [359]. In New Zealand, it is the second most common form of cancer amongst both men and women, with an incidence in 2008 of 44.1 and 37.5 per 100,000 people respectively [360]. These rates are more than double the worldwide incidence of 20.6 and 14.3 per 100,000 people in men and women respectively, and attribute to a mortality rate in New Zealand comparable to both breast and prostate cancer combined. This places New Zealand jointly with Australia at the highest rates of CRC in the world [359]. Although CRC is more common amongst non-Maori, Maori have a higher mortality rate upon diagnosis, possibly due to disparities in access to treatment. The incidence rate amongst Maori is also increasing, and may eventually equate their non-Maori counterparts if observed trends continue [361]. Notably, 49% of patients diagnosed with CRC in New Zealand are already at an advanced stage of the disease, with 24% presenting with distant metastases [362]. This gives New Zealand the highest rate of metastatic disease at presentation in the world [363]. This rate is even higher amongst Maori and Pacific patients, with 32% and 35% respectively.

1.3.1 Familial Colorectal Cancer Familial adenomatous polyposis (FAP) and hereditary non-polyposis colon cancer (HNPCC) are the two most common inheritable genetic conditions associated with increased risk of developing CRC. FAP is predominantly associated with mutations in APC, and manifests as the formation of multiple benign polyps throughout the colon. Each polyp has the potential to undergo malignant transformation, posing significant risk for recipients of the condition, and ensuring the development of CRC within third to fifth decade of life if the colon is not excised [35, 364]. As the mean age of CRC is only 36 years amongst patients with FAP [365], prophylactic removal of the colon is recommended. Germ-line APC mutations are the most common cause of FAP, with over 95% of FAP-associated APC mutations involving frameshift or nonsense mutations that prematurely truncate the APC protein during translation [365, 366]. Mutations located between codons 1250 and 1464 are associated with particularly profuse polyposis, while those N-terminal of codon 157 or near the C-terminus 54 give rise to attenuated polyposis. In addition to FAP-associated CRC, APC mutation or inactivation can be found in around 70-80% of sporadic colorectal adenomas and carcinomas [366-368]. APC mutations are also associated with several extra-colonic tumours, found in the brain in Turcot syndrome, and various connective tissues in Gardner syndrome [365].

HNPCC, also known as Lynch syndrome, results from defects in genes of the mismatch repair (MMR) system. These mutations induce microsatellite instability, promoting the rapid accumulation of somatic mutations [17, 18, 369, 370]. Genomic instability plays a significant role in the molecular pathogenesis of CRC, as individuals with germline mutations in DNA repair mechanisms such as the base excision repair (BER) system or the MMR system have increased risk of developing the disease [371-373]. In humans, the MMR system is largely dependent upon the integrity of two genes, human mutL homolog (hMLH) 1 and 2. Mutations in either hMLH1 or hMSH2 can account for over 90% of HNPCC cases [374]. The hMLH1 gene can be found at chromosome 3p21-23, a candidate region highlighted in studies of HNPCC [375, 376]. The protein product of hMLH1 forms a heterodimer with hMLH3, hPMS1 or hPMS2, recruiting additional DNA repair proteins to form the mismatch repair complex. Similarly, the location of the hMSH2 gene at chromosome 2p21 is also within an area associated with HNPCC [370, 377]. The protein product of hMSH2 can form a heterodimer with either hMSH3 or hMSH6, and binds to incorrectly matched nucleotide pairs produced during DNA replication. Upon binding, this complex recruits the mismatch repair complex to excise and replace the error in the daughter sequence. Pathogenic mutations associated with impairment of mismatch repair can be found throughout all coding exons of both hMLH1 and hMSH2, with the broadest range of variants found in exon 16 of hMLH1 and exon 3 of hMSH2 [374].

1.3.2 Spontaneous Mutagenesis The intervening benign adenomatous stage of colon adenocarcinoma is referred to as a polyp, often originating at the bottom of a colonic crypt and proliferating to fill the intestinal villi [368]. Early neoplastic transformation of colonic epithelial cells is thought to result from dysregulation of T cell factor (TCF)/β-catenin-induced gene expression [378], a key molecular pathway responsible for embryonic patterning [379]. Disruption of normal gene expression regulation may result from mutations in the glycogen synthase kinase 3β phosphorylation sites of β-catenin, preventing normal clearance of the TCF co-factor [380],

55 or through truncation of adenomatous polyposis coli (APC) [381, 382]. Genes targeted by TCF/β-catenin are associated with proliferation [383] and transformation of cells [384], although dysregulation of additional pathways is likely required [385, 386]. The genes that are most commonly mutated in association with CRC are illustrated in Figure 1.7.

KRAS is a member of the RAS family of genes, which encode guanine-5’-triphosphate (GTP)-binding proteins. KRAS signals primarily but not exclusively through the BRAF and MAPK pathways, and is a crucial effector of EGFR signalling. Between 32-40% of CRC cases can be found to contain mutations in KRAS, and around 85-90% of these mutations are in codons 12 or 13 [387-389]. Mutations in KRAS interfere with GTPase activity, resulting in accumulation of an activated GTP-bound form and persistent activation of downstream signalling pathways. Downstream of KRAS, BRAF is mutated in approximately 15% of CRC cases [390, 391]. At a frequency of over 95%, the most commonly reported BRAF mutation is the V600E mutation within the kinase activation domain. This mutation increases activation of the BRAF target, MAP2K, and subsequently MAPK1/3 [388]. Additional signalling pathway involvement is likely, as BRAF and KRAS mutations have differing histology and clinical characteristics [392]. BRAF mutant CRC is also associated with the CpG island methylator phenotype (CIMP) and microsatellite instability, while KRAS mutants tend to have low CIMP and are microsatellite stable [391, 392].

Figure 1.7 Signalling Pathways Commonly Mutated in Colorectal Cancer While combinations of mutations are likely involved in the pathology of the disease, CRC is commonly associated with mutations in several specific genes, the most common of which have been circled. Mutations in KRAS, BRAF, PIK3CA and PTEN downstream of EGFR promote proliferation of CRC cells, while constitutive activation of P53 effectively immortalises the cell by inhibiting apoptosis, and regulating the cell cycle. Mutations in APC can be found in cases of familial adenomatous polyposis and sporadic CRC, resulting in the accumulation of β-catenin and the dysregulated expression of various oncogenes [393-395].

PIK3CA mutations are present in between 15-18% of CRC cases [396, 397], with over 80% of PIK3CA mutations occurring in either exon 9 or 20 [388]. PIK3CA encodes the p110α

57 isoform of the catalytic subunit of a class IA PI3K heterodimer. Exon 9 mutations result in functional independence from the regulatory p85 subunit and constitutive activation dependant on interaction with RAS-GTP, while exon 20 results in functional independence from RAS-GTP but are dependent on interaction with p85 [398]. PI3K signalling is inhibited by phosphatase and tensin homologue (PTEN). PTEN activity can be impaired through direct mutation, chromosomal aberrations or promoter hypermethylation, resulting in elevated PI3K activity [399].

1.3.3 Additional Risk Factors As has been previously discussed, CRC is associated with predisposing conditions such as obesity and diabetes mellitus, medical procedures such as cholecystectomy, and lifestyle factors such as smoking, alcohol consumption and diet. Recently, the World Health Organisation (WHO) classified processed meats as a group 1 carcinogen due to epidemiological studies demonstrating positive associations with development of CRC, and to a lesser extent pancreatic and prostate cancer [359]. Mechanistically, this association may be due to the presence of preservatives and other chemical compounds used in processed meats. Red meat was also classified as a group 2A carcinogen, although the association between red meat and cancer may be accounted for by the production of polycyclic aromatic hydrocarbons (PAHs) and heterocyclic amines (HCAs) during open flame or high temperature cooking, rather than by the meat itself. It may be postulated that countries with an elevated incidence of CRC, such as New Zealand, may favour these types of cooking methods.

PAHs are organic compounds with two or greater fused aromatic rings, and tend to be lipophilic. Although they can be detected in the air, water and soil, most non-occupational exposure amongst non-smoking individuals is through food consumption [400-403]. PAHs are produced through a pyrolytic process during high temperature treatment of organic materials, and can even be found in alcoholic beverages due to toasting of raw materials in some production methods [404]. Several PAHs have been evaluated by chemical and food safety bodies including the International Program on Chemical Safety (IPCS), the Joint FAO/WHO Expert Committee on Food Additives (JECFA), and the Scientific Committee on Food (SCF). Both the SCF and JECFA identified a range of PAHs with mutagenic and genotoxic activity on somatic cells, and carcinogenicity in experimental animals [405]. HCAs

58 are produced by chemical reactions that occur at high temperatures, originating from amino acids, sugars and creatine/creatinine [406, 407]. They have demonstrated potent mutagenic properties in bacterial assays, and have induced tumourigenesis in the gastrointestinal tract in animal models [408-410]. Epidemiological association of HCAs with CRC in humans is inconsistent, with evidence both indicative of an association [411-413], or disparative [414- 416]. Variations in acetylation of HCAs and PAHs by N-Acetyltransferase (NAT) 2 have been suggested as a potential contributor to discrepancies, as individuals with a propensity for rapid acetylation may more readily convert HCAs and PAHs into carcinogenic forms, increasing CRC risk [417, 418].

Another important dietary factor that may contribute to elevated incidence of CRC in New Zealand is consumption of dietary fibre. Unlike previous examples, this risk factor elevates in potency as its consumption is reduced, presenting rather as a preventative factor. Unfortunately, average Western diets have seen a significant reduction in dietary fibre, and New Zealand is no exception. As far back as 430 BC, Hippocrates explored the association between whole grain flour and maintenance of gastrointestinal health. The term ‘dietary fibre’ was originally coined by Hipsley in 1953, defined as indigestible components of plant cell walls including cellulose, hemicelluloses and lignin [419]. Dietary fibre was further developed into a hypothesis that postulated an inverse relationship between consumption and incidence of CRC and coronary heart disease [420, 421]. The definition of dietary fibre was subsequently expanded to include indigestible polysaccharides such as gums, mucilages, oligosaccharides and pectin [422]. Dietary fibres are thought to reduce the contact time of carcinogenic molecules with the intestinal epithelia, and promote growth of commensal bacteria [423, 424]. For example, inulin, a polymer of fructose found in asparagus, garlic and onions, can promote growth of Bifidobacterium while hindering pathogenic E. coli, Listeria, and Salmonella [425, 426]. Low dietary fibre intake has been found to be associated with a reduced prevalence of Clostridium, Eubacterium and Roseburia spp., and an elevated prevalence of Enterococcus and Streptococcus spp. [427].

As originally postulated by Burkett, Painter and others, dietary fibre intake is inversely related to the incidence of CRC [428]. Fungal polysaccharides in particular have been linked with increased survival in advanced stages of gastric and CRC [429], and can enhance immune functional parameters such as natural killer cell activity, lymphocyte cytotoxicity, and alter proportions of effector and suppressor cells [430]. Dietary fibres have also demonstrated proficiency in counteracting the detrimental effects of HCAs and PAHs, and

59 consumption of cereals, whole grains, fibre-rich fruits and vegetables, and even marine algae can reduce the prevalence of CRC [431, 432]. The anti-carcinogenic and anti-mutagenic properties of dietary fibres have been predominantly attributed to the production of short- chain fatty acids (SCFAs), although dietary fibres may also bind mutagenic or carcinogenic compounds, or dilute them through fecal bulking [433]. A diet particularly rich in non-starch polysaccharides has a significant effect on gut bacteria such as Ruminococcus bromii, which can degrade starch and produce SCFAs [434]. Fucoidan, a polysaccharide found in brown algae, can induce apoptosis in human CRC cells, stimulating death receptor pathways, increasing cleavage of caspases, and triggering mitochondria-mediated apoptosis [435]. Fucoidan can also induce apoptosis by downregulating the ERK pathway, and can induce cleavage of PARP, a mechanism of DNA repair.

1.3.4 Angiogenesis in Colorectal Cancer Proliferation of tumour cells and the formation of solid tumour mass remains, like other tissues throughout the body, dependent upon adequate nutrient supply. Depletion of nutrients, such as the development of hypoxia within the tumour core, can result in tissue necrosis. Therefore, it becomes imperative for a growing tumour to maintain sufficient blood supply, especially as the mass of the tumour begins to restrict diffusion from surrounding vessels. Angiogenesis is the process of forming new blood vessels, extended from the endothelium of existing vasculature. This process is a balancing act, promoted by signalling molecules such as VEGF, which can be produced by supporting tissues or tumour cells. CRC cells can produce both IL-6 [436], a cytokine with proangiogenic properties, and VEGF [437]. Microvessel density (MVD), a surrogate measure of tumour vascularisation, and VEGF expression, are directly related with poor prognostic outcomes amongst patients with CRC [438]. The VEGF gene encodes eight individual exons interspaced by seven introns [439]. The pre-messenger RNA from VEGF can yield two families of proteins, VEGFxxx or VEGFxxxb. Each family contains multiple protein isoforms, with differences predominantly in the C-terminal domain, while the binding domains of the VEGF tyrosine kinase receptor are generally conserved [440]. The VEGFxxx family of variants are pro-angiogenic, with roles in promoting cell survival, migration and proliferation through stimulation of VEGF receptor (VEGFR) 2 [441, 442]. Major isoforms of VEGFxxx include VEGF121, VEGF165 and VEGF189 [443], with minor published variants including VEGF145, VEGF183 and VEGF206 [439, 444-446]. The VEGFxxxb family of variants are anti-angiogenic, and result

60 from distal splicing in exon 8, within the C-terminal region of the VEGF pre-mRNA [447,

448]. The predominant isoform of VEGFxxxb is VEGF165b, with several variants also mentioned in the literature including VEGF121b, VEGF145b, VEGF183b and VEGF189b [448, 449]. Expression of VEGFxxx isoforms are upregulated across a multitude of tumour types, while expression of VEGFxxxb isoforms are notably downregulated in colon carcinoma [450], kidney cell carcinoma [448], melanoma [451] and prostate carcinoma [447].

VEGF165b acts as a competitive inhibitor of VEGF165, with relatively similar binding affinities between these two variant proteins [447, 452]. Overexpression of VEGF165b can inhibit the growth of various tumours, and administration of recombinant VEGF165b has been shown to be particular effective at inhibiting growth of colon carcinoma [447, 450, 453]. A similar therapeutic effect can be observed from neutralising VEGF with an antibody, such as

Bevacizumab [454]. Bevacizumab binds with similar affinity to both VEGF165 and

VEGF165b, and as some tumours are associated with an elevation in expression of VEGF165b, this presents a subtype of tumours that may grow slowly when compared to VEGFxxx expressing tumours, but that would be resistant to anti-VEGF therapy [450]. The immune system can also influence the expression of VEGF. Stimulation of CRC cells in vitro with

IL-17, a cytokine predominantly released by TH17 cells, induces increased expression of both VEGF and IL-6, as well as other pro-tumourigenic genes such as plasminogen activator urokinase (PLAU), neuropilin (NRP) 2 and inhibitor of DNA binding (ID) 3 [455]. IL-17 producing cells are frequently present in CRC tumours, with high prevalence amongst those still proficient in MMR [456]. Although expression of IL-17 by TH17 cells and macrophages has been implicated with elevations in MVD and poor survival amongst CRC patients [455], the role of TH17 cells in CRC is complex. In some cases, such as in the presence of an intraepithelial infiltrate, TH17 cells can correlate positively with improved survival [457].

1.3.5 Colorectal Cancer Staging For some time the progression of CRC was divided based on Dukes’ method, originally a three-grade ABC classification system separating carcinoma contained within the rectal wall (A), extrarectal invasion without lymph node metastases (B) and with lymph node metastases (C) [458]. The Dukes’ classification was later updated to include a fourth classification, (D), denoting the presence of distant metastases [459]. This outdated system was replaced by the Tumor Node Metastasis (TNM) staging outlined by the Union for International Cancer

Control (UICC) and the American Joint Committee on Cancer (AJCC), the 7th edition of which was released in 2009. This staging provides comprehensive classification, ranging from carcinoma in situ (Stage 0) through to metastasis in multiple organ sites (Stage IVB) [460]; however, there are some notable discrepancies between the TNM staging and patient prognosis, and controversies over changes made between iterations [461, 462]. Histopathologic tumour grading is an alternative approach to predicting pathogenesis of CRC, based on various anatomical features such as tumour vascularity, differentiation, and margination [463, 464]. Proposed alternatives to the TNM staging system include the assessment of tumour aggression based on budding along the leading edge, which has demonstrated potential prognostic value for resected CRC [465], and poorly differentiated clusters (PDCs) as a measure of CRC metastatic potential [466-468].

The immunoscore of Galon and colleagues was proposed as another alternative for staging of CRC (Figure 1.8) [469]. The immunoscore considers the distribution of immune cell populations throughout the core tumour (CT) mass, along the invasive margin (IM), and within organised lymphoid follicles, in addition to levels of immune effector and regulatory molecules as a determinant of tumour staging. Immune cell infiltrates in both the CT and along the IM are significantly correlated with patient outcome, and improves the accuracy of predicted survival [470]. In particular, infiltrates of CD3+, CD8+ and CD45RO+ T cells are positively associated with better prognosis amongst CRC patients [470-473]. Additionally, infiltrates consisting of TH1 cells are associated with a good prognosis in CRC patients [474- 476], while B cells, NK cells, macrophages and other T helper cell subsets differ in value between different tumour types and stages [477].

The Immunoscore is a scoring system that involves the evaluation of immune cell infiltration at the CT and along the IM by two lymphocyte populations. These populations are selected from three subsets which are associated with improved prognosis amongst CRC patients, defined as CD3+/CD45RO+, CD3+/CD8+ or CD8+/CD45RO+. Results range from an Immunoscore of 0 (I0) with low densities of both populations, through to an Immunoscore of 4 (I4) with both at a high density [478]. The presence of these immune cell infiltrates are associated with improved prognosis amongst CRC patients, hence a higher immunoscore is preferable. The Immunoscore system is a prognostic factor for disease-free survival (DFS), disease-specific survival (DSS) and overall survival (OS), including in early-stage CRC.

Figure 1.8 The Immunoscore for Staging of Colorectal Cancer Calculating the Immunoscore is an alternative method for staging CRC that is significantly associated with patient prognosis. The immunoscore of a tumour is calculated by staining two immune cell populations, and assessing their distribution at the core of the tumour (CT) and along the invasive margin (IM) by immunohistochemistry. These populations are selected from three marker sets that define immune cell populations associated with good prognosis amongst CRC patients, CD3+/CD45RO+, CD3+/CD8+ or CD8+/CD45RO+ cells [469]. A high (Hi) distribution is indicative of an infiltrate consisting of the target immune cell population, while low (Lo) distribution indicates scant or absent distribution. A higher immunoscore with more immune cell infiltration of the selected populations is associated with an improved prognosis for the patient.

1.3.6 Approved Treatment Current treatment for CRC involves a combination of surgical resection and chemotherapy, and radiotherapy in rectal cancer. Effective agents include 5-fluorouracil (5-FU), irinotecan, oxaliplatin and capecitabine. Monoclonal antibodies (mAb) are a promising alternative to conventional therapies, capable of directly facilitating immune-mediated anti-tumour activity by inducing complement-dependent cytotoxicity (CDC), antibody-dependent cellular cytotoxicity (ADCC) or through conjugation with radioactive or cytotoxic molecules [479]. mAbs can also be used to block growth receptors overexpressed by tumour cells, such as EGFR, HER2, hepatocyte growth factor receptor precursor (MET) and insulin-like growth factor (IGF) 1 receptor (IGF1R) [480-483]. An extensive variety of mAbs have been developed for their potential to bind to the surface of tumour cells, with targets including haematopoietic differentiation glycoproteins (e.g. CD20, CD30, CD33 and CD52) [484-486],

63 angiogenic factors (e.g. VEGF, integrin αVβ3 and integrin α5β1) [487] and components of the stroma or extracellular matrix (e.g. fibroblast activation protein, tenascin) [488-490].

Development of a novel mAb must follow a series of key checkpoints for evaluating both clinical efficacy and safety. These checkpoints include: (1) Preclinical characterisation of the mAbs chemical and physical properties; (2) Analysis of specificity against a panel of malignant and normal tissues; (3) Assessment of immune effector activity; (4) Determination of in vivo localisation and distribution in relation to a transplanted tumour model; (5) Humanisation; and (6) Evaluation of in vivo therapeutic activity [484, 490, 491]. Regulatory approval of mAb therapy is primarily based on benefits to overall survival over conventional therapy in large Phase III clinical trials; however, when there are limited standard therapeutic options available, surrogate markers of anti-tumour immune responses may be sufficient [479]. mAbs approved for use by the United States (US) Food and Drug Administration (FDA) include HER2-specific trastuzumab (Roche) [492], EGFR-specific cetuximab (ImClone LLC) [493, 494] and CD20-specific rituximab (Genentech) [495]. mAbs such as bevacizumab (Genentech), targeting VEGF, and cetuximab and panitumumab (Amgen) targeting EGFR can be effective in colorectal patients that are negative for KRAS mutations [480, 496, 497]. Bevacizumab was the first monoclonal approved for treatment of metastatic CRC in 2004, followed by cetuximab also approved in 2004, and panitumumab approved in 2006. Panitumumab is the only completely human monoclonal antibody approved for treatment of CRC, while cetuximab is chimaeric and bevacizumab is humanised [480, 496, 497]. Bevacizumab is often used in first and second line treatments of metastatic CRC in combination with fluoropyrimidine-based chemotherapy. Cetuximab and panitumumab are only suitable for wild-type KRAS CRC, as constitutive activation of KRAS overrides effective blockade of EGFR, which lies upstream of the KRAS signalling pathway. Alternative monoclonal therapies have also been considered for treatment of CRC, including labetuzumab targeting carcinoembryonic antigen, pemtumomab and oregovomab targeting mucin, trastuzumab and pertuzumab targeting HER-2, mapatumumab targeting TRAILR1 and sibrotuzumab targeting fibroblast activation protein [479]. These monoclonal antibodies are yet to receive approval for use in patients, and primarily target CRC-associated surface receptors or molecules, with modes of action beyond signalling pathway blockade. The modalities of action of these monoclonals will be discussed further in a later section.

In New Zealand since 2011, 95% of patients with non-metastatic colon cancer underwent surgical resection of their primary tumour [362]. Amongst patients that had a stage III colon cancer resected, 59% received adjuvant chemotherapy. Current recommendations include the post-operative administration of the FOLFOX chemotherapy combination of 5-FU, leucovorin and oxaliplatin, or alternative combinations of oral capecitabine or 5-FU with or without leucovorin where FOLFOX is not appropriate [498]. 92% of patients with non- metastatic rectal cancer underwent surgical resection of their primary tumour, with 52% of patients also receiving either pre- or post-operative radiotherapy, predominantly with a long exposure course [362]. Only 36% of patients with non-metastatic rectal cancer received adjuvant chemotherapy. Rates of resection were much lower amongst patients with metastatic CRC at 52%, with treatment recommendations generally palliative in the most advanced cases. 49% of patients with metastatic disease received adjuvant chemotherapy. Although the New Zealand public health system does not include the use of modern immunotherapies such as bevacizumab, cetuximab and panitumumab for treatment of CRC, these agents and various others are available through the private sector or in clinical trials [362].

1.3.7 Monoclonal Antibodies in Clinical Trials for Colorectal Cancer CTLA4 and PD1 are T cell receptors that have been targeted with immunotherapeutic mAbs. CTLA4 is a primary immune-checkpoint receptor involved in early T cell activation, counteracting the activity of the T cell co-stimulatory receptor CD28 [499-501]. While CTLA4 and CD28 share the ligands CD80 and CD86, CTLA4 has a higher binding affinity [502]. This enables CTLA4 to induce signalling-independent T cell inhibition by outcompeting CD28, and even actively removing CD80 and CD86 from the surface of APCs [503-507]. CTLA4 may also block T cell activation through activation of protein phosphatases (e.g. Src homology region 2-containing protein tyrosine phosphatase 2 (SHP2) and protein phosphatase 2 (PP2A)), which counteract TCR and CD28 induced kinase signalling pathways; however the details of this mechanism are still under investigation [500]. Despite predictions of immune toxicity from the observed lethal hyperimmune and autoimmune phenotype of Ctla4-knockout mice [508, 509], partial blockade of CTLA4 was demonstrated to yield an achievable therapeutic window [510]. This enabled the production and testing of humanised anti-CTLA4 mAbs [511], with ipilimumab (Bristol-Meyers Squibb) subsequently 65 approved for treatment of metastatic melanoma [512]. Ipilimumab is currently in a phase I/II clinical trial for treatment metastatic colon cancer in combination with the anti-PD1 monoclonal nivolumab (Bristol-Meyers Squibb) [513]. A recent phase II clinical trial with an alternative anti-PD1 monoclonal, tremelimumab (Pfizer), for treatment of refractory metastatic CRC had a single partial responder amongst 47 patients [514]. Alternative anti- CTLA4 monoclonals in clinical trials for CRC include AMP224 (Amplimmune) [515], PDR001 (Novartis) [516], REGN2810 (Regeneron) [517], BGB-A317 (BeiGene) [518] and MEDI0680 (MedImmune/AstraZeneca) [519].

PD1 is primarily involved in limiting T cell activity in peripheral tissues during inflammation [520-522], significantly contributing to immunoregulation in the tumour microenvironment [523-525]. PD1 expression is induced during T cell activation [522], known to inhibit signalling pathways through activation of the phosphatase SHP2 when triggered by its respective ligand [526]. PD1 has two ligands, PDL1 (CD274) and PDL2 (CD273) [527], which are thought to also have a corresponding co-stimulatory T cell receptor with a relationship reminiscent of the interaction between CD80/CD86 and CD28/CTLA4 [528]. Initial clinical trials of Anti-PD1 treatment produced promising results, with a lower frequency of immune-related toxicities than observed with anti-CTLA4 treatment. Pembrolizumab (Merck) is an anti-PD1 monoclonal that has recently gained FDA approval for use in CRC based on the results of its phase II clinical trial [529]. In this trial, the immune- related objective response rate was 40% in MMR-deficient CRC patients, with an immune- related progression-free survival rate of 78%. These rates were comparably lower amongst non-MMR-deficient CRC patients, at 0% and 11% respectively, indicating clear differences in treatment efficacy respective to MMR state. Alternative anti-PD1/anti-PDL1 monoclonals in clinical trials for CRC include MEDI4736 (MedImmune/AstraZeneca) [530], nivolumab (Bristol-Meyers Squibb) [513], azetolizumab (Roche) [531] and avelumab (EMD Serono) [532].

In addition to CTLA-4 and PD-1/PDL-1 there are multiple alternative immune receptors that have been explored as potential targets for monoclonal therapy. These include targeting alternative inhibitory receptor pathways for blockade, or targeting stimulatory immune receptors to promote activation of immune cells. Alternative inhibitory receptor blockade monoclonals currently in trial for CRC include BMS-986016 (Bristol-Meyers Squibb), an anti-LAG-3 antibody [533], and ARGX-110 (arGEN-x), an anti-CD70 antibody [534]. LAG- 3 (CD223) is a member of the immunoglobulin superfamily expressed by activated T cells,

B cells, NK cells and plasmacytoid DCs [535]. LAG-3 binds to MHC-II, and has a role in the negative regulation of T cell proliferation. A member of the tumour necrosis factor family of receptors, CD70 is normally expressed by activated T and B cells, and mature DCs. Tumour cells can also express CD70, and through interactions with it complementary receptor CD27, CD70 drives T cell exhaustion, induces T cell apoptosis and increases the + amount of Treg cells [536]. CD70-expressing tumours can also recruit an infiltrate of CD27 + Treg cells [537, 538]. Only around 9% of CRC tumours are CD70 [539], so it is uncertain how effective targeting this pathway will be for treating this disease.

The repertoire of immunostimulatory receptors targetable by agonistic monoclonals includes OX40, 4-1BB, GITR, CD40 and CD27. OX40 (CD134) is a member of the tumour necrosis factor family of receptors, and is transiently expressed by CD4+ and CD8+ T cells following antigen-specific priming [540]. Its ligand, OX40L, can be found on the surface of activated

APCs, as well as endothelial and epithelial cells [541]. OX40 agonists directly regulate Treg cells, impairing their immunosuppressive effects, and subsequently promoting anti-tumour

TC cell responses [540]. OX40 agonist monoclonals currently in clinical trials for treatment of CRC include MEDI6469 (AgonOx) [542], MEDI6383 (AgonOx) [543] and MOXR0916 (Genentech) [544]. 4-1BB (CD137) is a member of the tumour necrosis factor family of receptors, expressed primarily by activated T cells [545]. Its ligand, 4-1BBL, is expressed by activated APCs and stimulates T cell proliferation and differentiation. Expression of 4-1BBL is suppressed in CRC, and impairment of this pathway is implicated in promoting immune escape [546]. Targeting 4-1BB with agonistic antibodies has already demonstrated efficacy against CRC in animal models [547, 548], and monoclonals currently in clinical trials include urelumab (Bristol-Meyers Squibb) [549] and PF-05082566 (Pfizer) [550].

GITR (CD357) is a costimulatory receptor constitutively expressed by Treg cells, and transiently expressed by activated CD4+ and CD8+ T cells [551]. Its ligand, GITRL, is expressed by activated APCs. GITR provides costimulation for effector T cells, while also impairing the suppressive activity of Treg cells [552]. Stimulation of GITR with agonistic antibodies can promote tumour regression, while inducing lineage instability in Treg cells [553]. GITR agonistic monoclonals currently in clinical trials include TRX518 (Tolerx) [554] and MK-4166 (Merck) [555]. CD27 is the ligand for CD70, and is a tumour necrosis factor family receptor involved in activation, proliferation and survival of T cells [556]. There is currently only one agonistic monoclonal in clinical trials for CRC, which is varlilumab (Celldex) [557]. There are several additional receptors that are also currently under

67 investigation as immunotherapeutic targets, including B- and T-lymphocyte attenuator (BTLA), TIM3 and adenosine A2A receptor (A2aR) [558, 559]. With a multitude of avenues to explore and promising clinical data, the future appears bright for monoclonal therapies.

1.4 Therapeutic Vaccination Prophylactic cancer vaccination has already demonstrated significant success in the form of Cervarix [560] and Gardasil [561]; however, an approved therapeutic vaccine suitable for treatment of established tumours is yet to surface from clinical trials. Therapeutic cancer vaccination is foreshadowed by several limiting factors that impair the induction of an effective antitumour response, including the non-ubiquitous expression of target tumour- specific or tumour-associated antigens throughout the heterogeneous tumour mass, or the presence of an overwhelmingly immunosuppressive tumour microenvironment. Despite these daunting obstacles, many new therapeutic vaccination methods are currently being developed and refined to improve upon the induction of an antitumour immunity.

Some of the earliest cancer vaccines were based on the administration of selected proteins or peptides, including heat-shock proteins (HSPs) [562], peptide epitopes/agonist peptides [563- 565], fusion proteins [566] and anti-idiotype antibodies [567, 568]. Suitable peptide vaccines can be produced with reverse immunology, using computer algorithms to predict epitopes present in tumour-specific antigens (TSAs) and TAAs. These predicted epitopes are subsequently validated in vitro for binding specificity and stability, prior to consideration for clinical trials. Alternatively, direct immunology involves isolation of patient-derived autologous tumour-specific CTLs, which can be screened to identify the corresponding peptide for vaccine validation [569]. While these vaccines are relatively simple to produce, they often suffer from poor immunogenicity, susceptibility to tumour escape and induction of an ineffective TH2 immune response [570, 571]. Vaccination efficiency can be improved by utilising a surrogate delivery system, such as preloading ex-vivo-derived DCs with the vaccine [572, 573]. Exposing DCs produced from haematopoietic progenitor cells to the vaccine ex vivo avoids interference from endogenous regulatory mechanisms that may impair activation, and enables controlled administration of stimulatory cytokines (e.g. GM-CSF). Sipuleucel-T (Provenge) is a prostatic acid phosphatase (PAP)-based DC vaccine which demonstrated particularly significant potential in clinical trials, now approved for treatment of metastatic prostate cancer [574, 575].

Unlike transient peptide vaccination, DNA-based vaccines have the potential to establish an endogenous supply of the protein antigen to prolong exposure and incite a potent immune response. While the therapy can be delivered directly, an enhanced delivery system (e.g. Gene Gun, cationic liposomes) may be used to circumvent degradation by DNases and enhance treatment efficiency [576, 577]. DNA-based vaccines can also be improved by 69 modification of the plasmid-encoded antigens [578, 579], the addition of biological adjuvant genes (e.g. hepatitis B surface antigen) [580] and simultaneous cytokine administration (e.g. granulocyte macrophage colony-stimulating factor (GM-CSF) or IL2) [581]. mRNA-based vaccines are another intriguing prospective immunotherapeutic for cancer treatment, based on the transient transfection of non-dividing cells [582, 583]. An mRNA-based vaccine can also be electroporated into DCs ex vivo to improve transfection efficiency, enhancing therapeutic delivery [584, 585]. As mRNA does not integrate into the host genome, mRNA- based vaccines are considered relatively safe [586]; however, antigen expression will be limited by mRNA degradation mechanisms. Genes encoding TSAs and TAAs can also be incorporated into recombinant vectors including viruses, bacteria and yeast. Poxviruses are the most commonly used vector for recombinant vaccine delivery, with a relatively large double-stranded DNA genome of 130 kb in mammalian strains allowing insertion of over 10 kb of foreign DNA [569]. This method harnesses the expression of poxvirus gene products to produce a potent immune response against recombinant antigen products, further enhanced by expression of T cell co-stimulatory molecules and cytokines induced by some modified poxviruses.

1.4.1 Virus-like Particles Virus-like particles (VLP) can be formed from self-assembling virus-derived capsid proteins, or viral proteins embedded in a phospholipid bilayer. VLP can be differentiated from their parent virus due to the absence of viral genomic material, and are roughly comparable to a hollow shell. This also renders VLP incapable of infection or replication, making them a safe subunit alternative to intact virus-based vaccines. The viral capsid proteins that form VLP retain their natural conformation, with respective antigenic motifs presented in a state more immunologically relevant than those present in an inactivated virus vaccine. VLP are a natural by-product that is formed during the infection cycle of some viruses, and were first identified in patients infected with hepatitis B virus (HBV) [587]. VLP can be manufactured in vitro using various protein expression systems, such as in bacteria, yeast, mammalian and insect cell lines, plants and cell-free expression systems. Expression of virus capsid proteins for production of VLP in vitro also enables the formation of VLP from viruses that cannot be routinely cultured, such as hepatitis C virus and human norovirus [588, 589]. VLP have also been found to be highly stable through prolonged storage [590-592], an important consideration in vaccine development. An HBV vaccine, Engerix (GlaxoSmithKline), was

70 the first VLP-based vaccine to be approved for use in humans [593]. Additional VLP-based vaccines that have been approved include Epaxal (Crucell), a vaccine for hepatitis A virus [594], Inflexal V (Crucell), a vaccine for influenza A [595], and Gardasil (Merck) and Cervarix (GlaxoSmithKline), vaccines previously mentioned for HPV [596].

The morphology of a VLP is defined by its parent virus, inheriting both its size and characteristic shape (e.g. spherical, icosahedral, rod-like). VLP can be produced with multiple degrees of complexity, ranging from a single type of capsid protein in a single layer, through to multi-layer and enveloped VLP (Figure 1.9) [597]. Some examples of simple VLP with a single layer include HBV VLP produced from the HBV surface antigen [593], norovirus VLP produced from the VP1 capsid protein [598], and Qβ VLP produced from Qβ bacteriophage coat protein [599-601]. Simultaneously expressing multiple viral structural proteins can facilitate the production of VLP with multiple distinct capsid layers, such as in double or triple layered rotavirus VLP [602, 603]. VLP can also be produced from multiple strains of the same virus, creating a mosaic VLP that can theoretically confer protection against each of the strains incorporated into the particle. Further complexity can be added through supporting the formation of a phospholipid envelope around VLP derived from enveloped viruses, such as influenza A virus. Co-expression of influenza A virus haemagglutinin, neuraminidase, matrix protein and ion channel protein results in the formation of influenza VLP encapsulated inside a phospholipid envelope [604]. Influenza A virus VLP retain the characteristic surface haemagglutinin and neuraminidase spikes of their parent virus, and can be produced from multiple influenza strains. Expression of only the membrane anchored influenza A virus proteins can also form a single-layered VLP called a virosome [605], VLP formed from virus proteins anchored in a phospholipid bilayer.

Figure 1.9 Structural conformation repertoire of VLP The structural variation amongst VLP is as diverse as the broad range of viruses from which they can be assembled. The simplest form, (A) Single Protein VLP, can be formed by expressing a single self-assembling capsid protein from their parent virus. Examples with increased complexity include multi-protein VLP arranged in (B) a Single Layer, (C) Multiple Layers, or (D) a Mosaic pattern. Some types can even include a phospholipid bilayer in their structure, forming (E) an Enveloped VLP or (F) a virosome particle. VLP can also be used as a scaffold for various molecules, incorporated into the particle through (G) Recombination or (H) Chemical Conjugation depending on the type of molecule [606].

VLP are a versatile vaccine vector due to their ability to support modification. These modifications include the recombinant insertion of peptides, or the chemical conjugation of peptides, proteins, lipoproteins, carbohydrates and entire cellular lysates. Modifications that incorporate heterologous antigens produce chimaeric VLP capable of inducing immune responses against antigens other than those derived from the parent virus of the VLP. VLP can support recombinant peptide insertion of considerable length without impairing structural integrity, variably depending on the type of VLP. For example, polyomavirus VLP can support the insertion of up to 120 amino acids in its VP1 capsid protein [607-610]. Polyomavirus VLP can also be produced containing an additional non-structural capsid protein, VP2, which can support insertion of large truncated proteins [611]. A wide variety

72 of commercially available vaccines are based on the recombinant insertion of epitopes into the HBV core antigen of HBV VLP, such as the pan-influenza A vaccine ACAM-FLU A [612], and the Plasmodium falciparum anti-malaria vaccine Malariavax [613]. VLP can also be used as a scaffold for chemical conjugation through standard protein conjugation methods, such as through carboxylic acid, sulfhydryl groups or amine groups of amino acids, azide- alkyne click chemistry [614], biotin-streptavidin complexes [615], and through heterobifunctional or biorthogonal cross-linkers.

The repetitive structure of VLP promotes the internalisation, processing and presentation of integrated antigens by various APCs, including DCs, macrophages and B cells [616]. The route of administration and the physical properties of the VLP predominantly determine which immune cells will preferentially internalise the particles, and how they will be processed. One of the most important determinants is VLP size, with particles larger than 200 nm in diameter requiring uptake into APCs for efficient transport into lymph nodes [617]. VLP can be internalised through various mechanisms including micropinocytosis, phagocytosis and receptor-mediated endocytosis. VLP that retain receptor ligands can gain entry into cells through the same mechanisms as their parent virus, such as influenza A VLP binding to sialic acid through haemagglutinin. VLP can also be modified to target specific receptors for receptor-mediated endocytosis, such as the conjugation of superantigen onto HBV VLP for internalisation through MHC-II [618]. VLP are potent activators of the humoral arm of the adaptive immune system, their repetitive surface promoting cross-linking of the BCR. The capacity of VLP to stimulate activation of B cells and the production of antibodies is particularly useful in prophylactic vaccines against the parent virus of the VLP, such as in vaccines for HBV [593], HPV [596] and influenza A virus [619]. Some VLP can also be cross-presented by specific APCs onto MHC-I for stimulation of the cell-mediated arm of the adaptive immune system [620, 621]. VLP that can be cross-presented include Qβ VLP, which are cross-presented through TAP-dependent pathways in DCs and TAP- independent pathways in macrophages [621], and rabbit haemorrhagic disease virus VLP, which are cross-presented through the MHC-I receptor recycling pathway [622]. Additional examples of VLP that are cross-presented include hepatitis C virus VLP [623], human immunodeficiency virus VLP [624], HPV VLP [625] and papaya mosaic virus VLP [626].

1.4.1.1 VLP as Therapeutic Cancer Vaccines VLP have a broad range of potential applications, including as potential vectors for cancer vaccines. Where the prophylactic HPV VLP vaccines Gardasil and Cervarix provide protection against the causative agent of cervical cancer, HPV, a vaccine has also been developed that can therapeutically treat existing cervical cancer. This vaccine is based on VLP formed from the infectious bursal disease virus (IBDV) VP2 capsid protein. These VLP are formed from 20 trimeric clusters of VP2 with a T=1 structural symmetry, and can support insertion at the N- and C-terminus of VP2, or in the hypervariable loops [627, 628]. Recombinant insertion of a C-terminal fragment of the HPV-16 E7 protein into IBDV VLP and subsequent vaccination of TC1 cervical cancer tumour-bearing mice induced complete remission of tumours in all vaccinated mice [629]. VLP can also deliver TAAs in an immunogenic manner, potentially stimulating immune responses against tolerised epitopes. For example, incorporation of three B cell epitopes derived from the extracellular domain of HER2/neu into influenza A virosome VLP produces a vaccine capable of inducing production of HER2/neu-specific antibodies. This vaccine has completed a phase I clinical trial, producing HER2/neu-specific antibodies in 80% of patients, and reducing the number of Treg cells [605]. The Qβ VLP has also been adapted for use as a cancer vaccine, through recombinant insertion of a peptide from the melanoma antigen Melan-A/Mart-1. In a phase I/II clinical trial, the Melan-A/Mart-1 Qβ VLP induced detectable Melan-A/Mart-1-specific T cell responses in 64% of patients [630]. RHDV VLP has also demonstrated promise in preclinical studies, capable of inducing potent cytotoxic immune responses that delay tumour growth and improve survival in subcutaneous murine tumour models [631-633].

1.4.1.2 Rabbit Hemorrhagic Disease Virus VLP RHDV is a member of the Caliciviridae family of the genus Lagovirus, consisting of non- enveloped single-stranded positive-sense RNA viruses, and is responsible for a contagious haemorrhagic disease in rabbits [634]. The RHDV viral capsid consists of 180 copies of the 60 kDa major capsid protein, VP60, arranged as 90 dimers into a T=3 icosahedral structure with a diameter of around 40 nm [635]. The VP60 protein can be divided into four individually identifiable domains, the N-terminal domain (amino acids 1-65) contained within the capsid, the shell domain (amino acids 66-229) that forms the backbone of the icosahedral capsid, a short hinge domain (amino acids 230-237) and the C-terminal protruding domain (amino acids 231-579) exposed on the surface [636-639]. Recombinant

74 expression of the RHDV VP60 protein results in the spontaneous formation of VLP that are structurally indistinguishable from the native virus [634, 635]. Expression of RHDV VLP has achieved in bacteria [640], yeast [641, 642], plants [643, 644] and in insect cells [633- 635, 638, 645].

RHDV VLP were originally developed for vaccination of domestic rabbits against RHDV, but have since proven to be particularly versatile as a vector for constructing cancer vaccines. Vaccination using VLP derived from non-human viruses are particularly desirable as vaccine vectors, as pre-existing immunity can be detrimental to the generation of anti-tumour immune responses for some VLP vaccines [646]. RHDV VLP are internalised by APCs through phagocytosis and macropinocytosis, and are cross-presented through the MHC-I receptor recycling pathway [622]. Conjugation onto the surface of RHDV VLP can enhance the immunogenicity of antigens, as evidenced by the significantly greater degree of ovalbumin- specific OT-II cell proliferation in OT-II splenocytes pulsed with ovalbumin conjugated on VLP in comparison to ovalbumin protein alone [647]. Several MHC-I restricted model peptide epitopes have also been recombinantly inserted at the N-terminus of RHDV VP60, or conjugated onto the surface of RHDV VLP, including the H-2Kb-restricted chicken ovalbumin epitope SIINFEKL [633, 648], and the H-2Db-restricted LCMV gp33 epitope SAVYNFATM [632, 649]. These model antigens are used due to the availability of mice transgenic for specificity of their TCR against these epitopes [650, 651], and are commonly used model targets genetically engineered for expression in tumour cell lines. Ovalbumin- expressing melanoma (B16-OVA)-bearing mice vaccinated with RHDV VLP containing the SIINFEKL epitope have delayed tumour growth and enhanced survival [633]. A similar response can also be observed in gp33-expressing Lewis’ lung carcinoma (LL-LCMV)- bearing mice vaccinated with RHDV VLP containing SAVYNFATM [632]. More recently, RHDV VLP have been developed containing epitopes derived from the melanoma antigen gp100, singularly or in multiple copies [652]. Vaccination of B16-bearing mice with RHDV VLP containing gp100 epitopes can also delay tumour growth and improve survival. These responses were further improved by vaccination with RHDV VLP containing two copies of the gp100 epitope, indicating that antigen dosage plays an important role in the potency of cancer vaccines. Improved functionality of the RHDV VLP vaccine has also been explored through the conjugation of adjuvants, such as CpG oligonucleotides and α- galactosylceramide [649, 653], or by targeting alternative internalisation pathways, such as through mannosylation of VLP [648].

1.5 Project Objectives RHDV VLP have already demonstrated themselves as an exceptional cancer vaccine vector in model antigen systems. They have the ability to naturally associate with various molecules and compounds, producing composite vaccines capable of stimulating targeted immune responses. This PhD project is composed to two interrelated objectives: (1) Investigating the interactions between RHDV VLP and nucleic acids for modification of immune responses to vaccines; and (2) The development of a VLP-based vaccine for CRC. These objectives combine to explore the potential for a CRC vaccine capable of inducing a targeted anti- tumour immune response, while simultaneously manipulating components of the immune system.

Modification of immune responses to vaccines can be achieved through the use of adjuvants that selectively induce the activation of particular immune cell subsets, or promote a specific type of immune response. In this project, DNA oligonucleotides containing unmethylated CpG islands (CpG) were investigated for their potential ability to be encapsulated inside RHDV VLP, based on the ability of VLP to encapsulate short nucleic acids during assembly. In addition, targeted manipulation of expression pathways in APCs was assessed by investigating the ability of RHDV VLP to encapsulate siRNA, with the potential for selective knockdown of suppressive pathways to promote activation. This PhD project concludes with the development of a VLP-based vaccine for CRC. In a recent study, epitopes derived from the melanoma-associated antigen gp100 were inserted at the N-terminus of RHDV VP60, forming VLP that delayed tumour growth and prolonged survival in a subcutaneous murine melanoma model [652]. Despite the induction of a targeted immune response, most mice in this study eventually succumbed to their tumours. It is possible that the tumours present in vaccinated mice consisted of populations of cells with low gp100 expression, through changes either in expression or through selective elimination of high gp100 expression cells. This study will investigate whether targeting multiple epitopes simultaneously in the same vaccine construct can provide an advantage in an immunotherapeutic vaccine by hindering tumour escape.

1.5.1 Hypotheses 1. Nucleic acids will associate with RHDV VLP, as has been previously observed for α- galactosylceramide and mannose. 2. Chimaeric RHDV VLP will support the insertion of multiple peptides derived from murine CRC-associated antigens. 3. Chimaeric RHDV VLP containing peptide epitopes derived from murine CRC- associated antigens will induce targeted immune responses against their inserted epitopes. 4. Chimaeric RHDV VLP containing peptide epitopes derived from murine CRC- associated antigens will be an effective immunotherapeutic vaccine in a murine model of CRC. 5. Chimaeric RHDV VLP containing multiple peptide epitopes derived from murine CRC-associated antigens will impair tumour growth, and enhance survival in a murine model of CRC compared to chimaeric RHDV VLP containing single epitopes.

2 Materials and Methods

2.1 Development of Chimaeric RHDV VLP This study involved the use of both previously constructed and newly developed chimaeric RHDV VLP. Each construct was based on the ‘Sugar Loaf’ variant of VP60 from a locally identified strain of RHDV. Both VP60, and the recombinant form containing the H-2Kb- restricted chicken ovalbumin epitope SIINFEKL at the N-terminus of VP60 (SIIN.VP60), were provided pre-constructed and packaged in verified recombinant baculovirus inoculums. The recombinant constructs containing either an octaarginine sequence (R8.VP60) or a DNA binding site derived from HPV-16 (DBS.VP60) at the N-terminus of VP60 were provided pre-cloned into pAcUW51(GUS) plasmids. Each of the constructs that contained epitopes derived from the CRC-associated antigens topoisomerase IIα and survivin were developed from custom synthetic sequences provided in pUC57 or pUC57-Simple plasmids (Genscript, New Jersey, USA). These constructs included insertions of either the epitopes from topoisomerase IIα (T.VP60), survivin (S.VP60) or both (TS.VP60) at the N-terminus of VP60, each of these epitopes at the N- or C-terminus (S.VP60.T), and a modified version of TS.VP60 with a C-terminus insertion of the topoisomerase IIα epitope (TS.VP60.T). These constructs and their respective sequences are summarised in Table 2.1.

2.1.1 Permits and Approvals The development, production and use of recombinant baculoviruses expressing each RHDV VP60 construct were undertaken under the Institutional Biological Safety Committee (IBSC) approval number GM010/UO006, including Environmental Risk Management Authority (ERMA) approval numbers GMD100300, GMD100301 and GMD100302, or BOK numbers GMD101715 and GMD101729.

Table 2.1 Summary of Chimaeric RHDV VLP Constructs Construct Sequence* SIIN.VP60 M.SIINFEKL.GGS.VP60 R8.VP60 M.RRRRRRRR.GGS.VP60 DBS.VP60 M.STTAKRKKRKL.GGS.VP60 T.VP60 M.DSDEDFSGL.GGS.VP60 S.VP60 M.TVSEFLKL.GGS.VP60 TS.VP60 M.DSDEDFSGL.ALL.TVSEFLKL.GGS.VP60 S.VP60.T M.TVSEFLKL.GGS.VP60.DSDEDFSGL TS.VP60.T M.DSDEDFSGL.ALL.TVSEFLKL.GGS.VP60.DSDEDFSGL * H-2Kb epitope sequences are highlighted in blue, DNA-binding sequences are highlighted in purple, and linker sequences are highlighted in red.

2.1.2 Construction of Expression Plasmids Each RHDV VLP construct that was developed completely for the purposes of this study (e.g. T.VP60, S.VP60, TS.VP60, S.VP60.T and TS.VP60.T) were ligated together through a multi-step process into pAcUW51 (GUS) supplied by the Ward Laboratory (Department of Microbiology and Immunology, University of Otago, NZ) using established methods [652]. The N-terminus of each recombinant VP60 was supplied pre-synthesised, and supplied in either pUC57 of pUC57-Simple (GeneScript, New Jersey, USA). Each construct was designed to insert the recombinant sequence between the initial methionine and glutamic acid residues, and included sequences corresponding to the N-terminus of VP60 up to and including its PstI restriction digestion cleavage site. These epitope sequences were obtained from the National Center for Biotechnology Information (NCBI) database (Mus musculus topoisomerase IIα Gene ID: 21973, Mus musculus baculoviral IAP repeat-containing 5 also known as survivin Gene ID: 11799). Each epitope sequence was manually codon optimised for expression in Spodoptera frugiperda using the Kazusa Codon Usage Database [654]. The sequences supplied are provided below in Table 2.2. Linker sequences incorporated into each construct included a spacer consisting of glycine-glycine-serine (GGS) between the recombinantly inserted sequences and the remainder of VP60, and an immunoprotease cleavage linker consisting of alanine-lysine-lysine (ALL) between epitopes. While the GGS spacer has been historically used in RHDV VLP constructs produced by our laboratory, the

ALL linker was only recently investigated for its ability to enhance the immunogenicity of multiple epitopes incorporated into the same construct [652].

Table 2.2 Pre-synthesised Recombinant VP60 N-terminus Sequences Construct Sequence* T.VP60 CACAGATCTAAAATGGACTCGGACGAAGACTTTTCGGGC TTGGGAGGTTCCGAGGGCAAAGCCCGTGCAGCGCCGCAAG GCGAAGCAGCGGGCACTGCCACCACAGCATCAGTTCCCGGA ACCACGACTGATGGCATGGATCCTGGCGTTGTGGCCACTAC CAGCGTGATCACTGCAGAAA S.VP60 CACAGATCTAAAATGACGGTGAGCGAATTTTTGAAATTG GGAGGTTCCGAGGGCAAAGCCCGTGCAGCGCCGCAAGGCG AAGCAGCGGGCACTGCCACCACAGCATCAGTTCCCGGAACC ACGACTGATGGCATGGATCCTGGCGTTGTGGCCACTACCAG CGTGATCACTGCAGAAA TS.VP60 CACAGATCTAAAATGGACTCGGACGAAGACTTTTCGGGC TTGGCCTTGTTGACGGTGAGCGAATTTTTGAAATTGGGAG GTTCCGAGGGCAAAGCCCGTGCAGCGCCGCAAGGCGAAGC AGCGGGCACTGCCACCACAGCATCAGTTCCCGGAACCACGA CTGATGGCATGGATCCTGGCGTTGTGGCCACTACCAGCGTG ATCACTGCAGAAA * H-2Kb epitope sequences are highlighted in blue and restriction enzyme target sequences are highlighted in red and purple for BglII and PstI respectively. The methionine initiation codon for each sequence has been underlined.

2.1.2.1 Plasmid Transformation and Production Stocks of calcium competent E. coli XL-1 Blue MRF cells were supplied at -80 °C by the Ward Laboratory for plasmid transformation, replication and extraction. These stocks were produced by streaking E. coli XL-1 Blue MRF cells out onto Luria-Bertani (LB) agar plates containing 12.5 μg/mL tetracycline, incubated overnight at 37 °C. A single colony was isolated from the plate, and was used to inoculate 5 mL of LB broth containing 12.5 μg/mL tetracycline. The inoculated broth was incubated at 37 °C for 16 hours under agitation. 1.2

80 mL of this culture was used to inoculate a flask containing 120 mL of LB broth, and was incubated as previously described until the optical density at 600 nm (OD600) reached approximately 0.4. Upon reaching the desired density, the flask culture was incubated on ice for 15 minutes, and then centrifuged at 3000 xg at 4 °C for 10 minutes. Upon removal of the media, the pelleted cells were resuspended in 40 mL of Transformation Buffer I pre-chilled at 4°C. The resuspension was then centrifuged at 3000 xg at 4 °C for 10 minutes, after which the pelleted cells were resuspended in 40 mL of Transformation Buffer II pre-chilled at 4°C. This resuspension was distributed into 100 uL aliquots that were subsequently stored at -80 °C.

Plasmids were transformed into these calcium competent E. coli cells by heat shock. For each transformation, a stock aliquot of cells was removed from -80 °C storage and was thawed on ice for 15 minutes. A 10 μL solution containing either the pre-synthesised plasmids diluted in Milli-Q H2O, or ligation reaction products, was then added to the cell aliquot and incubated for a further 15 minutes on ice. The cells were then heat shocked at 37 °C for 5 minutes, before returning to ice for 2 minutes. 100 uL of the resulting transformation solution was streaked onto a LB agar plate containing 50 μg/mL ampicillin, and incubated at 37 °C overnight. Selected colonies were used to inoculate 5 mL of LB broth containing 50 μg/mL ampicillin, and incubated at 37 °C for 16 hours. 3 mL of the resulting culture broth was centrifuged as two 1.5 mL aliquots at 12,000 x g for 1 minute for plasmid extraction from pelleted cells.

2.1.2.2 Plasmid Extraction Heat shock transformed, antibiotic-selected plasmid producing E. coli cells pelleted as previously described were processed for plasmid extraction using the AxyPrep Plasmid Miniprep kit (Axygen, California, USA) following the instructions provided by the manufacturer. Briefly, 2x 1.5 mL aliquots of plasmid containing E. coli cultured overnight at 37 °C in LB broth containing 50 μg/mL ampicillin were centrifuged at 12,000 xg for 1 minute. The pelleted bacteria were resuspended in 250 μL of Buffer S1 by briefly vortexing. 250 μL of Buffer S2 was added, and the tubes were inverted 6 times. 350 μL of Buffer S3 was added, and the tubes were inverted 8 times. The lysed material was centrifuged at 12,000 xg for 10 minutes, and the supernatant from both tubes was transferred into a Miniprep column loaded into a collection tube. The column was centrifuged at 12,000 xg for 1 minute.

As XL1-Blue E. coli are EndA deficient, Buffer W1 was skipped and 700 μL of Buffer W2 was added to the spin column. The column was centrifuged at 12,000 xg for 1 minute, before adding a further 700 μL of Buffer W2 and repeating centrifugation. The spin column was then centrifuged again at 12,000 xg for 1 minute, and transferred to a 1.5 mL microfuge tube.

50 μL of Milli-Q H2O was added, allowed to stand for 1 minute at room temperature, and the column was centrifuged at 12,000 xg for 1 minute. Purified plasmids were analysed for both quality and quantity using a NanoDrop 1000 (Thermo Scientific, Delaware, USA), and were then stored at -20 °C.

2.1.2.3 Restriction Digestion Restriction digests were performed using 1-5 μg of plasmid DNA, added to 2 μL of a 10X Restriction Buffer (Roche Diagnostics, Germany) appropriate for the restriction enzyme combination used, 1 μL of each appropriate restriction enzyme, and sufficient Milli-Q H2O as to bring the total reaction volume up to 20 μL. Restriction digestion reactions were incubated at 37 °C for 2 hours, prior to separation of digestion fragments by agarose gel electrophoresis.

2.1.2.4 Agarose Gel Electrophoresis For each restriction digestion reaction, 2 μL of 6X loading dye was added to a 10 μL of the reaction product. The resulting solution was loaded on a 1-2 % agarose gel containing 1X Tris acetate EDTA (TAE) buffer, and digestion fragments were separated by electrophoresis in 1X TAE buffer for 1 hour at 100 V. Samples were run alongside an aliquot of 1 Kb Plus DNA ladder (Invitrogen, California, USA) for size comparison. Upon completion of electrophoresis, gels were incubated in a solution of 1 μg/mL ethidium bromide in H2O for 30 minutes, before being visualised under ultraviolet light using a Bio-Rad ChemiDoc gel visualisation system (Bio-Rad, California, USA).

2.1.2.5 Gel Extraction of Digestion Fragments Following restriction digestion and separation by agarose gel electrophoresis, select digestion fragments were excised from the gel for extraction. These fragments were excised using a scalpel under reduced ultraviolet light exposure on a Bio-Rad ChemiDoc gel visualisation

82 system (Bio-Rad, California, USA). Excised agarose gel segments containing each digestion fragment were processed immediately for DNA extraction, or were frozen at -20 °C for later extraction. DNA extraction was performed using an AxyPrep DNA Gel Extraction kit (Axygen, California, USA) following the instructions provided by the manufacturer. Briefly, the extracted fragment of gel was weighed and assigned a volume as per the manufacturer’s instructions, based on 100 mg of gel equating roughly 100 μL sample volume (SV). 3X SV of Buffer DE-A was added to the gel and heated at 75 °C for 6 minutes. 1.5X SV of Buffer DE-B was added and mixed gently by inversion. 1X SV of isopropanol was then added if the DNA fragments were known to be less than 400 bp in length. The solubilised gel solution was transferred to a spin column loaded into a collection tube, and centrifuged at 12,000 xg for 1 minute. 500 μL of Buffer W1 was added, and the column was centrifuged at 12,000 xg for 30 seconds. 700 μL of Buffer W2 was added, and the column was centrifuged at 12,000 xg for 30 seconds. A further 700 μL of Buffer W2 was added, and centrifugation was repeated. The spin column was centrifuged for a further 1 minute at 12,000 xg, before 30 μL of MilliQ H2O was added to the spin column. The spin column was allowed to equilibriate for 1 minute at room temperature, then centrifuged at 12,000 xg for 1 minute. Extracted DNA fragments were analysed for both quality and quantity using a NanoDrop 1000, and were then stored at -20 °C.

2.1.2.6 DNA Ligation Ligation of appropriate combinations of DNA fragments was used to complete the formation of plasmids that could be amplified through transformation into E. coli. For each ligation reaction, 14 μL of a single or a 1:1 ratio of a double solution of digestion fragments was added to 3 μL of plasmid linearised through the restriction digestion and agarose gel extraction procedures previously described. Concentrations of gel-extracted digestion fragments varied between 50-200 μg/μL, while gel-extracted linearised plasmids varied between 200-300 μg/μL. 2 μL of 10X Ligation Buffer (Roche Diagnostics, Germany) and 1 μL of T4 DNA Ligase (Roche Diagnostics, Germany) were added, and the resulting ligation solution was incubated overnight at 4 °C. The ligation products were directly transformed into E. coli as previously described.

2.1.2.7 Expression Plasmid Construction Pre-synthesised recombinant VP60 N-terminus sequences provided in pUC57 or pUC57- Simple plasmids were transformed, amplified and purified from E. coli as previously described in sections 2.1.2.1 and 2.1.2.2. The N-terminus sequence was excised from each plasmid by restriction digestion with BglII (Roche Diagnostics, Germany) and PstI (Roche Diagnostics, Germany) restriction enzymes, separated by agarose gel electrophoresis and purified by gel extraction as previously described. The remainder of VP60 C-terminus was excised from a pAcpol-.VP60 expression plasmid purified from E. coli stocks supplied by the Ward Laboratory using PstI and EcoRI (Roche Diagnostics, Germany) restriction enzymes and the fragments were gel extracted. pAcUW51(GUS) expression plasmid purified from E. coli stocks supplied by the Ward Laboratory was linearised using BglII and EcoRI restriction enzymes and gel extracted as described in section 2.1.2.5. A combination of the recombinant N-terminus and VP60 C-terminus digestion fragments were ligated with the linearised pAcUW51(GUS) plasmids as previously described.

The expression plasmids for the dual N-terminus and C-terminus insertion constructs S.VP60.T and TS.VP60.T were derived from the expression plasmids of S.VP60 and TS.VP60 respectively, in addition to the expression plasmid of a C-terminus topoisomerase IIα construct (VP60.T) supplied by the Ward Laboratory. The N-terminus of each recombinant construct was excised by restriction digestion with BglII and PstI restriction enzymes, separated by agarose gel electrophoresis and purified by gel extraction as previously described. The C-terminus of VP60.T was excised by restriction digestion with PstI and EcoRI restriction enzymes, separated and purified. These digestion fragments were ligated together into the pAcUW51(GUS) expression plasmid, which had been linearised as previously described. Each of the resulting completed expression plasmids were transformed into E. coli, amplified and extracted for verification. A summary of the expression plasmid development process is provided in Figure 2.1.

Figure 2.1 Development of Recombinant VP60 Expression Plasmids Recombinant VP60 expression plasmids were constructed using the plasmid backbone of (A) pAcUW51(GUS) linearised with BglII and EcoRI, the modified N-terminus of VP60 (X.VP60) in (B) pUW57-Simple.X.VP60 extracted with BglII and PstI, and the C-terminus of VP60 in (C) pAcpol-.VP60 extracted with PstI and EcoRI. These fragments and the plasmid backbone were ligated together to form the VP60 expression plasmid, (D) pAcUW51(GUS).X.VP60.

2.1.2.7 Verification of Expression Plasmids The expression plasmid for each recombinant VP60 construct was verified initially by restriction digestion and gel electrophoresis, and subsequently by sequencing. The initial check was undertaken by digesting each expression plasmid with BglII and EcoRI, and separating the digestion fragments by agarose gel electrophoresis as previously described in section 2.1.2.4. This protocol was used to screen for plasmids containing fragments of an appropriate length prior to sequencing. For sequencing, 600 ng of each expression plasmid selected for positive fragment incorporation was diluted in 28 μL aliquots Milli-Q H2O containing 0.2 pmol/μL of either P10F, VP60intF or VP60intF3 sequencing primer supplied by the Ward Laboratory. The combination of sequence reads from these three primers provides full coverage of the VP60 gene, and their sequences are provided below in Table 2.3. These samples were sequenced by the Massey Genome Service (Massey University, NZ), and the results were analysed using the SeqMan Pro application of the Lasergene analysis software (DNASTAR, Wisconsin, USA).

Table 2.3 Expression Plasmid Sequencing Primers Primer Binding Location* Sequence P10F -217 to -230 CTAGAGTCGAGCAAGAAAATAAA VP60intF 487 to 503 CTCGAACCTGTTACCAT VP60intF3 1452 to 1475 GTACGGCACAGGCTCCCAACCACT * Binding location is relative to the first base of the VP60 gene.

2.1.3 Production of Recombinant Baculovirus Each recombinant Autographa californica multicapsid nucleopolyhedrovirus (AcMNPV), henceforth generically referred to as ‘baculovirus’, was produced through co-transfection of each expression plasmid into Spodoptera frugiperda IPLB-Sf21 (Sf21) cells along with the FlashBAC ULTRATM expression system (Oxford Expression Systems, Oxford, UK). This expression system was used as it negates the requirement to separate recombinant and unmodified parent virus produced, as the parent virus is non-viable in its unmodified form [655].

2.1.3.1 Sf21 Cell Culture Sf21 seed cultures were supplied by the Ward Laboratory, and were maintained in Spodoptera frugiperda 900-III (Sf900-III) serum-free media (Gibco Invitrogen, New York, USA). Cultures were maintained at 27 °C with agitation at 125 rpm, and were split routinely every 3-4 days after reaching a density of >5 x 106 cells/mL. New cultures were seeded at 8 or 10 x 105 cells/mL for a 4 or 3 day growth period respectively. Cell viability and density was monitored using a haemocytometer loaded with culture samples mixed 1:1 with trypan blue.

2.1.3.2 Sf21 Co-transfection Cultures of Sf21 cells were seeded into each well of a 6-well plate at 5 x 105 cells/mL in 3 mL of Sf900-III media, and incubated at 27 °C for 1 hour. During this time, a transfection solution was prepared that consisted of 6 μL of FuGENE 6 transfection reagent (Promega,

Wisconsin, USA) in 100 uL of Sf900-III media. This solution was added to another 100 μL aliquot of Sf900-III media containing 1 μg of a select recombinant VP60 expression plasmid, and 80 ng of the FlashBAC ULTRATM plasmid. The resulting transfection mixture was incubated at room temperature for 30 minutes. Upon completion of the co-ordinated incubation times of both the 6-well cell cultures and the transfection mixture, the media was carefully removed from the 6-well cell cultures. The transfection mixture was added dropwise onto the cell monolayer, rocked to ensure even distribution across the well, and incubated for a further 1 hour at room temperature with additional rocking every 20 minutes. After incubation, 3 mL of Sf900-III media containing 100 U/mL penicillin and 0.1 mg/mL streptomycin (1X Pen/Strep) (Roche Diagnostics, Germany) was added to each well, and incubated at 27 °C for 4 days. Afterwards, media containing budded baculovirus was removed and stored at 4 °C for screening. 10 μL of X-gluc was added to each well of the 6- well culture plate after complete removal of the media. Identification of a blue colour change was indicative of a successful transfection reaction, based on expression of the GUS reporter gene from each expression plasmid.

2.1.3.3 Baculovirus Plaque Assay Plaque assays were used to screen, purify and to determine the titres of recombinant baculovirus preparations. Cultures of Sf21 cells were seeded into each well of a 6-well plate at 5 x 105 cells/mL in 3 mL of Sf900-III media, and incubated at 27 °C for 1 hour. During this time, a 10-fold serial dilution ranging from 10-3 to 10-8 of the baculovirus preparation was prepared. Upon completion of the incubation time, the media was removed from each of the wells and replaced with 100 μL of the baculovirus dilutions assigned to respective wells. The plate was rocked to ensure even distribution over the cell monolayer, and then incubated at room temperature for 1 hour with additional rocking every 20 minutes. Afterwards, the baculovirus dilution solutions were removed from each well, and replaced with a 2 mL overlay of 1.5% SeaPlaque agarose (Lonza, Maine, USA) in Sf900-III media containing 10% fetal bovine serum (FBS) pre-warmed to 41 °C. The overlaid wells were incubated at room temperature for 2-3 minutes to allow the SeaPlaque agarose to set, after which 1.5 mL of Sf900-III media containing 10% FBS and 1X Pen/Strep was added to each well, and the plate was incubated at 27 °C for 4 days. Plaques were visualised by adding 10 μL of neutral red stain to each well, and incubating for 2 hours at room temperature. The viral titre was determined by counting the number of plaques that correlated with each well of the 10-fold

87 serial dilution. Suitable plaques were picked using a sterile Pasteur pipette to purify a selected baculovirus, and inoculated into 10 mL cultures of Sf21 cells to propagate the virus.

2.1.3.4 Baculovirus Inoculum Preparation Upon plaque purification, each recombinant baculovirus was propagated in a 10 mL culture of Sf21 cells grown over 4 days at 27 °C with agitation at 125 rpm. These cultures were subsequently spun at 500 xg for 5 minutes, with the supernatant inoculum retained and filtered through a 0.2 μm Minisart hydrophilic filter (Satorius Stedium Biotech GmbH, Gottingen, Germany). Inoculums were stored at 4 °C, and serially amplified through at least two rounds of inoculation of 1 mL of inoculum into 50 mL cultures of Sf21 cells. Viral titres in subsequent amplified inoculums were determined by plaque assay as previously described.

2.1.4 Production and Purification of RHDV VLP

2.1.4.1 Inoculation of Sf21 Cultures 400 mL cultures of Sf21 cells seeded at 1 x 106 cells/mL, and containing 1X Pen/Strep, were inoculated with recombinant baculovirus at a multiplicity of infection (MOI) of 1. Inoculated cultures were incubated for 3 days at 27 °C with agitation at 125 rpm. Cell viability was checked as previously described following these 3 days, with a proportion of approximately 40% dead cells desirable for optimal yields of RHDV VLP.

2.1.4.2 Purification of RHDV VLP RHDV VLP were harvested from inoculated Sf21 cell cultures through a combination of cell lysis and centrifugation. Triton-X 100 (Sigma-Aldrich, Montana, USA) was added to a concentration of 0.5% in each culture, and incubated at room temperature for 30 minutes to ensure complete cell lysis and inactivation of recombinant baculovirus. The lysed material was then centrifuged at 10,000 xg for 30 minutes to remove cellular debris. The supernatant was retained, and spun at 100,000 xg for 1.5 hours on an Optima L-90K Ultracentrifuge (Beckman Coulter, California, USA) to pellet the VLP and other remaining material. These pellets were resuspended in 1 mL of insect phosphate buffered saline (iPBS) at 4 °C overnight. Further debris was removed from these resuspensions with an additional spin at

10,000 xg for 30 minutes. The resulting supernatants were overlaid onto gradients of caesium chloride (CsCl), consisting of 3 mL of 1.4 g/cm3 CsCl carefully underlaid beneath 3 mL of 1.2 g/cm3 CsCl in a SW32.1 polyallomer tube. Each tube was topped up and balanced with additional iPBS carefully overlaid over the VLP containing supernatant. Gradients were spun at 100,000 xg for 18 hours at 4 °C. As RHDV VLP have a density of 1.34 g/cm3 [645], they form a band that can be harvested from the gradient interface. VLP harvested from CsCl gradients were stored at 4 °C, or dialysed for 30 minutes with 10,000 Da dialysis tubing into a sterile 50% glycerol solution for storage at -20 °C.

2.1.5 Analysis of RHDV VLP

2.1.5.1 Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis RHDV VLP production was monitored, screened and verified primarily using sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). 5 μL of each sample was diluted into 45 μL Milli-Q H2O, then added to 50 μL of 2X sample buffer and boiled for 5 minutes. Boiled samples, along with Broad Range protein marker (New England Biolabs, Maryland, USA), were loaded onto 10% polyacrylamide gels containing SDS, and were resolved at 170V for around 60 minutes in a Mini-PROTEAN III gel tank (Bio-Rad, California, USA). Gels were stained for 30 minutes in Coomassie blue stain and then incubated overnight in destain solution, with a submerged absorptive paper pad included to absorb any liberated dye. Stained gels were visualised using a Bio-Rad ChemiDoc system.

2.1.5.2 Western Blot The identity of the VP60 band separated by SDS-PAGE was verified for each chimaeric RHDV VLP construct by Western blot. Instead of staining with Coomassie blue, the resolved proteins in SDS-PAGE gels were transferred into Immobilon PVDF membrane with a TransBlot SD Semi-Dry Transfer System (Bio-Rad, California, USA). Each transfer arranged from anode plate to cathode plate consisted of two layers of Whatman 3 mm filter paper soaked in anode buffer I, one layer of filter paper soaked in anode buffer II, a section of PVDF membrane soaked in cathode buffer for 10 minutes, the gel soaked in cathode buffer for 10 minutes, and three layers of filter paper soaked in cathode buffer. 22 V was applied for 20 minutes to induce the transfer, after which the membrane was air dried on filter paper

89 for 20 minutes. The membrane was then blocked for 1 hour in a 20 mL solution of phosphate buffered saline (PBS) containing 1% casein alanate (Fonterra, NZ). Membranes were washed 5 times for 5 minutes in 20 mL of PBS containing 0.02% Tween-20 (Sigma-Aldrich, Montana, USA), before being probed for 30 minutes with rabbit anti-VP60 polyclonal antibody supplied by the Ward Laboratory, diluted 1 in 2000 in PBS containing 1% casein alanate and 0.02% Tween-20. Membranes were again washed 5 times for 5 minutes in 20 mL of PBS containing 0.02% Tween-20, before being probed for 30 minutes with DyLight 800- labelled donkey anti-rabbit monoclonal antibody (Clone SA5-10044, Lot QC1998302, Thermo Scientific, Delaware, USA) diluted 1 in 20,000 in PBS containing 1% casein alanate and 0.02% Tween-20. Membranes were washed a final time, 5 times for 5 minutes in 20 mL of PBS containing 0.02% Tween-20, and visualised at 800 nm using an Odyssey FC fluorescence imager (LI-COR, Nebraska, USA).

2.1.5.3 Mass Spectrometry Verification of the identity of each chimaeric VLP produced was performed using mass spectrometry, providing indications of masses corresponding to peptide sequences associated with the recombinant insertion. Resolved VP60 bands were excised using a scalpel blade from an SDS-PAGE gel, digested in trypsin, and analysed using MALDI-TOF/TOF or LTQ- Orbitrap hybrid mass spectrometry by the Centre for Protein Research (University of Otago, NZ).

2.1.5.4 Electron Microscopy Assembly of intact VLP derived from each recombinant VP60 construct was confirmed by visualising samples of VLP by electron microscopy. VLP samples were diluted to 0.2 mg/mL in Milli-Q H2O, and 10 μL of the diluted VLP was incubated for 1 minute on a plasma glowed carbon-coated grid. Excess sample was removed using absorptive filter paper, and replaced with 10 μL of a negative stain consisting of 1% phosphotungstic acid at pH 6.8. This negative stain was immediately removed using absorptive filter paper. Each grid was dried under a lamp for 5 minutes, before viewing on a CM100 BioTWIN transmission electron microscope (TEM) (Phillips/FEI Corporation, Netherlands). Images of negatively stained VLP were obtained using the MegaVIEW III Soft Imaging System, in combination with the iTEM

Universal Imaging Platform (Soft Imaging System, Germany) at the Otago Centre for Electron Microscopy (University of Otago, NZ).

2.2 Association of Nucleic Acids with RHDV VLP

2.2.1 Detection of Endogenous Nucleic Acids in RHDV VLP Endogenous nucleic acids were detected in purified RHDV VLP samples by electrophoresis on an agarose gel. 2 μL of 6X loading dye was added to 10 μL of RHDV VLP diluted to 1 mg/mL in iPBS, loaded onto a 2% agarose gel containing 1X TAE buffer, and electrophoresed at 100V in 1X TAE buffer. Samples were run alongside an aliquot of 1 Kb Plus DNA ladder for size comparison. Upon completion of electrophoresis, gels were incubated in a solution of 1 μg/mL ethidium bromide in H2O for 30 minutes, before being visualised under ultraviolet light using a Bio-Rad ChemiDoc gel visualisation system.

2.2.1.1 Benzonase Treatment of RHDV VLP Samples of RHDV VLP were treated with benzonase in an attempt to remove endogenous nucleic acids present in purified RHDV VLP samples that are not directly associated with and protected by the VLP. 100 IU of benzonase was added to 50 μL of RHDV VLP diluted to 1 mg/mL in iPBS, and incubated for 1 hour at 37 °C. Benzonase was inactivated by adding 25 mM EDTA and heat inactivation for 10 minutes at 65 °C. The sample was then incubated for 2 hours at 56°C with the addition of 3% SDS and 50 μg proteinase K.

2.2.2 Disassembly and Reassembly of RHDV VLP RHDV VLP were disassembled and reassembled using methods previously published by El Mehdaoui et al [656], with some minor modification. Samples of VLP were initially dialysed into disassembly buffer in 10,000 Da dialysis tubing over 18 hours, with the buffer exchanged after 2 and 6 hours. Disassembly was initiated by diluting VLP to 1 mg/mL in a solution consisting of disassembly buffer with the addition of 20 mM dithiothreitol (DTT) and 10 mM ethylene glycol tetraacetic acid (EGTA). This solution was incubated for 12 hours at room temperature to achieve maximal disassembly of VLP. Reassembly was induced by diluting disassembled VLP 1:5 in a reassembly solution consisting of disassembly buffer,

91 dimethylsulfoxide (DMSO) and calcium chloride (CaCl2), with a final concentration of 1%

DMSO and 5 mM CaCl2.

2.2.2.1 Differential Ultracentrifugation Gradients of CsCl were constructed by underlaying 1 mL of 1.4 g/cm3 CsCl solution beneath 1 mL of 1.2 g/cm3 CsCl in SW55 polypropylene tubes (Beckman Coulter, California, USA). 0.3 mL of untreated or treated VLP was overlaid over the gradient, and the tubes were topped up and balanced with disassembly buffer. Gradients were ultracentrifuged at 100,000 xg for 18 hours. 0.1 mL was harvested from the 1-1.2 g/cm3 interface, and the 1.2-1.4 g/cm3 interface. Samples from each of these interfaces were separated by SDS-PAGE, transferred onto nitrocellulose membrane and blotted by western blot as previously described in sections 2.1.5.1 and 2.1.5.2. VP60 was semi-quantified using Odyssey FC software based on band intensity. The proportion of VP60 detected at each interface was calculated as a percentage with respect to the total amount of VP60 retrieved.

2.2.2.2 Dynamic Light Scattering 0.5 mL of VLP at 2 mg/mL was stabilised at 25 °C in a disposable cuvette, prior to analysis on a Zetasizer ZS-90 (Malvern Instruments, Worcestershire, UK). VLP were analysed with settings including an RI of 1.45 and an absorption of 0.001, with PBS as a dispersant. PBS viscosity was set at 1.02 cP, with a refractive index of 1.33. These values were derived from previous analyses of RHDV VLP on this instrument. Upon reaching thermal stability within the Zetasizer, 0.5 mL of test solution was added to the cuvette. This test solution consisted of disassembly buffer for the untreated group, and disassembly buffer containing 40 mM DTT and 20 mM EGTA for the treated/disassembly group. Average particle diameter was monitored over time as determined from instrumental calculations of particle volume. Each measure was calculated from 12-15 independent automated measures undertaken by the Zetasizer software.

2.2.2.3 Spectrophotometry VLP were analysed spectrophotometrically on a NanoDrop 1000, using a custom protein analysis protocol that included settings of ε at 78000 M-1cm-1, and a molecular weight of 60

92 kDa. For disassembly samples, a disassembly blanking solution was used that consisted of disassembly buffer containing 20 mM DTT and 10 mM EGTA. This blanking solution was incubated under the same conditions and for the same period as the disassembly samples, accounting for any background changes in the absorbance profile of the solution. Measurements of absorbance at A250 and A280 were recorded at each time point for calculation of the A250/A280 ratio.

2.2.3 Loading/Association of CpGs with RHDV VLP Custom CpG oligonucleotides based on the ODN 1826 sequence were synthesised (GeneWorks, South Australia, Australia) for loading into RHDV VLP through disassembly and reassembly, or by association through direct incubation. A second, benzonase digestible CpG oligonucleotide was also supplied by the Ward Laboratory for select experiments involving benzonase. 25 μg of CpG was added to 100 μL of untreated or disassembled VLP at 1 mg/mL, and incubated for 30 minutes at room temperature. Disassembled VLP were subsequently reassembled, as previously described.

2.2.3.1 Dialysis to Remove Free CpGs Samples of RHDV VLP loaded or associated with CpG were dialysed to remove free CpG remaining in solution. Dialysis was performed using 1,000,000 Da dialysis tubing into coupling PBS (cPBS) at 4 °C over a period of 18 hours, with the buffer exchanged after 2 and 6 hours.

2.2.3.2 TAE Acrylamide Gel Electrophoresis CpG remaining in samples of VLP following dialysis or benzonase digestion was detected by acrylamide gel electrophoresis. 2 μL of 6X loading dye was added to 10 μL of each VLP sample, loaded onto a 20% acrylamide gel containing 1X TAE buffer, and electrophoresed at 100V in 1X TAE buffer. Samples were run alongside an aliquot of 1 Kb Plus DNA ladder for size comparison. Gels were stained with 1X Gel Green (Biotium, California, USA) at room temperature for 20 minutes, and visualised at 600 nm using an Odyssey FC fluorescence imager.

2.2.3.3 Quantification of VLP Associated CpGs The quantity of CpG associated with each VLP sample was determined using the Image Studio Software for the Odyssey Fc (LI-COR, Nebraska, USA) by comparing the CpG band isolated from each sample against a concentration curve derived from a 2-fold serial dilution of a sample with a known CpG concentration. Post-dialysis concentrations of VLP were determined by running each sample on an SDS-PAGE gel, using the aforementioned software to compare the VP60 band from each sample against a concentration curve derived from a 2- fold serial dilution of sample with a known VP60 concentration. Together, the concentration of CpG and VP60 was used to calculate the relative amount of CpG that associated with each chimaeric VLP tested.

2.2.4 RAW-Blue SEAP/NF-κB Reporter Cell Line

2.2.4.1 Culture of RAW-Blue Cells The RAW-Blue cell line was supplied by the Ward Laboratory, previous acquired from the American Type Culture Collection (ATCC, Virginia, USA), and cultured under ERMA approval NOC000707, BOK number GMD101729. RAW-Blue cells were maintained in Dulbecco’s modified eagle medium (DMEM) GlutaMax media (Gibco Invitrogen, New York, USA) containing 10% FBS, 1X Pen/Strep, 1X Normocin (InvivoGen, California, USA) and 1X Zeocin (Invitrogen, New York, USA) (RB-DMEM). Cells were grown in culture at 37 °C with 5% CO2, and split every 2-3 days upon reaching confluency of around 70-90%.

2.2.4.2 Treatment of RAW-Blue Cells RAW-Blue cells were seeded into a 24-well plate at a concentration of 4 x 105 cells/mL in

500 μL of RB-DMEM, and incubated overnight at 37 °C + 5% CO2. Assigned wells were treated with 25 μL of attributed treatment groups, consisting of cPBS and CpG treated cPBS control groups, or 1 mg/mL VLP samples pre- or post-dialysis. Treated plates were incubated for 24 hours at 37 °C + 5% CO2, after which 300 μL of media was harvested from each well and stored at -20 °C for quantification of secreted alkaline phosphatase (SEAP).

2.2.4.3 QUANTI-Blue Assay The QUANTI-Blue assay is a colourimetric enzyme assay for determining the concentration of SEAP, or any other alkaline phosphatase present in a solution. 20 μL aliquots of media isolated from treated RAW-Blue cells, or 20 μL samples from a 2-fold serial dilution of recombinant SEAP at a known concentration, were incubated with 180 μL of QUANTI-Blue reagent (InvivoGen, California, USA) at room temperature for 1 hour. Measurements were taken at 651 nm using a Multiskan microplate reader (MTX Lab Systems, Florida, USA) and concentrations of SEAP were determined relative to a concentration curve established by the 2-fold serially diluted recombinant SEAP.

2.2.5 Murine Bone Marrow-derived Dendritic Cells BMDCs compose a heterologous population of dendritic cells, monocyte-derived macrophages and various other cell populations [657], produced through maturation of haematopoietic progenitor cells under the influence of GM-CSF. This procedure produces a large number of mature DCs with a phenotype most closely resembling conventional DCs, making BMDCs a favoured model for emulating the potency of DC activation induced by vaccine and adjuvant combinations.

2.2.5.1 Ethical Approvals Tissues retrieved for use in vitro were obtained from male and female C57BL/6 mice supplied by the Hercus Taieri Research Unit (University of Otago, NZ), undertaken in accordance with ethical approvals and permits granted by the Animal Ethics Committee (University of Otago, NZ) (ET 22/11, ET 10/13). All animals were euthanised humanely by either cervical dislocation or through administration of carbon dioxide.

2.2.5.2 Generation of BMDCs BMDCs were generated following pre-established laboratory protocols. Hind legs were excised under strict sterile conditions from euthanised C57BL/6 mice, using tools frequently decontaminated with 70% ethanol. Muscle and other connective tissues were removed from the surface of the bones, which were subsequently washed in 70% ethanol followed by two washes in sterile Dulbecco’s PBS (DPBS) (Gibco Invitrogen, New York, USA) containing

5% foetal calf serum (FCS). Both ends of the cleaned bones were cut, and the marrow space was flushed with DPBS + 5% FCS. The flushed marrow material was passed through a 70 μm cell strainer, and centrifuged at 250 xg for 5 minutes. The pelleted cells were resuspended in 1 mL of red blood cell (RBC) lysis buffer pre-warmed to 37 °C, and incubated at 37 °C for 1 minute. 9 mL of DPBS + 5% FCS was added to dilute the RBC lysis buffer, and the cell solution was spun at 250 xg for 5 minutes. The pelleted cells were resuspended at 5 x 105 cells/mL in complete Iscove’s modified Dulbecco’s media (cIMDM) (Gibco Invitrogen, New York, USA) containing 5% FCS and 20 ng/mL recombinant murine GM-CSF (ProSpec, New Jersey, USA). 4 mL of the cell suspension was seeded into each well of a 6-well plate and incubated at 37 °C + 5% CO2 for 6 days, with 2 mL of the media replaced per well on day 3 with 2 mL of DMEM + 5% FCS containing 20 ng/mL murine GM-CSF. Upon day 6, BMDCs were harvested by gently resuspending clumps of proliferated non-adherent cells into the media, and transferring the combined suspension from multiple wells into an appropriate tube. These cells were then spun at 250 xg for 10 minutes, and resuspended in cIMDM + 5% FCS at concentrations determined by the specific down-stream application intended for each BMDC preparation.

2.2.5.3 BMDC Activation Assay BMDCs harvested from day 6 cultures as previously described were resuspended at 1 x 106 cells/mL in cIMDM + 5% FCS, and plated out at 200 μL per well in a rounded bottom 96- well culture plate. Assigned wells were treated with 10 μL of attributed treatment groups, consisting of cPBS and CpG treated cPBS control groups, 1 μg LPS, or 1 mg/mL VLP samples pre- or post-dialysis. Treated plates were incubated for 24 hours at 37 °C + 5% CO2, after which the cells were harvested for flow cytometry.

2.2.5.4 Flow Cytometry Harvested BMDCs from each treatment group were washed twice with 1 mL DPBS + 5% FCS, and then stained for 15 minutes with 50 μL DPBS containing near infra-red (IR) live/dead (L/D) stain (Life Technologies, California, USA) at 1:1000 dilution. The cells were then washed with 1 mL of fluorescence-activated cell sorting (FACS) buffer, before being stained for 30 minutes at 4 °C with 50 μL of a cocktail of fluorescently labelled antibodies, including anti-CD11c/APC (Clone N418), anti-CD40/PE-Cy7 (Clone 3123), anti-

CD80/Pacific Blue (Clone 16-10A1), anti-CD86/PE (Clone GL-1) and anti-I-A/I-E/FITC (Clone M5/114.15.2) (BioLegend, California, USA) at lot-specific dilutions of 1:300-1:500. The cells were then washed with 1 mL DPBS + 5% FCS, before being fixed for 15 minutes at 4 °C in 50 μL of 2% paraformaldehyde (PFA). The cells were washed a final time with 1 mL DPBS + 5% FCS, resuspended in 200 μL of FACS buffer, and stored at 4 °C until they were analysed by FACS. FACS was performed on a Gallios Flow Cytometer (Beckman Coulter, California, USA), and the data was analysed using Kaluza version 1.2 (Beckman Coulter, California, USA).

Figure 2.2 Flow Cytometry Gating Strategy An example of the gating strategy used for assessing DC activation markers. (A) Non-cellular debris was excluded from analysis by gating cells from a forward scatter area (INT) linear versus side scatter INT linear

97 plot. (B) Doublets were excluded from analysis by gating for single cells from a side scatter height (PEAK) linear versus side scatter INT linear plot. (C) Dead cells were excluded from analysis by gating for live cells, unstained with near-IR L/D stain under FL8 INT log. (D) DCs were selected for analysis by gating for CD11c positive cells under FL6 INT log. DC activation markers were assessed from the respective channel INT log plots for: (E) CD40; (F) CD80; (G) CD86; and (H) I-A/I-E.

2.2.6 Association of siRNA with RHDV VLP Synthetic mouse-specific SignalSilence Stat3 siRNA II (Cell Signalling Technology, Maryland, USA) and SignalSilence Control siRNA (Cell Signalling Technology, Maryland, USA) were obtained for experiments involving association of siRNA with RHDV VLP. 0.5 μg of siRNA was added to 100 μL of RHDV VLP at a concentration of 1 mg/mL, and incubated for 30 minutes at 4 °C. Non-associated siRNA was removed by dialysis as previously described.

2.2.6.1 Quantification of siRNA Associated with RHDV VLP The quantity of siRNA associated with each VLP sample was determined using the Image Studio Software for the Odyssey Fc by comparing the siRNA band isolated from each sample against a concentration curve derived from a 2-fold serial dilution of a sample with a known siRNA concentration. Post-dialysis concentrations of VLP were determined by running each sample on an SDS-PAGE gel, using the aforementioned software to compare the VP60 band from each sample against a concentration curve derived from a 2-fold serial dilution of sample with a known VP60 concentration. Together, the concentration of siRNA and VP60 was used to calculate the relative amount of siRNA that associated with each chimaeric VLP tested.

2.2.6.2 siRNA-associated RHDV VLP Delivery The ability of RHDV VLP to deliver an associated siRNA was tested in the RAW-Blue cell line. RAW-Blue cells were seeded into 96-well plates at a concentration of 4 x 105 cells/mL in 200 μL of RB-DMEM, and incubated overnight at 37 °C + 5% CO2. Assigned wells were treated with 10 μL of attributed treatment groups, consisting of cPBS and siRNA + FuGENE HD (Promega, Wisconsin, USA) transfection control groups, or 1 mg/mL VLP samples pre- or post-dialysis. Treated plates were incubated for 8 hours at 37 °C + 5% CO2, after which 1

μg/mL lipopolysaccharide (LPS) was added. The plates were incubated for a further 12 or 24 hours at 37 °C + 5% CO2, when the media from each well was removed and stored at -20 °C for an IL-6 enzyme-linked immunosorbent assay (ELISA). The cells were lysed with 60 μL 2X Laemmli buffer, and the resulting lysates were stored at -20 °C for Stat3 Western blots.

2.2.6.3 Stat3/Actin Western blot Knockdown of Stat3 was identified by Western blot, using Actinβ as a loading control. 30 μL of each lysate was loaded onto extra-thick 1.5 mm SDS-PAGE gels, and were separated by electrophoresis and transferred onto PVDF membrane as previously described. In this instance, the membranes were probed with rabbit anti-Stat3 (Clone 79D7, Lot 7, Cell Signalling Technology, Maryland, USA) and goat anti-β-Actin (Clone I-F1, Lot G3115, Santa Cruz Biotechnology, California, USA) antibodies diluted 1:1000 in PBS containing 1% casein alanate and 0.02% Tween-20. These where subsequently probed for using a combination of fluorescently labelled anti-rabbit mAb, and a DyLight 680-labelled anti-goat mAb (Clone SA5-10090, Lot QC1989644, Thermo Scientific, Delaware, USA) diluted 1:20,000 in PBS containing 1% casein alanate and 0.02% Tween-20. Following the washing steps previously described, the membranes were visualised at 700 and 800 nm using an Odyssey FC fluorescence imager. Relative quantification of STAT3 was calculated based on the fluorescence intensity of the Dylight 800 labelled secondary antibody, normalised by the fluorescence intensity of the Dylight 700 labelled secondary corresponding to actin. Quantities were then determined as a percentage in comparison to the average amongst untreated controls, with this average set as 100%.

2.3 In Vivo Assays and Trials

2.3.1 Ethical Approvals The C57BL/6 mice used throughout the in vivo assays and trials were supplied by the Hercus Taieri Resource Unit. Experiments were conducted in accordance with those outlined in the ethical approvals granted by the Animal Ethics Committee (AEC 39/13 and AEC 12/15). All animals were euthanised humanely by either cervical dislocation or through administration of carbon dioxide.

2.3.2 In Vivo Cytotoxicity Assay In vivo cytotoxicity assays were used to test the capability and specificity of each chimaeric RHDV VLP construct to induce a targeted immune response against their respective incorporated epitope(s). This process involves vaccinating mice with each VLP construct, and then testing the ability of the induced immune responses to selectively eliminate a target population of splenocytes laced with the same epitope.

2.3.1.1 Vaccination with RHDV VLP For each in vivo cytotoxicity experiment, 6 female C57BL/6 mice aged 6-8 weeks were assigned to each of 6 treatment groups, cPBS, VP60, T.VP60, S.VP60, TS.VP60 and T + S Peptides. At day 0, each mouse was vaccinated with their assigned vaccine, consisting of 100 μL cPBS, 100 μL cPBS containing 100 μg VLP with 25 μg CpG, or 100 μL cPBS containing an equimolar amount of synthetic topoisomerase IIα and survivin peptide (JPT Peptide Technologies, Germany) with 25 μg CpG. Each vaccine was administered subcutaneously into the left flank. At day 21, each mouse received a second dose of the same vaccine they had previously received.

2.3.1.2 Preparation of Target Cells The target cells used in these in vivo cytotoxicity assays consisted of splenocytes derived from naïve female C57BL/6 mice. Splenocytes were isolated by pressing excised spleens through a 70 μm cell strainer, washed with DPBS + 5% FCS. The resulting splenocyte solution was centrifuged at 300 xg for 10 minutes, and the pelleted cells were resuspended in 2 mL of RBC lysis buffer pre-warmed to 37 °C. This solution was incubated for 2 minutes at 37 °C. 18 mL of DPBS + 5% FCS was added to dilute the RBC lysis buffer, and the splenocyte solution was centrifuged at 300 xg for 10 minutes. The pelleted cells were resuspended at 2 x 107 cells/mL in cIMDM + 5% FCS, and separated into four equal populations corresponding with an unpulsed population, a population pulsed with 10 μM of topoisomerase IIα peptide, a population pulsed with 10 μM of survivin peptide, and a population pulsed with 10 μM of topoisomerase IIα peptide and 10 μM of survivin peptide. Each population of splenocytes was pulsed respectively, and incubated for 2 hours at 37 °C

+ 5% CO2. Following this incubation, each population was washed three times with DPBS containing no FCS, with centrifugation at 300 xg for 10 minutes between each wash, and

100 were resuspended at 2 x 108 cells/mL. These populations were assigned to one of four corresponding dye staining group, including 0.2 μM carboxyfluorescein succinimidyl ester (CFSE) (CFSELo), 2 μM CFSE (CFSEHi), 5 μM violet proliferation dye (VPD) (VPDHi), and 2 μM CFSE and 5 μM VPD (CFSEHi/VPDHi). Each assigned target cell population was stained according, incubated for 7 minutes at room temperature shielded from light. The staining reactions were quenched by adding an equal volume of FCS. Each population was washed three times DPBS containing no FCS, with centrifugation at 300 xg for 10 minutes between each wash, and were resuspended at 1 x 108 cells/mL. Equal volumes of each target cell population was then combined together to produce the target cell mixture suitable for infusion into vaccinated mice.

2.3.1.2 Target Cell Infusion and Detection At day 28, each vaccinated mouse was infused intravenously with 200 μL of the target cell mixture. Mice were culled 48 hours later at day 30, and their spleens were excised. Each spleen was processed to produce a single suspension of splenocytes using the same procedure previously described for production of the target cells. These single cell suspensions were spun at 300 xg for 5 minutes, resuspended in 100 μL of DPBS + 5% FCS containing Near- IR live/dead dye at 1:1000 dilution, and were incubated for 15 minutes at room temperature. The cells were then washed with 1 mL DPBS + 5% FCS, before being fixed for 15 minutes in 100 μL of 2% PFA. The cells were washed a final time with 1 mL DPBS + 5% FCS, resuspended in 500 μL FACS buffer, and stored at 4 °C until they were analysed by FACS. FACS was performed on a Gallios Flow Cytometer, and the data was analysed using Kaluza version 1.2. A summary of the in vivo cytotoxicity assay is provided in Figure 2.3.

101

Figure 2.3 In Vivo Cytotoxicity Assay An in vivo cytotoxicity assay was used to assess the ability of chimaeric RHDV VLP vaccines to induce epitope- specific immunity. A) Vaccines were administered subcutaneously into 6-8 week old C57BL/6 mice, with a prime vaccination at day 0 and a boost vaccination at day 21. B) Target splenocytes were prepared from the spleens of naïve C57BL/6 mice. C) Target cells were infused intravenous into vaccinated mice on day 28. D) Spleens were harvest from vaccinated mice on day 30, and processed to identify target cell populations for the calculation of percent specific lysis.

2.3.1.3 Calculation of Specific Lysis (%) The determination of relative cytotoxicity against each target cell population between vaccinated mice was determined using an equation, which includes a correction for background cytotoxicity against a control cell population.

2.3.3 In Vitro (VITAL) Cytotoxicity Assay An in vitro cytotoxicity assay was trialled as a rapid means to determining relative immunogenicity of VLP vaccines. The method used was derived from the VITAL assay originally developed by Hermans et al [658]. The principles behind the in vitro cytotoxicity assay are similar to those of the in vivo cytotoxicity assay, except that the targeted elimination of peptide pulsed cells by epitope specific CD8+ T cells is performed in an in vitro co-culture.

2.3.3.1 Vaccination with RHDV VLP For each in vitro cytotoxicity experiment, 9 female C57BL/6 mice aged 6-8 weeks were assigned to each of 3 treatment groups, VP60, SIIN.VP60 and S.VP60. At day 0, each mouse was vaccinated subcutaneously with their assigned vaccine, consisting of 100 μL cPBS containing 100 μg VLP with 25 μg CpG. For each treatment group, 6 mice were selected and received an additional dose of their respective vaccine on day 1. From these mice, 3 were selected from each group to receive a final dose of their respective vaccine on day 2. This provided 3 mice from each treatment group who had received 1 dose (1D), 2 doses (2D) or 3 doses (3D) of their assigned vaccine.

102

2.3.3.2 Isolation of CD8+ T Cells Each vaccinated mouse was culled on day 7, and the lymph nodes draining the area of subcutaneous vaccination were excised. These lymph nodes were pressed through a 70 μm cell strainer, washed with DPBS + 5% FCS. The resulting lymph node solution was centrifuged at 300 xg for 5 minutes at 4 °C, and the pelleted cells were resuspended in 1 mL of RBC lysis buffer pre-warmed to 37 °C. This solution was incubated for 2 minutes at 37 °C. 9 mL of DPBS + 5% FCS was added to dilute the RBC lysis buffer, and the splenocyte solution was centrifuged at 300 xg for 5 minutes at 4 °C. The pelleted cells were resuspended in 10 mL of DPBS + 5% FCS, counted, and centrifuged again at 300 xg for 5 minutes at 4 °C. The cells were resuspended in 90 μL of AutoMACS buffer per 1 x 107 cells, and 10 μL of CD8α (Ly-2) microbeads (Miltenyi Biotec, Germany) was added per 1 x 107 cells. This solution was incubated for 15 minutes at 4 °C, after which 2 mL of AutoMACS buffer was added per 1 x 107 cells. These samples were run through an AutoMACS Pro cell separator (Miltenyi Biotec, Germany) on a positive selection protocol. The resulting separated cell solution was spun at 300 xg for 5 minutes at 4 °C, and resuspended at 1 x 106 cells/mL in cIMDM + 5% FCS.

2.3.3.3 Co-culture and Analysis Peptide pulsed target splenocytes were prepared from the spleens of naïve C57BL/6 mice as previously described. In this instance, the unpulsed population was stained CFSEHi, the SIINFEKL peptide pulsed population was stained VPDHi, and the survivin peptide pulsed population was stained CFSEHi/VPDHi. These populations were combined in equal proportions, and resuspended at 1.5 x 105 cells/mL in cIMDM + 5% FCS. 100 μL of this mixed target cell solution was added to each well of a 96-well U bottom plate. CD8+ T cells isolated from the draining lymph nodes of vaccinated mice and separated as previously described were added to assigned wells, at an effector:target ratio of 0.1:1. Sufficient cIMDM + 5% FCS was added to bring these co-culture volumes up to 200 μL. These co-cultures were incubated for 4 hours at 37 °C, after which the cells were gently harvested from the wells. These co-cultured cell suspensions were spun at 300 xg for 5 minutes, resuspended in 100 μL of DPBS + 5% FCS containing Near-IR live/dead dye at 1:1000 dilution, and were incubated for 15 minutes at room temperature. The cells were then washed with 1 mL DPBS

103

+ 5% FCS, before being fixed for 15 minutes in 100 μL of 2% PFA. The cells were washed a final time with 1 mL DPBS + 5% FCS, resuspended in 500 μL of FACS buffer, and stored at 4 °C until they were analysed by FACS. FACS was performed on a Gallios Flow Cytometer, and the data was analysed using Kaluza version 1.2. A summary of the in vitro cytotoxicity assay is provided in figure 2.4. Calculation of Specific Lysis (%) was made using the equation outlined above for the in vivo cytotoxicity assay in section 2.3.1.3.

Figure 2.4 In Vitro Cytotoxicity Assay The in vitro cytotoxicity protocol used for assessing the potential benefit of sequential vaccination. A) Vaccines were administered subcutaneously into 6-8 week old C57BL/6 mice, with each mouse receiving 1, 2 or 3 doses over a 3-day period. B) At day 7, CD8+ T cells were isolated and sorted from the draining lymph node. C) Target splenocytes were prepared from the spleen of a naïve C57BL/6 mouse. D) CD8+ T cells from vaccinated mice were co-cultured with target splenocytes for 4 hours. E) Each target cell population was identified by flow cytometry, and the percent specific lysis was calculated.

2.3.4 MC38-OVA The MC38-OVA cell line is a derivative of the MC38 C57BL/6 murine colon adenocarcinoma cell line originating from the laboratories of James W. Hodge and Jeffrey Schlom (National Cancer Institute, National Institutes of Health, USA). The MC38-OVA variant has been modified to stably express chicken ovalbumin, and was provided for these studies by Nicole Haynes (Peter MacCallum Institute, Australia). Cells were split every 2-3 days by removing the media, and adding 0.5 mL, 2 mL or 3 mL of trypsin-EDTA (Gibco Invitrogen, New York, USA) to the T25, T75 or T150 flask respectively. Trypsinised cells were washed with 20 mL DMEM + 10% FCS, spun at 300 xg for 5 minutes, and resuspended 104 at 5 x 106 cells/mL in DMEM + 10% FCS. 0.5 mL, 2 mL or 5 mL of this cell suspension was used to seed a T25, T75 or T150 flask respectively, topped up with 4.5 mL, 18 mL or 45 mL of DMEM + 10% FCS respectively. Cells were cultured in DMEM + 10% FCS incubated at

37 °C + 5% CO2, and were monitored closely for adverse clumping.

2.3.3.1 Verification of Topoisomerase IIα and Survivin Expression During the modification of the original MC38 cell line to produce the MC38-OVA variant, the expression of some proteins may have been altered. The expression of topoisomerase IIα and survivin was confirmed by Western blot as previously described, comparing lysate prepared from MC38 and MC38-OVA cells. The probing antibodies used included rabbit anti-topoisomerase IIα (Clone 4733S, Lot 1, Cell Signalling Technology, Maryland, USA) and rabbit anti-survivin (Clone 7164B7, Lot 13, Cell Signalling Technology, Maryland, USA) antibodies, in combination with mouse anti-Actinβ antibody diluted 1:1000 in PBS containing 1% casein alanate and 0.02% Tween-20. The membranes were visualised at 700 and 800 nm using an Odyssey FC fluorescence imager.

2.3.3.2 Verification of Ovalbumin Expression The expression of ovalbumin by MC38-OVA cells was confirmed using an IFN-γ release assay. MC38-OVA cells were resuspended at 1x105 cells/mL in DMEM + 10% FCS, and 100 μL was added to each well of a 96-well U bottom plate. Splenocytes were prepared from the spleen of a naïve C57BL/6 mouse (Spl), resuspended at 1x105 cells/mL in DMEM + 10% FCS, and 100 μL was added to each well of a 96-well U bottom plate, separate from those wells containing MC38-OVA cells. 10 μM of SIINFEKL peptide was added to the Spl wells, and the plates were incubated for 2 hours at 37 °C. Splenocytes were prepared from the spleen of an OT-I mouse, consisting of a C57BL/6 mouse engineered to predominantly express TCRs specific for SIINFEKL on their CD8+ T cells (OTI). Further splenocytes were also prepared from another naïve C57BL/6 mouse, and each of these splenocyte preparations were resuspended at 1x105 cells/mL in DMEM + 10% FCS. 100 μL of either population was added to appropriate wells such that 4 different co-cultures were prepared, consisting of Spl + OTI cells, SIINFEKL pulsed Spl + OTI cells, MC38-OVA + Spl cells and MC38-OVA + OTI cells. These co-cultures were incubated for 48 hours at 37 °C, with media sampled at 24 and

105

48 hours. The concentration of IFN-γ present in these samples was determined with an IFN- γ ELISA.

2.3.3.3 IFN-γ ELISA For the IFN-γ ELISA, 50 μL of capture antibody consisting of purified anti-mouse IFN-γ (Clone RA-6A2, BD Biosciences, California, USA) diluted to 2 μg/mL in coating buffer was added to each well of a 96-well plate. The plate was incubated overnight at 4 °C, after which it was washed 5 times by filling the wells with ELISA wash buffer, and ejecting out the wash solution from the inverted plate. 200 μL of PBS + 1% BSA was added to each well, and incubated for 2 hours at room temperature. The plate was then washed 5 times as previously described. 50 μL of each sample was added to assigned wells, in addition to 50 μL in appropriate wells of each concentration of recombinant murine IFN-γ (BD Biosciences, California, USA) from a 2-fold serial dilution in PBS + 1% BSA. The plate was incubated for 4 hours at room temperature, and then washed 5 times as previously described. 50 μL of biotinylated anti-mouse IFN-γ (Clone XMG1.2, BD Biosciences, California, USA) diluted to 2 μg/mL in PBS + 1% BSA was added to each well, and incubated for 45 minutes at room temperature. The was washed 5 times as described above, and then 50 μL of Steptavidin- HRP (BioLegend, California, USA) diluted at 1/3000 in PBS + 1% BSA was added to each well. The plate was incubated for 30 minutes at room temperature, and then washed 5 times as previously described. 100 μL of TMB substrate (Invitrogen, California, USA) was added to each well, and incubated in the dark at room temperature for 5-10 minutes until colouration had sufficiently developed. The reaction was stopped with 100 μL of 1N H2SO4 added to each well. Absorbance at 450 nm was determined for each well using a Model 550 microplate reader (Bio-Rad, California, USA), and IFN-γ concentrations were determined using Microplate Manager Software version 5.0 (Bio-Rad, California, USA).

2.3.3.4 MC38-OVA Tumour Trials For each tumour trial, MC38-OVA cells were thawed from liquid nitrogen and cultured for at least 3 passages prior to use. These cells were washed 5 times by centrifuging at 300 xg for 10 minutes at 4 °C, and resuspending in 50 mL DPBS pre-chilled at 4 °C. Following their final centrifugation, the MC38-OVA cells were resuspended at 2 x 107 cells/mL. On day 0, 50 μL of these cells were infected subcutaneously on the left flank of an appropriate number

106 of mice, defined as the number of treatment groups multiplied by 10. At day 7, the developing tumours were palpated and measured. Tumour-bearing mice were assigned to each treatment group, such that the average tumour size and the distribution of tumour sizes were even between each treatment group. These treatment groups consisted of cPBS, VP60, SIIN.VP60, T.VP60, S.VP60 and TS.VP60. Each mouse was then vaccinated subcutaneously with their assigned treatment, consisting of 100 μL cPBS, 100 μL cPBS containing 100 μg VLP with 25 μg CpG, at a site separate and superior to the tumour site on the left flank. Each mouse also received a second and third vaccination of their assigned treatment on days 8 and 9. Tumours were measured every 2-3 days, and mice were culled when the tumours reached 150 mm2 in cross-sectional area. Mice were monitored for a period of 100 days, at which time the remaining mice received a rechallenge of 50 μL MC38-OVA cells resuspended at 2 x 107 cells/mL in DPBS injected subcutaneously on the right flank. A group of 10 naïve C57BL/6 mice were also engrafted with MC38-OVA tumours in the right flank as a control group. These mice were monitored for a further 100 days with tumour measurements every 2-3 days, and were culled when the tumours reached 150 mm2 in cross-sectional area. Mice that survived until the completion of the trial were culled at day 200.

2.4 Recombinant Vesicular Stomatitis Virus Vesicular stomatitis virus (VSV) is an enveloped member of the Rhabdoviridae family. Recombinant insertion of a gene at a specific site between the G and L genes of the VSV genome results in the production of a virus that can induce expression of the inserted gene in host cells. The protocol used in this study for the production of recombinant VSV is routinely used by the Vile Laboratory (Mayo Clinic, Minnesota, USA) [659]. For this study, the RHDV VLP constructs VP60, SIIN.VP60 and TS.VP60 were adapted for insertion into the recombinant VSV expression system.

2.4.1 Construction of Expression Plasmids The recombinant VSV expression plasmid, pVSV-NX2, contains an insertion site flanked by a XhoI and a NheI restriction site. As our RHDV VLP constructs are based on flanking BglII and EcoRI sites, new constructs were required to facilitate adaptation into this expression plasmid. A single synthetic plasmid containing the N-terminus of VP60, SIIN.VP60 and

107

TS.VP60 adapted with flanking XhoI and NheI sites (VSV-VP60-N, VSV-SIIN.VP60-N and VSV-TS.VP60-N respectively) was purchased (pUC57-Simple.VSV-VLP3X) (Genescript, New Jersey, USA). This plasmid construct was also designed with an interspacing HindIII site to facilitate the stepwise excision of specific constructs from the plasmid. The structure of the pUC57-Simple.VSV-VLP3X plasmid is summarised in figure 2.5.

Figure 2.5 Stucture of the pUC57-Simple.VSV-VLP3X Plasmid The pUC57-Simple.VSV-VLP3X plasmid was designed with three different modified VP60 N-termini, X1.VP60, X2.VP60 and X3.VP60. The proximal BglII site was replaced for each construct with XhoI, with the NheI VSV adaption site inserted distal of the PstI-EcoRI site used for insertion of the VP60 C-terminus. Additional HindIII sites were included proximal of X1.VP60, and between X1.VP60 and X2.VP60 to facilitate construction using the X2.VP60 component.

2.4.1.1 Production of Individual Adapter Plasmids The pUC57-Simple.VSV-VLP plasmid was transformed into calcium competent E.coli, amplified, and purified using methods previously described. pUC57-Simple.VSV-VLP3X was linearised with PstI, separated by electrophoresis, gel extracted and ligated using previously described methods. This process resulted in the production of the first adapter plasmid construct, pUC57-Simple.VSV-VP60-N. Separately, pUC57-Simple.VSV-VLP was also linearised with HindIII (Roche Diagnostics, Germany), separated by electrophoresis, gel

108 extracted and ligated, resulting in the excision of the VSV-VP60-N sequence. The resulting pUC57-Simple.VSV-VLP2X plasmid was linearised with either PstI or XhoI (Roche Diagnostics, Germany), separated by electrophoresis, gel extracted and ligated to produce the pUC57-Simple.VSV-SIIN.VP60-N and pUC57-Simple.VSV-TS.VP60-N plasmids respectively.

2.4.1.2 Incorporation of VP60 C-terminus The C-terminus of VP60 was excised from the pAcpol-.VP60 plasmid by restriction digestion with PstI and EcoRI, separation by electrophoresis, and purified by gel extraction. Each of the VSV-VLP adapter plasmids were linearised with PstI and EcoRI, separated by electrophoresis, purified by gel extraction and ligated with the purified VP60 C-terminus fragments. This resulted in the production of the completed VSV-VLP adapter plasmids, pUC57-Simple.VSV-VP60, pUC57-Simple.VSV-SIIN.VP60 and pUC57-Simple.VSV- TS.VP60. 50 μL of each plasmid at a concentration of around 500 ng/μL was absorbed onto strips of Whatman 3 mm filter paper, and air dried for 30 minutes at room temperature. These strips were placed inside individual 2 mL cryo vials for shipment to the Vile laboratory at the Mayo Clinc, Minnesota, USA.

2.4.1.3 Transformation of HB101 Cells Upon arrival, each completed VSV-VLP adapter plasmid was transformed into HB101 E. coli cells. Segments of the plasmid-laced filter paper were submerged into 100 μL of double distilled H2O (ddH2O), allowing the plasmid to dissolve into solution for 30 minutes at room temperature. 10 μL of each dissolved plasmid was added to aliquots of HB101 E. coli cells supplied by the Vile laboratory, that were then placed on ice for 20 minutes. Each aliquot was heat shocked at 42 °C for 1 minute, then returned to ice for 2 minutes. 5 mL of super optimal broth with catabolite repression (SOC) media (Sigma-Aldrich, Montana, USA) pre- warmed to 37 °C was added to each transformation reaction, and the resulting broths were incubated for 60 minutes at 37 °C. 200 μL of each broth was streaked onto a LB agar plate containing 50 μg/mL ampicillin. The plates were incubated overnight at 37 °C, and selected colonies were used to inoculate 5 mL of LB broth containing 50 μg/mL ampicillin. These broths were incubated for 16 hours at 37 °C, and amplified plasmids were extracting using a

109

QIAprep Spin Miniprep kit (Qiagen, Germany). Purified plasmids were quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Delaware, USA).

2.4.1.4 Adaptation into pVSV-XN2 Purified pVSV-XN2 expression plasmid was supplied by the Vile laboratory. This plasmid was linearised with XhoI and NheI, separated by electrophoresis, and purified by gel extraction as previously described. Each of the VSV-VLP adapter constructs were excised from their respective plasmids with XhoI and NheI (Roche Diagnostics, Germany), separated by electrophoresis, and purified by gel extraction. Each purified VSV-VLP adapter construct was ligated with linearised pVSV-XN2, and the resulting plasmids pVSV.VP60-XN2, pVSV-SIIN.VP60-XN2 and pVSV-TS.VP60-XN2 were transformed into HB101 E. coli. Each plasmid was amplified and purified from culture broths as previously described.

2.4.1.5 Verification of pVSV-XN2 Plasmids The presence of an inserted VP60 sequence into each construct expression plasmid was confirmed by digestion with XhoI and NheI, and separation by electrophoresis as previously described. The presence of an insert was confirmed based on the separation of bands corresponding to both the linearised plasmid, and the inserted sequence. The identity of each insert was confirmed by sequencing undertaken by the Molecular Biology Core facility (Mayo Clinic, Minnesota, USA).

2.4.2 Production of Recombinant VSV

2.4.2.1 Culture of Baby Hamster Kidney Cells Baby hamster kidney (BHK) cells were supplied by the Vile laboratory. BHK cells were split every 2-3 days by removing the media, and adding 0.5 mL, 2 mL or 3 mL of trypsin-EDTA to the T25, T75 or T150 flask respectively. Trypsinised cells were washed with 20 mL DMEM + 5% FCS, spun at 300 xg for 5 minutes, and resuspended at 5 x 106 cells/mL in DMEM + 5% FCS. 0.5 mL, 2 mL or 5 mL of this cell suspension was used to seed a T25, T75 or T150 flask respectively, topped up with 4.5 mL, 18 mL or 45 mL of DMEM + 5%

FCS respectively. Cells were cultured in DMEM + 5% FCS incubated at 37 °C + 5% CO2.

110

2.4.2.2 Co-transfection of Expression Plasmids The co-transfection protocol used to generate recombinant VSV requires 3 expression plasmids in addition to the pVSV-XN2 construct expression plasmid, in addition to provision of T7 RNA polymerase. BHK cells were resuspended at 1 x 106 cells/mL in DMEM + 5% FCS, and 5 mL was seeded in 10 cm culture plates. These plates were incubated overnight at

37 °C + 5% CO2, after which the cells were washed briefly with 5 mL PBS before adding 5 mL of DMEM with no FCS. 50 μL of stock inoculum containing replication defective vaccinia that produces T7 RNA polymerase (MVA-T7) was added to each dish, equating to a MOI of 10. These inoculated cultures were incubated for 1 hour at 37 °C + 5% CO2. During this time, a transfection mixture consisting of 10 μg pVSV-XN2 construct expression plasmid, 1 μg pVSV-L, 5 μg pVSV-P and 3 μg pVSV-N in 60 μL FuGENE HD transfection reagent, diluted in sufficient DMEM to bring the total volume to 500 μL. This transfection mixture was incubated for 30 minutes at room temperature. The media was removed from each MVA-T7 inoculated culture, and replaced with 10 mL DMEM + 5% FCS. 500 μL of transfection mixture was added to assigned plates. The plates were swirled to mix, and incubated for 48 hours at 37 °C + 5% CO2. Following this incubation the media was harvested, centrifuged at 350 xg for 5 minutes, and the supernatant was passed through a 0.2 μm filter to remove any residual MVA-T7. 3 mL of this supernatant was added to per well of a 6-well plate containing BHK cells at >95% confluency. This plate was incubated for 1 hour at 37 °C + 5% CO2, after which the crude inoculum was removed and replaced with 3 mL DMEM + 5% FCS. The plate was incubated for 48 hours at 37 °C + 5% CO2, when the media was harvested. The resulting solution represented the unpurified recombinant VSV inoculum, ready for plaque purification.

2.4.2.3 VSV Plaque Assay The VSV plaque assay was used for both purification and quantification of the recombinant viruses. A pair of 6-well plates were seeded with 3 mL of BHK cells at 1 x 106 cells/mL, and incubated at 37 °C + 5% CO2 until they reached a confluency of >90%. The media was then aspirated, and the cells were washed with 3 mL of PBS. A 10-fold serial dilution of VSV inoculum was prepared, beginning with 12 μL of inoculum in 1188 μL of DMEM, followed by transfer of 120 μL along further aliquots of 1080 μL of DMEM and yielding dilutions

111 corresponding to virus titres in the ranges of between 102 and 1012. The PBS was aspirated from the BHK cells, and 1000 μL of each virus dilution was added to assigned wells. 1000 μL of DMEM was added to the last well as a mock infection control. The plates were incubated for 1 hour at 37 °C, after which the media was removed. 1.5 mL of an overlay mixture consisting of 2% noble agar pre-warmed to 42 °C and mixed 1:1 with DMEM was added to each well, and allowed to set for 10 minutes at room temperature. The plates were incubated for 48 hours at 37 °C + 5% CO2, when visible plaques were counted or picked for purification of virus. Plaques were picked by excising a core of agar using a sterile pipette tip, which was then stored at -80 °C prior to inoculum preparation.

2.4.2.4 VSV Inoculum Preparation BHK cells were resuspended at 1 x 106 cells/mL in DMEM + 5% FCS, and 5 mL was seeded in 10 cm culture plates. These plates were incubated overnight at 37 °C + 5% CO2. Each plaque pick tip was incubated in 100 μL of DMEM + 5% FCS for 10 minutes, and 10 μL of this solution was added to an assigned culture plate. The plates were then incubated for 48 hours at 37 °C + 5% CO2, after which the media was harvested and centrifuged at 350 xg for 5 minutes. The resulting supernatant was considered purified VSV inoculum, and the VSV titre was determined by plaque assay. These inoculums were bulked up through successive rounds of infection in BHK cells seeded in T150 flasks, and inoculated at a MOI of 1.

2.4.2.5 Verification of RHDV VP60 Expression RHDV VP60 expressed by recombinant VSV was verified by purifying VLP from VSV infected BHK cell cultures, and identifying VP60 by Western blot. 5 mL of BHK cells at 5 x 106 cells/mL in DMEM + 5% FCS was seeded in a T150 flask, and topped up with 45 mL of

DMEM + 5% FCS. The flask was incubated for 1 hour at 37 °C + 5% CO2, then inoculated with VSV at a MOI of 1 and incubated for 24 hours at 37 °C + 5% CO2. 0.5% Triton-X 100 was added, and the culture was incubated for 30 minutes at room temperature to lyse the BHK cells, liberate VLP and inactivated VSV. The resulting solution was centrifuged at 10,000 xg for 30 minutes to pellet cellular debris, and the supernatant was centrifuged at 100,000 xg for 1.5 hours. The pellet was resuspended in 200 μL of iPBS. Samples of VLP expressed by VSV were separated by SDS-PAGE, transferred and identified by Western blot as previously described.

112

2.4.2.6 Visualisation of VSV-expressed VLP Samples of VLP produced by VSV centrifuged at 100,000 xg for 16 hours on a 1.2-1.4 g/cm3 CsCl gradient as previously described. Purified VLP was harvested from the gradient interface, diluted to 0.2 mg/mL in Milli-Q H2O, and 10 μL of the diluted VLP was incubated for 1 minute on a plasma glowed carbon-coated grid. Excess sample was removed using absorptive filter paper, and replaced with 10 μL of a negative stain consisting of 1% phosphotungstic acid at pH 6.8. This negative stain was immediately removed using absorptive filter paper. Each grid was dried under a lamp for 5 minutes, before viewing on a JEM-1400 TEM (JEOL, Maine, USA) at the Microscopy and Cell Analysis Core Facility (Mayo Clinic, Minnesota, USA).

2.5 Statistical Analyses Statistical analysis was performed using Prism version 6.0 (GraphPad, California, USA). Power calculations, statistical test selection and analysis were performed under advice provided by a biostatistician, Andrew Gray (University of Otago, NZ).

113

3 Nucleic Acids Associated with RHDV VLP for Immunomodulation

3.1 Introduction

3.1.1 Nucleic Acids and VLP Virus-like particles are considered devoid of an intact genome, rendering them incapable of self-replication. However, VLP may not necessary be devoid of all nucleic acids. For example, VLP derived from feline herpesvirus (FHV) have been found to contain heterologous RNA, predominantly ribosomal but also including host-specific RNA [660]. Host RNA packaging has also been reported in satellite tobacco necrosis virus (STNV), cowpea chlorotic mottle virus (CCMV) VLP, and MS2 VLP [661-663]. It has been proposed that secondary structures present in host RNA may actually facilitate or induce VLP formation in the absence of the virus genome, at least for STNV VLP [661]. Beyond the encapsulation of host RNA during VLP formation, some VLP also possess the capacity to encapsulate DNA plasmids for applications in gene delivery. The concept of using a VLP for gene delivery can be traced back to studies on polyoma pseudoviruses, or polyoma-like particles [664-666]. Several other VLP have since proven suitable candidates for plasmid encapsulation and gene delivery, including human papillomavirus [667], John Cunningham virus (JCV) [668], polyomavirus [669, 670], Hepatitis E virus VLP [671] and Hepatitis B virus VLP [672].

JCV VLP in particular has been extensively explored for their potential to deliver genes in vivo. JCV is a polyomavirus that was originally identified in a patient suffering from progressive multifocal leukoencephalopathy [673], and has been identified within multiple tumours including those of central nervous system [674], colon [675], lung [676], stomach [677] and oesophagus [678]. This propensity for infecting tumour tissue has made JCV and its VLP a potential candidate for tumour targeted gene delivery. Recombinant expression of the JCV VP1 major capsid protein results in the formation of JCV VLP that are indistinguishable from virions of their parent virus [679]. Plasmids can be packaged in JCV VLP post-production through osmotic shock [680] or disassembly and reassembly [681]; however, production of JCV VLP in E. coli that also express the plasmid intended for delivery has been found to yield enhanced plasmid loading [682]. Nude mice bearing COLO- 320 HSR human colon carcinoma vaccinated with JCV VLP loaded with a GFP expression

114 plasmid resulted in selective uptake of VLP into tumour nodules, and the expression of GFP in tumour cells [682]. The procedure was replicated with a plasmid encoding herpes simplex virus (HSV) thymidine kinase (TK), with the selective uptake and gene delivery facilitating targeted elimination of the tumour cells through subsequent administration of ganciclovir. The capacity of JCV VLP for plasmid encapsulation is also quite substantial, known to be capable of encapsulating plasmids up to 9.4 kbp in length [683]. With the ability to accommodate plasmids of this length, it may be possible to use JCV VLP to selectively deliver large genes, or even multiple genes simultaneously.

VLP that do not inherently possess the ability to encapsulate DNA plasmids can be afforded this ability through recombinant insertion of nucleic acid binding domains derived from other viruses. A reported example of this involved the recombinant insertion of the DNA binding domain from the L1 or L2 proteins of HPV type 16 onto the N-terminus of the RHDV VP60 gene [656]. The resulting VLP, containing the L1 binding site (VP60-L1BS) or the L2 binding site (VP60-L2BS), were evaluated for their capacity to encapsulate and deliver plasmids in vitro. Both VP60-L1BS and VP60-L2BS gained the ability to bind DNA as detected by south-western blotting, although the VP60-L1BS variant had markedly superior binding affinity. This difference was subsequently translated into measures of gene delivery efficacy, with the VP60-L1BS variant outperforming VP60-L2BS in delivery of plasmids encoding either β-galactosidase or luciferase. Amongst a panel of cell lines used, gene delivery by VP60-L1BS was most effective in RK13 rabbit kidney cells, R17 rat/mouse hybridoma cells, and Cos-7 monkey fibroblast cells [656].

Modification of VLP to enable encapsulation of a DNA plasmid has the potential for multiple applications in vaccine design. Potential applications may include target antigen expression, replacement of defective tumour suppressor genes in tumour cells, or the expression of various regulatory genes for alteration of cell phenotype. While selective targeting may not be necessary in all applications, those with potentially harmful off-target effects benefit from incorporation of some form of selective targeting. For example, DNA-binding VLP can be designed to target a specific cell population through conjugation of the Fv fragment of antibodies generated against surface antigens of those cells. This process was demonstrated with polyoma virus VLP, in which aggregates of purified VL and VH domains from a selected antibody were conjugated onto the surface of the VLP [684]. The native binding affinity of these VLP was impaired through insertion of a polyionic fusion peptide in the H1 loop of the polyoma capsid protein, VP1, inhibiting off-target delivery. An alternative means

115 of selective targeting includes the conjugation of proteins or peptides specific for receptors expressed by the target cells, such as the conjugation of a human hepatocellular carcinoma (HCC)-specific binding peptide (SP) 94, as demonstrated with bacteriophage MS2 VLP [685].

3.1.2 Immunomodulation with VLP Vaccines Immunomodulation involves the manipulation of the immune system, altering how it responds through the action of an immunomodulator. An immunomodulator is any molecule or compound capable of such a manipulation, and includes cytokines, adjuvants and various inflammatory or anti-inflammatory compounds. Depending on the type of immunomodulator, the action employed may involve the suppression of inflammation, such as in an autoimmune or hypersensitivity reaction, or may induce presentation of co- stimulatory molecules that promote a specific type of immune response, enhancing vaccination. Although some VLP have been reported as potentially self-adjuvanting [686, 687], the inclusion of an adjuvant can alter the responses generated by a VLP vaccine. For example, the adjuvant α-galactosylceramide can form a composite particle when combined with RHDV VLP, capable of inducing increased activation of antigen-specific T cells [649]. VLP derived from enveloped viruses can also support the insertion of immunomodulators with a transmembrane motif into their envelope, often mediated by special lipid rafts. This unique property has been used to coat enveloped VLP in membrane-bound adjuvants such as flagellin [688], or various ligands targeting co-stimulatory receptors such as CD40 [689]. Completely independent immunosomes capable of directly stimulating T cell activation have also been reportedly constructed from Moloney murine leukemia virus (MoMLV) VLP with a surrogate TCR/CD3 ligand and the co-stimulatory receptor CD80 anchored in its envelope [690].

DNA oligonucleotides containing CpGs have also been investigated extensively for their potential as immunomodulators in combination with VLP vaccines. CpGs come in three main types, dubbed A, B and C. Type A CpGs feature a palindromic sequence flanked with guanosine, usually constructed with phosphodiester bridges, and simulate pDCs to secrete IFN-α [691]. Type B CpGs contain phosphorothioate bridges rather than phosphodiester, and are potent activators of B cells [692, 693]. Type C CpGs represent a mixture of both the structures and functions of Type A and B CpGs [694, 695]. The applications of CpGs in VLP

116 vaccines has been investigated most extensively in Qβ VLP [696], but CpG has also been used as an adjuvant with HPV VLP [697] and RHDV VLP [698]. With Qβ VLP, CpGs have been encapsulated inside the VLP for both the stimulation of immune responses [696], and the suppression of allergic immune responses [699]. Encapsulation of CpG facilitates targeted delivery to antigen presenting cells that take up the VLP, preventing potential off- target pathological effects from systemic administration of CpG, such as splenomegaly.

RNA nucleic acids have only recently been explored for their potential applications in combination with VLP vaccines, ranging from messenger RNA (mRNA) [700, 701], micro- RNA (miRNA) [702], short hairpin RNA (shRNA) [703] and short interfering RNA (siRNA) [704]. Although the underlying concept of RNA delivery by VLP is the induction, suppression or alternation of gene expression in target cells, these applications potentially take on an immunomodulatory role when designed to manipulate the immune cells. For example, targeting components of immune cell signalling pathways, such as signal transducer and activator of transcription (STAT) 3, may alter cell phenotype and the immune response. STAT3 is activated through tyrosine phosphorylation by Janus kinase (JAK) 1, through gp130 signalling induced by IL-6. This pathway is of particular interest in CRC, as a self- sustained feedback loop between IL-6, STAT3 and miR-34a identified in CRC has been implicated in the induction of an EMT transition, invasion and metastasis [705]. This feedback loop is completed by the identification of IL-6 expression induced by STAT3 activation in macrophages and human endothelial cells [706-708]. Targeted knock down of STAT3 was found to impair this feedback loop, inhibiting IL-6 induced transformation in a murine model of metastatic CRC [705]. The combination of VLP vaccination and siRNA delivery has significant potential for immunomodulation by targeting and modifying both stimulatory and inhibitory signalling pathways.

117

3.2 Objectives The purpose of this chapter was to investigate the potential for RHDV VLP to encapsulate and deliver short nucleic acids, including CpGs and siRNA, for the purpose of immunomodulation. Although CpGs have already been used in conjunction with RHDV VLP vaccines, the potential interaction between CpGs and RHDV VLP has not been investigated. To promote the encapsulation of CpGs, chimaeric VLP were developed that contained either a string of positively charged residues or a known DNA binding site recombinantly inserted at the N-terminus of VP60. CpGs were associated with RHDV VLP post-production through a combination of disassembly and reassembly, and assessed for their ability to induce NF-κB activation, and subsequent activation of DCs. The potential for RHDV VLP to deliver encapsulated siRNA was also assessed in this chapter, focusing on STAT3 as a model system for targeted immunomodulation.

3.2.1 Aims 1. Assess the encapsulation and delivery of CpGs inside RHDV VLP as a composite adjuvant vaccine particle. 2. Evaluate the potential for RHDV VLP-mediated delivery of siRNA.

118

3.3 Results

3.3.1 Production of Chimaeric RHDV VLP Designed for Nucleic Acid Association Chimaeric RHDV VLP with recombinant VP60 containing the L1BS DNA binding site at its internally located N-terminus (DBS.VP60) has previously been described for its ability to encapsulate and deliver plasmids [656]. The production of this construct has previously been replicated in our laboratory, and was explored for its natural affinity to bind both DNA and RNA [709]. An alternative construct consisting of an octaarginine residue at the N-terminus of VP60 (R8.VP60) was also previously developed in our laboratory, but its properties had not been investigated. The concept behind this variant was to generate a positively charged interior surface within the VLP, potentially attracting and facilitating the encapsulation of negatively charged nucleic acids. The availability of these binding or attraction motifs would be enhanced by disassembling the VLP into subunits, with encapsulation of molecules mediated by reassembly into intact VLP. VP60, R8.VP60 and DBS.VP60 were provided pre- cloned in pAcUW51(GUS) plasmids, and were expressed using the FlashBAC Ultra system as has been described.

VP60, R8.VP60 and DBS.VP60 purified from inoculated Sf21 cultures were identified by SDS-PAGE (Figure 3.1 A). Purified preparations of both R8.VP60 and DBS.VP60 VLP consistently contain additional protein bands. It is unknown whether these bands represent the use of alternate start codons, premature termination of translation, or are indicative of post-translational degradation; however, the arginine residues inserted at the N-terminus of VP60 in R8.VP60 may act as a cleavage site for various intracellular proteases. DBS.VP60 VLP has also been identified as relatively unstable in comparison to unmodified RHDV VLP [709], and may also be susceptible to some form of degradation process. Formation of intact VP60, R8.VP60 and DBS.VP60 VLP was confirmed by transmission electron microscopy (Figure 3.1 B, C and D). Both R8.VP60 VLP and DBS.VP60 VLP appear similar to VP60 VLP in structure and morphology, commonly described as cog-shaped due to its surface projections visible on the surface of the icosahedron cross-section. While R8.VP60 VLP appeared relatively homogeneous, DBS.VP60 VLP have marked variability in appearance with an apparent increase in partially disassembled VLP. This provides some evidence in support of the previously reported instability of this VLP construct. It is unknown at this stage what effects this instability may have on the functionality.

119

Figure 3.1 Production of VP60, R8.VP60 and DBS.VP60 (A) SDS-PAGE analysis of RHDV VLP comparing 1) VP60, 2) R8.VP60 and 3) DBS.VP60 VLP. RHDV VLP have distinct cog-like morphology as identified by electron microscopy, features shared by (B) VP60, (C) R8.VP60 and (D) DBS.VP60 VLP. (L) broad range protein ladder. Magnified images inset to demonstrate cog- like morphology.

3.3.2 Endogenous Nucleic Acids in RHDV VLP The presence of endogenous nucleic acids in VP60 and DBS.VP60 VLP has previously been investigated [709]. In this study, VP60 VLP were found to contain small amounts of short nucleic acids, presumably consisting of fragments encapsulated from their expression cell line during assembly. DBS.VP60 demonstrated an enhanced ability to encapsulate endogenous nucleic acids, identified to consist of both short RNA fragments, and a distinct DNA fragment that could only be dislodged through addition of SDS. These findings indicated that while RHDV VLP may be devoid of an intact virus genome, they might still contain various genetic materials. The presence of small amounts of endogenous short

120 nucleic acids present in preparations of VP60 VLP was confirmed by agarose gel electrophoresis (Figure 3.2 A). R8.VP60 VLP were found to contain an increased amount of endogenous nucleic acids, with fragments of much longer length in comparison to VP60 VLP (Figure 3.2 B).

Figure 3.2 Endogenous Nucleic Acids in VP60 and R8.VP60 VLP Identification of endogenous nucleic acids present in 20 µL samples of 1 mg/mL RHDV VLP by staining with ethidium bromide. (A) Endogenous nucleic acids present in samples of untreated (UT) VP60 VLP, and benzonase (Benz) treated VP60 VLP. (B) Endogenous nucleic acids present in samples of UT R8.VP60 VLP, and Benz treated R8.VP60 VLP. (L) 1 kb plus Ladder. These gels are representative of three independent repeats, each including triplicate samples.

Benzonase was used to remove any endogenous nucleic acids available for enzymatic digestion in the VLP sample, leaving nucleic acids encapsulated within VLP untouched. Although the effects of benzonase digestion were difficult to determine in VP60 VLP due the low levels of nucleic acids prior to digestion, in R8.VP60 VLP there was a notable partial reduction in the endogenous nucleic acids present. This suggests that some of these endogenous nucleic acids may be present on or around the surface of VLP, or in solution. However, as these nucleic acids are not completely ablated by benzonase treatment, a proportion of these nucleic acids are assumed to be physically inaccessible to the benzonase

121 enzyme. These nucleic acids may be contained within the VLP, or held in sufficiently tight association as to impair their accessibility to enzymatic degradation. This benzonase resistance appeared to be skewed towards shorter nucleic acids, suggesting that there may be some kind of size limit, exclusion or preference upon the resistance conferred.

3.3.3 Disassembly and Reassembly of RHDV VLP Based on published reports, the efficient encapsulation of exogenous nucleic acids within VLP is reliant upon disassembly and reassembly [696]. Disassembly of VLP involves the temporary disruption of bonds between the virus capsid proteins, most commonly through a combination of disulphide bond reduction and divalent cation chelation. The method for disassembly and reassembly of RHDV VLP has been previously published [656]; however, direct replication of this method proved insufficient for efficient disassembly of our VLP. This discrepancy was rectified following optimisation, identified to be primarily due to insufficient EGTA. This may be due to an increased amount of divalent cations associated with our VLP, possible acquired from expression cells, growth media, or buffers used. In comparison to untreated RHDV VLP (Figure 3.3 A), disassembly with the published method including 1 mM EGTA produced few of the capsomere-like particles noted by this publication as indicative of disassembly (Figure 3.3 B). These capsomere-like particles are morphologically distinct from intact VLP, and vary in size and shape. While the distribution of VLP across the surface of the carbon-coated grids is unlikely to be even, it appeared that the vast majority of VLP were unaffected by the treatment. When the concentration of EGTA was increased ten-fold to 10 mM, there was a notable increase in the number and size of the clumps of capsomere-like particles (Figure 3.3 C). Upon reassembly from either treatment, RHDV VLP can be restored to their uniform, cog-like appearance (Figure 3.3 D).

122

Figure 3.3 Disassembly and Reassembly of RHDV VLP RHDV VLP were disassembled and reassembled using a previously published method [656]. Electron microscopy was used to assess the morphology of RHDV VLP under different treatment conditions, including (A) untreated, (B) disassembled using a method including 1 mM EGTA, and (C) disassembled using a method including 10 mM EGTA. (D) Reassembled VLP following disassembly, in this example from a disassembly method that included 10 mM EGTA. Arrows indicate potential capsomere-like particles, thought to be formed during disassembly of RHDV VLP. **** p < 0.0001.

3.3.4 Qualification and Detection of RHDV VLP Disassembly During the process of optimising the disassembly and reassembly of RHDV VLP, several methods were considered for determining the efficacy of each modified treatment. Methods considered for assessing the disassembly and reassembly of VLP included native gel electrophoresis, differential ultracentrifugation and dynamic light scattering (DLS). Amongst these, differential ultracentrifugation proved a particularly valuable tool for quantification of disassembly. Initial attempts at differential ultracentrifugation involving sucrose cushions were ineffective; however, the use of CsCl gradients proved suitable. The resulting method selected is identical to that used during the purification of RHDV VLP, involving overlaying

123 treated and untreated VLP over a gradient of 1.2-1.4 g/cm3 CsCl, and ultracentrifuging the samples at 100,000 xg for 18 hours (Figure 3.4 A).

Figure 3.4 Quantification of RHDV VLP Disassembly Disassembly of RHDV VLP was quantified using a combination of differential ultracentrifugation and VP60 western blot. (A) Diagrammatic representation of the differential ultracentrifugation method, with a) CPBS, overlaid over b) VLP sample, which was overlaid over a gradient of c) 1.2 g/cm3 CsCl and d) 1.4 g/cm3 CsCl. 100 μL samples were carefully extracted from the interface between the sample and 1.2 g/cm3 CsCl fraction (Int 1), and at the 1.2-1.4 g/cm3 CsCl interface (Int 2). (B) VP60 was quantified at Int 1 and Int 2 for untreated and disassembled VLP samples by western blot, based on the fluorescence intensity of the Dylight 800 labelled secondary antibody used. The provided western blot is representative of three independent repeats. (C) Quantification of VLP disassembly with either 1 mM or 10 mM EGTA. Statistical analysis was performed using unpaired t-tests. **** p < 0.0001.

Intact, non-disassembled VLP accumulate at the CsCl gradient interface (Int 2), forming a diffuse visible band. When disassembled VLP are applied to this gradient, they do not all accumulate at this interface, with VP60 also identifiable at the interface between the sample and the CsCl gradient (Int 1). Solutions harvested from each of these interfaces were analysed by SDS-PAGE and western blot, facilitating the detection of VP60 at each interface (Figure 3.4 B). Relative quantification based on band density facilitated the estimation of disassembly efficacy, based on the percentage of VP60 identified at each interface. This process enabled the optimisation of the disassembly protocol, increasing the concentration of EGTA used

124 from 1 mM to 10 mM (Figure 3.4 C). Disassembly achieved was estimated using this method at around 25% with 1 mM EGTA, and around 75% with 10 mM EGTA. It may be possible to further increase the efficacy of disassembly by altering the concentration of DTT used; however, this may risk impairing reassembly of VLP.

A suitable method was also sought for the rapid detection of disassembly and reassembly of RHDV VLP, as quantification by differential ultracentrifugation and western blot consumed the entire sample. DLS can fill this role, involving the calculation of particle diameter based on how those particles scatter light when in solution. The average diameter of RHDV VLP as determined by DLS was around 37.1 nm (Figure 3.5 A), and does not change following dialysis into disassembly buffer. When monitored over time by DLS, the average diameter of RHDV VLP gradually decreases following disassembly treatment, stabilising at around 23.5 nm after 8-12 hours. This state was maintained over the entire monitored period, up to 24 hours. Particles detected by DLS at around 23.5 nm may represent the capsomere-like particles previously identified by electron microscopy. It is unknown whether individual or dimerised VP60 are liberated into solution during the disassembly process, and alternative methods would be required to detect this occurrence. Following reassembly, RHDV VLP were found to return to their normal size, with a post-reassembly average diameter of around 36.7 nm (Figure 3.5 A).

Although DLS is suitable for detection and monitoring of RHDV VLP disassembly and reassembly, it also consumes the entire sample or exposes it to non-sterile conditions. An alternative method was instead proposed, based on the light absorption profile of RHDV VLP. It was hypothesised that the process of VLP disassembly and reassembly would result in the exposure of previously buried amino acids, some of which possess specific light absorption properties, like tyrosine and tryptophan [710, 711]. These aromatic amino acids have a characteristic spectrophotometric absorbance over a ratio of A250/A280, and were proposed to potentially provide a surrogate determination of VLP assembly state. RHDV VLP were found to have a A250/A280 ratio of very close to 1 (Figure 3.5B), which was unaffected by dialysis into disassembly buffer. When monitored over time, this ratio gradually decreased following disassembly treatment, stabilising at around 0.4 after 8-12 hours. This ratio was maintained over the entire monitored period, up to 24 hours. Following reassembly, this ratio gradually returned to around 1. The kinetics of this analysis closely matched that generated by DLS, suggesting that spectrophotometric analysis of absorbance at A250/A280 may be a suitable surrogate measure of RHDV VLP assembly state. This

125 method holds significant practical advantages over DLS, as it requires a relatively small amount of sample to analyse, and can be performed on commonly available laboratory equipment. As a result, the A250/A280 ratio was used as a means of checking for batch variability between experiments involving disassembly and reassembly of RHDV VLP.

Figure 3.5 Detection of RHDV VLP Disassembly and Reassembly A comparison of methods used for monitoring RHDV VLP assembly state following dialysis, disassembly and reassembly. (A) Measurement of average particle diameter by DLS. (B) Measurement of A250/A280 light absorbance ratio by spectrophotometry. Statistical analysis was performed using unpaired t-tests at each time point, with a Bonferonni post hoc adjustment for multiple measures. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

126

3.3.5 Encapsulation of CpGs into RHDV VLP The capacity for RHDV VLP to encapsulate CpGs was investigated with and without disassembly and reassembly; using dialysis to remove non-encapsulated CpGs. Unexpectedly, VP60, R8.VP60 and DBS.VP60 all possessed the ability to retain CpGs in solution following dialysis, both with and without disassembly and reassembly (Figure 3.6 A). This finding may suggest that RHDV VLP possesses a natural affinity for short DNA oligonucleotides, and that this affinity may not be reliant on the exposure of internal N- terminus residues. In addition to retention following dialysis, RHDV VLP were also found to resist degradation of CpGs with benzonase (Figure 3.6 B). It is unknown by what mechanism this resistance is conferred, but possible sources include steric hindrance of benzonase, inhibition of the enzyme by molecules associated with VLP, or disassembly- independent encapsulation of CpG.

The ability to separate, detect and relatively quantify CpG on TBE gels using the Odyssey FC system also provided the potential for quantification of CpG associated with RHDV VLP. Triplicates of VP60, R8.VP60 and DBS.VP60 VLP samples associated with CpG and dialysed were electrophoresed on 20% TBE acrylamide gels, and stained with Gel Green (Figure 3.7 A). The fluorescence intensity of each Gel Green labelled CpG band was compared to a concentration curve of CpG, calculating the relative quantity of CpG associated with each VLP sample (Figure 3.7 B and C). Due to the loss of VLP during dialysis, each sample was also assessed by western blot and compared to a 1 mg/mL VP60 VLP standard. Correcting for the concentration of VLP present in each sample, the amount loaded onto the TBE acrylamide gels, and calculating up to a 1 mg VLP sample yielded estimated quantities of CpG associated with 1 mg of each VLP tested (Figure 3.7 D). While DBS.VP60 VLP was found to be associated with significantly more CpG relative to the amount of VLP present in comparison to VP60 VLP, R8.VP60 VLP was associated with significantly less. This suggests that the octaarginine domain present in R8.VP60 is not promoting the association of CpG with RHDV VLP, and may actually be detrimental for this phenomenon.

127

Figure 3.6 Encapsulation of CpGs into RHDV VLP (A) Retention of CpGs following dialysis in the presence of VP60, R8.VP60 or DBS.VP60 VLP with or without disassembly/reassembly (Dis/Reas), as assessed by TBE acrylamide gel stained with gel green. (B) Resistance of CpGs to benzonase digestion in the presence of RHDV VLP, as assessed by TBE acrylamide gel stained with gel green. L1 = Hyperladder V, L2 = Low Molecule Weight Ladder, Dial = Dialysis treatment, Benz = Benzonase treatment.

128

Figure 3.7 Quantification of CpG Associated with RHDV VLP (A) TBE acrylamide gel electrophoresis of triplicate samples, indicating the retention of CpGs in dialysed samples in the presence of VP60, R8.VP60 or DBS.VP60. (B) TBE acrylamide gel electrophoresis of a concentration curve of CpGs, representative of three independent repeats. (C) Concentration curve of CpGs graphed respective of their band intensity. (D) Quantification of CpGs associated with VP60, R8.VP60 and DBS.VP60 VLP following dialysis, with CpG quantified by TBE acrylamide gel electrophoresis, and VLP quantified by western blot. Statistical analysis was performed using unpaired t-tests, comparing each recombinant VP60 VLP to control VP60 VLP. * p < 0.05, ** p < 0.01

3.3.5 CpGs Associated with RHDV VLP Enhances NF-κB Activation CpGs can stimulate the activation of NF-κB through endosomal TLR-9 recognition, with CpGs naturally present in bacterial DNA. The ability for CpGs associated with RHDV VLP to stimulate NF-κB was investigated using the reporter RAW-Blue cell-line (Invivogen, California, USA). In this reporter cell line, based on the murine RAW 264.7 macrophage cell line, NF-κB activation induces the expression of SEAP. The amount of SEAP secreted into

129 the culture media is linked to NF-κB activation, and can be quantified using a colorimetric assay. Using this reporter system, we can see that CpG is a potent activator of NF-κB, and that dialysis completely removes all detectable CpG from a solution (Figure 3.8). VP60, R8.VP60 and DBS.VP60 all retained the ability to induce activation of NF-κB following dialysis to remove free CpGs, although activation was most pronounced with DBS.VP60. R8.VP60 associated with CpG was impaired in its ability to induce NF-κB activation, with significantly reduced SEAP secretion. These results matched the trends observed in the quantification of CpG associated with each chimaeric VLP.

Figure 3.8 NF-κB Activation in RAW-blue Cells Activation of NF-κB was compared based on SEAP production. SEAP production was compared between VP60, R8.VP60 and DBS.VP60 associated with CpGs, before and after dialysis. CPBS, CpG dialysed and 25 ug of CpG were used as controls. Statistical analysis was performed using unpaired t-tests between groups. ** p < 0.01, **** p < 0.0001. Results are representative of three independent repeats.

3.3.6 CpGs Associated with RHDV VLP Enhances BMDC Activation Upon stimulation with potent adjuvants such as LPS, APCs such as BMDCs become activated and upregulate the expression of co-stimulatory molecules including CD40, CD80

130 and CD86. The relative expression of these co-stimulatory molecules is linked to the magnitude of BMDC stimulation. While LPS stimulated potent activation of BMDCs with upregulation of all three co-stimulatory molecules, the CpGs used did not appear to stimulate the upregulation of CD80, with increases observed in surface presentation of CD40, CD86 and I-A/I-E (Figure 3.9 A-D). However, when these CpGs were administered in association with RHDV VLP, BMDCs upregulated all three co-stimulatory molecules to levels similar to those induced by LPS.

Figure 3.9 BMDC Activation with CpGs Associated with RHDV VLP

131

Median fluorescent intensity (MFI) of the activation markers (A) CD40, (B) CD80, (C) CD86 and (D) I-A/I-E on the surface of murine BMDCs, following stimulation with CpGs associated with RHDV VLP. Dialysed CpG, 25 μg CpG and 1 μg LPS treatments were used as controls. Statistical analysis was performed using unpaired t-tests between groups. NS = Non-significant, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Results are representative of three independent repeats.

For CD80, CD86 and I-A/I-E, the surface presentation observed was significantly increased when the CpG was co-delivered in association with RHDV VLP. These results suggest that when CpGs are delivered in association with RHDV VLP, the resulting composite particle induces enhanced activation of BMDCs over each component administered individually. While RDHV VLP alone did not appear to upregulate CD40, CD80 or CD86, stimulated cells had increased surface levels of I-A/I-E. This may represent the background effect of RHDV VLP when administered in the absence of a suitable adjuvant.

3.3.7 Association of siRNA with RHDV VLP Based on the ability of RHDV VLP to naturally associate with CpGs, it was hypothesised that they would also possess the ability to associate with siRNA. The capability of RHDV VLP to associate with siRNA was investigated using dialysis to remove free siRNA. RHDV VLP were confirmed to possess the ability to naturally associate with siRNA (Figure 3.10 A). VP60 VLP was found to associate with around 1.6 μg siRNA per 1 mg VLP, while R8.VP60 and DBS.VP60 associated with around 0.8 and 2.5 μg siRNA respectively (Figure 3.10 B). These values are likely an under-representation of this association, as siRNAs are highly labile and prone to degradation in solution. The proportions of siRNA associated with each type of VLP followed the same trend established by the association of CpGs, with decreased association with R8.VP60 and increased association with DBS.VP60.

132

Figure 3.10 Association of siRNA with RHDV VLP (A) Retention of siRNA following dialysis in the presence of RHDV VLP, as assessed by TBE acrylamide gel electrophoresis. (B) Quantification of siRNA associated with VP60, R8.VP60 and DBS.VP60 VLP following dialysis, with siRNA detected by TBE acrylamide gel electrophoresis, and VLP quantified by western blot as previously used for CpG quantification. ** p < 0.01

3.3.8 Knockdown of Stat3 with siRNA associated with RHDV VLP The capacity for RHDV VLP to deliver siRNA to cells in vitro was investigated in the RAW- Blue cell line. RAW-Blue cells were treated with VP60, R8.VP60 and DBS.VP60 VLP associated with STAT3 siRNA. 8 hours later, these cells were stimulated with LPS, which should induce the upregulation of Stat3 expression. The media was removed and the cells were lysed 12 hours after the addition of LPS. Stat3 expression was assessed by separating lysates by SDS-PAGE, transfer onto nitrocellulose membrane, and probing by western blot with an anti-Stat3 mAb. Actin expression was used as a control. VP60, R8.VP60 and DBS.VP60 VLP were each capable of inducing a targeted knockdown in Stat3 expression in RAW-Blue cells, with dose-dependent efficacy observed in both VP60 and DBS.VP60 VLP (Figure 3.11 A-D). The extent of the knockdown achieved varied between each VLP, with Stat3 expression reduced down to around 70% with 10 μg of either VP60 or R8.VP60 VLP, and down to around 60% with DBS.VP60 VLP. As was previously identified with CpG association, DBS.VP60 VLP performed slightly better than both VP60 and R8.VP60, possibly as a reflection of an increased quantity of siRNA associated with DBS.VP60 VLP. Transfection of the STAT3 siRNA using FuGene 6 resulted in a knockdown of Stat3 133 expression to around 20%, demonstrating the limited efficacy of RHDV VLP-mediated delivery of siRNA with respect to standard methods.

Figure 3.11 Knockdown of STAT3 by siRNA associated with RHDV VLP A) Expression of STAT3 and actin in RAW-Blue cells 24 hours after stimulation with LPS, as detected by western blot. STAT3 expression was normalised relative to actin expression for cells treated with B) VP60, C) R8.VP60 and D) DBS.VP60 delivered in combination with STAT3 siRNA. Untreated and FuGene 6 transfected cells were used as controls.

134

3.4 Discussion

3.4.1 Nucleic Acids Associated with RHDV VLP RHDV VLP can associate with various nucleic acids, accumulated endogenously during production or added exogenously following purification. While this ability is possessed by VP60 VLP, it appears to be enhanced by recombinant insertion of DNA-binding or charge associated motifs. R8.VP60 VLP was found to have an increased amount of endogenous nucleic acids in preparations, with the length of benzonase resistant fragments reaching up to around 400-500 bp. These fragments may be of cytoplasmic origin, potentially consisting of mRNA, micro RNA (miRNA) or various other forms of RNA. Alternatively, these endogenously encapsulated nucleic acids may include extrachromosomal or extranuclear DNA present in the cytoplasm. As the cells producing these VLP are also infected with replicating AcMNPV, they may also theoretically encapsulate nucleic acids encoding components of their own expression system. The short <100 bp fragments that appeared to be associated with all RHDV VLP assessed may consist of building block materials, such as transfer RNA (tRNA) or other small nucleic acid structures.

Although the most likely location of endogenous nucleic acids associated with RHDV VLP would be within its intraparticulate space, the ability for exogenously applied fragments to be afforded resistance to enzymatic digestion may challenge this assumption. The size restriction of this resistance may be due to either: A) The size of gaps between VP60 dimers for diffusion between the intraparticulate and extraparticulate spaces; B) The limit of steric hindrance of enzymatic digestion on nucleic acids tightly associated to the surface of VLP; or C) The distance that nucleic acids can insert themselves within the gaps between VP60 dimers. While the first two possibilities represent inherent functionality of a hollow particle, encapsulating molecules or simply facilitating surface attraction through various motifs, the final theory is based on an observation made some time ago with BK virus (BKV) VLP [712]. In this study, Touze et al found that BKV VLP could bind to the ends of a linearised plasmid, with the DNA strands visibly connecting between two VLP (Figure 3.12). If this phenomenon resulted solely from surface interaction, the linearised plasmid should be coiled around the VLP surface. Instead, this interaction appears to involve only the terminus of DNA strand, with its terminal motifs interacting with the VLP surface, or by partially inserting itself into the space between capsid subunits. It would be interesting to investigate whether this phenomenon also occurs in RHDV VLP through the replication of these experiments.

135

Figure 3.12 Association Between BKV VLP and a Linearised Plasmid [712] An electron micrograph from Touze et al [712], identifying a strand of DNA linking between two BKV VLP. Preparations were observed by positive staining with 1.5% aqueous uranyl acetate over a dark field. The white bar represents 100 nm.

3.4.2 CpGs associated with RHDV VLP CpGs have been used as an adjuvant in a variety of VLP-based vaccines, including with HPV VLP [697], Qβ VLP [696] and RHDV VLP [698]. Co-delivery of CpG and VLP together as a composite particle through encapsulation, chemical conjugation or direct association ensures that APCs that receive the vaccine are also receiving sufficient danger signals as to promote activation, and surface presentation of co-stimulatory molecules. As RHDV VLP has been found to naturally associate with CpGs, previous studies involving vaccination with this VLP and adjuvant combination may have unwittingly benefitted from this effect. The association between RHDV VLP and CpGs is reminiscent of its reported affinity for α- galactosylceramide [649] and mannose [652]. Each of these molecules shares a common chemical backbone, consisting of a carbohydrate ring domain. In other caliciviruses, most notably norovirus, the P2 domain of the major capsid protein contains a conserved carbohydrate-binding domain [713-715]. Several caliciviruses are known to use these domains to bind to the carbohydrate moieties that form some of the blood group antigens present on the surface of host red blood cells [716-719].

136

RHDV has been proposed to potentially share this characteristic, with a reduction in disease susceptibility following a low virus titre challenge identified amongst adult rabbits with low amounts of related blood group antigen ligands [720]. However, hepatocytes, the main target for RHDV infection and replication do not possess these blood group antigen ligands, indicating that this is unlikely the only mechanism employed by RHDV for infection of its host cells [717]. It is intriguing that RHDV VLP may retain the affinity for carbohydrate moieties possessed by its parent virus. This further expands the functionality of RHDV VLP as a versatile vaccine vector, and potentially opens multiple additional avenues for research. Even with CpGs, it would be fascinating to explore whether the delivery of each class of CpG in association with RHDV VLP retains or overrides their respective class-defined affinities. It would also be interesting to pinpoint the specific location where associated CpGs can be found in RHDV VLP through cryo-electron microscopy, though it seems likely that the implicated P2 domain is responsible. It is worth noting that there are already patents held that restrict some commercial applications of CpG associated with VLP, including a patent covering the process of CpG encapsulation (US 8691209 B2), and on the actual vaccination applications (WO 2015089114 A1); however, it may be possible to avoid these patents through the use of novel vaccine constructs, composites or combinations. Some examples of potential alternative constructs that could be explored are discussed in Section 5.1.2.1.

3.4.3 siRNA associated with RHDV VLP As with the association between CpGs and VLP, the potential for siRNA associated with VLP has already been recognised through the publishing of a patent (US8729038). Interestingly, this patent specifically covers siRNA internalised within the interior of the VLP, but makes no mention of conjugation or association throughout the remaining structure of VLP. As was identified with CpGs, RHDV VLP have a natural affinity for siRNA. Although siRNA associated with RHDV VLP was able to induce a detectable targeted knockdown in treated cells, the efficacy was such that it is unlikely that VLP will be a viable in vitro alternative to traditional transfection or electroporation techniques. Instead, it is possible that in vivo is where VLP-mediated delivery of siRNA may yield improved results; however, the lability of siRNA may present as a significant limitation. The ability to selectively manipulate the expression of target proteins in antigen presenting cells in vivo would be a powerful tool, capable of both stimulation and suppression in different disease contexts.

137

The APC-specific avenue of siRNA associated with VLP is unique compared to similar therapies involving viruses for the treatment of cancer, such as the use of an oncolytic virus to induce the expression of siRNA or similar RNA interference molecules [721, 722]. The ability to target RHDV VLP to specific immune cell populations through the chemical conjugation of various receptor ligands could open up further potential for VLP-mediated siRNA delivery, with potential applications in combination with prophylactic and therapeutic vaccines, therapies for autoimmunity, prevention of graft-versus-host disease or related conditions, and even in commercial diagnostic techniques. Before endeavouring on these potential applications, it remains that the delivery of siRNA associated with RHDV VLP must be confirmed with primary cells in vitro, and then translate in vivo. This represents a significant body of additional research required to facilitate the investigation of these potential applications.

3.4.3 Conclusions The addition of nucleic acids further expands the list of molecules known to naturally associate with RHDV VLP. Collectively, these adjuvant-vaccine composites represent an intriguing novel approach to vaccine formulation, taking advantage of the natural affinity of virus capsid proteins for specific chemical moieties. It would be interesting to see whether other immunogenic compounds with similar moieties, such as glycolipids, glycoproteins, or even glycosylated/glycated proteins and peptides would also naturally associate with VLP. Likewise, it would also be interesting to further explore the role this binding affinity may play in the infection and replication of host cells with RHDV and other related Caliciviridae. These and other potential future applications of RHDV VLP will be discussed further in Chapter 5. For the purposes of this study, it was interesting to highlight a mechanistic advantage associated with the use of CpGs as an adjuvant, and the potential for formation of a composite vaccine particle through the direct association between CpG and RHDV VLP. Similarly, if the ability to deliver siRNA in association with RHDV VLP is confirmed in vivo, this could further broaden the avenues for the development of new VLP-based vaccines.

138

4 Development of a Virus-based Vaccine for Colorectal Cancer

4.1 Introduction

4.1.1 Incorporating Multiple Colorectal Cancer Epitopes into RHDV VLP Tumour escape and recurrence is a prominent potential adverse outcome amongst immunotherapies for cancer, with progression-free survival remaining an important prognostic marker of therapeutic efficacy. Tumour escape following immunotherapy may result from a multitude of factors, including ineffective vaccination, tolerisation, or due to immunoediting of the tumour to suppress or eliminate expression of vaccine-targeted TAAs. Although the design of the vaccine itself, incorporation of adjuvants, and an appropriate dosage regimen should restrict the former two limitations, tumour immunoediting remains particularly challenging to counter. Counteracting this process may be achieved through several potential mechanisms. These include immunomodulation, which has been previously investigated and discussed in Chapter 3, and broadening the repertoire of the immune response induced. As a heterogeneous mass of cells, solid tumours may possess multiple subpopulations or variations in TAA expression profiles. This provides redundancies within the tumour, with the elimination of populations expressing a specific TAA giving rise to populations with low or absent expression. Development of a vaccine targeting and eliciting immune responses against multiple epitopes should limit these redundancies, and promote the elimination of a broader range of tumour cells. This may include potentially resistant subpopulations that do not express all of the individual antigens targeted, and hence may survive one or several mono-target therapies.

Although the concepts behind multi-epitope targeting should provide a significant advantage in vaccine development, this construct also yields further complications that may restrict the functionality if improperly designed. An important factor for consideration in the development of a multi-target vaccine is epitope immunodominance. Although an antigenic protein will contain a broad repertoire of potential immune epitopes that may be targeted in a vaccine, immunodominance may result in only a select few inducing an epitope-specific immune response. Even in the context of a viral infection involving multiple viral genes products, cytotoxic immune responses tend to be predominated by a single MHC-I haplotype

139

[723, 724], and may even be restricted to only a single peptide of 8-10 residues [725-727]. A similar situation is mirrored in MHC-II, where these predominant immunogenic peptides were characterised as immunodominant determinants [728]. Within each antigen, there are usually also multiple peptide sequences with variable immunogenicity under specific circumstances, including subdominant and cryptic determinants. Subdominant determinants generate relatively weak immune responses, while cryptic determinants are generally only immunogenic in a synthetic form [728]. Beyond detection of the epitopes themselves, mechanisms that can influence immunodominance include cross-reactivity from a prior exposure [729], differential thymic selection of T cells [730], and competition between activated CTLs during expansion [731].

Within the scope of multi-epitope cancer vaccines there are several examples of promising candidates, ranging from peptide pools [732], DNA vaccines [733], synthetic polypeptides [734], and DC-based vaccines [735]. For example, a triplicate peptide vaccine for treatment of esophageal squamous cell carcinoma (ESCC) completed a multicentre phase II clinical Trial [736]. This triplicate vaccine consisted of HLA-A24-restricted peptides derived from the Cancer-Testis antigens TKK protein kinase (TKK), lymphocyte antigen 6 complex locus K (LY6K) and insulin-like growth factor (IGF)-II mRNA binding protein 3 (IMP-3), administered subcutaneously with incomplete Freund’s adjuvant (IFA). Progression free survival was significantly improved amongst HLA-A24+ ESCC patients in comparison to HLA-A24-, with patients that induced CTL response against multiple peptides yielding better clinical responses. This group has also completed a phase II clinical trial for a similar triplicate peptide vaccine consisting of HLA-A24-restricted peptides from LY6K, IMP-3 and cadmium-specific carbonic anhydrase 1 (CDCA1) with IFA, for treatment of head and neck squamous cell cancer (HNSCC) [737]. In this trial, they observed a significant improvement in overall survival amongst HLA-A24+ HNSCC patients in comparison to HLA-A24-, and reported one case of complete remission.

A peptide pool-based vaccine has also completed a phase II clinical trial for treatment of melanoma, consisting of twelve melanoma peptides (12MP) restricted to HLA-A1, A2 or A3 [738]. This 12MP vaccine was administered with IFA and GM-CSF by itself, or in combination with either tetanus peptide or with a six melanoma helper peptide (6MHP) pool. Although cytotoxic immune responses against the 12MP vaccine were detected, there was no significant improvement in clinical outcome associated with this vaccine. Interestingly, while the 6MHP vaccine failed to augment the 12MP vaccine when administered in combination,

140 immune responses generated against the 6MHP vaccine were associated with significant improvements in both clinical response and overall survival [738]. Multi-epitope cancer vaccines in clinical trials are not limited solely to peptide pools, as a recent phase I trial for a multi-epitope peptide-pulsed DC vaccine has also demonstrated promising results in patients with glioblastoma multiforme (GBM) [735]. This vaccine consisted of HLA-A1-restricted peptides derived from MAGE1 and AIM-2, and HLA-A2-restricted peptides from TRP-2, gp100, HER2 and IL13Rα2. These peptides were pulsed onto DCs cultured ex vivo, and infused into GBM patients. Expression of MAGE1, absent in melanoma 2 (AIM2), gp100 and HER2 in patients prior to vaccination were each associated with improved survival. Recurrent tumours from 4 HLA-A2 patients had downregulated expression of gp100, HER2 and IL13Rα2, indicating that the vaccine had induced immunoediting of these tumours [735].

4.1.2 Prime-Boost and Primary-Recurrent Vaccination Prime-boost vaccination involves pairing an initial priming vaccine with a secondary booster, during or following the refractory phase of the initial B or T cell activation. Use of a booster vaccine significantly enhances vaccination efficacy, and promotes the establishment of immunological memory [310]. Prime-boost vaccines can be either homologous, with both the prime and boost consisting of the same vaccine, or heterologous, pairing different vaccine vectors carrying the same target antigen. Homologous prime-boost vaccination can be particularly good at inducing a potent humoral immune response, and is traditionally used for the majority of childhood vaccines deployed in developed countries. This form of vaccination is particularly useful when the desired outcome is the production of antibodies that neutralise the vaccine itself, as the vector is usually a homologue of the invading pathogen or its toxic products. However, when the desired effect of the vaccine is the production of a potent cell-mediated immune response, homologous vaccination using traditional vaccine types can fall short. The exception to this is when a vaccine vector remains effective despite the presence of vector neutralising antibodies, a function that has been notably observed with some VLP-based vaccine vectors [739, 740]. Heterologous vaccination can also be used to avoid the issue of pre-existing immunity established by the prime vaccine, and is particularly efficient for establishing robust populations of memory T cells [741]. Synergistic stimulation targeting the same antigen can also select for T cells with higher avidity, enriching for a more specific and reactive population [742, 743]. This form of vaccination has been used against intracellular pathogens that are particularly resistant to

141 traditional vaccines, such as Mycobacterium tuberculosis [744-747], immunodeficiency viruses [748-750] and Plasmodium falciparum [751-753].

The compatibility of prime-boost vaccination with therapeutic cancer vaccines is primarily dependent upon the vaccine vector used. When the vaccine involves the use of a live virus, alternating vectors in a heterologous prime-boost vaccination strategy generates superior anti-tumour responses in comparison to a homologous regimen [754]. An alternative approach is to adoptively transfer virally infected cells, effectively smuggling the virus past the protective anti-vector immunity [755-757]. This approach can also be used to combat an undesirable phenomenon observed with therapeutic vaccination, the emergence of treatment- resistant tumours [758-761]. These recurrent tumours no longer express antigens targeted by the original vaccine, potentially through either selective elimination of susceptible tumour cell populations, or evolutionary development of the tumour. This change in expression profile can sometimes represent a shift in tumour state, resembling an epithelial-to- mesenchymal transition [760, 762-764]. While recurrent tumours may be resistant to the vaccine that induced their development, they can be targeted with a second vaccine specific for their new antigen expression profile. The resulting combination primary-recurrent vaccination schedule uses a vaccine targeting the primary tumour to induce production of recurrent tumours, which can be targeted with a second vaccine. This method has been proven effective in mouse models of prostate cancer, with 50% overall survival with combination vaccination in comparison to no survival amongst mice vaccinated with the primary vaccine alone [760]; However, these experiments required adoptive cell pre-loading for the secondary vaccine due to the induction of anti-vector immunity from the primary vaccine. Substitution with an alternative secondary vector would alleviate the complication of using adoptive cell transfer in this vaccination strategy.

4.1.3 Colorectal Cancer-associated Epitopes Although there have been many immunotherapeutic vaccines designed for treatment of CRC, targeting a multitude of different antigens, the vast majority have predominantly focused on CEA and mucin 1 (MUC1). CEA is a glycoprotein of the immunoglobulin superfamily expressed during oncofetal development, predominantly in cells of the normal colonic mucosa, and is commonly overexpressed in CRC. It has been attributed with multiple roles, including cell adhesion, inhibition of extracellular matrix detachment-associated cell death

142

(anoikis), promoting of entrance into a G0-like cell cycle state, and in processes of cellular transformation [765-769]. The closest potential analogue of human CEA in mice is the biliary glycoprotein Bgp1, which has a similar expression profile and a similarly late onset of expression in the intestine during embryogenesis [770]. The murine analogue of MUC1 is strikingly different from its human counterpart, with significant homology only observed at the O-glycosylation sites, and in the transmembrane and cytoplasmic domains [771]. MUC1 is expressed on the luminal surface of polarised epithelial cells, and is overexpressed across the surface of adenocarcinomas [772]. Due to the lack of homology, studies that assess the efficacy of immunotherapeutic vaccines targeting either CEA or MUC1 tend to involve the use of transgenic mice and CRC cells lines expressing the human form of these proteins [773- 777]. In an attempt to limit the complexity of this study, alternative CRC-associated targets were sought, preferably that shared sequence homology, expression profile, and function between humans and mice.

The first target antigen selected for this study was survivin. Survivin is overexpressed in a range of solid tumours including CRC, and is notably absent in normal differentiated tissues [778]. It inhibits apoptosis through mitochondria-dependent caspases [779], with overexpression in tumour cells attributing a reduction in apoptotic index in vivo [780]. Survivin also plays an important role in angiogenesis, as its expression is increased in endothelial cells during vascular remodelling [781, 782], and the targeted suppression of survivin during angiogenesis can induce capillary involution [783]. Survivin has been heralded as a potential pan-tumour antigen, capable of eliciting targeted anti-tumour immunity against a broad range of tumour types [784]. A phase I clinical trial investigating the potential of survivin-derived peptides as an immunotherapeutic vaccine for CRC concluded that targeting survivin was safe for use in humans [785]. More recently, a peptide pool vaccine consisting of five survivin epitopes restricted for HLA-A1, HLA-A2, HLA-A3, HLA-A24 and HLA-B7 administered with Montanide completed a phase I clinical trial [786]. Montanide is a proprietary water in oil emulsion adjuvant, an improvement over transitional emulsion adjuvants such as IFA [787]. Peptide-specific T cell responses were detected in 63% of vaccinated patients, who possessed a broad range of different solid tumours, including cancers of the breast, colon, kidney, lung, ovary, rectum, skin and testicle. Again, this vaccine was well tolerated amongst recipients, with the best clinical outcomes reported as stable disease in 28% of patients. Importantly, survivin has been previously studied in mice in the context of H-2Kb-restricted CTL epitopes, and immunotherapeutic vaccination

143 against CRC. The murine survivin epitopes Msur57 (CFFCFKEL), Msur82 (SGCAFLSV), and Msur97 (TVSEFLKL) were screened as candidate peptides by Hofmann et al [788]. Msur82 was eliminated early in the study due to an inability to bind stably to H-2Kb. Both Msur57 and Msur97 were capable to inducing target specific lysis in in vitro restimulation microcytotoxicity assays, and also delayed the growth of B78D14 murine melanoma when the peptides were administered loaded on DCs [788]. Of these two survivin epitopes, Msur97 was selected for use in the chimaeric VLP constructs developed in this study due to a marginal advantage in binding affinity for H-2Kb, and a desire to avoid potential issues with repeats that might be encountered during cloning of the Msur57 sequence.

The second target antigen selected for this study was another potential pan-tumour antigen, topoisomerase II alpha (topIIα). TopIIα is expressed in proliferating tissues [789, 790], and is overexpressed in a range of solid tumours including CRC [791], prostate cancer [792] and glioma [793], as well as lymphoma [794]. Topoisomerases are involved in regulating the high order structural state of DNA, altering superhelices and disentangling interlinked chromosomes [795]. TopIIα is involved in chromosome segregation and DNA replication [796, 797]; hence, its predominant association with proliferating cells. Although topIIα has previously been used as a prognostic marker in prostate and liver cancer [798, 799], and as a target of chemotherapeutic agents [800], immunotherapies targeting this potential antigen are scarce. Fortunately, candidate H-2Kb-restricted epitopes derived from topIIα have already been identified and studied in mice by Park et al [801]. The topIIα peptides p1327 (DSDEDFSGL), p1339 (DEDEDFLPL), p1470 (SDSDFERAI), p1490 (EEQDFPVDL) and p1509 (RARKPIKYI) were identified as potential candidate epitopes using two online prediction tools. Amongst these peptides, Park et al found only p1327 was capable of inducing IFN-γ production when pulsed on splenocytes from mice previously vaccinated with topIIα RNA-loaded DCs [801]. When loaded onto DCs, p1327 was also capable of inducing topIIα-specific lysis of target cells, and delayed the growth of MC-38 tumours. In accordance with these results, p1327 was selected as the H-2Kb-restricted topIIα epitope used in the chimaeric VLP constructs developed for this study.

4.1.4 Spacer and Linker Sequences used in Chimaeric RHDV VLP RHDV VLP has previously been engineered to contain various model epitopes at the N- terminus of the VP60 capsid protein, including the ovalbumin antigen epitope SIINFEKL

144

(SIIN.VP60) [633], and the gp33 antigen epitope SAVYNFATM (gp33.VP60) [632]. These VLP included a glycine-glycine-serine (GGS) spacer sequence between the recombinant peptide and the N-terminus of VP60, as glycine and serine residues provide flexibility [802- 804]. In these VLP constructs, this GGS sequence is predicted to limit the interference of recombinantly inserted peptides with folding of the VP60 protein, or with interactions involved in the formation of stable VLP. More recently, chimaeric RHDV VLP have also been developed containing the human melanoma-associated gp100 antigen epitope KVPRNQDWL singularly, in multiple copies, and in multiple copies interspaced by linker sequences [652]. The largest of these constructs, a triplicate copy human gp100 VLP interspaced with linkers sequences, encompassed a 36 amino acid insertion at the N-terminus of VP60. This construct was only 6 amino acids shorter than the longest reported stable insertion into VP60 at 42 amino acids [805]. Other than the aforementioned GGS spacer linker, further linkers used in this construct include an alanine-leucine-leucine (ALL) sequence inserted between each individual epitope. This ALL linker is an immunoprotease cleavage sequence [806], and its incorporation into the multi-epitope construct was predicted to facilitate optimal processing of epitopes amongst a selection of candidate linker sequences.

145

4.2 Objectives The purpose of this chapter is to adapt the RHDV VLP vaccine vector for targeting murine CRC, providing justification for humanisation of the resulting vaccine construct. Specifically, this will be attempted through the recombinant insertion of murine H-2kb epitopes derived from CRC TAAs. This chapter will explore the hypothesis that the incorporation of multiple epitopes into a single VLP construct will enhance anti-tumour immune responses. This enhancement may be immediately apparent following therapeutic vaccination in tumour-bearing mice, or it may be represented by the ability to prevent the emergence of recurrent tumours. The resulting candidate VLP vaccine will then be explored in a combination vaccine therapy with the oncolytic virus, vesicular stomatitis virus (VSV), with the objective of exploring this combination to facilitate the development of a primary- recurrent vaccination strategy.

4.2.1 Aims 1. Development of chimaeric RHDV VLP containing single and multiple murine H-2kb restricted epitopes derived from CRC TAAs. 2. Investigation of the efficacy of these chimaeric VLP to induce epitope-specific immune responses, and anti-tumour immunity in a murine model of CRC. 3. Exploration of the potential of primary-recurrent vaccination using a combination of chimaeric RHDV VLP and VSV.

146

4.3 Results

4.3.1 Chimaeric RHDV VLP containing Murine CRC Epitopes The CRC epitope containing RHDV VLP constructs used in this study were designed to build upon the foundation of previously developed VLP. To that effect, the N-terminus of VP60 was engineered to incorporate a single copy of either the topIIα p1327 peptide (T.VP60) or the survivin Msur97 peptide (S.VP60), or a combination of both peptides in series separated by an ALL linker (TS.VP60) (Figure 4.1). Each of these constructs also included the GGS spacer linker separating the recombinantly inserted peptide from VP60. The additional linker used in the multi-epitope TS.VP60 construct was selected by repeating the immunoprotease cleavage analyses previously employed during development of the gp100 VLP construct [652]. This analysis was performed using the three online prediction tools, the Immune Epitope Database (IEDB) Analysis Resource Consensus tool [807], the Proteasome Cleavage Prediction Server (PCPS) [808] and NetChop [809]. The assessed list of linker sequences included the GGS spacer sequence and the ALL cleavage sequence, in addition to two alternative predicted immunoprotease cleavage sequences serine-serine-leucine (SSL) [806] and alanine-alanine-tyrosine (AAY) [806, 810]. Each prediction tool utilises its own proprietary algorithms and database for prediction of cleavage sites, resulting in differing rankings for each sequence between programs (Table 4.1 and 4.2).

Figure 4.1 CRC Epitope RHDV VP60 Constructs Chimaeric RHDV VP60 constructs developed containing murine CRC-associated TAAs included (A) T.VP60, (B) S.VP60 and (C) TS.VP60. Each construct included the spacer linker GGS between the inserted sequence and the N-terminus of VP60. TS.VP60 also included an immunoproteasome cleavage linker ALL between its two TAA epitopes.

The results from these prediction tools indicated that the incorporation of a linker was unnecessary for the processing of the N-terminal topIIα epitope from the TS.VP60 construct, but was advantageous for the processing of the survivin epitope. The ALL linker was the 147 optimal linker based on these predictions, and was selected for use in the construct. It is also noteworthy that the incorporation of a GGS spacer linker between the epitopes significantly impaired processing of the survivin epitope in 2 out of 3 prediction tools, corroborating published observations that incorporation of inappropriate linkers may be detrimental to epitope processing [811, 812]. This illustrates the importance of cross-referencing between multiple processing prediction tools, as each utilises its own set of algorithms to provide unique analysis.

Table 4.1 Linker Comparison for TopIIα Processing

Ranked Score: A rank of 1 indicates optimal predicted linker

Table 4.2 Linker Comparison for Survivin Processing

Ranked Score: A rank of 1 indicates optimal predicted linker, Nil indicates predicted incompatibility

148

4.3.2 Production of Chimaeric RHDV VLP pUC57 DNA plasmids containing sequences encoding N-term-T.VP60, -S.VP60 and - TS.VP60 were obtained for ligation with the remainder of the VP60 gene. Each sequence was manually codon optimised for expression in Spodoptera frugiperda (Sf21) cells by silent mutation of each codon, using the Kazusa Codon Usage Database as a reference [813]. N- term insertion sequences were excised from their pUC57 vector plasmids by digestion with BglII and PstI (Figure 4.2 A and B). Upon gel purification, each fragment was ligated with the remainder of the VP60 gene excised from a pAcpol-.VP60 plasmid by digestion with PstI and EcoRI (Figure 4.2 C), and a pAcUW51(GUS) plasmid linearised by digestion with BglII and EcoRI (Figure 4.2 D). Successful formation of the pAcUW51(GUS).T.VP60, pAcUW51(GUS).S.VP60 and pAcUW51(GUS).TS.VP60 was confirmed through identification of each VP60 sequence excised by digestion with BglII and EcoRI (Figure 4.2 E). The predicted difference in fragment size between each construct can be difficult to interpret on an agarose gel when the comparative fragments are greater than 1 kb; however, the construct fragment for TS.VP60 appears to migrate slightly above both T.VP60 and S.VP60 (Figure 4.2 E).

Upon confirmation of construct integrity by sequencing, each plasmid was co-transfected into Sf21 cells using the FlashBAC ULTRA transfection system. FlashBAC ULTRA negates the requirement to separate recombinant and parent baculovirus, as the parent strain is non- viable [655]. Viable baculovirus for each construct was picked from plaque assays, and bulked up through repeated culture in Sf21 cells until at least their third passage. Baculovirus inoculums were harvested directly from the growth media, with Sf21 separated through pelleting by centrifugation. Expression of each VP60 construct was confirmed by SDS- PAGE in a small culture volume prior to upscaling to batch VLP production. This production involves infection of Sf21 cultures with the recombinant baculovirus over a period of 3 days, followed by lysis with the detergent Triton-X to liberate intracellular VLP, and inactivate the recombinant baculovirus. VLP were then separated from cellular debris, before being concentrated and purified by differential ultracentrifugation. As RHDV VLP have a density of 1.34 g cm-3 [645], they band at the interface between two caesium chloride layers at 1.2 and 1.4 g/cm3. VLP harvested from this band can be stored short-term at 4 °C in CsCl, or can be dialysed or diluted into 50% glycerol for long-term storage at -20 °C. Yields of T.VP60 and S.VP60 from a 400 mL Sf21 culture varied between 6 and 12 mg total protein as

149 determined by spectrophotometry, while TS.VP60 had slightly lower yields averaging around 6 mg total protein.

Figure 4.2 Production of T.VP60, S.VP60 and TS.VP60 Expression Plasmids DNA electrophoresis on 2% agarose gels stained with ethidium bromide. (A) Isolation of the N-term fragment from 1) pUC57.T.VP60 + BglII + PstI and 2) pUC57.S.VP60 + BglII + PstI; (B) Isolation of the N-term fragment from 1) pUC57.TS.VP60 + BglII + PstI; (C) Isolation of the remainder of the VP60 gene from 1) pAcpol-.VP60-SL + PstI + EcoRI; (D) Linearisation of the delivery vector, 1) pAcUW51(GUS), 2) pAcUW51(GUS) + BglII + EcoRI. Constructs produced through ligation were checked prior to co-transfection with the baculovirus expression system FlashBac Ultra (E) 1) pAcUW51(GUS).T.VP60 + BglII + EcoRI, 2) pAcUW51(GUS).S.VP60 + BglII + EcoRI, 3) pAcUW51(GUS).TS.VP60, based on the production of a DNA fragment corresponding to the length of the VP60 gene with the additional epitope sequence insertion. The arrows indicate the desired band position for each digest. 1 kb plus DNA Ladder is provided as a reference.

Samples of T.VP60, S.VP60 and TS.VP60 were separated by SDS-PAGE, transferred onto nitrocellulose and probed for by western blot with anti-RHDV VP60 (Figure 4.3 A). The assembly of these constructs into intact VLP was confirmed by electron microscopy (Figure 4.3 B-D). Each chimaeric VLP produced possessed structural similarities comparable to

150 native RHDV virions [639], including an icosahedral shape with indentations around the circumference often referred to as cog-like in appearance.

Figure 4.3 Production and Purification of T.VP60, S.VP60 and TS.VP60 VLP (A) Western blot of CsCl purified chimaeric VLP following separation by SDS-PAGE and transfer onto nitrocellulose film. Carbon-coated grids were loaded and negatively stained with phosphotungstic acid for visualisation by transmission electron microscopy of (B) T.VP60, (C) S.VP60 and (D) TS.VP60. NEB broad protein ladder was used as a marker. The scale bar represents 200 nm.

Although the size of the VLP was consistent between each construct at between 36-40 nm in diameter, TS.VP60 VLP did appear to possess a visible increase in intraparticulate density (Figure 4.3 D). This may be attributable to the length of the recombinantly inserted peptide sequence in this construct, as a similar appearance has been observed in preparations of chimaeric RHDV VLP with insertions of similar lengths. Chimaeric VLP identify was also confirmed by mass spectrometry, with the 60 kDa band excised for each VLP construct from SDS-PAGE gels, digested with trypsin and analysed by MALDI-TOF. Trypsin can cleave peptides at the site of a lysine (K) or arginine (R), in the absence of a subsequent proline (P). Mass spectrometric analysis confirmed the presence of N-terminal insertions in each construct, with a predicted peptide fragment identified corresponding with 151

MDSDEDFSGLGGSEGK for T.VP60, TVSEFLKLGGSEGK for S.VP60, and MDSDEDFSGLALLTVSEFLK for TS.VP60 (Figure 4.4).

Figure 4.4 Mass Spectrometric Analysis of T.VP60, S.VP60 and TS.VP60 SDS-PAGE gel extracted, trypsinised samples were analysed by MALDI-TOF mass spectrometry to confirm the presence of an N-terminal insertion in CsCl purified samples of (A) T.VP60, (B) S.VP60 and (C) TS.VP60. Emboldened and underlined sequences represent individual peptide sequences detected by mass spectrometry.

4.3.3 Simultaneous N- and C-terminus Insertion into RHDV VP60 In addition to supporting recombinant insertion of epitopes at its N-terminus RHDV VP60 can also tolerate insertion at alternative sites such as the C-terminus. Originally intended as a potential site for insertion of B cell epitopes, the C-terminus of VP60 has previously been modified to incorporate a model six-residue epitope tag derived from bluetongue virus capsid protein VP7 (Btag), and detectable with the mAb 20D11 [814]. The Btag was successfully inserted at the C-terminus of VP60, producing stable VLP, and could be detected with the 20D11 mAb by western blot. Recently, our laboratory experimented with insertion of an H- 2kb-restrict epitope from murine topIIα at the C-terminus of VP60 (VP60.T). Based on the successful formation of this VLP containing an MHC-I epitope at its C-terminus, we decided to briefly explore whether it was possible use simultaneous insertion at both the N- and C- terminus to increase the number of epitopes that could be incorporated into a single chimaeric RHDV VLP. The VP60.T construct was used as a base scaffold, excised from its plasmid by digestion with PstI and EcoRI (Figure 4.5 A). This C-terminus VP60 fraction was combined with the N-terminus of the S.VP60 and TS.VP60, which were liberated from each of their own plasmid constructs by digestion with BglII and PstI, and ligated into pAcUW51(GUS) linearised with BglII and EcoRI. Each resulting plasmid construct was screened for inserts by digestion with BglII and EcoRI (Figure 4.5 B). Both combinations produced detectable inserts, represented as two new RHDV VLP constructs, one with a single epitope at each the

152

N- and C- terminus (S.VP60.T) (Figure 4.5 C), and one with both epitopes at the N-terminus and a second topIIα epitope at the C-terminus (TS.VP60.T) (Figure 4.5 D).

Figure 4.5 Production of S.VP60.T and TS.VP60.T Expression Plasmids DNA electrophoresis on 2% agarose gels stained with ethidium bromide. (A) Excision of the C-terminus from VP60.T in a pAcUW51(pol-) plasmid by digestion with PstI and EcoRI. (B) Screening of BglII-EcoRI insert size in 1) pAcUW51(GUS).S.VP60, 2) pAcUW51(GUS).TS.VP60, 3) pAcUW51(GUS).S.VP60.T and 4) pAcUW51(GUS).TS.VP60.T. A diagrammatic representation is provided illustrating the structure of (C) S.VP60.T and (D) TS.VP60.T. The arrows indicate the desired band position for each digest. 1 kb plus DNA Ladder was used as a marker.

The pAcUW51(GUS).S.VP60.T and pAcUW51(GUS).TS.VP60.T plasmids were co- transfected into Sf21 cells along with the FlashBac Ultra vector as previously described. The resulting baculovirus was bulked up to generate a concentrated inoculum prior to inoculation of Sf21 cell cultures, and production of VLP. S.VP60.T and TS.VP60.T expression was confirmed through separation by SDS-PAGE, transfer onto nitrocellulose film, and probing with western blot with anti-RHDV VP60 mAb (Figure 4.6 A). The S.VP60.T band was notably higher than the S.VP60 control band, while TS.VP60.T was higher than both S.VP60 and S.VP60.T. Intact VLP were also visualised by electron microscopy in preparations of S.VP60.T and TS.VP60.T (Figure 4.6 B & C), with each construct forming VLP with the

153 characteristic cog-like morphology and size. Unfortunately, the sequence structure of VP60 is such that these constructs could not be confirmed by mass spectrometry. Although the size shift identified between these chimaeric VLP predicted that these constructs were likely to be of correct identity, the inability to provide confirmation halted development of chimaeric VLP with simultaneous N- and C-terminus insertion. Additionally, due to the inward tertiary folding at the C-terminus of VP60, recombinant insertion at this site may have additional restrictions. Insertions made at this site may be particularly limited in size, and may be more significantly influenced by specific amino acid sequences that interact with or disrupt the formation of other motifs within the VP60 protein structure.

Figure 4.6 Production and Purification of S.VP60.T and TS.VP60.T VLP (A) Western blot of 1) S.VP60, 2) S.VP60.T and 3) TS.VP60.T. Electron microscopy and negative staining was used to visualise (B) S.VP60.T and (C) TS.VP60.T VLP. NEB broad protein ladder was used as a marker. The scale bar represents 200 nm.

154

4.3.4 Vaccination with chimaeric RHDV VLP The ability to stimulate target epitope-specific immune responses with T.VP60, S.VP60 and TS.VP60 was assessed using an in vivo cytotoxicity assay. This method has been used previously for assessing new chimaeric VLP constructs, and is effective for the determination of epitope-specific immunogenicity. In order to evaluate the ability to generate simultaneous immune responses against multiple targets, the standard method was modified to accommodate multiple target cell populations by adding the dye VPD to combinations of CFSE at different concentrations. Vaccines were administered subcutaneously on days 0 and 21, consisting of 100 μg of VLP or an equimolar amount of peptides, combined with 25 μg of CpG adjuvant. Peptide-pulsed, fluorescently-labelled target cells were administered intravenously on day 28. Spleens were harvested 48 hours later, and splenocyte preparations were analysed by flow cytometry (Figure 4.7 A). Gating was used to exclude cell doublets by comparing side scatter height and area, and to exclude dead cells with near-IR Live/Dead dye (Figure 4.7 B).

Each peptide pulsed target splenocyte population was then assessed for % specific lysis relative to an unpulsed population, compensating for background non-specific lysis. Both the T.VP60 and TS.VP60 vaccinated groups mounted an epitope-specific immune response against target cells laced with the topIIα p1327 peptide, while S.VP60 vaccinated mice did not (Figure 4.7 C). Similarly, S.VP60 and TS.VP60 vaccinated groups mounted an epitope- specific immune response against target cells laced with the survivin Msur97 peptide, while T.VP60 vaccinated mice did not (Figure 4.7 D). T.VP60, S.VP60 and TS.VP60 each mounted epitope-specific immune responses against target cells laced with equimolar amounts of both the topIIα p1327 peptide and the survivin Msur97 peptide (Figure 4.7 E). The combination of these results indicates that TS.VP60 successfully vaccinated mice against both epitopes, without apparent detriment. This may provide an indication that any potential immunodominance that exists between these epitopes does not impair the ability to induce simultaneous epitope-specific responses when delivered in equimolar quantities.

155

Figure 4.7 In vivo Cytotoxicity Assay In vivo cytotoxicity assays were used to assess the ability of each chimaeric VLP to induce epitope-specific immune responses. (A) Diagrammatic representation of the modified in vivo cytotoxicity assay used, explained in Section 2.3.2. (B) Flow cytometry gating strategy, excluding doublets [Gate A] and dead cells [Gate B]. % Specific lysis was determined by comparing the proportion of each peptide pulsed target cell population, topIIα pulsed [Gate D] and survivin pulsed [Gate G] to an unpulsed population [Gate C]. Percent (%) specific lysis was determined relative to (C) topIIα, (D) survivin, and (E) topIIα/survivin peptide pulsed populations. Statistical analysis was performed using one-way ANOVA, with multiple comparisons and Dunnett correction for multiple comparisons. *** p < 0.001, **** p < 0.0001. These results are representative of two independent repeats.

156

Vaccination with a combination of topIIα p1327 and survivin Msur97 synthetic peptide did not induce significant cytotoxicity against either target, demonstrating the superior cytotoxicity induced by these epitopes when recombinantly inserted into RHDV VLP (Figure 4.7 C-E). Although the difference between the specific lysis between the peptide vaccinated and control VP60 vaccinated mice was not significant, this population did display and interesting trend. Across each target population, the percent specific lysis was tightly grouped marginally above the control population. This may be indicative of an immune response induced under impairment, perhaps in uptake, presentation or priming.

4.3.5 Vaccination with Chimaeric RHDV VLP using a Multi-dosage Regimen Although there is significant variability in the infection kinetics amongst different species and strains of viruses, RNA viruses tend to reach a peak post-infection titre at around 72 hrs post infection [815-817]. As a potent cytotoxic immune response is preferential for tumour clearance, it logically follows that mimicking an active viral infection may enhance cytotoxic immune responses and improve tumour vaccine efficacy. A multi-dosage regimen with daily injections over a 3 day period simulates the natural ability of an RNA virus to replicate and accumulate at the infection site. Vaccination efficacy induced by different dosage regimens of S.VP60 and SIIN.VP60 were compared to VP60 VLP using a modified version of the in vitro VITAL assay, a cytotoxicity assay developed by Hermans et al [658]. The original VITAL assay involved isolation of CD8+ T cells from pre-vaccinated mice, and in vitro co- culture with peptide pulsed Jurkat/A2Kb cells labelled with either CFSE or CMTMR. The modified form used in this study included replacing Jurkat/A2Kb cells with whole splenocytes due to the unavailability of the cell line, and substituted CMTMR dye for VPD. An in vitro cytotoxicity assay holds the advantage of requiring only a week post vaccination before assessing CTL efficacy against target cells. This provides an efficient platform for screening multiple VLP constructs, while providing justification for an in vivo cytotoxicity assay, or direct transition into a tumour trial. The primary flaw with this shortened timespan is a marked reduction in cytotoxic activity in the absence of boosting, especially when vaccinating against self-antigens.

Mice were vaccinated subcutaneously with 100 μg of VLP in combination with 25 μg of CpG adjuvant on day 0 (1D), days 0 and 1 (2D) or days 0, 1 and 2 (3D), and culled at day 7. CD8+

157

T cells were isolated from the draining lymph node of each mouse, and co-cultured with peptide pulsed, fluorescently labelled splenocytes for 4 hours prior to analysis by flow cytometry (Figure 4.8 A). Gating was used to exclude cell doublets by comparing side scatter height and area, and to exclude dead cells unable to process near-IR Live/Dead dye (Figure 4.8 B). Each peptide pulsed target splenocyte population was then assessed for % specific lysis relative to an unpulsed population, compensating for background non-specific lysis. VP60 was used as a negative control, and analysis of its inability to induce cytotoxicity against SIINFEKL is provided as an example (Figure 4.8 C). Cytotoxicity induced by the model VLP SIIN.VP60 was significantly enhanced with both 2D and 3D vaccination over 1D (Figure 4.8 D). The effect on the cytotoxicity induced by S.VP60 was less striking, but did result in a statistically significant enhancement with 3D vaccination over 1D (Figure 4.8 E). These results indicate that while multiple doses appear to enhance the potency of VLP vaccination, the effect is seemingly impaired against tolerised antigens. Theoretically, this may represent the comparative responses generated in a primary naïve T cell population (SIINFEKL), and a secondary chronically tolerised T cell population (survivin). If this is the case, then it is intriguing that the chimaeric VLP can stimulate effector cytotoxicity in this tolerised population at all.

158

Figure 4.8 Multi-dosage Regimens for Enhancing Epitope-specific Cytotoxicity In vitro cytotoxicity assays were used to assess the immunogenicity of multi-dosage regimens with chimaeric VLP. (A) Diagrammatic representation of the modified VITAL assay, explained in section 2.3.3. (B) Flow cytometry gating strategy, excluding doublets [Gate A] and dead cells [Gate B]. % Specific lysis was determined by comparing the proportion of each peptide pulsed target cell population, survivin pulsed [Gate D], and SIINFEKL pulsed [Gate F] to an unpulsed population [Gate C]. (C) Percent specific lysis relative to the survivin pulsed group for mice vaccinated with VP60, calculated for each dosage regimen including vaccination on day 0 (1D), days 0 and 1 (2D) and days 0, 1 and 2 (3D). (D) Cytotoxicity relative to SIINFEKL pulsed group for mice vaccinated with SIIN.VP60. (E) Cytotoxicity relative to survivin pulsed group for mice vaccinated with S.VP60. Statistical analysis was performed using an unpaired t-test comparing between dosage groups. NS = non-significant, * p < 0.05, **** p < 0.0001. Results are combined from three independent repeats.

159

4.3.6 Kinetics of MC38-OVA Tumours in B57BL/6 Mice MC38 is a murine colon adenocarcinoma cell line that was induced through injection of dimethyl-hydrazine into the colonic lining of C57BL/6 mice [818]. As a model of CRC, MC38 is commonly engrafted subcutaneously into CD57BL/6 mice at amounts ranging from 1x105-1x106 cells per mouse [819-821]. The growth rate of MC38 tumours is comparatively slow relative to other commonly used tumour cell lines, reaching 1 cm3 in an average of 31.3 days [819]; however, the growth rate can be notably variable, reaching 1 cm3 anywhere between around 24.4 and 44.1 days. The use of a tumour model with a longer growth window is advantageous in studies of immunotherapeutic vaccines, as it provides sufficient time for the induction of immune responses to the tumour cells, the infiltration of the tumour with immune cell populations, and the subsequent exhaustion and establishment of an immunosuppressive tumour microenvironment. Additionally, an extended growth window provides sufficient time to investigate longer vaccination schedules, such as a prime-boost regimen. The MC38-OVA used in this study was produced by infection of the MC38 parent line with a retrovirus based on the pMIG/MSCV-IRES-eGFP plasmid, engineered to express membrane-bound ovalbumin [822]. MC38-OVA is notably more immunogenic than its parent cell line. This may attributable to the natural potency of ovalbumin as a model antigen, the high expression of ovalbumin from incorporation of multiple gene copies, or the potential release of ovalbumin into the tumour microenvironment.

During initial experiments with MC38-OVA, dosages corresponding to protocols established for the parent cell line proved incapable of guaranteeing subcutaneous engraftment (Figure 4.9 A). This issue was seemingly resolved by increasing the dosage of MC38-OVA cells injected, and may have been influenced by the length of time the cells spent in culture post- thaw. MC38-OVA tumours engrafted into control C57BL/6 mice were palpable by day 7-10 post injection, and usually grew slowly over the first 2-3 weeks, before expanding rapidly over the remaining 1-2 weeks. Striking variability in growth kinetics matched reports of the parent strain [819], on average taking 27.5 days to reach the ethically approved maximum of 150 mm2 in area, ranging between 19-35 days (Figure 4.9 B and C). It is important to note that the previously mentioned ethical limit of 1 cm3 used by Pajtasz-Piasecka et al in their study of MC38 is smaller than the ethical limit of 150 mm2 used in these MC38-OVA studies. This means that the aforementioned kinetics for MC38 were for a smaller tumour, yet with a longer growth window, further identifying the differences in the growth kinetics of the parent cell line and its derivative.

160

Figure 4.9 MC38-OVA Engraftment and Growth Rate Kinetics Although subcutaneously engrafted MC38-OVA tumours share the spread of variation in growth kinetics of its parent strain, there are differences that can distinguish between them. (A) Engraftment rates of MC38-OVA tumours with different numbers of subcutaneously injected cells. (B) The average tumour sizes from individual mice in negative control populations of one VLP vaccination trial. (C) Kaplan-Meier curve plotting the survival of control mice. Groups X and Y represent tumours with seemingly different growth kinetics.

Counterintuitively, these results suggest that MC38-OVA tumours grow faster than tumours established from MC38 cells. It is possible that prolonged passage in combination with retroviral infection has selected for a more robust or aggressive population of cells, while the predicted increase in immunogenicity contributed by the expression of ovalbumin manifests as increased rejection early during tumour establishment. Engrafted MC38-OVA tumours also appear to follow patterns of growth kinetics that can be tentatively clustered into two groups, X and Y (Figure 4.9 B and C). These groups are particularly distinguishable based on monitoring tumour growth, with group X tumours appearing to grow faster than group Y tumours. Group X tumours reached 150 mm2 within 26 days, while group Y tumours required between 31-35 days to grow to this size. The separation of these tumour groups may be due

161 to the presence of subpopulations within the MC38-OVA cell line, possibly from differing numbers of proviral transduction events between each cell.

As the MC38-OVA cell line demonstrated differences in growth kinetics in comparison to its parent strain, it became necessary to confirm that the modified cell line had retained the expression of the selected CRC-associated antigens. Cultures of MC38 and MC38-OVA cells were lysed, and the resulting lysate was separated by SDS-PAGE, transferred onto nitrocellulose membrane, and probed with anti-topIIα or anti-survivin by western blot. MC38-OVA was found to express both topIIα and survivin, though possibly at slightly reduced levels relative to the parent strain (Figure 4.10 A and B). Expression of ovalbumin by MC38-OVA was also confirmed by co-culture of live MC38-OVA cells with splenocytes isolated from OT-I mice. Co-culture of MC38-OVA cells with OT-I splenocytes induced production of IFN-γ to levels comparable to co-culture with a SIINFEKL peptide pulsed C57BL/6 splenocyte population, as determined by murine IFN-γ ELISA (Figure 4.10 C).

162

Figure 4.10 Confirmation TopIIα, Survivin and Ovalbumin Expression in MC38-OVA The MC38-OVA cell line was screened for expression of (A) topIIα and (B) survivin by western blot. Expression was compared to the MC38 parent strain, with actin as a control. (C) Expression of OVA was confirmed based on secretion of IFN-γ at 24 and 48 hours following co-culture of MC38-OVA cells with splenocytes isolated from OTI mice (OTI), using naive C57BL/6 splenocytes as a negative control (Spl). NEB broad protein ladder was used as a marker.

4.3.7 Chimaeric RHDV VLP as an Immunotherapy for CRC Induction of an epitope-specific cytotoxic immune response in an in vivo cytotoxicity assay is indicative that each chimaeric VLP was capable of successfully vaccinating against their target epitope; however, the efficacy of these responses may be impaired when combined with the predominantly immunosuppressive microenvironment of an established tumour. A

163 murine model of CRC was used to investigate the ability of these vaccines to induce a anti- tumour immune responses. MC38-OVA tumours were engrafted subcutaneously on the left flank of C57BL/6 mice. On day 7, the mice were sorted according to tumour size and distributed evenly between each treatment group, ensuring that the average tumour size and distribution was consistent between each group. VLP vaccines were administered on days 7, 8 and 9 using the 3D protocol, injected subcutaneously proximal to but away from the tumour site. Vaccination was specifically administered at this site to facilitate drainage of VLP to the lymph node draining the tumour site, which may contain T cell populations primed for anti-tumour immunity. Tumour growth was monitored by measuring the cross-sectional area of each tumour, and mice were culled once these tumours reached the ethically approved maximum of 150 mm2 (Figure 4.11 A).

Vaccination with either T.VP60 or S.VP60 delayed tumour growth, with 60% overall survival (Figure 4.11 B and C). These vaccines matched the efficacy of the SIIN.VP60 vaccine, with complete remission of tumours in mice that responded to these vaccines. Vaccination with TS.VP60 further delayed tumour growth, inducing regression that came close to complete remission across all mice at around days 19-21 (Figure 4.11 B and C). Tumours in 30% of the mice in this group continued to grow from around day 21 onward following this prolonged regression period. Interestingly, tumour escape was also observed in an additional mouse in this group at day 68, over two months after tumour cell engraftment (Figure 4.12 F). This mouse last had a palpable tumour on day 16, suggesting that its tumour must have regressed down to a small resilient population, possibly under prolonged immunoediting. The emergence of this tumour illustrates the importance of a prolonged monitoring period following complete tumour remission in animal tumour models, as recurrent tumours can develop even after an extended period.

164

Figure 4.11 Efficacy of VLP constructs against engrafted MC38-OVA tumours Subcutaneously engrafted MC38-OVA tumours were used as a model for investigating efficacy of candidate vaccines. (A) Diagrammatic representation of tumour trials, with injection of 1x106 MC38-OVA cells on the left flank of each mouse, followed by vaccination with 100 μg of VLP and 25 μg CpG adjuvant on days 7, 8 and 9. Mice were monitored for a period of 100 days, with survivors rechallenged with a second subcutaneous dose of 1x106 MC38-OVA cells in the opposing flank. (B) Tumour size was monitored based on cross-sectional area, and mice were culled upon reaching the ethically approved maximum of 150 mm2. (C) Survival was plotted on a Kaplan-Meier curve. Statistical analysis was performed using the Log-rank (Mantel-Cox) test, ** p < 0.01, *** p < 0.001, NS = Non-significant. These results are representative of two independent repeats.

165

This was the only significantly delayed tumour recurrence observed amongst the treatment groups, with all other mice either succumbing to their tumour, or achieving complete remission (Figure 4.12 A-F). The TS.VP60 group achieved an overall survival of 70%, which was not a statistically significant benefit over either T.VP60 or S.VP60; however, the trend of both improvement in overall survival and further delayed tumour growth was replicated in a repeat tumour trial.

Figure 4.12 Individual tumour growth curves Growth curves were plotted for individual mice within each group, (A) CPBS, (B) VP60, (C) SIIN.VP60, (D) T.VP60, (E) S.VP60, and (F) TS.VP60. Mice were culled after their tumour reached the ethically approved maximum of 150 mm2.

166

4.3.8 Rechallenge of Tumour-free Mice Following a period of 100 days of monitoring post tumour engraftment, mice in complete remission were rechallenged with a second dose of MC38-OVA cells. This rechallenge was injected subcutaneously into the mice, on the opposite flank to their original tumour site (Figure 4.11 A). A control group of naïve C57BL/6 mice were also engrafted to ensure the viability of the injected cells. The control group developed tumours that grew unhindered, reaching the ethically approved maximum size of 150 mm2 between days 20-37 post engraftment (Figure 4.13 A and B). None of the mice previously cured of their primary MC38-OVA tumour following vaccination with SIIN.VP60, T.VP60, S.VP60 or TS.VP60 were susceptible to rechallenge, with no palpable tumours identified in any of the groups.

Figure 4.13 Rechallenge with MC38-OVA Rechallenge of mice in complete remission from their primary tumour with a secondary injection of 1x106 MC38-OVA cells. A group of naïve C57BL/6 mice were also challenged as a control. (A) Tumour size was monitored based on cross-sectional area in mm2. (B) Survival was plotted on a Kaplan-Meier curve. Statistical analysis was performed using the Log-rank (Mantel-Cox) test. ** p < 0.01, *** p < 0.001.

167

As the rechallenge was injected on the opposing flank, this represents the establishment of systemic immunity capable of selectively targeting the MC38-OVA cells. While this may be attributable to the immune response directly induced by each vaccine, there may be contributions from additional responses generated during periods of tumour regression. Targeted elimination of tumour cells through induction of apoptosis by cytotoxic immune cells would liberate additional TAAs for APCs to process and present, potentially triggering multiple responses that compound to promote tumour rejection. While the development of these bystander immune responses may potentially occur from any tumour cell line, the expression of ovalbumin by MC38-OVA represents a potent immunogenic target that may exacerbate this effect.

4.3.9 Effect of Pre-existing Immunity to RHDV VLP Pre-existing immunity in the form of vaccine vector neutralising antibodies can impair the ability of some vaccines to induce an immune response against their target antigen. The effect of pre-existing immunity on the functionality of VLP-based vaccines is mixed, with some reports of deleterious effects [646], while others have reported no adverse effect [739, 740]. We have not previously observed any noteworthy interference from pre-existing immunity with our RHDV VLP vaccines, such as when using RHDV VLP in a homologous prime- boost vaccination schedule for in vivo cytotoxicity assays. In order to confirm that establishing pre-existing immunity against RHDV VLP does not impair their ability to induce a cytotoxic immune response against recombinantly inserted epitopes, we set up a prime- boost vaccination schedule in a MC38-OVA tumour trial. MC38-OVA tumours were engrafted subcutaneously, injected into the left flank of C57BL/6 mice. Prime vaccinations were delivered on days 7, 8 and 9 by subcutaneous injection proximal to but separate from the tumour site. Boost vaccinations were delivered on days 21, 22 and 23. Mice were monitored for a period of 50 days, and culled when their tumours reached the ethically approved maximum of 150 mm2 (Figure 4.14 A). Delayed vaccination with SIIN.VP60 at days 21, 22 and 23 (2’ SIIN.VP60) noticeably disrupts growth of MC38-OVA tumours (Figure 4.14 B), but does not provide a statistically significant survival advantage over the control group (Figure 4.14 C).

168

Figure 4.14 Prime-boost Vaccination with RHDV VLP Establishment of pre-existing immunity against the RHDV VLP vaccine vector was emulated using a prime- boost vaccination schedule, administered in the context of a MC38-OVA tumour trial. (A) Diagrammatic representation of the tumour trial, with injection of 2x105 MC38-OVA cells on the left flank of each mouse, followed by primary vaccination with 100 μg VLP and 25 μg CpG adjuvant on days 7, 8 and 9, and/or boost vaccination with 100 μg VLP and 25 μg CpG adjuvant on days 21, 22 and 23. Mice were monitored for a period of 50 days. (B) Tumour size was monitored based on cross-sectional area in mm2. (C) Survival was plotted on a Kaplan-Meier curve. Statistical analysis was performed using the Log-rank (Mantel-Cox) test. NS = Non- significant. Results are representative of two independent repeats.

169

Establishment of pre-existing immunity through vaccination with VP60 prior to administration of SIIN.VP60 (1’ VP60, 2’ SIIN.VP60) impairs growth to the same degree as the 2’ SIIN.VP60 group, with matching 40% overall survival. These findings indicate that an RHDV VLP vaccine used as a booster retains its ability to induce an anti-tumour immune response even if the primary vaccine is also based on the RDHV VLP vector. This means that it should be theoretically possible to pair RHDV VLP vaccines containing epitopes derived from primary tumour antigens, and recurrent tumour antigens, in a primary-recurrent vaccination schedule for preventing tumour escape. Unfortunately, this particular trial was undertaken prior to the optimisation of MC38-OVA engraftment and growth kinetics, as identified by the incomplete engraftment of MC38-OVA amongst the CPBS treated group.

4.3.10 Expression of RHDV VLP by Vesicular Stomatitis Virus Oncolytic viruses, such as VSV, preferentially infect and replicate in tumour cells. The combination of this infection preference with a lytic replication cycle allows oncolytic viruses to be used as a therapeutic virotherapy for cancer, homing in on and destroying tumour cells. VSV is a member of the Rhabdoviridae family, with a negative sense genome that encodes for five individual proteins, the surface glycoprotein (G), nucleocapsid (N), polymerase proteins (L and P), and the peripheral matrix protein (M). Together these proteins are arranged to produce the characteristic Rhabdoviridae bullet shape, with the VSV genome encased within a protective shell consisting of nucleocapsid (N) and polymerase proteins (L and P). The matrix (M) protein binds the nucleocapsid-enclosed RNA genome to the surface glycoproteins (G), and is encapsulated within a lipid bilayer (Figure 4.15) [823, 824]. The exposed portions of the surface glycoproteins facilitates infection of target host cells through an undefined receptor [825], and also serves as the primary target for neutralisation of the virus by VSV-specific antibodies [826, 827].

170

Figure 4.15 Structure of Vescicular Stomatitis Virus (VSV) VSV is a bullet shaped member of Rhabdoviridae family, consisting of a negative-sense RNA genome encased within a nucleocapsid (N) and polymerase (L/P) protein shell. This nucleocapsid is linked to the surface glycoproteins (G) through matrix proteins (M), enveloped within a lipid bilayer [823, 824].

VSV can tolerate the insertion of foreign genes at a specific site between the G and L genes, with functional transcription and translation along with the rest of the VSV genome [659]. Recombinant VSV expressing libraries of TAAs have been developed as an effective therapeutic cancer vaccine vector for both melanoma and prostate cancer [760]; however, the use of a live virus makes this form of vaccination susceptible to neutralisation with vector- specific antibodies. The adaptation of VSV to express our RHDV VLP represents a potential combination vaccine strategy, and may facilitate the development of a heterologous primary- recurrent vaccination regimen. The combination of VSV and RHDV VLP should theoretically provide the potency of a live virus vaccine and the tumour homing properties of an oncolytic virus, with both the epitope cross-presentation capabilities and the repeated vaccination potential of RHDV VLP. The first step in formulating this potential vaccination system is evaluating the viability of using an oncolytic virus to express a heterologous VLP. Sequences derived from the RHDV VP60 constructs VP60, SIIN.VP60 and TS.VP60 were adapted to facilitate insertion of these sequences into the primary VSV expression plasmid, pVSV-XN2 (Figure 4.16). Each subsequent plasmid construct was co-transfected along with the pVSV-L, pVSV-N and pVSV-P expression plasmids, into BHK cells pre-infected with the vaccinia virus MVA-T7. The pre-infection with MVA-T7 provides the T7 RNA polymerase required for the initial construction of the VSV genome.

171

Figure 4.16 Adaptation of RHDV VP60 Constructs for the VSV Expression system The RHDV VP60 gene was adapted for insertion into the VSV genome by introducing the flanking endonuclease cleavage sites XhoI and NheI at the N- and C-terminus respectively. Each of the three candidate constructs, (A) VP60, (B) SIIN.VP60 and (C) TS.VP60, were recombinantly inserted between the G and L genes of VSV.

Following plaque purification, the resulting viruses VSV-VP60, VSV-SIIN.VP60 and VSV- TS.VP60 were amplified through replication in BHK cells. Expression of each VP60 construct was confirmed by western blot (Figure 4.17 A). Each recombinant VSV expressed VP60, with sizes corresponding with the recombinant insertion of one or two epitopes at the N-terminus relative to unmodified VP60. Additional bands were also picked up by western blot with anti-RHDV VP60, which may represent a truncated form of VP60, degraded VP60, or cross-reactivity with VSV products. Intact VLP were subsequently isolated with the same purification methods used for production of RHDV VLP in Sf21 insect cells. VLP produced by each recombinant VSV were identified by transmission electron microscopy, with characteristic cog-like morphology indicating that the expressed VP60 protein was capable of forming intact RHDV VLP (Figure 4.17 B-D). The background material in these TEM images consists of an unidentified stringy substance, which may be comprised of remnants of VSV. TS.VP60 VLP expressed by VSV also appear to have a dense core, a feature that was previously noted for TS.VP60 VLP expressed in insect cells.

172

Figure 4.17 Detection of RHDV VP60 VLP Produced by Recombinant VSV (A) Expression of each VP60 construct was confirmed by western blot, comparing 1) VSV-VP60, 2) VSV- TS.VP60 and 3) VSV-SIIN-VP60. Formation of intact VLP was identified by transmission electron microscopy for (B) VSV-VP60, (C) VSV-SIIN.VP60 and (D) VSV-TS.VP60. The scale bar represents 200 nm.

4.3.11 VSV Expressing RHDV VLP as an Immunotherapy for CRC Production of the recombinant VSV constructs VSV-VP60, VSV-SIIN.VP60 and VSV- TS.VP60 produced titres roughly equivalent to that of a control VSV expressing GFP (VSV- GFP). The predominant cytopathic effects (CPE) observed included detachment and rounding of cells, and were clearly identifiable within virally infected plaques and across virally infected cultures (Figure 4.18). The VSV constructs did not demonstrate any distinguishable growth characteristics or induce CPE different from those observed with VSV-GFP. Based on these observations, it was deemed unlikely that the expression of VP60 or its variants in virally infected cells was having an effect on VSV viability, infectivity or replicative capacity in vitro.

173

Figure 4.18 Cytopathic effect of recombinant VSV Cytopathic effects were used as a visual determinant of VSV infection when examining plaque assays. (A) Mock infected BHK cells. (B) BHK cells inoculated with VSV-VP60.

A screening trial was undertaken to assess the efficacy of these candidate constructs against subcutaneous MC38-OVA tumours. Vaccinations were administered using the 3D regimen on days 7, 8 and 9 following tumour engraftment. Vaccination with VSV constructs was performed intravenously, while VLP vaccines were administered subcutaneously with CpG adjuvant (Figure 4.19 A). Neither vaccination with VSV-SIIN.VP60 or VSV-TS.VP60 induced significant effects on tumour growth rate or overall survival, with only SIIN.VP60 vaccination providing any notable effect (Figure 4.19 B and C). These findings suggest that the in vitro efficacy of RHDV VLP expression by VSV may not translate in vivo, or that the quantity of VLP produced is insufficient for inducing anti-tumour immune responses. Further experiments are required to elucidate the specific cause of this discrepancy, representing a significant avenue for further investigation into the potential for a combination vaccine therapy including VSV and RHDV VLP.

174

Figure 4.19 Screening VSV constructs for efficacy against MC38-OVA tumours VSV constructs VSV-VP60, VSV-SIIN.VP60 and VSV-TS.VP60 were screened for efficacy against subcutaneous MC38-OVA tumours. (A) Diagrammatic representation of the tumour trial, with injection of 2x105 MC38-OVA cells on the left flank of each mouse, followed by vaccination with 100 μg VLP and 25 μg CpG adjuvant subcutaneously, or 5x106 PFU of VSV intravenously, on days 7, 8 and 9. Mice were monitored for a period of 50 days. (B) Tumour size was monitored based on cross-sectional area, and mice were culled upon reaching the ethically approved maximum of 150 mm2. (C) Survival was plotted on a Kaplan-Meier curve. Statistical analysis was performed using the Log-rank (Mantel-Cox) test. * p < 0.05, NS = Non-significant.

175

4.4 Discussion

4.4.1 Chimaeric RHDV VLP Vaccine for Colorectal Cancer Chimaeric RHDV VLP have previously been developed to incorporate a variety of immune epitopes, as immunotherapeutic vaccines for LL-LCMV and melanoma models in mice [632, 633, 652]. In this study, new constructs were developed containing epitopes derived from the CRC TAAs topIIα and survivin as an immunotherapeutic vaccine for a CRC model in mice. These constructs proved capable of inducing targeted immune responses against their incorporated epitopes, which subsequently translated into the development of anti-tumour immune responses in murine CRC tumour trials of these vaccines; however, there remain several issues arising from these results that require further discussion. While the immune responses induced were epitope specific, the magnitude of these responses does not appear to translate directly with their in vivo efficacy. In previous work, the magnitude of target- specific immune responses induced by chimaeric VLP as represented by percent specific lysis in in vivo cytotoxicity assays may be close to 100% with the model epitope SIINFEKL [633, 828]. When targeting a cross-species epitope from human gp100, percent specific lysis was still around a respectable 50% in vaccinated mice; however, when an epitope from murine gp100 was used, these responses dropped to around 10% [652]. These values are consistent with those achieved against murine topIIα and survivin in this study, but appear comparatively low in comparison to those achieved against model or xenogeneic epitopes.

The impaired cytotoxicity induced may be the result of potent inherent immunotolerance for these epitopes. In this scenario, the ability to induce responses against these epitopes at all is testament to the capacity of chimaeric VLP vaccines to challenge immunotolerance. What remains unexplained is how these apparently low levels of cytotoxicity translate into anti- tumour immune responses, especially those achieved in this study. One possible explanation is that the ability to break tolerance with chimaeric VLP vaccines may differ between vaccination of naïve mice, and tumour-bearing mice. The engraftment of a tumour in mice will include immunoediting, potentially priming responses against select TAAs. While the incompetence of these responses can be observed through the ability for tumours to establish and grow in naïve mice, when coupled with chimaeric VLP vaccination these responses may be magnified. Vaccination of mice with chimaeric VLP in the absence of this tumour immunoediting process may lack this initial priming, limiting the observed cytotoxicity of the responses induced in in vivo or in vitro cytotoxicity assays. This potential phenomenon

176 may be investigated by repeating these assays in tumour bearing mice, providing a more relevant representation of responses that may be induced in tumour bearing patients.

4.4.2 Targeting Multi-epitopes with a Chimaeric VLP Vaccine In an ideal world, tumour cells would express a core antigen that is immunogenic, expressed in abundance, and critical for tumour cell survival. Such a situation does exist for some malignancies, such as in select cases of chronic myelogenous leukemia (CML) that have arisen from the balanced reciprocal translocation of chromosomes 9 and 22, forming the BCR-ABL1 fusion gene [829, 830]. Similarly, the majority of HER2 mutant breast cancers respond well to therapies targeted to block HER2 [831, 832]. However, in the absence of such a critical central mutation, the complex genetic background of tumour cells may include multiple redundancies, or such redundancies may develop during immunoediting of the tumour. These redundancies may enable tumour cells to survive, evolve, and resist selective targeting of a single antigen. Alternatively, amongst the heterogeneous mass of tumour cells there may dwell a resistant subpopulation capable of regenerating the tumour completely following vaccination.

Through careful epitope selection and design, multi-epitope vaccines can be produced with the ability to elicit simultaneous cytotoxic immune responses, designed to limit the ability for inherent redundancies to facilitate tumour escape. However, multi-epitope vaccines pose additional complications in design and development, ensuring that immune responses are generate simultaneously against each target. An important consideration is the relative abundance of each epitope, and the requirement for epitopes to be processed efficiently. When delivered in equal proportion, epitopes of relatively similar processing compatibility should be presented equally and stimulate simultaneous immune responses, as demonstrated in studies of vaccines consisting of equimolar peptide pool [833]. The use of chimaeric VLP as a multi-epitope vaccine construct holds an advantage with regards to this concept over other design forms, such as those involving chemical conjugation, as the number of epitopes incorporated is consistent. For each VP60 capsid protein amongst the 180 copies that form each RHDV VLP, there is one of each epitope, providing equimolar delivery. Assuming that the processing and binding affinity of each epitope is relatively equal, this provides an ideal delivery platform for a multi-epitope vaccine. This concept is supported by this study, which provides the foundation for the use of chimaeric RHDV VLP as a multi-epitope vaccine for

177 cancer by demonstrating the ability to induce simultaneous immune responses against multiple CRC TAAs.

Although the multi-epitope chimaeric VLP TS.VP60 appeared to induce some benefit over its mono-epitope counterparts T.VP60 and S.VP60 with regards to delaying tumour growth, and improving survival, these benefits were not found to be statistically significant. This outcome was an unexpected scenario, and resulted from the efficacy achieved with these mono-target therapies. Although it is possible that targeting two epitopes simultaneously is indeed not enough to counteract the redundancies present within a heterogeneous tumour, it is also possible that the model used was inappropriate for detecting the intended difference. As the cell line used in this model of CRC included the expression of the potent model antigen, ovalbumin, the induction of any targeted anti-tumour cytotoxicity may have been sufficient to liberate this protein and induce anti-ovalbumin immune responses. This bystander response may have been additive with the mono-target vaccines, blurring the potential differences that may exist between each therapy. While such bystander responses are likely to be induced as a side-effect of many targeted anti-tumour vaccines, the presence of ovalbumin as a potent antigen may have exacerbated an otherwise minimally significant effect. This potential issue would be rectified by replicating these trials using the MC38 parent strain. Otherwise, additional targets could be added through recombinant insertion to further expand the repertoire of the multi-target chimaeric VLP.

4.4.3 Combination Vaccine Therapies The concept behind combination vaccine therapies is to administer two or more vaccines and/or therapeutic agents that would theoretically cooperate, resulting in additive or complementary effects. Examples include the use to chemotherapy or radiotherapy to increase the production and release of DAMPs or other immunogenic molecules in tumour cells prior to targeted vaccination, or as a means to enhance the potency of whole cell vaccines. An example of this is the use of irradiated, α-galactosylceramide laced autologous tumour cells as a vaccine in a murine model of glioma [834]. The approach used with this vaccine also utilised another therapeutic in its most effective combination formulation, administering anti-CTLA mAbs as an immune checkpoint inhibitor in support of the vaccine. The use of checkpoint inhibitors in combination with vaccines is becoming increasingly common [835-837], attempting to enhance marginal or minimal clinical responses induced

178 by vaccines alone through immunomodulation. Other potential vaccine therapy combinations include the use of therapeutic vaccines or checkpoint blockade to boost the efficacy of adoptively transferred T cells [838], or targeting cytotoxic or radioactive compounds to tumour cells aboard anti-TAA antibodies [491, 839].

The combination vaccine therapy investigated in this study involved the use of chimaeric RHDV VLP, and the oncolytic virus VSV. Rather than simply exploring the potential cumulative effects of co-delivery or sequential delivery of these vaccines, these vaccine constructs were combined to produce an oncolytic virus expressing a virus-like particle. This concept was previously reported by Ma et al, who developed a recombinant VSV that expressed the VP1 protein (VSV-VP1) of human norovirus (HuNoV) VLP [840]. VSV-VP1 were found to induce the formation of intact HuNoV VLP through expression of VP1 in vitro, and induced comparable and in some instances elevated titres of anti-VP1 antibodies in comparison to baculovirus expressed HuNoV VLP. Although the RHDV VLP expressing VSV developed for this study were likewise capable of inducing the formation of intact RHDV VLP in vitro, this did not translate into anti-tumour activity. In fact, there was some evidence that suggested that the inherent oncolytic activity of VSV might have been impaired in these combination constructs. While these results may seem contradictory, there is a major factor that may contribute to this discrepancy, that of dosage. While the expression of VLP by VSV in vivo may result in sufficient production and release of VLP for induction of anti- vector humoral immune responses, these same expressed levels may be insufficient for induction of cell-mediated immune responses against recombinantly inserted epitopes. Alternatively, VSV expressed RHDV VLP may be immunologically inferior in comparison to baculovirus expressed RHDV VLP, or the VSV constructs produced may be have possessed some detrimental flaw. Despite these complications, the underlying concept of using oncolytic viruses in combination with VLP vaccines remains intriguing and merits further investigation, perhaps through synergistic combination or prime-boost vaccination strategies.

4.4.4 Conclusions The successful incorporation of murine H-2kb epitopes derived from CRC TAAs into recombinant RHDV VP60 further expands the list of chimaeric VLP developed as potential cancer vaccines. In this study, the focus was predominantly on the comparison between

179 targeting multiple epitopes in comparison to targeting individual epitopes. Although the distinction between TS.VP60 and both T.VP60 and S.VP60 was not statistically significant, the trends in both reduced tumour growth rate and improved survival with the multi-epitope vaccine suggest that this concept is worth further investigation. The ability to express the TS.VP60 chimaeric VLP with recombinant VSV also provides multiple options for combination vaccine therapies, although further work and optimisation is required for completion of translation from in vitro to in vivo efficacy of these vaccines. The lack of efficacy with the recombinant VSV constructs also impaired the ability to assess the combination of VSV and RHDV VLP in a primary-recurrent vaccination strategy, a concept that still has the potential to alleviate the issues plaguing VSV vaccination with regards to anti-vector immunity.

180

5 Future Directions and Conclusion 5.1 Delivery of Molecules Associated with RHDV VLP The versatility of RHDV VLP as a vaccine platform is profound, possessing the capacity to support the recombinant insertion of peptides and the chemical conjugation of various molecules. Natural affinity of these VLP for specific compounds provides an additional level of complexity, facilitating the formation of composite particles from a relatively simple preparation. This functionality adds to the list of benefits attributable to VLP over synthetic vaccine nanoparticles, which already includes the immunogenicity of repetitive surface structures and the ability to induce cross-presentation of recombinantly inserted epitopes. In addition to adding as a vaccine scaffold, together these components may facilitate the development of complex VLP-based combination vaccine therapies. Some of the potential components that may be incorporated into future VLP vaccines may include different types of adjuvants with different forms of binding or association with VLP, the use of other forms of immunomodulation and gene manipulation with VLP vaccines, and the development of personalised vaccines for cancer.

5.1.1 Adjuvant Delivery An adjuvant can be considered any molecule, compound or formulation that can act as either a non-specific, or specific immunostimulant. A non-specific adjuvant would include most commonly used commercial adjuvants, such as hydrated aluminium sulfate-based compounds (Alum), squalene-based emulsions (e.g. MF59, AS03), or monophosphoryl lipid A (MPL-A). These compounds promote general immune cell activation; however, they can still favour activation of a particular type of immune response above others. For example, Alum is known to stimulate high antibody titres, while the liposomal adjuvant CAF01 induces a mixed TH1/TH17 phenotype [841]. In contrast, specific immunostimulants, such as the different types of CpG or α-galactosylceramide, target relatively narrow populations of immune cells. These adjuvants induce reasonably predictable responses, and can help tailor a vaccine towards achieving a specific outcome. The role of a vaccine adjuvant is to provide the PAMP/DAMP signals that usually accompany infection or inflammation, stimulating their respective receptors and priming APCs. This results in the increased expression and presentation of co-stimulatory molecules, and the release of cytokines. The modality of adjuvant delivery is another important consideration. Co-delivery of the vaccine and adjuvant

181 together as a composite provides compounding benefits, enhancing vaccination efficacy over the delivery of each component separately [842]. This compounding effect may be responsible for the benefits observed when CpG was associated with RHDV VLP.

In addition to co-delivery of adjuvants that naturally associate with RHDV VLP, this interaction may also be utilised for alternative vaccine enhancement application. The ability for these molecules to associate with RHDV VLP may be harnessed as an alternative, non- covalent linkage to associate other molecules with VLP. For example, it would be interesting to see if peptides chemically conjugated to CpG can associate with RHDV VLP. Any such potential interaction is likely to be restricted by the specific location that nucleic acids associate with RHDV VLP, as while surface association may allow conjugated molecules to eject from the surface of VLP, internal association may not. Similarly to this conjugated peptide concept, it would be interesting to see if naturally glycosylated peptides can associate with RHDV VLP. The flexibility of this interaction is particularly intriguing, as the formation of the composite particle need only be induced prior to administration. This may allow each component of the vaccine to be stored under their individually ideal storage conditions, and allow on site interchanging of the vaccine components used for specific formulations. Possessing a repertoire of adjuvants, glycosylated/conjugated receptor ligands, or peptides that alter intracellular processing to add to a range of vaccine constructs could provide the foundation for tailored VLP-based vaccines.

5.1.2 Manipulation of Gene Expression Multiple molecular techniques are available for manipulation of gene expression, both in vitro and in vivo. Examples include the transformation of cells through crude biolistic projectiles, transfection with liposomal particles, infection with genetic modifying viruses and utilisation of the clustered regularly interspaced short palindromic repeats (CRISPR) / CRISPR-associated protein (Cas) system. Each of these tools may be used to introduce xenogeneic genes, additional copies of allogeneic genes, or gene expression regulatory elements such as siRNA, shRNA or miRNA. While each of these techniques have broad biotechnical applications in vitro, in vivo applications for most have been hampered by significant safety concerns, predominantly regarding off-target effects. The relatively modern CRISPR/Cas system for gene editing is unique in this regard, as this technique does contain some inherent guidance and control in its delivery. This tool can be delivered in vivo using

182 conventional virus-based gene delivery techniques, and has already proven effective for inducing expression of target genes in mice [843-846].

Manipulation of gene expression using siRNA associated with VLP is not likely to be as momentous as the utilisation of the CRISPR/Cas system, but does represent an alternative delivery system with some unique potential applications. As VLP are predominantly consumed by APCs, VLP as a vector for siRNA inherently possess a form of targeted delivery. Additionally, conjugation of peptides, ligands and various other molecules onto the surface of VLP that promote uptake into a particular cell population could be used to further specify delivery. In comparison to reported knock-down rates achieved though JCV VLP- mediated delivery of siRNA at around 60-90% [704], variable rates of 10-60% across different dosages of siRNA delivered with RHDV VLP may not be as practically efficacious; however, delivery and knockdown rates may be enhanced through improved chimaeric VLP designs. Alternatively, RHDV VLP-based delivery may have applications in the manipulation of genes in pathways that are sensitive to minor changes in expression.

5.1.2.1 R8.VP60 and DBS.VP60 Despite the intention of establishing a net positive charge at the intraparticulate surface of VLP, R8.VP60 did not appear to have an increased affinity for negatively charged nucleic acids over the base association with unmodified VP60. This may have been due to the degradation or cleavage of this octaarginine domain, or truncation due to a mutation introducing an alternative start codon. Either of these scenarios may provide an explanation for both the duel bands identified by western blot in preparations of R8.VP60, and its apparent inability to benefit from its proposed innate positive charge. If degradation of the octaarginine domain does occur in R8.VP60, the liberation of poly-arginine peptides may also interfere with the natural association between nucleic acids and the R8.VP60 VLP. An alternative explanation may be that the polarised charge introduced at the N-terminus of VP60 may alter the position of the N-terminus within the tertiary structure of the protein. Prediction modelling using the I-TASSER server (Zhang Laboratory, University of Michigan, Michigan, USA) predicts that in comparison to VP60 (Figure 5.1 A), the N- terminus of R8.VP60 may be folded into the protruding (P) domain (Figure 5.1 B) [847]. Although server modelling is predictive only, such an orientation may mask the charged octaarginine motif within the P domain. By contrast, the I-TASSER server predicted that the

183

N-terminus of DBS.VP60 would retain its orientation below the shell (S) domain, remaining at the interparticulate surface (Figure 5.1 C). This might explain why DBS.VP60 VLP appears to have an increased affinity for nucleic acids over unmodified VP60 VLP.

Figure 5.1 Modelling Predictions of VP60, R8.VP60 and DBS.VP60 3D models generated using the I-TASSER server from sequences of (A) VP60, (B) R8.VP60 and (C) DBS.VP60, using the catalogued reference model of the RHDV VP60 conformational structure 3ZUE as a reference [805]. Portions highlighted in red indicate recombinantly inserted sequences. P = Protruding, S = Shell.

The next step for research into these current constructs would be to confirm their predicted structural differences through crystallography, and to link how this structure functionally applies to concepts such as the association of nucleic acids with RHDV VLP. As has been previously mentioned, it would be interesting to investigate how and where these nucleic acids are associating with the VLP; however, it may be of greater interest to investigate how the N-terminus DBS insertion affects this association. If the association is peripheral, perhaps differences in charge can account for the increased association, or perhaps the DBS motif provides a cumulative effect, with its binding interface additive to the natural association. Should the association be internal, or involve projection through gaps in the shell domain as suggested by Touze et al in BKV VLP [712], perhaps the internal DNA binding site is merely functioning as intended. These concepts represent only a fraction of the research that would be required to bring the investigation of these constructs closer to completion.

184

5.1.2.2 Potential Construct Designs Although DBS.VP60 possessed an enhanced ability to associate with nucleic acids, it also suffered from significant instability and low production yields. The L1 binding site of human papilloma virus type-16 is only one of many naturally occurring DNA binding sites that may be incorporated at the N-terminus of RHDV VP60, and it may be prudent to consider alternative constructs with potentially superior stability. For example, protamines are a group of short nuclear proteins that are involved in the packaging of DNA during spermatogenesis [848], containing arginine-rich regions that are highly conserved between different species. Based on its ability to shuttle DNA, incorporation of sequences derived from protamine into RHDV VLP may enhance its ability to associate with nucleic acids.

Figure 5.2 Charge Maps of Proposed Recombinant VP60 for Enhanced siRNA Association Charge density maps generated in Protean 3D, Lasergene 10 (DNAStar, Wisconsin, USA), comparing the relative positive charge at the N-terminus of (A) VP60, (B) R8.VP60, (C) DBS.VP60, (D) PRM.VP60 and (E) DPS.VP60.

In comparison to VP60, R8.VP60 and DBS.VP60, insertion of protamine at the N-terminus of VP60 (PRM.VP60) results in the formation of a domain with a positive charge of a relatively higher magnitude than R8.VP60 (Figure 5.2 A-D). Due to this excessive charge 185 level, it is unknown whether a theoretical PRM.VP60 would possess the same functionality issues as R8.VP60. Another alternative construct may be developed based on the DNA- binding proteins from starved cells (DPS) proteins, derived from bacteria. These DPS proteins are thought to confer protection from damage induced due to the accumulation of reactive oxygen species by binding directly to DNA, and preventing ferroxidation through sequestration of iron [849].

Insertion of the hypothesised DNA-binding site of a Dps protein derived from Mycobacterium smegmatis (UniProt Accession #P0C558) at the N-terminus of VP60 (DPS.VP60) yields the formation a domain with relatively low positive charge in magnitude compared to either R8.VP60 or DBS.VP60 (Figure 5.2 E); however, this positive charge is spread over a longer sequence. 3D models generated for both PRM.VP60 and DPS.VP60 predict that these N-terminus insertions may not adversely interact with the P domain, and hence may functionally more closely resemble DBS.VP60 rather than R8.VP60 (Figure 5.3 A and B). It would be interesting to find out whether either of these constructs are suitable candidates for promoting association or binding of nucleic acids with RHDV VLP, especially as there are multiple isoforms of each DNA-binding domain available for assessment across a broad range of species and strains.

Figure 5.3 Predicted 3D Models of PRM.VP60 and DPS.VP60 Predicted 3D models generated for (A) PRM.VP60 and (B) DPS.VP60 using the I-TASSER server [847], with 3ZUE used as a scaffold [805]. Portions highlighted in red indicate recombinantly inserted sequences.

186

5.1.2.2 Potential Immunomodulation Pathways While further optimisation is required for efficiency delivery of siRNA associated with RHDV VLP, potential appropriate target genes may still be theorised. There are multiple key regulatory, signal transduction, cell surface receptor or cytokines that may serve as appropriate targets amongst APCs. Amongst these, two candidates were selected for further discussion, namely suppressor of cytokine signalling 1 (SOCS1), and forkhead box O3 (FOXO3). These genes were selected based on their central roles in immune cell regulation, and from published evidence manipulation of these genes in immune responses. SOCS1 and SOCS3 are unique amongst the various members of the SOCS protein family, in that they contain an N-terminus kinase inhibitory domain. This domain serves to suppress cytokine- induced signalling pathways, compounding with the inhibition of JAK/STAT signal transduction pathways through the neighbouring SH2 domain shared between SOCS proteins [850, 851]. While deletion of Socs1 in mice can induce a severe inflammatory syndrome that eventually results in polycystic kidney disease and other inflammatory lesions [852], deficiency in Socs3 causes failure to thrive entirely, with mice incapable of completing embryogenesis due to excessive leukaemia inhibitory factor signalling [853]. Socs1 has been previously explored as a potential siRNA target in combination with vaccines, including in combination with a polymer nanoparticle vaccine for cancer [854], and a DC-based vaccine for lymphoma [855].

Amongst the FoxO orthologs expressed in mice, FoxO1 and FoxO3 are the primary isoforms expressed in immune cells such as dendritic cells [856]. FoxO3 is an important regulator of APC function, with dysregulation of cytokine expression in response to infection when mice are deficient in FoxO3 [857]. While FoxO1 has been implicated in both primary CD8+ T cell responses and memory formation [858, 859], FoxO3 appears to be restricted to mediating primary responses [860]. FoxO3 siRNA have also been investigated in combination with an immunotherapeutic vaccine for cancer, using a chimaeric peptide nanoparticle [861]. Silencing Socs1 and/or FoxO3 with siRNA in combination with an RHDV VLP-based vaccine may enhance responses to the vaccine. Broad dampening of inhibitory signalling in APCs could also theoretically facilitate the emergence of suppressed inherent anti-tumour responses, similar in functionality to checkpoint inhibition with monoclonal antibodies. Immunomodulation through manipulation of gene expression in APCs may become an important component of immunotherapeutic vaccine design.

187

5.2 RHDV VLP Vaccines for Cancer Amongst subunit vaccines, VLP possess a unique balance of immunogenicity, versatility and scalability. RHDV VLP have been identified to possess each of these qualities, serving as a potentially ideal vaccine scaffold. Although significant progress has already been made to bring RHDV VLP-based vaccines for cancer closer towards clinical applications, further work is necessary to realise this goal. These remaining hurdles include: 1) identifying tumour-associated antigen targets that can be translated into human tumours; 2) finalising the optimal RHDV VLP vaccine design and delivery regimen; and 3) manufacture of the vaccine to a standard suitable for use in humans. While the first and third hurdles are limited predominantly by infrastructure, it is the second that may be more immediately solvable. RHDV VLP are amenable to recombinant insertion, chemical conjugation, and naturally associate with an expanding list of molecules. Although in previous studies each of these properties have been explored individually, perhaps it is time to consider how each of these properties may be utilised together in one coherent vaccine construct.

5.2.1 Chimaeric RHDV VLP RHDV VLP can tolerate insertion of peptides at the C-terminus, N-terminus, and at additional sites such as at AA306 of the P domain [862]. Inherent restrictions in insertion length and sequence lends towards conservative approaches, pitting quality against quantity. On the one hand, insertion of a single peptide corresponding to a ubiquitously expressed tumour-specific antigen may produce a robust vaccine, but such targets are rare amongst cancers. More epitopes equates to longer insertions, increasing the risk of VLP instability, and the looming threat of immunodominance. Add in the complexity of providing epitopes specific for multiple HLA isoforms to produce a vaccine relevant across a diverse human population, and the demands for recombinant insertion begin to stretch thin the restrictions imposed by the particle.

One potential solution that has been investigated is to develop truncated variants of VP60 [636], potentially providing additional capacity for insertion. An alternative approach involving simultaneous insertion at both the N- and C-terminus was also explored in Chapter 4. This approach may be suitable for further research into expanding the recombinant insertion capacity of RHDV VLP. Alternatively, the AA306 insertion site may be a potential target. While this site was previously investigated for the insertion of immune domains

188 suitable for inducing humoral immune responses due to its surface orientation, this site may also accommodate insertion of epitopes for cytotoxic responses. Minimisation of the effect this insertion has on the P domain of VP60 would be imperative to the formation of intact, stable VLP. A proposed solution to this issue would be to project the inserted peptide away from the P domain, with the α-helical coiled coil structure of keratin suggested as a possible candidate Cytokeratin (C) domain. For this concept, a portion of the α-helical sequences from human keratin type I cytoskeletal (KRT) 10 (UniProt Accession #P13645) and KRT1 (UniProt Accession #P04264) were obtained for prediction modelling of a potential construct, CYK306.VP60, based on model 4ZRY from x-ray crystallography work by Bunick et al (unpublished data). The resulting model of CYK306.VP60 suggests that the proposed C domain would extend away from the protruding domain, and hence away from the surface of the VLP (Figure 5.4).

Figure 5.4 Proposed VP60 Construct for Long Peptide Insertion A Predicted model of CYK306.VP60 using the I-TASSER server [847], with 3ZUE used as a scaffold [805]. Portions highlighted blue indicate KRT1 and KRT10, while portions in red indicate the model epitope SIINFEKL.

189

This model also predicts that the central epitope, in this example SIINFEKL, would be exposed at the tip of the C domain. The potential elongation of this domain and its incorporated epitope away from the surface of the resulting VLP also harkens back to the initial research done using the AA306 insertion site [862]. While the use of the AA306 site for direct insertion of epitopes proved ineffective at inducing epitope-specific antibody production, it is possible the elongation of such an epitope away from the surface of VLP would increase its availability for detection by B cells. Finally, this elongated domain may also be suitable for insertion of peptides that facilitate receptor-mediated endocytosis, potentially providing an additional pathway for VLP uptake and processing.

5.2.2 Conjugated RHDV VLP Although recombinant insertion of epitopes into RHDV VLP is a robust means of vaccine design, each construct must be developed individually with a comparatively significant initial investment of time and resources. An alternative, rapid approach is to chemically conjugate the epitope peptides onto the surface of RHDV VLP. In a direct comparison, VLP conjugated with SIINFEKL peptide were found to be significantly less immunogenic than chimaeric SIIN.VP60 [828]; however, the presence of the conjugation motifs on the end of the SIINFEKL peptide may have affected its ability to be processed and presented. This may be rectified by including an immunoprotease cleavage domain in between the peptide and the conjugation motif, potentially facilitating the release an unmodified version of the peptide (Figure 5.5 A).

Figure 5.5 Proposed Peptide Conjugation Constructs Proposed modified peptide conjugation chemistries based on NHS-phosphine/Azide bond formation, and separation of the peptide epitope from the conjugation chemistry with either (A) ALL or (B) GGG.

Alternatively, a general spacer such as a glycine triplicate may be sufficient to facilitate processing of the epitope (Figure 5.5 B). These proposed conjugation systems may

190 theoretically bring chemically conjugated RHDV VLP in line with chimaeric VLP with regards to immunogenicity. Chemical conjugation also facilitates the incorporation of additional functionality, such as targeted delivery through the conjugation of receptor ligands, receptors or mAbs onto the surface of VLP.

5.2.3 Personalised RHDV VLP Vaccines for Cancer Personalised cancer vaccines are based on the ability to tailor each vaccine to target specific antigens expressed by a particular patient’s tumour. This vaccine would then be redesigned for the next patient based on their specific antigen expression profile, and so forth. This concept deals with heterogeneity between patients with tumours of the same type, providing a vaccine with the best chance of inducing potent anti-tumour immune responses. The primary limiting factor with this vaccine model is the identification of the specific antigens expressed by each patient’s tumour. This may be rectified by arranging the known expression profiles for a particular tumour line into a form of antigen library, which may be screened against for each individual patient. While theoretically possible, current techniques are limited in their ability to provide the desired functionality. For example, RNAseq could be used to assess the transcriptome of individual patient’s tumours, providing the capacity to establish both a transcriptome library and to screen for candidate vaccine targets within in individual transcriptomes of patients; however, gene expression does not necessarily result in protein translation, which may result in discrepancies between the transcriptome and proteome.

Alternatively, the proteome or similar protein-based markers may be screened by mass spectrometry or other techniques; however, the multitude of proteins that would be present in low concentrations relative to cytoskeletal and other structural proteins would make detection relatively difficult or tedious. While each of these approaches are individually flawed, a combination of each may provide sufficient compensation. The proposed method, referred to as the peptide library method, would involve a combination of both RNAseq for transcriptome analysis (Figure 5.6 A, B), and mass spectrometry for analysing a portion of the proteome. Rather than analysing the entire proteome amongst tumours, peptides loaded in MHC-I on the surface of tumour cells may be isolated, with peptides eluted from the MHC- I cleft, and the resulting peptide pool analysed by Triple-TOF (Figure 5.6 A, C).

191

Figure 5.6 Establishment of a Tumour-associated Peptide Library Establishment of a peptide library for personalised vaccines in a murine tumour model. (A) Engrafted tumours and normal tissue harvested from mice would be isolated. (B) RNA extracted from samples would be analysed by RNAseq, providing a transcriptome data. (C) Peptides eluted from MHC-I isolated from the surface of tumour cells would be analysed by mass spectrometry. (D) Bioinformatic analysis of the resulting data correlating the transcriptome expression with peptide presentation would result in a list of candidate peptide targets for potential cancer vaccines.

As with the original proteome analysis, this MHC-I restricted peptide pool would contain many different peptides in potentially low concentrations. This issue can be rectified by screening for a list of genes selected based on the RNAseq data, using this transcriptome data as guidance for the peptide pool mass spectrometry (Figure 5.6 D). Through the analysis of tumour samples across a comprehensive patient cohort, the result would be a list of MHC-I restricted peptides that are overexpressed for the particular tumour type. Repetition of the peptide elution and guided mass spectrometry from tumour samples resected from each subsequent patient would allow the selection of potential vaccine targets from this peptide library. When combined with the rapid chemical conjugation of peptides on the surface of RHDV VLP, which may be undertaken over a period of around 48 hours, this may represent a rapid method for providing personalised vaccines for cancer.

192

5.2.4 Combination Vaccination The concept of combination vaccination involving VLP has been mentioned previously, and includes the use of VLP with other vaccines in both prime-boost or primary-recurrent vaccination strategies, or synergy between two homologous vaccines. Examples of the latter include the use of immunomodulatory agents such as checkpoint blockade mAbs in combination with VLP, or the use of VLP to stimulate adoptively transferred cells in vivo. One of the concepts explored in this study included the use of recombinant VSV to express RHDV VLP, forming a combination vaccine construct that should theoretically combine the effector functions of both the host oncolytic virus and the derivative VLP vaccine. While the constructs produced to explore this combination proved ineffective, the use of alternative oncolytic viruses or optimisation of the vaccination strategy used may improve the potential synergy between these vaccine components. A similar concept that was not explored in this study is the expression of VLP induced by VLP. Essentially, VLP that can be used as a gene delivery tool should theoretically be capable of delivering a gene encoding their capsid protein, potentially inducing the formation of VLP in vivo. This concept may be stretched further by using VLP to deliver a gene encoding the capsid protein or a different VLP, bypassing anti-vector immunity against the encoded VLP. These ideas represent only a few of the potential concepts that could utilise VLP as a component of a combination vaccine, already representing a significant body of work. The potential for VLP-based vaccines is truly extensive.

193

5.3 Conclusion RHDV VLP are a promising vector for development of immunotherapeutic cancer vaccines. This thesis covered a broad range of applications related to improving the efficacy and versatility of RHDV VLP-based vaccines, including through the formation of a composite particle with CpG, the delivery of associated siRNA, simultaneously induction of targeted immune responses in a multi-epitope vaccine, and the expression of RHDV VLP by VSV. Immunomodulation with adjuvants and other regulatory molecules is an intriguing concept; probably most predominantly exhibited clinically using checkpoint inhibitors and other immunomodulatory mAbs. In this study, both CpG and siRNA was found to associate with RHDV VLP, forming composite particles capable of delivery of these molecules into cells in vitro, as well as in vivo for CpG. This portion of work confirmed that there is something unique about the combination of RHDV VLP and CpG, allowing the formation of a potent vaccine construct.

Chimaeric RHDV VLP were also found to be capable of inducing targeted immune responses against murine epitopes derived from the CRC TAAs topIIα and survivin. These responses also translated in a murine model of CRC, significantly improving survival amongst vaccinated mice. Although the multi-target vaccine TS.VP60 did not provide a statistically significant advantage over the individual mono-target vaccines, the trends observed suggest that the incorporation of additional epitopes may be significantly advantageous. Finally, RHDV VLP were successfully expressed by recombinant VSV in vitro; however, these recombinant VSV constructs proved ineffective in vivo. There were multiple potential issues relating to the expression of RHDV VLP by recombinant VSV, with multiple potential avenues for further research. Together this body of work provides another step along the progression towards the use of RHDV VLP as a cancer vaccine platform, though as with many research projects it proposes almost as many questions as it answered.

194

6 References 1. Ahmad, A.S., N. Ormiston-Smith, and Study. Cancer Epidemiol Biomarkers P.D. Sasieni, Trends in the lifetime risk Prev, 2005. 14(1): p. 133-7. of developing cancer in Great Britain: 11. Limburg, P.J., et al., Cigarette smoking comparison of risk for those born from and colorectal cancer: long-term, 1930 to 1960. Br J Cancer, 2015. subsite-specific risks in a cohort study 112(5): p. 943-7. of postmenopausal women. Clin 2. Cancer: New Registrations and Gastroenterol Hepatol, 2003. 1(3): p. Deaths 2012, M.o. Health, Editor. 202-10. 2015, Ministry of Health: Wellington. 12. Alley, P.G. and S.P. Lee, The increased 3. Bulliard, J.L., B. Cox, and J.M. Elwood, risk of proximal colonic cancer after Latitude gradients in melanoma cholecystectomy. Dis Colon Rectum, incidence and mortality in the non- 1983. 26(8): p. 522-4. Maori population of New Zealand. 13. Vernick, L.J. and L.H. Kuller, A case- Cancer Causes Control, 1994. 5(3): p. control study of cholecystectomy and 234-40. right-side colon cancer: the influence 4. Center, M.M., A. Jemal, and E. Ward, of alternative data sources and International trends in colorectal differential interview participation cancer incidence rates. Cancer proportions on odds ratio estimates. Epidemiol Biomarkers Prev, 2009. Am J Epidemiol, 1982. 116(1): p. 86- 18(6): p. 1688-94. 101. 5. Parkin, D.M., et al., Global cancer 14. Ferrari, P., et al., Lifetime and baseline statistics, 2002. CA Cancer J Clin, alcohol intake and risk of colon and 2005. 55(2): p. 74-108. rectal cancers in the European 6. Friedenreich, C., et al., Physical prospective investigation into cancer activity and risk of colon and rectal and nutrition (EPIC). Int J Cancer, cancers: the European prospective 2007. 121(9): p. 2065-72. investigation into cancer and 15. Wallace, K., et al., The association of nutrition. Cancer Epidemiol lifestyle and dietary factors with the Biomarkers Prev, 2006. 15(12): p. risk for serrated polyps of the 2398-407. colorectum. Cancer Epidemiol 7. Moradi, T., et al., Occupational Biomarkers Prev, 2009. 18(8): p. physical activity and risk for cancer of 2310-7. the colon and rectum in Sweden 16. Bufill, J.A., Colorectal cancer: among men and women by anatomic evidence for distinct genetic subsite. Eur J Cancer Prev, 2008. categories based on proximal or distal 17(3): p. 201-8. tumor location. Ann Intern Med, 8. Harriss, D.J., et al., Lifestyle factors 1990. 113(10): p. 779-88. and colorectal cancer risk (1): 17. Thibodeau, S.N., G. Bren, and D. systematic review and meta-analysis Schaid, Microsatellite instability in of associations with body mass index. cancer of the proximal colon. Science, Colorectal Dis, 2009. 11(6): p. 547-63. 1993. 260(5109): p. 816-9. 9. Hu, F.B., et al., Prospective study of 18. Ionov, Y., et al., Ubiquitous somatic adult onset diabetes mellitus (type 2) mutations in simple repeated and risk of colorectal cancer in sequences reveal a new mechanism women. J Natl Cancer Inst, 1999. for colonic carcinogenesis. Nature, 91(6): p. 542-7. 1993. 363(6429): p. 558-61. 10. Limburg, P.J., et al., Diabetes mellitus 19. Araten, D.J., et al., A quantitative and subsite-specific colorectal cancer measurement of the human somatic risks in the Iowa Women's Health mutation rate. Cancer Res, 2005. 65(18): p. 8111-7.

195

20. Quammen, D., Contagious Cancer: among Ashkenazi Jews. N Engl J Med, The Evolution of a Killer, in Harper's. 1997. 336(20): p. 1401-8. 2008. 32. Ford, D., et al., Genetic heterogeneity 21. United Nations. Scientific Committee and penetrance analysis of the BRCA1 on the Effects of Atomic Radiation., and BRCA2 genes in breast cancer Sources and effects of ionizing families. The Breast Cancer Linkage radiation : United Nations Scientific Consortium. Am J Hum Genet, 1998. Committee on the Effects of Atomic 62(3): p. 676-89. Radiation : UNSCEAR 2008 report to 33. Smyrk, T.C., et al., Tumor-infiltrating the General Assembly, with scientific lymphocytes are a marker for annexes. 2010, New York: United microsatellite instability in colorectal Nations. carcinoma. Cancer, 2001. 91(12): p. 22. Lijinsky, W., The formation and 2417-22. occurrence of polynuclear aromatic 34. Lynch, H.T. and A. de la Chapelle, hydrocarbons associated with food. Genetic susceptibility to non- Mutat Res, 1991. 259(3-4): p. 251-61. polyposis colorectal cancer. J Med 23. Knize, M.G., et al., Food heating and Genet, 1999. 36(11): p. 801-18. the formation of heterocyclic 35. Lynch, H.T. and A. de la Chapelle, aromatic amine and polycyclic Hereditary colorectal cancer. N Engl J aromatic hydrocarbon Med, 2003. 348(10): p. 919-32. mutagens/carcinogens. Adv Exp Med 36. Bakkenist, C.J. and M.B. Kastan, DNA Biol, 1999. 459: p. 179-93. damage activates ATM through 24. Dayan, A.D., Carcinogenicity and intermolecular autophosphorylation drinking water. Pharmacol Toxicol, and dimer dissociation. Nature, 2003. 1993. 72 Suppl 1: p. 108-15. 421(6922): p. 499-506. 25. Cantor, K.P., Drinking water and 37. Mu, D., et al., Reconstitution of cancer. Cancer Causes Control, 1997. human DNA repair excision nuclease 8(3): p. 292-308. in a highly defined system. J Biol 26. Koivusalo, M., et al., Drinking water Chem, 1995. 270(6): p. 2415-8. chlorination and cancer-a historical 38. Mu, D., D.S. Hsu, and A. Sancar, cohort study in Finland. Cancer Reaction mechanism of human DNA Causes Control, 1997. 8(2): p. 192- repair excision nuclease. J Biol Chem, 200. 1996. 271(14): p. 8285-94. 27. Manz, A., et al., Cancer mortality 39. Evans, E., et al., Mechanism of open among workers in chemical plant complex and dual incision formation contaminated with dioxin. Lancet, by human nucleotide excision repair 1991. 338(8773): p. 959-64. factors. EMBO J, 1997. 16(21): p. 28. Saracci, R., et al., Cancer mortality in 6559-73. workers exposed to chlorophenoxy 40. D'Amours, D. and S.P. Jackson, The herbicides and chlorophenols. Lancet, Mre11 complex: at the crossroads of 1991. 338(8774): p. 1027-32. dna repair and checkpoint signalling. 29. Miki, Y., et al., A strong candidate for Nat Rev Mol Cell Biol, 2002. 3(5): p. the breast and ovarian cancer 317-27. susceptibility gene BRCA1. Science, 41. Ferguson, D.O. and F.W. Alt, DNA 1994. 266(5182): p. 66-71. double strand break repair and 30. Lancaster, J.M., et al., BRCA2 chromosomal translocation: lessons mutations in primary breast and from animal models. Oncogene, ovarian cancers. Nat Genet, 1996. 2001. 20(40): p. 5572-9. 13(2): p. 238-40. 42. Khanna, K.K. and S.P. Jackson, DNA 31. Struewing, J.P., et al., The risk of double-strand breaks: signaling, cancer associated with specific repair and the cancer connection. Nat mutations of BRCA1 and BRCA2 Genet, 2001. 27(3): p. 247-54.

196

43. Petrini, J.H., The mammalian Mre11- interaction between transcription Rad50-nbs1 protein complex: factor E2F and the retinoblastoma integration of functions in the cellular gene product. Proc Natl Acad Sci U S DNA-damage response. Am J Hum A, 1992. 89(10): p. 4549-53. Genet, 1999. 64(5): p. 1264-9. 53. Lowy, D.R. and J.T. Schiller, 44. Rouse, J. and S.P. Jackson, Interfaces Prophylactic human papillomavirus between the detection, signaling, and vaccines. J Clin Invest, 2006. 116(5): p. repair of DNA damage. Science, 2002. 1167-73. 297(5581): p. 547-51. 54. Wang, F., et al., Epstein-Barr virus 45. Werness, B.A., A.J. Levine, and P.M. latent membrane protein (LMP1) and Howley, Association of human nuclear proteins 2 and 3C are papillomavirus types 16 and 18 E6 effectors of phenotypic changes in B proteins with p53. Science, 1990. lymphocytes: EBNA-2 and LMP1 248(4951): p. 76-9. cooperatively induce CD23. J Virol, 46. Dyson, N., et al., The human 1990. 64(5): p. 2309-18. papilloma virus-16 E7 oncoprotein is 55. Grossman, S.R., et al., The Epstein- able to bind to the retinoblastoma Barr virus nuclear antigen 2 gene product. Science, 1989. transactivator is directed to response 243(4893): p. 934-7. elements by the J kappa 47. Dyson, N., et al., Homologous recombination signal binding protein. sequences in adenovirus E1A and Proc Natl Acad Sci U S A, 1994. 91(16): human papillomavirus E7 proteins p. 7568-72. mediate interaction with the same set 56. Johannsen, E., et al., Epstein-Barr of cellular proteins. J Virol, 1992. virus nuclear protein 2 transactivation 66(12): p. 6893-902. of the latent membrane protein 1 48. Davies, R., et al., Human promoter is mediated by J kappa and papillomavirus type 16 E7 associates PU.1. J Virol, 1995. 69(1): p. 253-62. with a histone H1 kinase and with 57. Parker, G.A., R. Touitou, and M.J. p107 through sequences necessary Allday, Epstein-Barr virus EBNA3C can for transformation. J Virol, 1993. disrupt multiple cell cycle checkpoints 67(5): p. 2521-8. and induce nuclear division divorced 49. Scheffner, M., et al., The E6 from cytokinesis. Oncogene, 2000. oncoprotein encoded by human 19(5): p. 700-9. papillomavirus types 16 and 18 58. Jha, H.C., et al., EBNA3C-mediated promotes the degradation of p53. regulation of aurora kinase B Cell, 1990. 63(6): p. 1129-36. contributes to Epstein-Barr virus- 50. Huibregtse, J.M., M. Scheffner, and induced B-cell proliferation through P.M. Howley, A cellular protein modulation of the activities of the mediates association of p53 with the retinoblastoma protein and apoptotic E6 oncoprotein of human caspases. J Virol, 2013. 87(22): p. papillomavirus types 16 or 18. EMBO 12121-38. J, 1991. 10(13): p. 4129-35. 59. Maruo, S., et al., Epstein-Barr virus 51. Crook, T., J.A. Tidy, and K.H. Vousden, nuclear antigens 3C and 3A maintain Degradation of p53 can be targeted lymphoblastoid cell growth by by HPV E6 sequences distinct from repressing p16INK4A and p14ARF those required for p53 binding and expression. Proc Natl Acad Sci U S A, trans-activation. Cell, 1991. 67(3): p. 2011. 108(5): p. 1919-24. 547-56. 60. Floettmann, J.E. and M. Rowe, 52. Chellappan, S., et al., Adenovirus E1A, Epstein-Barr virus latent membrane simian virus 40 tumor antigen, and protein-1 (LMP1) C-terminus human papillomavirus E7 protein activation region 2 (CTAR2) maps to share the capacity to disrupt the the far C-terminus and requires

197

oligomerisation for NF-kappaB 70. Guo, W., J.L. Lasky, 3rd, and H. Wu, activation. Oncogene, 1997. 15(15): Cancer stem cells. Pediatr Res, 2006. p. 1851-8. 59(4 Pt 2): p. 59R-64R. 61. Izumi, K.M. and E.D. Kieff, The 71. Jordan, C.T., M.L. Guzman, and M. Epstein-Barr virus oncogene product Noble, Cancer stem cells. N Engl J latent membrane protein 1 engages Med, 2006. 355(12): p. 1253-61. the tumor necrosis factor receptor- 72. Shipitsin, M. and K. Polyak, The cancer associated death domain protein to stem cell hypothesis: in search of mediate B lymphocyte growth definitions, markers, and relevance. transformation and activate NF- Lab Invest, 2008. 88(5): p. 459-63. kappaB. Proc Natl Acad Sci U S A, 73. Held, M.A., et al., Characterization of 1997. 94(23): p. 12592-7. melanoma cells capable of 62. Danesh, J., Helicobacter pylori propagating tumors from a single cell. infection and gastric cancer: Cancer Res, 2010. 70(1): p. 388-97. systematic review of the 74. Hoek, K.S., et al., Stemming the flood. epidemiological studies. Aliment Pigment Cell Melanoma Res, 2009. Pharmacol Ther, 1999. 13(7): p. 851- 22(1): p. 6-7. 6. 75. Hoek, K.S. and C.R. Goding, Cancer 63. Huang, J.Q., et al., Meta-analysis of stem cells versus phenotype- the relationship between switching in melanoma. Pigment Cell Helicobacter pylori seropositivity and Melanoma Res, 2010. 23(6): p. 746- gastric cancer. Gastroenterology, 59. 1998. 114(6): p. 1169-1179. 76. Scheel, C., et al., Paracrine and 64. Weinstein, I.B., The origins of human autocrine signals induce and maintain cancer: molecular mechanisms of mesenchymal and stem cell states in carcinogenesis and their implications the breast. Cell, 2011. 145(6): p. 926- for cancer prevention and treatment- 40. -twenty-seventh G.H.A. Clowes 77. Vermeulen, L., et al., The developing memorial award lecture. Cancer Res, cancer stem-cell model: clinical 1988. 48(15): p. 4135-43. challenges and opportunities. Lancet 65. Pitot, H.C., T. Goldsworthy, and S. Oncol, 2012. 13(2): p. e83-9. Moran, The natural history of 78. Chen, J., et al., A restricted cell carcinogenesis: implications of population propagates glioblastoma experimental carcinogenesis in the growth after chemotherapy. Nature, genesis of human cancer. J Supramol 2012. 488(7412): p. 522-6. Struct Cell Biochem, 1981. 17(2): p. 79. Driessens, G., et al., Defining the 133-46. mode of tumour growth by clonal 66. Nowell, P.C., The clonal evolution of analysis. Nature, 2012. 488(7412): p. tumor cell populations. Science, 1976. 527-30. 194(4260): p. 23-8. 80. Schepers, A.G., et al., Lineage tracing 67. Foulds, L., The experimental study of reveals Lgr5+ stem cell activity in tumor progression: a review. Cancer mouse intestinal adenomas. Science, Res, 1954. 14(5): p. 327-39. 2012. 337(6095): p. 730-5. 68. Weissman, I.L., Stem cells: units of 81. Iliopoulos, D., et al., Inducible development, units of regeneration, formation of breast cancer stem cells and units in evolution. Cell, 2000. and their dynamic equilibrium with 100(1): p. 157-68. non-stem cancer cells via IL6 69. Clarke, M.F. and M. Fuller, Stem cells secretion. Proc Natl Acad Sci U S A, and cancer: two faces of eve. Cell, 2011. 108(4): p. 1397-402. 2006. 124(6): p. 1111-5. 82. Chaffer, C.L., et al., Poised chromatin at the ZEB1 promoter enables breast cancer cell plasticity and enhances

198

tumorigenicity. Cell, 2013. 154(1): p. 93. Anastas, J.N., et al., WNT5A enhances 61-74. resistance of melanoma cells to 83. Chaffer, C.L., et al., Normal and targeted BRAF inhibitors. J Clin Invest, neoplastic nonstem cells can 2014. 124(7): p. 2877-90. spontaneously convert to a stem-like 94. Zipser, M.C., et al., A proliferative state. Proc Natl Acad Sci U S A, 2011. melanoma cell phenotype is 108(19): p. 7950-5. responsive to RAF/MEK inhibition 84. Hoek, K.S., et al., Metastatic potential independent of BRAF mutation status. of melanomas defined by specific Pigment Cell Melanoma Res, 2011. gene expression profiles with no BRAF 24(2): p. 326-33. signature. Pigment Cell Res, 2006. 95. Sanchez-Laorden, B., et al., BRAF 19(4): p. 290-302. inhibitors induce metastasis in RAS 85. Hoek, K.S., et al., In vivo switching of mutant or inhibitor-resistant human melanoma cells between melanoma cells by reactivating MEK proliferative and invasive states. and ERK signaling. Sci Signal, 2014. Cancer Res, 2008. 68(3): p. 650-6. 7(318): p. ra30. 86. Quintana, E., et al., Efficient tumour 96. Yao, Z., et al., TGF-beta IL-6 axis formation by single human mediates selective and adaptive melanoma cells. Nature, 2008. mechanisms of resistance to 456(7222): p. 593-8. molecular targeted therapy in lung 87. Quintana, E., et al., Phenotypic cancer. Proc Natl Acad Sci U S A, 2010. heterogeneity among tumorigenic 107(35): p. 15535-40. melanoma cells from patients that is 97. Yauch, R.L., et al., Epithelial versus reversible and not hierarchically mesenchymal phenotype determines organized. Cancer Cell, 2010. 18(5): p. in vitro sensitivity and predicts clinical 510-23. activity of erlotinib in lung cancer 88. de Visser, K.E. and L.M. Coussens, The patients. Clin Cancer Res, 2005. 11(24 inflammatory tumor Pt 1): p. 8686-98. microenvironment and its impact on 98. Thomas, R., et al., In vitro binding cancer development. Contrib evaluation of 177Lu-AMBA, a novel Microbiol, 2006. 13: p. 118-37. 177Lu-labeled GRP-R agonist for 89. Thomas, A.J. and C.A. Erickson, The systemic radiotherapy in human making of a melanocyte: the tissues. Clin Exp Metastasis, 2009. specification of melanoblasts from 26(2): p. 105-19. the neural crest. Pigment Cell 99. Zhang, Z., et al., Activation of the AXL Melanoma Res, 2008. 21(6): p. 598- kinase causes resistance to EGFR- 610. targeted therapy in lung cancer. Nat 90. Adameyko, I., et al., Schwann cell Genet, 2012. 44(8): p. 852-60. precursors from nerve innervation are 100. Mani, S.A., et al., The epithelial- a cellular origin of melanocytes in mesenchymal transition generates skin. Cell, 2009. 139(2): p. 366-79. cells with properties of stem cells. 91. Seftor, E.A., et al., Epigenetic Cell, 2008. 133(4): p. 704-15. transdifferentiation of normal 101. Liu, S., et al., Breast cancer stem cells melanocytes by a metastatic transition between epithelial and melanoma microenvironment. Cancer mesenchymal states reflective of their Res, 2005. 65(22): p. 10164-9. normal counterparts. Stem Cell 92. Weeraratna, A.T., et al., Wnt5a Reports, 2014. 2(1): p. 78-91. signaling directly affects cell motility 102. Biddle, A., et al., Cancer stem cells in and invasion of metastatic squamous cell carcinoma switch melanoma. Cancer Cell, 2002. 1(3): p. between two distinct phenotypes that 279-88. are preferentially migratory or

199

proliferative. Cancer Res, 2011. 115. Dunn, G.P., L.J. Old, and R.D. 71(15): p. 5317-26. Schreiber, The immunobiology of 103. O'Connor, M.L., et al., Cancer stem cancer immunosurveillance and cells: A contentious hypothesis now immunoediting. Immunity, 2004. moving forward. Cancer Lett, 2014. 21(2): p. 137-48. 344(2): p. 180-7. 116. Schreiber, R.D., L.J. Old, and M.J. 104. Burnet, M., Cancer; a biological Smyth, Cancer immunoediting: approach. I. The processes of control. integrating immunity's roles in cancer Br Med J, 1957. 1(5022): p. 779-86. suppression and promotion. Science, 105. Thomas, L., in Cellular and Humoral 2011. 331(6024): p. 1565-70. Aspects of the Hypersensitive States, 117. Donin, N., et al., Complement H.S. Lawrence, Editor. 1959, Hoeber- resistance of human carcinoma cells Harper: New York. depends on membrane regulatory 106. Stutman, O., Tumor development proteins, protein kinases and sialic after 3-methylcholanthrene in acid. Clin Exp Immunol, 2003. 131(2): immunologically deficient athymic- p. 254-63. nude mice. Science, 1974. 183(4124): 118. Gorter, A. and S. Meri, Immune p. 534-6. evasion of tumor cells using 107. Stutman, O., Immunodepression and membrane-bound complement malignancy. Adv Cancer Res, 1975. regulatory proteins. Immunol Today, 22: p. 261-422. 1999. 20(12): p. 576-82. 108. Matzinger, P., Tolerance, danger, and 119. Markiewski, M.M., et al., Modulation the extended family. Annu Rev of the antitumor immune response by Immunol, 1994. 12: p. 991-1045. complement. Nat Immunol, 2008. 109. Pardoll, D., Does the immune system 9(11): p. 1225-35. see tumors as foreign or self? Annu 120. Alexopoulou, L., et al., Recognition of Rev Immunol, 2003. 21: p. 807-39. double-stranded RNA and activation 110. Dighe, A.S., et al., Enhanced in vivo of NF-kappaB by Toll-like receptor 3. growth and resistance to rejection of Nature, 2001. 413(6857): p. 732-8. tumor cells expressing dominant 121. Hayashi, F., et al., The innate immune negative IFN gamma receptors. response to bacterial flagellin is Immunity, 1994. 1(6): p. 447-56. mediated by Toll-like receptor 5. 111. Kaplan, D.H., et al., Demonstration of Nature, 2001. 410(6832): p. 1099- an interferon gamma-dependent 103. tumor surveillance system in 122. Xiao, T., Innate immune recognition of immunocompetent mice. Proc Natl nucleic acids. Immunol Res, 2009. Acad Sci U S A, 1998. 95(13): p. 7556- 43(1-3): p. 98-108. 61. 123. Javid, B., P.A. MacAry, and P.J. 112. Shankaran, V., et al., IFNgamma and Lehner, Structure and function: heat lymphocytes prevent primary tumour shock proteins and adaptive development and shape tumour immunity. J Immunol, 2007. 179(4): p. immunogenicity. Nature, 2001. 2035-40. 410(6832): p. 1107-11. 124. Feng, H., et al., Stressed apoptotic 113. Dunn, G.P., et al., Cancer tumor cells stimulate dendritic cells immunoediting: from and induce specific cytotoxic T cells. immunosurveillance to tumor escape. Blood, 2002. 100(12): p. 4108-15. Nat Immunol, 2002. 3(11): p. 991-8. 125. Shi, Y., J.E. Evans, and K.L. Rock, 114. Dunn, G.P., C.M. Koebel, and R.D. Molecular identification of a danger Schreiber, Interferons, immunity and signal that alerts the immune system cancer immunoediting. Nat Rev to dying cells. Nature, 2003. Immunol, 2006. 6(11): p. 836-48. 425(6957): p. 516-21.

200

126. Tesniere, A., et al., Immunogenic 138. Li, H., A. Ambade, and F. Re, Cutting death of colon cancer cells treated edge: Necrosis activates the NLRP3 with oxaliplatin. Oncogene, 2010. inflammasome. J Immunol, 2009. 29(4): p. 482-91. 183(3): p. 1528-32. 127. Apetoh, L., et al., The interaction 139. Ghiringhelli, F., et al., Activation of the between HMGB1 and TLR4 dictates NLRP3 inflammasome in dendritic the outcome of anticancer cells induces IL-1beta-dependent chemotherapy and radiotherapy. adaptive immunity against tumors. Immunol Rev, 2007. 220: p. 47-59. Nat Med, 2009. 15(10): p. 1170-8. 128. Akira, S., Toll-like receptors and 140. Martinon, F., K. Burns, and J. Tschopp, innate immunity. Adv Immunol, 2001. The inflammasome: a molecular 78: p. 1-56. platform triggering activation of 129. Schroder, K. and J. Tschopp, The inflammatory caspases and inflammasomes. Cell, 2010. 140(6): p. processing of proIL-beta. Mol Cell, 821-32. 2002. 10(2): p. 417-26. 130. Terra, J.K., et al., Cutting edge: 141. Allen, I.C., et al., The NLRP3 resistance to Bacillus anthracis inflammasome functions as a infection mediated by a lethal toxin negative regulator of tumorigenesis sensitive allele of Nalp1b/Nlrp1b. J during colitis-associated cancer. J Exp Immunol, 2010. 184(1): p. 17-20. Med, 2010. 207(5): p. 1045-56. 131. Boyden, E.D. and W.F. Dietrich, 142. Kato, H., et al., Differential roles of Nalp1b controls mouse macrophage MDA5 and RIG-I helicases in the susceptibility to anthrax lethal toxin. recognition of RNA viruses. Nature, Nat Genet, 2006. 38(2): p. 240-4. 2006. 441(7089): p. 101-5. 132. Suzuki, T. and G. Nunez, A role for 143. Saito, T., et al., Regulation of innate Nod-like receptors in autophagy antiviral defenses through a shared induced by Shigella infection. repressor domain in RIG-I and LGP2. Autophagy, 2008. 4(1): p. 73-5. Proc Natl Acad Sci U S A, 2007. 104(2): 133. Franchi, L., et al., Critical role for Ipaf p. 582-7. in Pseudomonas aeruginosa-induced 144. Ma, Y., et al., Autophagy and cellular caspase-1 activation. Eur J Immunol, immune responses. Immunity, 2013. 2007. 37(11): p. 3030-9. 39(2): p. 211-27. 134. Amer, A., et al., Regulation of 145. Zhang, Q., et al., Circulating Legionella phagosome maturation mitochondrial DAMPs cause and infection through flagellin and inflammatory responses to injury. host Ipaf. J Biol Chem, 2006. 281(46): Nature, 2010. 464(7285): p. 104-7. p. 35217-23. 146. Edme, N., et al., Ras induces NBT-II 135. Muruve, D.A., et al., The epithelial cell scattering through the inflammasome recognizes cytosolic coordinate activities of Rac and MAPK microbial and host DNA and triggers pathways. J Cell Sci, 2002. 115(Pt 12): an innate immune response. Nature, p. 2591-601. 2008. 452(7183): p. 103-7. 147. Salcedo, R., et al., MyD88 and its 136. Kanneganti, T.D., et al., Critical role divergent toll in carcinogenesis. for Cryopyrin/Nalp3 in activation of Trends Immunol, 2013. 34(8): p. 379- caspase-1 in response to viral 89. infection and double-stranded RNA. J 148. Trinchieri, G., Cancer and Biol Chem, 2006. 281(48): p. 36560-8. inflammation: an old intuition with 137. Gross, O., et al., Syk kinase signalling rapidly evolving new concepts. Annu couples to the Nlrp3 inflammasome Rev Immunol, 2012. 30: p. 677-706. for anti-fungal host defence. Nature, 149. Schetter, A.J., N.H. Heegaard, and C.C. 2009. 459(7245): p. 433-6. Harris, Inflammation and cancer: interweaving microRNA, free radical,

201

cytokine and p53 pathways. melanoma. J Clin Oncol, 2009. 27(20): Carcinogenesis, 2010. 31(1): p. 37-49. p. 3330-7. 150. Bayne, L.J., et al., Tumor-derived 159. Makitie, T., et al., Tumor-infiltrating granulocyte-macrophage colony- macrophages (CD68(+) cells) and stimulating factor regulates myeloid prognosis in malignant uveal inflammation and T cell immunity in melanoma. Invest Ophthalmol Vis Sci, pancreatic cancer. Cancer Cell, 2012. 2001. 42(7): p. 1414-21. 21(6): p. 822-35. 160. Balkwill, F.R. and A. Mantovani, 151. Chai, C.Y., et al., Hypoxia-inducible Cancer-related inflammation: factor-1alpha expression correlates common themes and therapeutic with focal macrophage infiltration, opportunities. Semin Cancer Biol, angiogenesis and unfavourable 2012. 22(1): p. 33-40. prognosis in urothelial carcinoma. J 161. Qian, B.Z. and J.W. Pollard, Clin Pathol, 2008. 61(5): p. 658-64. Macrophage diversity enhances 152. Hanada, T., et al., Prognostic value of tumor progression and metastasis. tumor-associated macrophage count Cell, 2010. 141(1): p. 39-51. in human bladder cancer. Int J Urol, 162. Pollard, J.W., Tumour-educated 2000. 7(7): p. 263-9. macrophages promote tumour 153. Leek, R.D., et al., Association of progression and metastasis. Nat Rev macrophage infiltration with Cancer, 2004. 4(1): p. 71-8. angiogenesis and prognosis in 163. DeNardo, D.G., et al., CD4(+) T cells invasive breast carcinoma. Cancer regulate pulmonary metastasis of Res, 1996. 56(20): p. 4625-9. mammary carcinomas by enhancing 154. Tsutsui, S., et al., Macrophage protumor properties of macrophages. infiltration and its prognostic Cancer Cell, 2009. 16(2): p. 91-102. implications in breast cancer: the 164. Coussens, L.M., L. Zitvogel, and A.K. relationship with VEGF expression Palucka, Neutralizing tumor- and microvessel density. Oncol Rep, promoting chronic inflammation: a 2005. 14(2): p. 425-31. magic bullet? Science, 2013. 155. Rao, G., et al., Reciprocal interactions 339(6117): p. 286-91. between tumor-associated 165. Gocheva, V., et al., IL-4 induces macrophages and CD44-positive cathepsin protease activity in tumor- cancer cells via osteopontin/CD44 associated macrophages to promote promote tumorigenicity in colorectal cancer growth and invasion. Genes cancer. Clin Cancer Res, 2013. 19(4): Dev, 2010. 24(3): p. 241-55. p. 785-97. 166. Su, S., et al., A positive feedback loop 156. Erreni, M., A. Mantovani, and P. between mesenchymal-like cancer Allavena, Tumor-associated cells and macrophages is essential to Macrophages (TAM) and breast cancer metastasis. Cancer Cell, Inflammation in Colorectal Cancer. 2014. 25(5): p. 605-20. Cancer Microenviron, 2011. 4(2): p. 167. Lin, E.Y., et al., The macrophage 141-54. growth factor CSF-1 in mammary 157. Hamada, I., et al., Clinical effects of gland development and tumor tumor-associated macrophages and progression. J Mammary Gland Biol dendritic cells on renal cell carcinoma. Neoplasia, 2002. 7(2): p. 147-62. Anticancer Res, 2002. 22(6C): p. 4281- 168. Liu, J., et al., Tumor-associated 4. macrophages recruit CCR6+ 158. Jensen, T.O., et al., Macrophage regulatory T cells and promote the markers in serum and tumor have development of colorectal cancer via prognostic impact in American Joint enhancing CCL20 production in mice. Committee on Cancer stage I/II PLoS One, 2011. 6(4): p. e19495.

202

169. Adeegbe, D.O. and H. Nishikawa, 180. Shortman, K. and W.R. Heath, The Natural and induced T regulatory cells CD8+ dendritic cell subset. Immunol in cancer. Front Immunol, 2013. 4: p. Rev, 2010. 234(1): p. 18-31. 190. 181. Belz, G.T. and S.L. Nutt, 170. Denning, T.L., et al., Lamina propria Transcriptional programming of the macrophages and dendritic cells dendritic cell network. Nat Rev differentially induce regulatory and Immunol, 2012. 12(2): p. 101-13. interleukin 17-producing T cell 182. Valladeau, J. and S. Saeland. responses. Nat Immunol, 2007. 8(10): Cutaneous dendritic cells. in Semin p. 1086-94. Immunol. 2005. Elsevier. 171. Savage, N.D., et al., Human anti- 183. Swiecki, M. and M. Colonna, inflammatory macrophages induce Unraveling the functions of Foxp3+ GITR+ CD25+ regulatory T plasmacytoid dendritic cells during cells, which suppress via membrane- viral infections, autoimmunity, and bound TGFbeta-1. J Immunol, 2008. tolerance. Immunol Rev, 2010. 181(3): p. 2220-6. 234(1): p. 142-62. 172. Oh, S.A. and M.O. Li, TGF-beta: 184. Gabrilovich, D., Mechanisms and guardian of T cell function. J Immunol, functional significance of tumour- 2013. 191(8): p. 3973-9. induced dendritic-cell defects. Nat Rev 173. Ng, T.H., et al., Regulation of adaptive Immunol, 2004. 4(12): p. 941-52. immunity; the role of interleukin-10. 185. Elia, A.R., et al., Human dendritic cells Front Immunol, 2013. 4: p. 129. differentiated in hypoxia down- 174. Sharda, D.R., et al., Regulation of modulate antigen uptake and change macrophage arginase expression and their chemokine expression profile. J tumor growth by the Ron receptor Leukoc Biol, 2008. 84(6): p. 1472-82. tyrosine kinase. J Immunol, 2011. 186. Mancino, A., et al., Divergent effects 187(5): p. 2181-92. of hypoxia on dendritic cell functions. 175. Doedens, A.L., et al., Macrophage Blood, 2008. 112(9): p. 3723-34. expression of hypoxia-inducible 187. Yang, M., et al., HIF-dependent factor-1 alpha suppresses T-cell induction of adenosine receptor A2b function and promotes tumor skews human dendritic cells to a Th2- progression. Cancer Res, 2010. stimulating phenotype under hypoxia. 70(19): p. 7465-75. Immunol Cell Biol, 2010. 88(2): p. 176. Rodriguez, P.C., et al., L-arginine 165-71. consumption by macrophages 188. Novitskiy, S.V., et al., Adenosine modulates the expression of CD3 zeta receptors in regulation of dendritic chain in T lymphocytes. J Immunol, cell differentiation and function. 2003. 171(3): p. 1232-9. Blood, 2008. 112(5): p. 1822-31. 177. Rodriguez, P.C., et al., Arginase I 189. Norian, L.A., et al., Tumor-infiltrating production in the tumor regulatory dendritic cells inhibit CD8+ microenvironment by mature myeloid T cell function via L-arginine cells inhibits T-cell receptor metabolism. Cancer Res, 2009. 69(7): expression and antigen-specific T-cell p. 3086-94. responses. Cancer Res, 2004. 64(16): 190. Lee, J.R., et al., Pattern of recruitment p. 5839-49. of immunoregulatory antigen- 178. Idoyaga, J. and R.M. Steinman, presenting cells in malignant SnapShot: Dendritic Cells. Cell, 2011. melanoma. Lab Invest, 2003. 83(10): 146(4): p. 660-660 e2. p. 1457-66. 179. Liu, K. and M.C. Nussenzweig, Origin 191. Munn, D.H., et al., Expression of and development of dendritic cells. indoleamine 2,3-dioxygenase by Immunol Rev, 2010. 234(1): p. 45-54. plasmacytoid dendritic cells in tumor-

203

draining lymph nodes. J Clin Invest, 202. Gabrilovich, D.I., S. Ostrand- 2004. 114(2): p. 280-90. Rosenberg, and V. Bronte, 192. Baban, B., et al., IDO activates Coordinated regulation of myeloid regulatory T cells and blocks their cells by tumours. Nat Rev Immunol, conversion into Th17-like T cells. J 2012. 12(4): p. 253-68. Immunol, 2009. 183(4): p. 2475-83. 203. Dolcetti, L., et al., Hierarchy of 193. Li, Y.W., et al., Intratumoral immunosuppressive strength among neutrophils: a poor prognostic factor myeloid-derived suppressor cell for hepatocellular carcinoma subsets is determined by GM-CSF. Eur following resection. J Hepatol, 2011. J Immunol, 2010. 40(1): p. 22-35. 54(3): p. 497-505. 204. Mandruzzato, S., et al., IL4Ralpha+ 194. Jensen, H.K., et al., Presence of myeloid-derived suppressor cell intratumoral neutrophils is an expansion in cancer patients. J independent prognostic factor in Immunol, 2009. 182(10): p. 6562-8. localized renal cell carcinoma. J Clin 205. Movahedi, K., et al., Identification of Oncol, 2009. 27(28): p. 4709-17. discrete tumor-induced myeloid- 195. Ilie, M., et al., Predictive clinical derived suppressor cell outcome of the intratumoral CD66b- subpopulations with distinct T cell- positive neutrophil-to-CD8-positive T- suppressive activity. Blood, 2008. cell ratio in patients with resectable 111(8): p. 4233-44. nonsmall cell lung cancer. Cancer, 206. Youn, J.I., et al., Subsets of myeloid- 2012. 118(6): p. 1726-37. derived suppressor cells in tumor- 196. Donskov, F. and H. von der Maase, bearing mice. J Immunol, 2008. Impact of immune parameters on 181(8): p. 5791-802. long-term survival in metastatic renal 207. Haile, L.A., et al., CD49d is a new cell carcinoma. J Clin Oncol, 2006. marker for distinct myeloid-derived 24(13): p. 1997-2005. suppressor cell subpopulations in 197. Sparmann, A. and D. Bar-Sagi, Ras- mice. J Immunol, 2010. 185(1): p. 203- induced interleukin-8 expression plays 10. a critical role in tumor growth and 208. Brandau, S., et al., Myeloid-derived angiogenesis. Cancer Cell, 2004. 6(5): suppressor cells in the peripheral p. 447-58. blood of cancer patients contain a 198. Shojaei, F., et al., Bv8 regulates subset of immature neutrophils with myeloid-cell-dependent tumour impaired migratory properties. J angiogenesis. Nature, 2007. Leukoc Biol, 2011. 89(2): p. 311-7. 450(7171): p. 825-31. 209. Youn, J.I., et al., Characterization of 199. Nozawa, H., C. Chiu, and D. Hanahan, the nature of granulocytic myeloid- Infiltrating neutrophils mediate the derived suppressor cells in tumor- initial angiogenic switch in a mouse bearing mice. J Leukoc Biol, 2012. model of multistage carcinogenesis. 91(1): p. 167-81. Proc Natl Acad Sci U S A, 2006. 210. Srivastava, M.K., et al., Myeloid- 103(33): p. 12493-8. derived suppressor cells inhibit T-cell 200. Houghton, A.M., et al., Neutrophil activation by depleting cystine and elastase-mediated degradation of cysteine. Cancer Res, 2010. 70(1): p. IRS-1 accelerates lung tumor growth. 68-77. Nat Med, 2010. 16(2): p. 219-23. 211. Mazzoni, A., et al., Myeloid 201. Fridlender, Z.G., et al., Polarization of suppressor lines inhibit T cell tumor-associated neutrophil responses by an NO-dependent phenotype by TGF-beta: "N1" versus mechanism. J Immunol, 2002. 168(2): "N2" TAN. Cancer Cell, 2009. 16(3): p. p. 689-95. 183-94. 212. Nagaraj, S., et al., Altered recognition of antigen is a mechanism of CD8+ T

204

cell tolerance in cancer. Nat Med, 223. Pena, J., et al., Natural killer 2007. 13(7): p. 828-35. susceptibility is independent of HLA 213. Hanson, E.M., et al., Myeloid-derived class I antigen expression on cell lines suppressor cells down-regulate L- obtained from human solid tumors. selectin expression on CD4+ and CD8+ Eur J Immunol, 1990. 20(11): p. 2445- T cells. J Immunol, 2009. 183(2): p. 8. 937-44. 224. Leiden, J.M., et al., Susceptibility to 214. Sakuishi, K., et al., Emerging Tim-3 natural killer cell-mediated cytolysis is functions in antimicrobial and tumor independent of the level of target cell immunity. Trends Immunol, 2011. class I HLA expression. J Immunol, 32(8): p. 345-9. 1989. 142(6): p. 2140-7. 215. Pan, P.Y., et al., Immune stimulatory 225. Smyth, M.J., et al., New aspects of receptor CD40 is required for T-cell natural-killer-cell surveillance and suppression and T regulatory cell therapy of cancer. Nat Rev Cancer, activation mediated by myeloid- 2002. 2(11): p. 850-61. derived suppressor cells in cancer. 226. Crowe, N.Y., M.J. Smyth, and D.I. Cancer Res, 2010. 70(1): p. 99-108. Godfrey, A critical role for natural 216. Watanabe, S., et al., Tumor-induced killer T cells in immunosurveillance of CD11b+Gr-1+ myeloid cells suppress T methylcholanthrene-induced cell sensitization in tumor-draining sarcomas. J Exp Med, 2002. 196(1): p. lymph nodes. J Immunol, 2008. 119-27. 181(5): p. 3291-300. 227. Smyth, M.J., et al., Differential tumor 217. Nagaraj, S., et al., Antigen-specific surveillance by natural killer (NK) and CD4(+) T cells regulate function of NKT cells. J Exp Med, 2000. 191(4): p. myeloid-derived suppressor cells in 661-8. cancer via retrograde MHC class II 228. Lantz, O. and A. Bendelac, An signaling. Cancer Res, 2012. 72(4): p. invariant T cell receptor alpha chain is 928-38. used by a unique subset of major 218. Sinha, P., et al., Cross-talk between histocompatibility complex class I- myeloid-derived suppressor cells and specific CD4+ and CD4-8- T cells in macrophages subverts tumor mice and humans. J Exp Med, 1994. immunity toward a type 2 response. J 180(3): p. 1097-106. Immunol, 2007. 179(2): p. 977-83. 229. Wu, D.Y., et al., Cross-presentation of 219. Hu, C.E., et al., Up-regulated myeloid- disialoganglioside GD3 to natural derived suppressor cell contributes to killer T cells. J Exp Med, 2003. 198(1): hepatocellular carcinoma p. 173-81. development by impairing dendritic 230. Van Kaer, L., V.V. Parekh, and L. Wu, cell function. Scand J Gastroenterol, Invariant natural killer T cells: 2011. 46(2): p. 156-64. bridging innate and adaptive 220. Long, E.O., Regulation of immune immunity. Cell Tissue Res, 2011. responses through inhibitory 343(1): p. 43-55. receptors. Annu Rev Immunol, 1999. 231. Moreno, M., et al., IFN-gamma- 17: p. 875-904. producing human invariant NKT cells 221. Vivier, E., et al., Functions of natural promote tumor-associated antigen- killer cells. Nat Immunol, 2008. 9(5): specific cytotoxic T cell responses. J p. 503-10. Immunol, 2008. 181(4): p. 2446-54. 222. Gasser, S., et al., The DNA damage 232. Terabe, M. and J.A. Berzofsky, The pathway regulates innate immune role of NKT cells in tumor immunity. system ligands of the NKG2D Adv Cancer Res, 2008. 101: p. 277- receptor. Nature, 2005. 436(7054): p. 348. 1186-90. 233. Semmling, V., et al., Alternative cross- priming through CCL17-CCR4-

205

mediated attraction of CTLs toward 244. Ling, L., et al., Circulating and tumor- NKT cell-licensed DCs. Nat Immunol, infiltrating mucosal associated 2010. 11(4): p. 313-320. invariant T (MAIT) cells in colorectal 234. Ambrosino, E., et al., Cross-regulation cancer patients. Sci Rep, 2016. 6: p. between type I and type II NKT cells in 20358. regulating tumor immunity: a new 245. Won, E.J., et al., Clinical relevance of immunoregulatory axis. J Immunol, circulating mucosal-associated 2007. 179(8): p. 5126-36. invariant T cell levels and their anti- 235. Berzofsky, J.A. and M. Terabe, A novel cancer activity in patients with immunoregulatory axis of NKT cell mucosal-associated cancer. subsets regulating tumor immunity. Oncotarget, 2016. Cancer Immunol Immunother, 2008. 246. Srivastava, N. and P.K. Srivastava, 57(11): p. 1679-83. Modeling the repertoire of true 236. Le Bourhis, L., et al., Antimicrobial tumor-specific MHC I epitopes in a activity of mucosal-associated human tumor. PLoS One, 2009. 4(7): invariant T cells. Nat Immunol, 2010. p. e6094. 11(8): p. 701-8. 247. Weir, B.A., et al., Characterizing the 237. Walker, L.J., et al., Human MAIT and cancer genome in lung CD8alphaalpha cells develop from a adenocarcinoma. Nature, 2007. pool of type-17 precommitted CD8+ T 450(7171): p. 893-8. cells. Blood, 2012. 119(2): p. 422-33. 248. Antonia, S.J., et al., Combination of 238. Kjer-Nielsen, L., et al., MR1 presents p53 cancer vaccine with microbial vitamin B metabolites to chemotherapy in patients with MAIT cells. Nature, 2012. 491(7426): extensive stage small cell lung cancer. p. 717-23. Clin Cancer Res, 2006. 12(3 Pt 1): p. 239. Reantragoon, R., et al., Structural 878-87. insight into MR1-mediated 249. Bioley, G., et al., HLA class I - recognition of the mucosal associated associated immunodominance affects invariant T cell receptor. J Exp Med, CTL responsiveness to an ESO 2012. 209(4): p. 761-74. recombinant protein tumor antigen 240. Kawachi, I., et al., MR1-restricted V vaccine. Clin Cancer Res, 2009. 15(1): alpha 19i mucosal-associated p. 299-306. invariant T cells are innate T cells in 250. Cloosen, S., et al., Expression of the gut lamina propria that provide a tumor-associated differentiation rapid and diverse cytokine response. J antigens, MUC1 glycoforms and CEA, Immunol, 2006. 176(3): p. 1618-27. in human thymic epithelial cells: 241. Serriari, N.E., et al., Innate mucosal- implications for self-tolerance and associated invariant T (MAIT) cells are tumor therapy. Cancer Res, 2007. activated in inflammatory bowel 67(8): p. 3919-26. diseases. Clin Exp Immunol, 2014. 251. Schietinger, A., M. Philip, and H. 176(2): p. 266-74. Schreiber, Specificity in cancer 242. Chua, W.J., et al., Polyclonal mucosa- immunotherapy. Semin Immunol, associated invariant T cells have 2008. 20(5): p. 276-85. unique innate functions in bacterial 252. Steer, H.J., et al., Harnessing the infection. Infect Immun, 2012. 80(9): immune response to treat cancer. p. 3256-67. Oncogene, 2010. 29(48): p. 6301-13. 243. Leeansyah, E., et al., Activation, 253. Bioley, G., et al., Vaccination with a exhaustion, and persistent decline of recombinant protein encoding the the antimicrobial MR1-restricted tumor-specific antigen NY-ESO-1 MAIT-cell population in chronic HIV-1 elicits an A2/157-165-specific CTL infection. Blood, 2013. 121(7): p. repertoire structurally distinct and of 1124-35. reduced tumor reactivity than that

206

elicited by spontaneous immune 262. Limmer, A., et al., Efficient responses to NY-ESO-1-expressing presentation of exogenous antigen by Tumors. J Immunother, 2009. 32(2): liver endothelial cells to CD8+ T cells p. 161-8. results in antigen-specific T-cell 254. Bundell, C.S., et al., Functional tolerance. Nat Med, 2000. 6(12): p. endogenous cytotoxic T lymphocytes 1348-54. are generated to multiple antigens 263. Basha, G., et al., A CD74-dependent co-expressed by progressing tumors; MHC class I endolysosomal cross- after intra-tumoral IL-2 therapy these presentation pathway. Nat Immunol, effector cells eradicate established 2012. 13(3): p. 237-45. tumors. Cancer Immunol 264. Groothuis, T.A. and J. Neefjes, The Immunother, 2006. 55(8): p. 933-47. many roads to cross-presentation. J 255. Wortzel, R.D., et al., Independent Exp Med, 2005. 202(10): p. 1313-8. immunodominant and 265. Rodriguez, A., et al., Selective immunorecessive tumor-specific transport of internalized antigens to antigens on a malignant tumor: the cytosol for MHC class I antigenic dissection with cytolytic T presentation in dendritic cells. Nat cell clones. J Immunol, 1983. 130(5): Cell Biol, 1999. 1(6): p. 362-8. p. 2461-6. 266. Huang, A.Y., et al., In vivo cross- 256. Nelson, D., C. Bundell, and B. priming of MHC class I-restricted Robinson, In vivo cross-presentation antigens requires the TAP of a soluble protein antigen: kinetics, transporter. Immunity, 1996. 4(4): p. distribution, and generation of 349-55. effector CTL recognizing dominant 267. Cone, R.E., J. Sprent, and J.J. and subdominant epitopes. J Marchalonis, Antigen-binding Immunol, 2000. 165(11): p. 6123-32. specificity of isolated cell-surface 257. Stebbins, C.C., et al., DM-mediated immunoglobulin from thymus cells release of a naturally occurring activated to histocompatibility invariant chain degradation antigens. Proc Natl Acad Sci U S A, intermediate from MHC class II 1972. 69(9): p. 2556-60. molecules. J Immunol, 1996. 157(11): 268. Joly, E. and D. Hudrisier, What is p. 4892-8. trogocytosis and what is its purpose? 258. van Lith, M., et al., Regulation of MHC Nat Immunol, 2003. 4(9): p. 815. class II antigen presentation by 269. Davis, D.M., Intercellular transfer of sorting of recycling HLA-DM/DO and cell-surface proteins is common and class II within the multivesicular body. can affect many stages of an immune J Immunol, 2001. 167(2): p. 884-92. response. Nat Rev Immunol, 2007. 259. Van Hateren, A., et al., The cell 7(3): p. 238-43. biology of major histocompatibility 270. Schiller, C., et al., Tunneling complex class I assembly: towards a nanotubes enable intercellular molecular understanding. Tissue transfer of MHC class I molecules. Antigens, 2010. 76(4): p. 259-75. Hum Immunol, 2013. 74(4): p. 412-6. 260. Kovacsovics-Bankowski, M., et al., 271. Ahmed, K.A., et al., Intercellular Efficient major histocompatibility trogocytosis plays an important role complex class I presentation of in modulation of immune responses. exogenous antigen upon Cell Mol Immunol, 2008. 5(4): p. 261- phagocytosis by macrophages. Proc 9. Natl Acad Sci U S A, 1993. 90(11): p. 272. Rechavi, O., I. Goldstein, and Y. Kloog, 4942-6. Intercellular exchange of proteins: the 261. Heit, A., et al., CpG-DNA aided cross- immune cell habit of sharing. FEBS priming by cross-presenting B cells. J Lett, 2009. 583(11): p. 1792-9. Immunol, 2004. 172(3): p. 1501-7.

207

273. Zheng, R. and S. Shu, Immune keratinocytes. J Invest Dermatol, response to cancer and its regulation 1998. 111(4): p. 645-9. in regional lymph nodes. J Surg Oncol, 284. Acosta-Rodriguez, E.V., et al., Surface 2011. 103(6): p. 550-4. phenotype and antigenic specificity of 274. Dardalhon, V., et al., IL-4 inhibits TGF- human interleukin 17-producing T beta-induced Foxp3+ T cells and, helper memory cells. Nat Immunol, together with TGF-beta, generates IL- 2007. 8(6): p. 639-46. 9+ IL-10+ Foxp3(-) effector T cells. Nat 285. Duhen, T., et al., Production of Immunol, 2008. 9(12): p. 1347-55. interleukin 22 but not interleukin 17 275. Veldhoen, M., et al., Transforming by a subset of human skin-homing growth factor-beta 'reprograms' the memory T cells. Nat Immunol, 2009. differentiation of T helper 2 cells and 10(8): p. 857-63. promotes an interleukin 9-producing 286. Eyerich, S., et al., Th22 cells represent subset. Nat Immunol, 2008. 9(12): p. a distinct human T cell subset involved 1341-6. in epidermal immunity and 276. Purwar, R., et al., Robust tumor remodeling. J Clin Invest, 2009. immunity to melanoma mediated by 119(12): p. 3573-85. interleukin-9-producing T cells. Nat 287. Trifari, S., et al., Identification of a Med, 2012. 18(8): p. 1248-53. human helper T cell population that 277. Lu, Y., et al., Th9 cells promote has abundant production of antitumor immune responses in vivo. interleukin 22 and is distinct from J Clin Invest, 2012. 122(11): p. 4160- T(H)-17, T(H)1 and T(H)2 cells. Nat 71. Immunol, 2009. 10(8): p. 864-71. 278. Faulkner, H., et al., Interleukin-9 288. Yamane, H. and W.E. Paul, Early enhances resistance to the intestinal signaling events that underlie fate nematode Trichuris muris. Infect decisions of naive CD4(+) T cells Immun, 1998. 66(8): p. 3832-40. toward distinct T-helper cell subsets. 279. Schlapbach, C., et al., Human TH9 Immunol Rev, 2013. 252(1): p. 12-23. cells are skin-tropic and have 289. Ueno, H., J. Banchereau, and C.G. autocrine and paracrine Vinuesa, Pathophysiology of T proinflammatory capacity. Sci Transl follicular helper cells in humans and Med, 2014. 6(219): p. 219ra8. mice. Nat Immunol, 2015. 16(2): p. 280. Yao, W., R.S. Tepper, and M.H. 142-52. Kaplan, Predisposition to the 290. Kaplan, M.H., Th9 cells: development of IL-9-secreting T cells differentiation and disease. Immunol in atopic infants. J Allergy Clin Rev, 2013. 252(1): p. 104-15. Immunol, 2011. 128(6): p. 1357-1360 291. Eyerich, K. and S. Eyerich, Th22 cells in e5. allergic disease. Allergo J Int, 2015. 281. Kotake, S., et al., IL-17 in synovial 24(1): p. 1-7. fluids from patients with rheumatoid 292. Constant, S.L. and K. Bottomly, arthritis is a potent stimulator of Induction of Th1 and Th2 CD4+ T cell osteoclastogenesis. J Clin Invest, responses: the alternative 1999. 103(9): p. 1345-52. approaches. Annu Rev Immunol, 282. Langrish, C.L., et al., IL-23 drives a 1997. 15: p. 297-322. pathogenic T cell population that 293. Yamane, H., J. Zhu, and W.E. Paul, induces autoimmune inflammation. J Independent roles for IL-2 and GATA- Exp Med, 2005. 201(2): p. 233-40. 3 in stimulating naive CD4+ T cells to 283. Teunissen, M.B., et al., Interleukin-17 generate a Th2-inducing cytokine and interferon-gamma synergize in environment. J Exp Med, 2005. the enhancement of proinflammatory 202(6): p. 793-804. cytokine production by human 294. Bettelli, E., et al., Reciprocal developmental pathways for the

208

generation of pathogenic effector Proc Natl Acad Sci U S A, 2011. TH17 and regulatory T cells. Nature, 108(33): p. E488-97. 2006. 441(7090): p. 235-8. 305. Chen, L. and D.B. Flies, Molecular 295. Ivanov, II, et al., The orphan nuclear mechanisms of T cell co-stimulation receptor RORgammat directs the and co-inhibition. Nat Rev Immunol, differentiation program of 2013. 13(4): p. 227-42. proinflammatory IL-17+ T helper cells. 306. Greenwald, R.J., G.J. Freeman, and Cell, 2006. 126(6): p. 1121-33. A.H. Sharpe, The B7 family revisited. 296. Mangan, P.R., et al., Transforming Annu Rev Immunol, 2005. 23: p. 515- growth factor-beta induces 48. development of the T(H)17 lineage. 307. Choi, S. and R.H. Schwartz, Molecular Nature, 2006. 441(7090): p. 231-4. mechanisms for adaptive tolerance 297. Veldhoen, M., et al., TGFbeta in the and other T cell anergy models. Semin context of an inflammatory cytokine Immunol, 2007. 19(3): p. 140-52. milieu supports de novo 308. Harrington, L.E., et al., Interleukin 17- differentiation of IL-17-producing T producing CD4+ effector T cells cells. Immunity, 2006. 24(2): p. 179- develop via a lineage distinct from the 89. T helper type 1 and 2 lineages. Nat 298. Turner, M.S., L.P. Kane, and P.A. Immunol, 2005. 6(11): p. 1123-32. Morel, Dominant role of antigen dose 309. Mosmann, T.R., et al., Differentiation in CD4+Foxp3+ regulatory T cell of subsets of CD4+ and CD8+ T cells. induction and expansion. J Immunol, Ciba Found Symp, 1995. 195: p. 42- 2009. 183(8): p. 4895-903. 50; discussion 50-4. 299. Molinero, L.L., et al., High TCR stimuli 310. Murphy, K., et al., Janeway's prevent induced regulatory T cell immunobiology. 7th ed. 2008, New differentiation in a NF-kappaB- York: Garland Science. xxi, 887 p. dependent manner. J Immunol, 2011. 311. Janeway Jr, C.A., et al., T cell- 186(8): p. 4609-17. mediated cytotoxicity. 2001. 300. Iezzi, G., et al., CD40-CD40L cross-talk 312. Sallusto, F., et al., Two subsets of integrates strong antigenic signals memory T lymphocytes with distinct and microbial stimuli to induce homing potentials and effector development of IL-17-producing CD4+ functions. Nature, 1999. 401(6754): T cells. Proc Natl Acad Sci U S A, 2009. p. 708-12. 106(3): p. 876-81. 313. Masopust, D., et al., Preferential 301. Yu, D., et al., The transcriptional localization of effector memory cells repressor Bcl-6 directs T follicular in nonlymphoid tissue. Science, 2001. helper cell lineage commitment. 291(5512): p. 2413-7. Immunity, 2009. 31(3): p. 457-68. 314. Reinhardt, R.L., et al., Visualizing the 302. Johnston, R.J., et al., Bcl6 and Blimp-1 generation of memory CD4 T cells in are reciprocal and antagonistic the whole body. Nature, 2001. regulators of T follicular helper cell 410(6824): p. 101-5. differentiation. Science, 2009. 315. Gebhardt, T., et al., Memory T cells in 325(5943): p. 1006-10. nonlymphoid tissue that provide 303. Arnold, C.N., et al., The germinal enhanced local immunity during center response is impaired in the infection with herpes simplex virus. absence of T cell-expressed CXCR5. Nat Immunol, 2009. 10(5): p. 524-30. Eur J Immunol, 2007. 37(1): p. 100-9. 316. Masopust, D., et al., Dynamic T cell 304. Bentebibel, S.E., et al., Human tonsil migration program provides resident B-cell lymphoma 6 (BCL6)-expressing memory within intestinal epithelium. CD4+ T-cell subset specialized for B- J Exp Med, 2010. 207(3): p. 553-64. cell help outside germinal centers. 317. Veiga-Fernandes, H., et al., Response of naive and memory CD8+ T cells to

209

antigen stimulation in vivo. Nat 327. Sharma, M.D., et al., Plasmacytoid Immunol, 2000. 1(1): p. 47-53. dendritic cells from mouse tumor- 318. Zimmermann, C., et al., Kinetics of the draining lymph nodes directly activate response of naive and memory CD8 T mature Tregs via indoleamine 2,3- cells to antigen: similarities and dioxygenase. J Clin Invest, 2007. differences. Eur J Immunol, 1999. 117(9): p. 2570-82. 29(1): p. 284-90. 328. Fallarino, F., et al., The combined 319. Mellor, A.L. and D.H. Munn, Creating effects of tryptophan starvation and immune privilege: active local tryptophan catabolites down- suppression that benefits friends, but regulate T cell receptor zeta-chain protects foes. Nat Rev Immunol, and induce a regulatory phenotype in 2008. 8(1): p. 74-80. naive T cells. J Immunol, 2006. 320. Zitvogel, L., A. Tesniere, and G. 176(11): p. 6752-61. Kroemer, Cancer despite 329. Peggs, K.S., et al., Blockade of CTLA-4 immunosurveillance: on both effector and regulatory T cell immunoselection and compartments contributes to the immunosubversion. Nat Rev antitumor activity of anti-CTLA-4 Immunol, 2006. 6(10): p. 715-27. antibodies. J Exp Med, 2009. 206(8): 321. Inman, B.A., et al., PD-L1 (B7-H1) p. 1717-25. expression by urothelial carcinoma of 330. Serafini, P., I. Borrello, and V. Bronte, the bladder and BCG-induced Myeloid suppressor cells in cancer: granulomata: associations with recruitment, phenotype, properties, localized stage progression. Cancer, and mechanisms of immune 2007. 109(8): p. 1499-505. suppression. Semin Cancer Biol, 2006. 322. Dong, H. and L. Chen, B7-H1 pathway 16(1): p. 53-65. and its role in the evasion of tumor 331. Bronte, V., et al., L-arginine immunity. J Mol Med (Berl), 2003. metabolism in myeloid cells controls 81(5): p. 281-7. T-lymphocyte functions. Trends 323. Derre, L., et al., Expression and Immunol, 2003. 24(6): p. 302-6. release of HLA-E by melanoma cells 332. Zajac, A.J., et al., Viral immune and melanocytes: potential impact on evasion due to persistence of the response of cytotoxic effector activated T cells without effector cells. J Immunol, 2006. 177(5): p. function. J Exp Med, 1998. 188(12): p. 3100-7. 2205-13. 324. McKallip, R., R. Li, and S. Ladisch, 333. Gallimore, A., et al., Induction and Tumor gangliosides inhibit the tumor- exhaustion of lymphocytic specific immune response. J Immunol, choriomeningitis virus-specific 1999. 163(7): p. 3718-26. cytotoxic T lymphocytes visualized 325. Kawamura, K., et al., Secretion of using soluble tetrameric major interleukin-10 from murine colon histocompatibility complex class I- carcinoma cells suppresses systemic peptide complexes. J Exp Med, 1998. antitumor immunity and impairs 187(9): p. 1383-93. protective immunity induced against 334. Wherry, E.J., T cell exhaustion. Nat the tumors. Cancer Gene Ther, 2002. Immunol, 2011. 12(6): p. 492-9. 9(1): p. 109-15. 335. Wherry, E.J., et al., Viral persistence 326. Gabrilovich, D.I., et al., Antibodies to alters CD8 T-cell immunodominance vascular endothelial growth factor and tissue distribution and results in enhance the efficacy of cancer distinct stages of functional immunotherapy by improving impairment. J Virol, 2003. 77(8): p. endogenous dendritic cell function. 4911-27. Clin Cancer Res, 1999. 5(10): p. 2963- 336. Wherry, E.J., et al., Antigen- 70. independent memory CD8 T cells do

210

not develop during chronic viral immunity. Cancer Res, 2005. 65(3): p. infection. Proc Natl Acad Sci U S A, 1089-96. 2004. 101(45): p. 16004-9. 347. Iwai, Y., et al., Involvement of PD-L1 337. Shin, H., et al., Viral antigen and on tumor cells in the escape from host extensive division maintain virus- immune system and tumor specific CD8 T cells during chronic immunotherapy by PD-L1 blockade. infection. J Exp Med, 2007. 204(4): p. Proc Natl Acad Sci U S A, 2002. 99(19): 941-9. p. 12293-7. 338. Obar, J.J., et al., Gammaherpesvirus 348. Staron, M.M., et al., The transcription persistence alters key CD8 T-cell factor FoxO1 sustains expression of memory characteristics and enhances the inhibitory receptor PD-1 and antiviral protection. J Virol, 2006. survival of antiviral CD8(+) T cells 80(17): p. 8303-15. during chronic infection. Immunity, 339. Muenst, S., et al., The presence of 2014. 41(5): p. 802-14. programmed death 1 (PD-1)-positive 349. Shin, H., et al., A role for the tumor-infiltrating lymphocytes is transcriptional repressor Blimp-1 in associated with poor prognosis in CD8(+) T cell exhaustion during human breast cancer. Breast Cancer chronic viral infection. Immunity, Res Treat, 2013. 139(3): p. 667-76. 2009. 31(2): p. 309-20. 340. Zhang, Y., et al., Programmed death-1 350. Oestreich, K.J., et al., NFATc1 upregulation is correlated with regulates PD-1 expression upon T cell dysfunction of tumor-infiltrating activation. J Immunol, 2008. 181(7): CD8+ T lymphocytes in human non- p. 4832-9. small cell lung cancer. Cell Mol 351. Kao, C., et al., Transcription factor T- Immunol, 2010. 7(5): p. 389-95. bet represses expression of the 341. Matsuzaki, J., et al., Tumor-infiltrating inhibitory receptor PD-1 and sustains NY-ESO-1-specific CD8+ T cells are virus-specific CD8+ T cell responses negatively regulated by LAG-3 and during chronic infection. Nat PD-1 in human ovarian cancer. Proc Immunol, 2011. 12(7): p. 663-71. Natl Acad Sci U S A, 2010. 107(17): p. 352. Chen, K., et al., Immunoglobulin D 7875-80. enhances immune surveillance by 342. Sfanos, K.S., et al., Human prostate- activating antimicrobial, infiltrating CD8+ T lymphocytes are proinflammatory and B cell– oligoclonal and PD-1+. Prostate, stimulating programs in basophils. 2009. 69(15): p. 1694-703. Nature immunology, 2009. 10(8): p. 343. Fourcade, J., et al., PD-1 is a regulator 889-898. of NY-ESO-1-specific CD8+ T cell 353. Mizoguchi, A., et al., Chronic expansion in melanoma patients. J intestinal inflammatory condition Immunol, 2009. 182(9): p. 5240-9. generates IL-10-producing regulatory 344. Ahmadzadeh, M., et al., Tumor B cell subset characterized by CD1d antigen-specific CD8 T cells upregulation. Immunity, 2002. 16(2): infiltrating the tumor express high p. 219-30. levels of PD-1 and are functionally 354. Fillatreau, S., et al., B cells regulate impaired. Blood, 2009. 114(8): p. autoimmunity by provision of IL-10. 1537-44. Nat Immunol, 2002. 3(10): p. 944-50. 345. Barber, D.L., et al., Restoring function 355. Olkhanud, P.B., et al., Tumor-evoked in exhausted CD8 T cells during regulatory B cells promote breast chronic viral infection. Nature, 2006. cancer metastasis by converting 439(7077): p. 682-7. resting CD4(+) T cells to T-regulatory 346. Hirano, F., et al., Blockade of B7-H1 cells. Cancer Res, 2011. 71(10): p. and PD-1 by monoclonal antibodies 3505-15. potentiates cancer therapeutic

211

356. Mauri, C., et al., Prevention of arthritis 370. Leach, F.S., et al., Mutations of a mutS by interleukin 10-producing B cells. J homolog in hereditary nonpolyposis Exp Med, 2003. 197(4): p. 489-501. colorectal cancer. Cell, 1993. 75(6): p. 357. Duddy, M.E., A. Alter, and A. Bar-Or, 1215-25. Distinct profiles of human B cell 371. Boland, C.R., et al., The biochemical effector cytokines: a role in immune basis of microsatellite instability and regulation? J Immunol, 2004. 172(6): abnormal immunohistochemistry and p. 3422-7. clinical behavior in Lynch syndrome: 358. Xiao, S., et al., Tim-1 is essential for from bench to bedside. Fam Cancer, induction and maintenance of IL-10 in 2008. 7(1): p. 41-52. regulatory B cells and their regulation 372. Sampson, J.R., et al., Autosomal of tissue inflammation. J Immunol, recessive colorectal adenomatous 2015. 194(4): p. 1602-8. polyposis due to inherited mutations 359. Stewart, B.W., et al., World cancer of MYH. Lancet, 2003. 362(9377): p. report 2014. xiv, 630 pages. 39-41. 360. Ministry of Health, Cancer: New 373. Sieber, O.M., et al., Multiple registrations and deaths 2008. 2011, colorectal adenomas, classic Ministry of Health: Wellington. adenomatous polyposis, and germ- 361. Shah, A.B., et al., Trends in colorectal line mutations in MYH. N Engl J Med, cancer incidence rates in New 2003. 348(9): p. 791-9. Zealand, 1981-2004. ANZ J Surg, 374. Mitchell, R.J., et al., Mismatch repair 2012. 82(4): p. 258-64. genes hMLH1 and hMSH2 and 362. Jackson, C., et al., The PIPER Project. colorectal cancer: a HuGE review. Am 2015, Health Research Council of New J Epidemiol, 2002. 156(10): p. 885- Zealand. 902. 363. Maringe, C., et al., Stage at diagnosis 375. Nystrom-Lahti, M., et al., Close and colorectal cancer survival in six linkage to chromosome 3p and high-income countries: a population- conservation of ancestral founding based study of patients diagnosed haplotype in hereditary nonpolyposis during 2000-2007. Acta Oncol, 2013. colorectal cancer families. Proc Natl 52(5): p. 919-32. Acad Sci U S A, 1994. 91(13): p. 6054- 364. Rustgi, A.K., The genetics of 8. hereditary colon cancer. Genes Dev, 376. Lindblom, A., et al., Genetic mapping 2007. 21(20): p. 2525-38. of a second locus predisposing to 365. Galiatsatos, P. and W.D. Foulkes, hereditary non-polyposis colon Familial adenomatous polyposis. Am J cancer. Nat Genet, 1993. 5(3): p. 279- Gastroenterol, 2006. 101(2): p. 385- 82. 98. 377. Peltomaki, P., et al., Genetic mapping 366. Segditsas, S. and I. Tomlinson, of a locus predisposing to human Colorectal cancer and genetic colorectal cancer. Science, 1993. alterations in the Wnt pathway. 260(5109): p. 810-2. Oncogene, 2006. 25(57): p. 7531-7. 378. Polakis, P., Wnt signaling and cancer. 367. Polakis, P., The many ways of Wnt in Genes Dev, 2000. 14(15): p. 1837-51. cancer. Curr Opin Genet Dev, 2007. 379. Cadigan, K.M. and R. Nusse, Wnt 17(1): p. 45-51. signaling: a common theme in animal 368. Kinzler, K.W. and B. Vogelstein, development. Genes Dev, 1997. Lessons from hereditary colorectal 11(24): p. 3286-305. cancer. Cell, 1996. 87(2): p. 159-70. 380. Fodde, R., R. Smits, and H. Clevers, 369. Aaltonen, L.A., et al., Clues to the APC, signal transduction and genetic pathogenesis of familial colorectal instability in colorectal cancer. Nat cancer. Science, 1993. 260(5109): p. Rev Cancer, 2001. 1(1): p. 55-67. 812-6.

212

381. Korinek, V., et al., Constitutive colon cancers. Cancer Res, 2005. transcriptional activation by a beta- 65(14): p. 6063-9. catenin-Tcf complex in APC-/- colon 391. Di Nicolantonio, F., et al., Wild-type carcinoma. Science, 1997. 275(5307): BRAF is required for response to p. 1784-7. panitumumab or cetuximab in 382. Morin, P.J., et al., Activation of beta- metastatic colorectal cancer. J Clin catenin-Tcf signaling in colon cancer Oncol, 2008. 26(35): p. 5705-12. by mutations in beta-catenin or APC. 392. Roth, A.D., et al., Prognostic role of Science, 1997. 275(5307): p. 1787-90. KRAS and BRAF in stage II and III 383. Tetsu, O. and F. McCormick, Beta- resected colon cancer: results of the catenin regulates expression of cyclin translational study on the PETACC-3, D1 in colon carcinoma cells. Nature, EORTC 40993, SAKK 60-00 trial. J Clin 1999. 398(6726): p. 422-6. Oncol, 2010. 28(3): p. 466-74. 384. He, T.C., et al., Identification of c-MYC 393. Berg, M. and K. Soreide, EGFR and as a target of the APC pathway. downstream genetic alterations in Science, 1998. 281(5382): p. 1509-12. KRAS/BRAF and PI3K/AKT pathways 385. You, Z., et al., Wnt signaling promotes in colorectal cancer: implications for oncogenic transformation by targeted therapy. Discov Med, 2012. inhibiting c-Myc-induced apoptosis. J 14(76): p. 207-14. Cell Biol, 2002. 157(3): p. 429-40. 394. Cabrera, C.M. and M.A. Lopez-Nevot, 386. Morin, P.J., B. Vogelstein, and K.W. APC and chromosome instability in Kinzler, Apoptosis and APC in colorectal cancer. Rev Esp Enferm Dig, colorectal tumorigenesis. Proc Natl 2005. 97(10): p. 738-43. Acad Sci U S A, 1996. 93(15): p. 7950- 395. Zhang, W.M., F. Yang, and Q.M. Zhan, 4. Dynamic analysis of DNA damage 387. Allegra, C.J., et al., American Society induced miRNAs in colon cancer cells. of Clinical Oncology provisional Chinese Science Bulletin, 2014. clinical opinion: testing for KRAS gene 59(19): p. 2254-2265. mutations in patients with metastatic 396. Nosho, K., et al., PIK3CA mutation in colorectal carcinoma to predict colorectal cancer: relationship with response to anti-epidermal growth genetic and epigenetic alterations. factor receptor monoclonal antibody Neoplasia, 2008. 10(6): p. 534-41. therapy. J Clin Oncol, 2009. 27(12): p. 397. Barault, L., et al., Mutations in the 2091-6. RAS-MAPK, PI(3)K 388. De Roock, W., et al., Effects of KRAS, (phosphatidylinositol-3-OH kinase) BRAF, NRAS, and PIK3CA mutations signaling network correlate with poor on the efficacy of cetuximab plus survival in a population-based series chemotherapy in chemotherapy- of colon cancers. Int J Cancer, 2008. refractory metastatic colorectal 122(10): p. 2255-9. cancer: a retrospective consortium 398. Zhao, L. and P.K. Vogt, Helical domain analysis. Lancet Oncol, 2010. 11(8): p. and kinase domain mutations in 753-62. p110alpha of phosphatidylinositol 3- 389. Samowitz, W.S., et al., Relationship of kinase induce gain of function by Ki-ras mutations in colon cancers to different mechanisms. Proc Natl Acad tumor location, stage, and survival: a Sci U S A, 2008. 105(7): p. 2652-7. population-based study. Cancer 399. Goel, A., et al., Frequent inactivation Epidemiol Biomarkers Prev, 2000. of PTEN by promoter 9(11): p. 1193-7. hypermethylation in microsatellite 390. Samowitz, W.S., et al., Poor survival instability-high sporadic colorectal associated with the BRAF V600E cancers. Cancer Res, 2004. 64(9): p. mutation in microsatellite-stable 3014-21.

213

400. Benford, D., et al., Application of the 411. Barbir, A., et al., Effects of phenotypes Margin of Exposure (MOE) approach in heterocyclic aromatic amine (HCA) to substances in food that are metabolism-related genes on the genotoxic and carcinogenic. Food association of HCA intake with the risk Chem Toxicol, 2010. 48 Suppl 1: p. S2- of colorectal adenomas. Cancer 24. Causes Control, 2012. 23(9): p. 1429- 401. Harris, K.L., et al., Bioaccessibility of 42. polycyclic aromatic hydrocarbons: 412. Cross, A.J., et al., A large prospective relevance to toxicity and study of meat consumption and carcinogenesis. Expert Opin Drug colorectal cancer risk: an Metab Toxicol, 2013. 9(11): p. 1465- investigation of potential 80. mechanisms underlying this 402. Ma, Y. and S. Harrad, Spatiotemporal association. Cancer Res, 2010. 70(6): analysis and human exposure p. 2406-14. assessment on polycyclic aromatic 413. Fu, Z., et al., Association of meat hydrocarbons in indoor air, settled intake and meat-derived mutagen house dust, and diet: A review. exposure with the risk of colorectal Environ Int, 2015. 84: p. 7-16. polyps by histologic type. Cancer Prev 403. Phillips, D.H., Polycyclic aromatic Res (Phila), 2011. 4(10): p. 1686-97. hydrocarbons in the diet. Mutat Res, 414. Gilsing, A.M., et al., Meat-related 1999. 443(1-2): p. 139-47. mutagen exposure, xenobiotic 404. Garcia-Falcon, M.S. and J. Simal- metabolizing gene polymorphisms Gandara, Determination of polycyclic and the risk of advanced colorectal aromatic hydrocarbons in alcoholic adenoma and cancer. Carcinogenesis, drinks and the identification of their 2012. 33(7): p. 1332-9. potential sources. Food Addit 415. Ollberding, N.J., et al., Meat Contam, 2005. 22(9): p. 791-7. consumption, heterocyclic amines 405. EFSA, Annual Report 2008. 2008, and colorectal cancer risk: the European Food Safety Authority: Multiethnic Cohort Study. Int J Italy. Cancer, 2012. 131(7): p. E1125-33. 406. Cross, A.J. and R. Sinha, Meat-related 416. Voutsinas, J., et al., Heterocyclic mutagens/carcinogens in the etiology amine intake, smoking, cytochrome of colorectal cancer. Environ Mol P450 1A2 and N-acetylation Mutagen, 2004. 44(1): p. 44-55. phenotypes, and risk of colorectal 407. Sinha, R. and N. Rothman, Exposure adenoma in a multiethnic population. assessment of heterocyclic amines Gut, 2013. 62(3): p. 416-22. (HCAs) in epidemiologic studies. 417. Hein, D.W., Molecular genetics and Mutat Res, 1997. 376(1-2): p. 195- function of NAT1 and NAT2: role in 202. aromatic amine metabolism and 408. Gooderham, N.J., et al., Food-derived carcinogenesis. Mutat Res, 2002. 506- heterocyclic amine mutagens: 507: p. 65-77. variable metabolism and significance 418. Hein, D.W., et al., Molecular genetics to humans. Drug Metab Dispos, 2001. and epidemiology of the NAT1 and 29(4 Pt 2): p. 529-34. NAT2 acetylation polymorphisms. 409. Sugimura, T., Overview of Cancer Epidemiol Biomarkers Prev, carcinogenic heterocyclic amines. 2000. 9(1): p. 29-42. Mutat Res, 1997. 376(1-2): p. 211-9. 419. Hipsley, E.H., Dietary "fibre" and 410. Sugimura, T., et al., Heterocyclic pregnancy toxaemia. Br Med J, 1953. amines: Mutagens/carcinogens 2(4833): p. 420-2. produced during cooking of meat and 420. Burkitt, D.P., A.R. Walker, and N.S. fish. Cancer Sci, 2004. 95(4): p. 290-9. Painter, Effect of dietary fibre on stools and the transit-times, and its

214

role in the causation of disease. adenomas: a systematic review and Lancet, 1972. 2(7792): p. 1408-12. meta-analysis of epidemiological 421. Painter, N.S. and D.P. Burkitt, studies. Cancer Causes Control, 2013. Diverticular disease of the colon, a 24(4): p. 611-27. 20th century problem. Clin 432. Santarelli, R.L., F. Pierre, and D.E. Gastroenterol, 1975. 4(1): p. 3-21. Corpet, Processed meat and 422. Trowell, H., et al., Letter: Dietary fibre colorectal cancer: a review of redefined. Lancet, 1976. 1(7966): p. epidemiologic and experimental 967. evidence. Nutr Cancer, 2008. 60(2): p. 423. Zhu, Y., et al., Gut microbiota and 131-44. probiotics in colon tumorigenesis. 433. Perrin, P., et al., Only fibres promoting Cancer Lett, 2011. 309(2): p. 119-27. a stable butyrate producing colonic 424. Van Der Kamp, J.W., et al., Dietary ecosystem decrease the rate of Fibre. 2010: Wageningen Academic aberrant crypt foci in rats. Gut, 2001. Publishers. 48(1): p. 53-61. 425. Bosscher, D., et al., Food-based 434. Abell, G.C., et al., Phylotypes related strategies to modulate the to Ruminococcus bromii are abundant composition of the intestinal in the large bowel of humans and microbiota and their associated increase in response to a diet high in health effects. J Physiol Pharmacol, resistant starch. FEMS Microbiol Ecol, 2009. 60 Suppl 6: p. 5-11. 2008. 66(3): p. 505-15. 426. Gibson, G.R., et al., Selective 435. Kim, E.J., et al., Fucoidan present in stimulation of bifidobacteria in the brown algae induces apoptosis of human colon by oligofructose and human colon cancer cells. BMC inulin. Gastroenterology, 1995. Gastroenterol, 2010. 10: p. 96. 108(4): p. 975-82. 436. Chung, Y.C. and Y.F. Chang, Serum 427. Chen, H.M., et al., Decreased dietary interleukin-6 levels reflect the disease fiber intake and structural alteration status of colorectal cancer. J Surg of gut microbiota in patients with Oncol, 2003. 83(4): p. 222-6. advanced colorectal adenoma. Am J 437. Calviello, G., et al., n-3 PUFAs reduce Clin Nutr, 2013. 97(5): p. 1044-52. VEGF expression in human colon 428. Lattimer, J.M. and M.D. Haub, Effects cancer cells modulating the COX- of dietary fiber and its components on 2/PGE2 induced ERK-1 and -2 and HIF- metabolic health. Nutrients, 2010. 1alpha induction pathway. 2(12): p. 1266-89. Carcinogenesis, 2004. 25(12): p. 429. Mitomi, T., et al., Randomized, 2303-10. controlled study on adjuvant 438. Des Guetz, G., et al., Microvessel immunochemotherapy with PSK in density and VEGF expression are curatively resected colorectal cancer. prognostic factors in colorectal The Cooperative Study Group of cancer. Meta-analysis of the Surgical Adjuvant literature. Br J Cancer, 2006. 94(12): Immunochemotherapy for Cancer of p. 1823-32. Colon and Rectum (Kanagawa). Dis 439. Houck, K.A., et al., The vascular Colon Rectum, 1992. 35(2): p. 123-30. endothelial growth factor family: 430. Ye, S. and X. Zhou, Purification, initial identification of a fourth molecular characterization and immune species and characterization of activities of polysaccharides from the alternative splicing of RNA. Mol fungus, Polyporus umbellatus. Food Endocrinol, 1991. 5(12): p. 1806-14. Science and Human Wellness, 2014. 440. Ladomery, M.R., S.J. Harper, and D.O. 3(2): p. 73-78. Bates, Alternative splicing in 431. Aune, D., et al., Red and processed angiogenesis: the vascular meat intake and risk of colorectal

215

endothelial growth factor paradigm. 450. Varey, A.H., et al., VEGF 165 b, an Cancer Lett, 2007. 249(2): p. 133-42. antiangiogenic VEGF-A isoform, binds 441. Ferrara, N., Role of vascular and inhibits bevacizumab treatment endothelial growth factor in in experimental colorectal carcinoma: physiologic and pathologic balance of pro- and antiangiogenic angiogenesis: therapeutic VEGF-A isoforms has implications for implications. Semin Oncol, 2002. 29(6 therapy. Br J Cancer, 2008. 98(8): p. Suppl 16): p. 10-4. 1366-79. 442. Neufeld, G., et al., Vascular 451. Pritchard-Jones, R.O., et al., endothelial growth factor (VEGF) and Expression of VEGF(xxx)b, the its receptors. FASEB J, 1999. 13(1): p. inhibitory isoforms of VEGF, in 9-22. malignant melanoma. Br J Cancer, 443. Ferrara, N. and T. Davis-Smyth, The 2007. 97(2): p. 223-30. biology of vascular endothelial 452. Cebe-Suarez, S., A. Zehnder-Fjallman, growth factor. Endocr Rev, 1997. and K. Ballmer-Hofer, The role of 18(1): p. 4-25. VEGF receptors in angiogenesis; 444. Jingjing, L., et al., Human Muller cells complex partnerships. Cell Mol Life express VEGF183, a novel spliced Sci, 2006. 63(5): p. 601-15. variant of vascular endothelial 453. Rennel, E.S., et al., Recombinant growth factor. Invest Ophthalmol Vis human VEGF165b protein is an Sci, 1999. 40(3): p. 752-9. effective anti-cancer agent in mice. 445. Poltorak, Z., et al., VEGF145, a Eur J Cancer, 2008. 44(13): p. 1883- secreted vascular endothelial growth 94. factor isoform that binds to 454. Hurwitz, H., et al., Bevacizumab plus extracellular matrix. J Biol Chem, irinotecan, fluorouracil, and 1997. 272(11): p. 7151-8. leucovorin for metastatic colorectal 446. Whittle, C., et al., Heterogeneous cancer. N Engl J Med, 2004. 350(23): vascular endothelial growth factor p. 2335-42. (VEGF) isoform mRNA and receptor 455. Liu, J., et al., IL-17 is associated with mRNA expression in human glomeruli, poor prognosis and promotes and the identification of VEGF148 angiogenesis via stimulating VEGF mRNA, a novel truncated splice production of cancer cells in variant. Clin Sci (Lond), 1999. 97(3): p. colorectal carcinoma. Biochem 303-12. Biophys Res Commun, 2011. 407(2): 447. Woolard, J., et al., VEGF165b, an p. 348-54. inhibitory vascular endothelial 456. Le Gouvello, S., et al., High prevalence growth factor splice variant: of Foxp3 and IL17 in MMR-proficient mechanism of action, in vivo effect on colorectal carcinomas. Gut, 2008. angiogenesis and endogenous protein 57(6): p. 772-9. expression. Cancer Res, 2004. 64(21): 457. Amicarella, F., et al., Dual role of p. 7822-35. tumour-infiltrating T helper 17 cells in 448. Bates, D.O., et al., VEGF165b, an human colorectal cancer. Gut, 2015. inhibitory splice variant of vascular 458. Dukes, C.E., The classification of endothelial growth factor, is down- cancer of the rectum. Journal of regulated in renal cell carcinoma. Pathology and Bacteriology, 1932. Cancer Res, 2002. 62(14): p. 4123-31. 35(3): p. 323-332. 449. Perrin, R.M., et al., Diabetic 459. Turnbull, R.B., Jr., et al., Cancer of the retinopathy is associated with a colon: the influence of the no-touch switch in splicing from anti- to pro- isolation technic on survival rates. angiogenic isoforms of vascular Ann Surg, 1967. 166(3): p. 420-7. endothelial growth factor. 460. Sobin, L.H., M.K. Gospodarowicz, and Diabetologia, 2005. 48(11): p. 2422-7. C. Wittekind, TNM classification of

216

malignant tumours. 2011: John Wiley cancers are associated with the state & Sons. of the local immune reaction. J Clin 461. Quirke, P., et al., Evidence-based Oncol, 2011. 29(6): p. 610-8. medicine: the time has come to set 472. Pages, F., et al., In situ cytotoxic and standards for staging. J Pathol, 2010. memory T cells predict outcome in 221(4): p. 357-60. patients with early-stage colorectal 462. Quirke, P., et al., The future of the cancer. J Clin Oncol, 2009. 27(35): p. TNM staging system in colorectal 5944-51. cancer: time for a debate? Lancet 473. Zlobec, I., et al., TIA-1 cytotoxic Oncol, 2007. 8(7): p. 651-7. granule-associated RNA binding 463. Newland, R.C., et al., Pathologic protein improves the prognostic determinants of survival associated performance of CD8 in mismatch with colorectal cancer with lymph repair-proficient colorectal cancer. node metastases. A multivariate PLoS One, 2010. 5(12): p. e14282. analysis of 579 patients. Cancer, 474. Pages, F., et al., Immune infiltration in 1994. 73(8): p. 2076-82. human tumors: a prognostic factor 464. Compton, C.C., Pathology report in that should not be ignored. colon cancer: what is prognostically Oncogene, 2010. 29(8): p. 1093-102. important? Dig Dis, 1999. 17(2): p. 67- 475. Ascierto, M.L., et al., An immunologic 79. portrait of cancer. J Transl Med, 2011. 465. Schmoll, H.J., et al., ESMO Consensus 9: p. 146. Guidelines for management of 476. Angell, H. and J. Galon, From the patients with colon and rectal cancer. immune contexture to the a personalized approach to clinical Immunoscore: the role of prognostic decision making. Ann Oncol, 2012. and predictive immune markers in 23(10): p. 2479-516. cancer. Curr Opin Immunol, 2013. 466. Ueno, H., et al., Histological grading 25(2): p. 261-7. of colorectal cancer: a simple and 477. Fridman, W.H., et al., The immune objective method. Ann Surg, 2008. contexture in human tumours: impact 247(5): p. 811-8. on clinical outcome. Nat Rev Cancer, 467. Ueno, H., et al., Proposed objective 2012. 12(4): p. 298-306. criteria for "grade 3" in early invasive 478. Galon, J., et al., Towards the colorectal cancer. Am J Clin Pathol, introduction of the 'Immunoscore' in 2010. 134(2): p. 312-22. the classification of malignant 468. Barresi, V., et al., Colorectal tumours. J Pathol, 2014. 232(2): p. carcinoma grading by quantifying 199-209. poorly differentiated cell clusters is 479. Scott, A.M., J.D. Wolchok, and L.J. more reproducible and provides more Old, Antibody therapy of cancer. Nat robust prognostic information than Rev Cancer, 2012. 12(4): p. 278-87. conventional grading. Virchows Arch, 480. Van Cutsem, E., et al., Cetuximab and 2012. 461(6): p. 621-8. chemotherapy as initial treatment for 469. Galon, J., et al., Cancer classification metastatic colorectal cancer. N Engl J using the Immunoscore: a worldwide Med, 2009. 360(14): p. 1408-17. task force. J Transl Med, 2012. 10: p. 481. Scartozzi, M., et al., Dalotuzumab, a 205. recombinant humanized mAb 470. Galon, J., et al., Type, density, and targeted against IGFR1 for the location of immune cells within treatment of cancer. Curr Opin Mol human colorectal tumors predict Ther, 2010. 12(3): p. 361-71. clinical outcome. Science, 2006. 482. Hudis, C.A., Trastuzumab-- 313(5795): p. 1960-4. mechanism of action and use in 471. Mlecnik, B., et al., Histopathologic- clinical practice. N Engl J Med, 2007. based prognostic factors of colorectal 357(1): p. 39-51.

217

483. Canadas, I., et al., C-MET as a new 493. Lenz, H.J., et al., Multicenter phase II therapeutic target for the and translational study of cetuximab development of novel anticancer in metastatic colorectal carcinoma drugs. Clin Transl Oncol, 2010. 12(4): refractory to irinotecan, oxaliplatin, p. 253-60. and fluoropyrimidines. J Clin Oncol, 484. Weiner, L.M., R. Surana, and S. Wang, 2006. 24(30): p. 4914-21. Monoclonal antibodies: versatile 494. Cunningham, D., et al., Cetuximab platforms for cancer immunotherapy. monotherapy and cetuximab plus Nat Rev Immunol, 2010. 10(5): p. 317- irinotecan in irinotecan-refractory 27. metastatic colorectal cancer. N Engl J 485. Cheson, B.D. and J.P. Leonard, Med, 2004. 351(4): p. 337-45. Monoclonal antibody therapy for B- 495. Maloney, D.G., et al., IDEC-C2B8 cell non-Hodgkin's lymphoma. N Engl (Rituximab) anti-CD20 monoclonal J Med, 2008. 359(6): p. 613-26. antibody therapy in patients with 486. Chan, A.C. and P.J. Carter, relapsed low-grade non-Hodgkin's Therapeutic antibodies for lymphoma. Blood, 1997. 90(6): p. autoimmunity and inflammation. Nat 2188-95. Rev Immunol, 2010. 10(5): p. 301-16. 496. Karapetis, C.S., et al., K-ras mutations 487. Schliemann, C. and D. Neri, Antibody- and benefit from cetuximab in based vascular tumor targeting. advanced colorectal cancer. N Engl J Recent Results Cancer Res, 2010. 180: Med, 2008. 359(17): p. 1757-65. p. 201-16. 497. Amado, R.G., et al., Wild-type KRAS is 488. Welt, S., et al., Antibody targeting in required for panitumumab efficacy in metastatic colon cancer: a phase I patients with metastatic colorectal study of monoclonal antibody F19 cancer. J Clin Oncol, 2008. 26(10): p. against a cell-surface protein of 1626-34. reactive tumor stromal fibroblasts. J 498. Ministry of Health, Management of Clin Oncol, 1994. 12(6): p. 1193-203. Early Colorectal Cancer. 2011, 489. Scott, A.M., et al., A Phase I dose- Ministry of Health: Wellington. escalation study of sibrotuzumab in 499. Lenschow, D.J., T.L. Walunas, and J.A. patients with advanced or metastatic Bluestone, CD28/B7 system of T cell fibroblast activation protein-positive costimulation. Annu Rev Immunol, cancer. Clin Cancer Res, 2003. 9(5): p. 1996. 14: p. 233-58. 1639-47. 500. Rudd, C.E., A. Taylor, and H. 490. Deckert, P.M., Current constructs and Schneider, CD28 and CTLA-4 targets in clinical development for coreceptor expression and signal antibody-based cancer therapy. Curr transduction. Immunol Rev, 2009. Drug Targets, 2009. 10(2): p. 158-75. 229(1): p. 12-26. 491. Herbertson, R.A., et al., Phase I 501. Schwartz, R.H., Costimulation of T biodistribution and pharmacokinetic lymphocytes: the role of CD28, CTLA- study of Lewis Y-targeting 4, and B7/BB1 in interleukin-2 immunoconjugate CMD-193 in production and immunotherapy. Cell, patients with advanced epithelial 1992. 71(7): p. 1065-8. cancers. Clin Cancer Res, 2009. 502. Linsley, P.S., et al., Human B7-1 15(21): p. 6709-15. (CD80) and B7-2 (CD86) bind with 492. Baselga, J., et al., Recombinant similar avidities but distinct kinetics to humanized anti-HER2 antibody CD28 and CTLA-4 receptors. (Herceptin) enhances the antitumor Immunity, 1994. 1(9): p. 793-801. activity of paclitaxel and doxorubicin 503. Egen, J.G. and J.P. Allison, Cytotoxic T against HER2/neu overexpressing lymphocyte antigen-4 accumulation human breast cancer xenografts. in the immunological synapse is Cancer Res, 1998. 58(13): p. 2825-31.

218

regulated by TCR signal strength. from: Immunity, 2002. 16(1): p. 23-35. https://clinicaltrials.gov/ct2/show/N 504. Parry, R.V., et al., CTLA-4 and PD-1 CT02060188?term=NCT02060188&r receptors inhibit T-cell activation by ank=1 NLM Identifier: NCT02060188. distinct mechanisms. Mol Cell Biol, 514. Chung, K.Y., et al., Phase II study of 2005. 25(21): p. 9543-53. the anti-cytotoxic T-lymphocyte- 505. Riley, J.L., et al., Modulation of TCR- associated antigen 4 monoclonal induced transcriptional profiles by antibody, tremelimumab, in patients ligation of CD28, ICOS, and CTLA-4 with refractory metastatic colorectal receptors. Proc Natl Acad Sci U S A, cancer. J Clin Oncol, 2010. 28(21): p. 2002. 99(18): p. 11790-5. 3485-90. 506. Schneider, H., et al., Reversal of the 515. Greten, T.F. AMP-224, a PD-1 TCR stop signal by CTLA-4. Science, Inhibitor, in Combination With 2006. 313(5795): p. 1972-5. Stereotactic Body Radiation Therapy 507. Schneider, H., et al., Cutting edge: in People With Metastatic Colorectal CTLA-4 (CD152) differentially Cancer. [cited 2016 04/02]; Available regulates mitogen-activated protein from: kinases (extracellular signal- https://clinicaltrials.gov/ct2/show/N regulated kinase and c-Jun N-terminal CT02298946?term=NCT02298946&r kinase) in CD4+ T cells from ank=1 NLM Identifier: NCT02298946. receptor/ligand-deficient mice. J 516. Novartis Pharmaceuticals. Phase I/II Immunol, 2002. 169(7): p. 3475-9. Study of PDR001 in Patients With 508. Tivol, E.A., et al., Loss of CTLA-4 leads Advanced Malignancies. [cited 2016 to massive lymphoproliferation and 04/02]; Available from: fatal multiorgan tissue destruction, https://clinicaltrials.gov/ct2/show/N revealing a critical negative CT02404441?term=NCT02404441&r regulatory role of CTLA-4. Immunity, ank=1 NLM Identifier: NCT02404441. 1995. 3(5): p. 541-7. 517. Regeneron Pharmaceuticals. Study of 509. Waterhouse, P., et al., REGN2810 (Anti-PD-1) in Patients Lymphoproliferative disorders with With Advanced Malignancies. [cited early lethality in mice deficient in Ctla- 2016 04/02]; Available from: 4. Science, 1995. 270(5238): p. 985-8. https://clinicaltrials.gov/ct2/show/N 510. Leach, D.R., M.F. Krummel, and J.P. CT02383212?term=NCT02383212&r Allison, Enhancement of antitumor ank=1 NLM Identifier: NCT02383212. immunity by CTLA-4 blockade. 518. Yang, J. Study of the Safety, Science, 1996. 271(5256): p. 1734-6. Pharmacokinetics and Antitumor 511. Hodi, F.S., et al., Biologic activity of Activities of BGB-A317 in Subjects cytotoxic T lymphocyte-associated With Advanced Tumors. [cited 2016 antigen 4 antibody blockade in 04/02]; Available from: previously vaccinated metastatic https://clinicaltrials.gov/ct2/show/N melanoma and ovarian carcinoma CT02407990?term=NCT02407990&r patients. Proc Natl Acad Sci U S A, ank=1 NLM Identifier: NCT02407990. 2003. 100(8): p. 4712-7. 519. MedImmune LLC. A Phase 1, Open- 512. Hodi, F.S., et al., Improved survival label Study to Evaluate the Safety and with ipilimumab in patients with Tolerability of MEDI0680 (AMP-514) metastatic melanoma. N Engl J Med, in Combination With MEDI4736 in 2010. 363(8): p. 711-23. Subjects With Advanced 513. Bristol-Myers Squibb. A Study of Malignancies. [cited 2016 04/02]; Nivolumab and Nivolumab Plus Available from: Ipilimumab in Recurrent and https://clinicaltrials.gov/ct2/show/N Metastatic Colon Cancer (CheckMate CT02118337?term=NCT02118337&r 142). [cited 2016 04/02]; Available ank=1 NLM Identifier: NCT02118337.

219

520. Nishimura, H., et al., Development of 530. Segal, N. and L. Saltz. Evaluate the lupus-like autoimmune diseases by Efficacy of MEDI4736 in disruption of the PD-1 gene encoding Immunological Subsets of Advanced an ITIM motif-carrying Colorectal Cancer. [cited 2016 immunoreceptor. Immunity, 1999. 04/02]; Available from: 11(2): p. 141-51. https://clinicaltrials.gov/ct2/show/N 521. Keir, M.E., et al., Tissue expression of CT02227667?term=NCT02227667&r PD-L1 mediates peripheral T cell ank=1 NLM Identifier: NCT02227667. tolerance. J Exp Med, 2006. 203(4): p. 531. Genentech Inc. A Phase 1 Study of 883-95. Atezolizumab (an Engineered Anti- 522. Ishida, Y., et al., Induced expression of PDL1 Antibody) in Patients With PD-1, a novel member of the Locally Advanced or Metastatic Solid immunoglobulin gene superfamily, Tumors. [cited 2016 04/02]; Available upon programmed cell death. EMBO from: J, 1992. 11(11): p. 3887-95. https://clinicaltrials.gov/ct2/show/N 523. Taube, J.M., et al., Colocalization of CT01375842?term=NCT01375842&r inflammatory response with B7-h1 ank=1 NLM Identifier: NCT01375842. expression in human melanocytic 532. EMD Serono. Avelumab in Metastatic lesions supports an adaptive or Locally Advanced Solid Tumors resistance mechanism of immune (JAVELIN Solid Tumor). [cited 2016 escape. Sci Transl Med, 2012. 4(127): 04/02]; Available from: p. 127ra37. https://clinicaltrials.gov/ct2/show/N 524. Dong, H., et al., Tumor-associated B7- CT01772004?term=NCT01772004&r H1 promotes T-cell apoptosis: a ank=1 NLM Identifier: NCT01772004. potential mechanism of immune 533. Bristol-Myers Squibb. Safety Study of evasion. Nat Med, 2002. 8(8): p. 793- Anti-LAG-3 With and Without Anti- 800. PD-1 in the Treatment of Solid 525. Blank, C., et al., PD-L1/B7H-1 inhibits Tumors. [cited 2016 04/02]; Available the effector phase of tumor rejection from: by T cell receptor (TCR) transgenic https://clinicaltrials.gov/ct2/show/N CD8+ T cells. Cancer Res, 2004. 64(3): CT01968109?term=NCT01968109&r p. 1140-5. ank=1 NLM Identifier: NCT01968109. 526. Freeman, G.J., et al., Engagement of 534. Novartis Pharmaceuticals. Safety and the PD-1 immunoinhibitory receptor Efficacy of LAG525 Single Agent and by a novel B7 family member leads to in Combination With PDR001 in negative regulation of lymphocyte Patients With Advanced activation. J Exp Med, 2000. 192(7): p. Malignancies. [cited 2016 04/02]; 1027-34. Available from: 527. Latchman, Y., et al., PD-L2 is a second https://clinicaltrials.gov/ct2/show/N ligand for PD-1 and inhibits T cell CT02460224?term=NCT02460224&r activation. Nat Immunol, 2001. 2(3): ank=1 NLM Identifier: NCT02460224. p. 261-8. 535. Goldberg, M.V. and C.G. Drake, LAG-3 528. Shin, T., et al., In vivo costimulatory in Cancer Immunotherapy. Curr Top role of B7-DC in tuning T helper cell 1 Microbiol Immunol, 2011. 344: p. and cytotoxic T lymphocyte 269-78. responses. J Exp Med, 2005. 201(10): 536. Jacobs, J., et al., CD70: An emerging p. 1531-41. target in cancer immunotherapy. 529. Le, D.T., et al., PD-1 Blockade in Pharmacol Ther, 2015. 155: p. 1-10. Tumors with Mismatch-Repair 537. Claus, C., et al., CD27 signaling Deficiency. N Engl J Med, 2015. increases the frequency of regulatory 372(26): p. 2509-20. T cells and promotes tumor growth. Cancer Res, 2012. 72(14): p. 3664-76.

220

538. Jacobs, J., et al., Unlocking the 546. Dimberg, J., A. Hugander, and D. potential of CD70 as a novel Wagsater, Expression of CD137 and immunotherapeutic target for non- CD137 ligand in colorectal cancer small cell lung cancer. Oncotarget, patients. Oncol Rep, 2006. 15(5): p. 2015. 6(15): p. 13462-75. 1197-200. 539. Ryan, M.C., et al., Targeting 547. Chen, S.H., et al., Rejection of pancreatic and ovarian carcinomas disseminated metastases of colon using the auristatin-based anti-CD70 carcinoma by synergism of IL-12 gene antibody-drug conjugate SGN-75. Br J therapy and 4-1BB costimulation. Mol Cancer, 2010. 103(5): p. 676-84. Ther, 2000. 2(1): p. 39-46. 540. Linch, S.N., M.J. McNamara, and W.L. 548. Mazzolini, G., et al., Immunotherapy Redmond, OX40 Agonists and and immunoescape in colorectal Combination Immunotherapy: cancer. World J Gastroenterol, 2007. Putting the Pedal to the Metal. Front 13(44): p. 5822-31. Oncol, 2015. 5: p. 34. 549. Bristol-Myers Squibb. Study of the 541. Cepowicz, D., et al., Assessment of Safety and Tolerability of Urelumab activity of an adhesion molecule Administered in Combination With CD134 and CD137 in colorectal cancer Nivolumab in Solid Tumors and B-cell patients. Pol Przegl Chir, 2011. 83(12): Non-Hodgkins Lymphoma. [cited p. 641-5. 2016 04/02]; Available from: 542. MedImmune LLC. A Phase 1b/2 Safety https://clinicaltrials.gov/ct2/show/N and Tolerability of MEDI6469 in CT02253992?term=NCT02253992&r Combination With Therapeutic ank=1 NLM Identifier: NCT02253992. Immune Agents or Monoclonal 550. Pfizer. A Study of PF-05082566 In Antibodies. [cited 2016 04/02]; Combination With Mogamulizumab Available from: In Patients With Advanced Solid https://clinicaltrials.gov/ct2/show/N Tumors. [cited 2016 04/02]; Available CT02205333?term=NCT02205333&r from: ank=1 NLM Identifier: NCT02205333. https://clinicaltrials.gov/ct2/show/N 543. MedImmune LLC. A Phase 1 Study to CT02444793?term=NCT02444793&r Evaluate MEDI6383 Alone and in ank=1 NLM Identifier: NCT02444793. Combination With MEDI4736 in Adult 551. Schaer, D.A., D. Hirschhorn- Subjects With Select Advanced Solid Cymerman, and J.D. Wolchok, Tumors. [cited 2016 04/02]; Available Targeting tumor-necrosis factor from: receptor pathways for tumor https://clinicaltrials.gov/ct2/show/N immunotherapy. J Immunother CT02221960?term=NCT02221960&r Cancer, 2014. 2: p. 7. ank=1 NLM Identifier: NCT02221960. 552. Pedroza-Gonzalez, A., et al., Activated 544. Genentech Inc. A Dose-Escalation tumor-infiltrating CD4+ regulatory T Study of the Safety and cells restrain antitumor immunity in Pharmacokinetics of MOXR0916 and patients with primary or metastatic Atezolizumab in Patients With Locally liver cancer. Hepatology, 2013. 57(1): Advanced or Metastatic Solid Tumors. p. 183-94. [cited 2016 04/02]; Available from: 553. Schaer, D.A., et al., GITR pathway https://clinicaltrials.gov/ct2/show/N activation abrogates tumor immune CT02410512?term=NCT02410512&r suppression through loss of ank=1 NLM Identifier: NCT02410512. regulatory T cell lineage stability. 545. Vinay, D.S. and B.S. Kwon, 4-1BB Cancer Immunol Res, 2013. 1(5): p. (CD137), an inducible costimulatory 320-31. receptor, as a specific target for 554. GITR Inc. Trial of TRX518 (Anti-GITR cancer therapy. BMB Rep, 2014. mAb) in Stage III or IV Malignant 47(3): p. 122-9. Melanoma or Other Solid Tumors

221

(TRX518-001). [cited 2016 04/02]; quadrivalent human papillomavirus Available from: vaccine (HPV4, Gardasil) for use in https://clinicaltrials.gov/ct2/show/N males and guidance from the CT01239134?term=NCT01239134&r Advisory Committee on Immunization ank=1 NLM Identifier: NCT01239134. Practices (ACIP). MMWR Morb Mortal 555. Merck Sharp & Dohme Corp. Study of Wkly Rep, 2010. 59(20): p. 630-2. MK-4166 and MK-4166 in 562. Bolhassani, A. and S. Rafati, Heat- Combination With Pembrolizumab shock proteins as powerful weapons (MK-3475) in Participants With in vaccine development. Expert Rev Advanced Solid Tumors (MK-4166- Vaccines, 2008. 7(8): p. 1185-99. 001). [cited 2016 04/02]; Available 563. Cereda, V., et al., New gene expressed from: in prostate: a potential target for T https://clinicaltrials.gov/ct2/show/N cell-mediated prostate cancer CT02132754?term=NCT02132754&r immunotherapy. Cancer Immunol ank=1 NLM Identifier: NCT02132754. Immunother, 2010. 59(1): p. 63-71. 556. Thomas, L.J., et al., Targeting human 564. Palena, C., et al., The human T-box CD27 with an agonist antibody mesodermal transcription factor stimulates T-cell activation and Brachyury is a candidate target for T- antitumor immunity. cell-mediated cancer Oncoimmunology, 2014. 3(1): p. immunotherapy. Clin Cancer Res, e27255. 2007. 13(8): p. 2471-8. 557. Celldex Therapeutics. A Dose 565. Thomas, A.M., et al., Mesothelin- Escalation and Cohort Expansion specific CD8(+) T cell responses Study of Anti-CD27 (Varlilumab) and provide evidence of in vivo cross- Anti-PD-1 (Nivolumab) in Advanced priming by antigen-presenting cells in Refractory Solid Tumors. [cited 2016 vaccinated pancreatic cancer 04/02]; Available from: patients. J Exp Med, 2004. 200(3): p. https://clinicaltrials.gov/ct2/show/N 297-306. CT02335918?term=NCT02335918&r 566. So-Rosillo, R. and E.J. Small, ank=1 NLM Identifier: NCT02335918. Sipuleucel-T (APC8015) for prostate 558. Fourcade, J., et al., CD8(+) T cells cancer. Expert Rev Anticancer Ther, specific for tumor antigens can be 2006. 6(9): p. 1163-7. rendered dysfunctional by the tumor 567. Hernandez, A.M., et al., microenvironment through Characterization of the antibody upregulation of the inhibitory response against NeuGcGM3 receptors BTLA and PD-1. Cancer Res, ganglioside elicited in non-small cell 2012. 72(4): p. 887-96. lung cancer patients immunized with 559. Ngiow, S.F., M.W. Teng, and M.J. an anti-idiotype antibody. J Immunol, Smyth, Prospects for TIM3-Targeted 2008. 181(9): p. 6625-34. Antitumor Immunotherapy. Cancer 568. Ai, W.Z., et al., Anti-idiotype antibody Res, 2011. 71(21): p. 6567-71. response after vaccination correlates 560. Centers for Disease, C. and with better overall survival in Prevention, FDA licensure of bivalent follicular lymphoma. Blood, 2009. human papillomavirus vaccine (HPV2, 113(23): p. 5743-6. Cervarix) for use in females and 569. Vergati, M., et al., Strategies for updated HPV vaccination cancer vaccine development. J recommendations from the Advisory Biomed Biotechnol, 2010. 2010. Committee on Immunization 570. Ostrand-Rosenberg, S., Immune Practices (ACIP). MMWR Morb Mortal surveillance: a balance between Wkly Rep, 2010. 59(20): p. 626-9. protumor and antitumor immunity. 561. Centers for Disease, C. and Curr Opin Genet Dev, 2008. 18(1): p. Prevention, FDA licensure of 11-8.

222

571. Kalinski, P., et al., Dendritic cell-based 581. Pasquini, S., et al., Cytokines and therapeutic cancer vaccines: what we costimulatory molecules as genetic have and what we need. Future adjuvants. Immunol Cell Biol, 1997. Oncol, 2009. 5(3): p. 379-90. 75(4): p. 397-401. 572. Lesterhuis, W.J., et al., Dendritic cell- 582. Kyte, J.A. and G. Gaudernack, based vaccines in cancer Immuno-gene therapy of cancer with immunotherapy: an update on clinical tumour-mRNA transfected dendritic and immunological results. Ann cells. Cancer Immunol Immunother, Oncol, 2004. 15 Suppl 4: p. iv145-51. 2006. 55(11): p. 1432-42. 573. Cerundolo, V., I.F. Hermans, and M. 583. Weide, B., et al., Plasmid DNA- and Salio, Dendritic cells: a journey from messenger RNA-based anti-cancer laboratory to clinic. Nat Immunol, vaccination. Immunol Lett, 2008. 2004. 5(1): p. 7-10. 115(1): p. 33-42. 574. Higano, C.S., et al., Integrated data 584. Van Tendeloo, V.F., et al., Highly from 2 randomized, double-blind, efficient gene delivery by mRNA placebo-controlled, phase 3 trials of electroporation in human active cellular immunotherapy with hematopoietic cells: superiority to sipuleucel-T in advanced prostate lipofection and passive pulsing of cancer. Cancer, 2009. 115(16): p. mRNA and to electroporation of 3670-9. plasmid cDNA for tumor antigen 575. Kantoff, P.W., et al., Sipuleucel-T loading of dendritic cells. Blood, 2001. immunotherapy for castration- 98(1): p. 49-56. resistant prostate cancer. N Engl J 585. Ponsaerts, P., V.F. Van Tendeloo, and Med, 2010. 363(5): p. 411-22. Z.N. Berneman, Cancer 576. Best, S.R., et al., Administration of immunotherapy using RNA-loaded HPV DNA vaccine via electroporation dendritic cells. Clin Exp Immunol, elicits the strongest CD8+ T cell 2003. 134(3): p. 378-84. immune responses compared to 586. Yamamoto, A., et al., Current intramuscular injection and prospects for mRNA gene delivery. intradermal gene gun delivery. Eur J Pharm Biopharm, 2009. 71(3): p. Vaccine, 2009. 27(40): p. 5450-9. 484-9. 577. U'Ren, L., R. Kedl, and S. Dow, 587. Bayer, M.E., B.S. Blumberg, and B. Vaccination with liposome--DNA Werner, Particles associated with complexes elicits enhanced antitumor Australia antigen in the sera of immunity. Cancer Gene Ther, 2006. patients with leukaemia, Down's 13(11): p. 1033-44. Syndrome and hepatitis. Nature, 578. Binder, R.J. and P.K. Srivastava, 1968. 218(5146): p. 1057-9. Peptides chaperoned by heat-shock 588. Baumert, T.F., et al., Hepatitis C virus proteins are a necessary and structural proteins assemble into sufficient source of antigen in the viruslike particles in insect cells. J cross-priming of CD8+ T cells. Nat Virol, 1998. 72(5): p. 3827-36. Immunol, 2005. 6(6): p. 593-9. 589. Green, K.Y., et al., Comparison of the 579. Segal, B.H., et al., Heat shock proteins reactivities of baculovirus-expressed as vaccine adjuvants in infections and recombinant Norwalk virus capsid cancer. Drug Discov Today, 2006. antigen with those of the native 11(11-12): p. 534-40. Norwalk virus antigen in serologic 580. Conry, R.M., et al., Safety and assays and some epidemiologic immunogenicity of a DNA vaccine observations. J Clin Microbiol, 1993. encoding carcinoembryonic antigen 31(8): p. 2185-91. and hepatitis B surface antigen in 590. Le Tallec, D., et al., Cervarix, the GSK colorectal carcinoma patients. Clin HPV-16/HPV-18 AS04-adjuvanted Cancer Res, 2002. 8(9): p. 2782-7. cervical cancer vaccine, demonstrates

223

stability upon long-term storage and icosahedrons to rods. FEBS Lett, 2000. under simulated cold chain break 482(3): p. 261-4. conditions. Hum Vaccin, 2009. 5(7): p. 602. Sabara, M., et al., Assembly of double- 467-74. shelled rotaviruslike particles by 591. Pincus, S., et al., Release and Stability simultaneous expression of Testing Programs for a Novel Virus- recombinant VP6 and VP7 proteins. J Like Particle Vaccine. Biopharm Virol, 1991. 65(12): p. 6994-7. International, 2010: p. 26-+. 603. Crawford, S.E., et al., Characterization 592. Shank-Retzlaff, M.L., et al., Evaluation of virus-like particles produced by the of the thermal stability of Gardasil. expression of rotavirus capsid Hum Vaccin, 2006. 2(4): p. 147-54. proteins in insect cells. J Virol, 1994. 593. Keating, G.M. and S. Noble, 68(9): p. 5945-52. Recombinant hepatitis B vaccine 604. Latham, T. and J.M. Galarza, (Engerix-B): a review of its Formation of wild-type and chimeric immunogenicity and protective influenza virus-like particles following efficacy against hepatitis B. Drugs, simultaneous expression of only four 2003. 63(10): p. 1021-51. structural proteins. J Virol, 2001. 594. Bovier, P.A., Epaxal: a virosomal 75(13): p. 6154-65. vaccine to prevent hepatitis A 605. Wiedermann, U., et al., A virosomal infection. Expert Rev Vaccines, 2008. formulated Her-2/neu multi-peptide 7(8): p. 1141-50. vaccine induces Her-2/neu-specific 595. Herzog, C., et al., Eleven years of immune responses in patients with Inflexal V-a virosomal adjuvanted metastatic breast cancer: a phase I influenza vaccine. Vaccine, 2009. study. Breast Cancer Res Treat, 2010. 27(33): p. 4381-7. 119(3): p. 673-83. 596. Romanowski, B., Long term 606. Al-Barwani, F., et al., Antigen delivery protection against cervical infection by virus-like particles for with the human papillomavirus: immunotherapeutic vaccination. Ther review of currently available vaccines. Deliv, 2014. 5(11): p. 1223-40. Hum Vaccin, 2011. 7(2): p. 161-9. 607. Middelberg, A.P., et al., A microbial 597. Donaldson, B., et al., Virus-Like platform for rapid and low-cost virus- Particles, a Versatile Subunit Vaccine like particle and capsomere vaccines. Platform, in Subunit Vaccine Delivery, Vaccine, 2011. 29(41): p. 7154-62. C. Foged, et al., Editors. 2015, 608. Mazeike, E., A. Gedvilaite, and U. Springer: New York. p. 159-180. Blohm, Induction of insert-specific 598. Jiang, X., et al., Baculovirus expression immune response in mice by hamster and antigenic characterization of the polyomavirus VP1 derived virus-like capsid proteins of three Norwalk-like particles carrying LCMV GP33 CTL viruses. Arch Virol, 2002. 147(1): p. epitope. Virus Res, 2012. 163(1): p. 2- 119-30. 10. 599. Storni, T., et al., Critical role for 609. Lasickiene, R., et al., The use of activation of antigen-presenting cells recombinant pseudotype virus-like in priming of cytotoxic T cell responses particles harbouring inserted target after vaccination with virus-like antigen to generate antibodies particles. J Immunol, 2002. 168(6): p. against cellular marker p16INK4A. 2880-6. ScientificWorldJournal, 2012. 2012: 600. Kozlovska, T.M., et al., RNA phage Q p. 263737. beta coat protein as a carrier for 610. Eriksson, M., et al., Murine foreign epitopes. Intervirology, 1996. polyomavirus virus-like particles 39(1-2): p. 9-15. carrying full-length human PSA 601. Cielens, I., et al., Mutilation of RNA protect BALB/c mice from outgrowth phage Qbeta virus-like particles: from

224

of a PSA expressing tumor. PLoS One, by dendritic cells or macrophages 2011. 6(8): p. e23828. prime murine MHC class I-restricted 611. Tegerstedt, K., et al., A single cytotoxic T lymphocytes in vivo. J vaccination with polyomavirus Immunol, 1995. 155(7): p. 3313-21. VP1/VP2Her2 virus-like particles 621. Ruedl, C., et al., Cross-presentation of prevents outgrowth of HER-2/neu- virus-like particles by skin-derived expressing tumors. Cancer Res, 2005. CD8(-) dendritic cells: a dispensable 65(13): p. 5953-7. role for TAP. Eur J Immunol, 2002. 612. Fiers, W., et al., M2e-based universal 32(3): p. 818-25. influenza A vaccine. Vaccine, 2009. 622. Win, S.J., et al., Cross-presentation of 27(45): p. 6280-3. epitopes on virus-like particles via the 613. Gregson, A.L., et al., Phase I trial of an MHC I receptor recycling pathway. alhydrogel adjuvanted hepatitis B Immunol Cell Biol, 2011. 89(6): p. core virus-like particle containing 681-8. epitopes of Plasmodium falciparum 623. Barth, H., et al., Scavenger receptor circumsporozoite protein. PLoS One, class B is required for hepatitis C virus 2008. 3(2): p. e1556. uptake and cross-presentation by 614. Patel, K.G. and J.R. Swartz, Surface human dendritic cells. J Virol, 2008. functionalization of virus-like particles 82(7): p. 3466-79. by direct conjugation using azide- 624. Mizukoshi, F., et al., Activation of HIV- alkyne click chemistry. Bioconjug 1 Gag-specific CD8+ T cells by yeast- Chem, 2011. 22(3): p. 376-87. derived VLP-pulsed dendritic cells is 615. Chackerian, B., M.R. Durfee, and J.T. influenced by the level of mannose on Schiller, Virus-like display of a neo-self the VLP antigen. Microbes Infect, antigen reverses B cell anergy in a B 2009. 11(2): p. 191-7. cell receptor transgenic mouse model. 625. Fausch, S.C., D.M. Da Silva, and W.M. J Immunol, 2008. 180(9): p. 5816-25. Kast, Differential uptake and cross- 616. Jegerlehner, A., et al., A molecular presentation of human assembly system that renders papillomavirus virus-like particles by antigens of choice highly repetitive for dendritic cells and Langerhans cells. induction of protective B cell Cancer Res, 2003. 63(13): p. 3478-82. responses. Vaccine, 2002. 20(25-26): 626. Leclerc, D., et al., Proteasome- p. 3104-12. independent major histocompatibility 617. Manolova, V., et al., Nanoparticles complex class I cross-presentation target distinct dendritic cell mediated by papaya mosaic virus-like populations according to their size. particles leads to expansion of specific Eur J Immunol, 2008. 38(5): p. 1404- human T cells. J Virol, 2007. 81(3): p. 13. 1319-26. 618. McIntosh, J.D., et al., An engineered 627. Coulibaly, F., et al., The birnavirus non-toxic superantigen increases crystal structure reveals structural cross presentation of hepatitis B virus relationships among icosahedral nucleocapsids by human dendritic viruses. Cell, 2005. 120(6): p. 761-72. cells. PLoS One, 2014. 9(4): p. e93598. 628. Garriga, D., et al., The 2.6-Angstrom 619. Lopez-Macias, C., et al., Safety and structure of infectious bursal disease immunogenicity of a virus-like particle virus-derived T=1 particles reveals pandemic influenza A (H1N1) 2009 new stabilizing elements of the virus vaccine in a blinded, randomized, capsid. J Virol, 2006. 80(14): p. 6895- placebo-controlled trial of adults in 905. Mexico. Vaccine, 2011. 29(44): p. 629. Caballero, J.M., et al., Chimeric 7826-34. Infectious Bursal Disease Virus-Like 620. Bohm, W., et al., Exogenous hepatitis Particles as Potent Vaccines for B surface antigen particles processed Eradication of Established HPV-16 E7-

225

Dependent Tumors. PLoS One, 2012. 639. Wang, X., et al., Atomic model of 7(12). rabbit hemorrhagic disease virus by 630. Speiser, D.E., et al., Memory and cryo-electron microscopy and effector CD8 T-cell responses after crystallography. PLoS Pathog, 2013. nanoparticle vaccination of 9(1): p. e1003132. melanoma patients. J Immunother, 640. Boga, J.A., et al., Molecular cloning, 2010. 33(8): p. 848-58. sequencing and expression in 631. Jemon, K., et al., An enhanced Escherichia coli of the capsid protein heterologous virus-like particle for gene from rabbit haemorrhagic human papillomavirus type 16 disease virus (Spanish isolate AST/89). tumour immunotherapy. PLoS One, J Gen Virol, 1994. 75 ( Pt 9): p. 2409- 2013. 8(6): p. e66866. 13. 632. Li, K., et al., Antigen incorporated in 641. Boga, J.A., et al., A single dose virus-like particles is delivered to immunization with rabbit specific dendritic cell subsets that haemorrhagic disease virus major induce an effective antitumor capsid protein produced in immune response in vivo. J Saccharomyces cerevisiae induces Immunother, 2013. 36(1): p. 11-9. protection. J Gen Virol, 1997. 78 ( Pt 633. Peacey, M., et al., Virus-like particles 9): p. 2315-8. from rabbit hemorrhagic disease virus 642. Farnos, O., et al., High-level can induce an anti-tumor response. expression and immunogenic Vaccine, 2008. 26(42): p. 5334-7. properties of the recombinant rabbit 634. Laurent, S., et al., Recombinant rabbit hemorrhagic disease virus VP60 hemorrhagic disease virus capsid capsid protein obtained in Pichia protein expressed in baculovirus self- pastoris. J Biotechnol, 2005. 117(3): assembles into viruslike particles and p. 215-24. induces protection. J Virol, 1994. 643. Castanon, S., et al., Immunization 68(10): p. 6794-8. with potato plants expressing VP60 635. Gromadzka, B., et al., Recombinant protein protects against rabbit VP60 in the form of virion-like hemorrhagic disease virus. J Virol, particles as a potential vaccine 1999. 73(5): p. 4452-5. against rabbit hemorrhagic disease 644. Fernandez-Fernandez, M.R., et al., virus. Acta Biochim Pol, 2006. 53(2): Protection of rabbits against rabbit p. 371-6. hemorrhagic disease virus by 636. Barcena, J., et al., The coat protein of immunization with the VP60 protein Rabbit hemorrhagic disease virus expressed in plants with a potyvirus- contains a molecular switch at the N- based vector. Virology, 2001. 280(2): terminal region facing the inner p. 283-91. surface of the capsid. Virology, 2004. 645. Wilson, S., Vaccine peptide delivery by 322(1): p. 118-34. virus particles. 2007, University of 637. Capucci, L., et al., Antigenicity of the Otago. rabbit hemorrhagic disease virus 646. Da Silva, D.M., et al., Effect of studied by its reactivity with preexisting neutralizing antibodies on monoclonal antibodies. Virus Res, the anti-tumor immune response 1995. 37(3): p. 221-38. induced by chimeric human 638. Laurent, S., et al., Structural, papillomavirus virus-like particle antigenic and immunogenic vaccines. Virology, 2001. 290(2): p. relationships between European 350-60. brown hare syndrome virus and rabbit 647. Peacey, M., et al., Versatile RHDV haemorrhagic disease virus. J Gen virus-like particles: incorporation of Virol, 1997. 78 ( Pt 11): p. 2803-11. antigens by genetic modification and

226

chemical conjugation. Biotechnol 658. Hermans, I.F., et al., The VITAL assay: Bioeng, 2007. 98(5): p. 968-77. a versatile fluorometric technique for 648. Al-Barwani, F., et al., Mannosylation assessing CTL- and NKT-mediated of virus-like particles enhances cytotoxicity against multiple targets internalization by antigen presenting in vitro and in vivo. J Immunol cells. PLoS One, 2014. 9(8): p. Methods, 2004. 285(1): p. 25-40. e104523. 659. Fernandez, M., et al., Genetically 649. McKee, S.J., et al., Virus-like particles engineered vesicular stomatitis virus and alpha-galactosylceramide form a in gene therapy: application for self-adjuvanting composite particle treatment of malignant disease. J that elicits anti-tumor responses. J Virol, 2002. 76(2): p. 895-904. Control Release, 2012. 159(3): p. 338- 660. Routh, A., T. Domitrovic, and J.E. 45. Johnson, Host RNAs, including 650. Hermans, I.F., et al., Impaired ability transposons, are encapsidated by a of MHC class II-/- dendritic cells to eukaryotic single-stranded RNA virus. provide tumor protection is rescued Proc Natl Acad Sci U S A, 2012. 109(6): by CD40 ligation. J Immunol, 1999. p. 1907-12. 163(1): p. 77-81. 661. Bunka, D.H., et al., Degenerate RNA 651. Burgdorf, S., V. Lukacs-Kornek, and C. packaging signals in the genome of Kurts, The mannose receptor Satellite Tobacco Necrosis Virus: mediates uptake of soluble but not of implications for the assembly of a T=1 cell-associated antigen for cross- capsid. J Mol Biol, 2011. 413(1): p. 51- presentation. J Immunol, 2006. 65. 176(11): p. 6770-6. 662. Legendre, D. and J. Fastrez, 652. Al-Barwani, F., Enhancing the Production in functionality of RHDV VLP, PhD Saccharomycescerevisiae of MS2 Thesis. 2015, University of Otago. virus-like particles packaging 653. Pelham, S.J., Coupling the Adjuvant functional heterologous mRNAs. J CpG Oligonucleotides to RHDV VLP. Biotechnol, 2005. 117(2): p. 183-94. 2014, University of Otago. 663. Phelps, J.P., et al., Expression and self- 654. Nakamura, Y. Kazusa Codon Usage assembly of cowpea chlorotic mottle Database. Available from: virus-like particles in Pseudomonas http://www.kazusa.or.jp/codon/. fluorescens. J Biotechnol, 2007. 655. Hitchman, R.B., R.D. Possee, and L.A. 128(2): p. 290-6. King, High-throughput baculovirus 664. Oppenheim, A., et al., Efficient expression in insect cells. Methods introduction of plasmid DNA into Mol Biol, 2012. 824: p. 609-27. human hemopoietic cells by 656. El Mehdaoui, S., et al., Gene transfer encapsidation in simian virus 40 using recombinant rabbit pseudovirions. Proc Natl Acad Sci U S hemorrhagic disease virus capsids A, 1986. 83(18): p. 6925-9. with genetically modified DNA 665. Osterman, J.V., A. Waddell, and H.V. encapsidation capacity by addition of Aposhian, DNA and gene therapy: packaging sequences from the L1 or uncoating of polyoma pseudovirus in L2 protein of human papillomavirus mouse embryo cells. Proc Natl Acad type 16. J Virol, 2000. 74(22): p. Sci U S A, 1970. 67(1): p. 37-40. 10332-40. 666. Slilaty, S.N. and H.V. Aposhian, Gene 657. Helft, J., et al., GM-CSF Mouse Bone transfer by polyoma-like particles Marrow Cultures Comprise a assembled in a cell-free system. Heterogeneous Population of Science, 1983. 220(4598): p. 725-7. CD11c(+)MHCII(+) Macrophages and 667. Touze, A. and P. Coursaget, In vitro Dendritic Cells. Immunity, 2015. gene transfer using human 42(6): p. 1197-211.

227

papillomavirus-like particles. Nucleic 678. Del Valle, L., et al., Detection of JC Acids Res, 1998. 26(5): p. 1317-23. virus DNA sequences and expression 668. Ou, W.C., et al., The major capsid of viral T antigen and agnoprotein in protein, VP1, of human JC virus esophageal carcinoma. Cancer, 2005. expressed in Escherichia coli is able to 103(3): p. 516-27. self-assemble into a capsid-like 679. Goldmann, C., et al., Molecular particle and deliver exogenous DNA cloning and expression of major into human kidney cells. J Gen Virol, structural protein VP1 of the human 1999. 80 ( Pt 1): p. 39-46. polyomavirus JC virus: formation of 669. Krauzewicz, N., et al., Virus-like gene virus-like particles useful for transfer into cells mediated by immunological and therapeutic polyoma virus pseudocapsids. Gene studies. J Virol, 1999. 73(5): p. 4465- Ther, 2000. 7(24): p. 2122-31. 9. 670. Forstova, J., et al., Polyoma virus 680. Wang, M., et al., Inhibition of simian pseudocapsids as efficient carriers of virus 40 large tumor antigen heterologous DNA into mammalian expression in human fetal glial cells by cells. Hum Gene Ther, 1995. 6(3): p. an antisense oligodeoxynucleotide 297-306. delivered by the JC virus-like particle. 671. Takamura, S., et al., DNA vaccine- Hum Gene Ther, 2004. 15(11): p. encapsulated virus-like particles 1077-90. derived from an orally transmissible 681. Stokrova, J., et al., Interactions of virus stimulate mucosal and systemic heterologous DNA with polyomavirus immune responses by oral major structural protein, VP1. FEBS administration. Gene Ther, 2004. Lett, 1999. 445(1): p. 119-25. 11(7): p. 628-35. 682. Chen, L.S., et al., Efficient gene 672. Brandenburg, B., et al., A novel transfer using the human JC virus-like system for efficient gene transfer into particle that inhibits human colon primary human hepatocytes via cell- adenocarcinoma growth in a nude permeable hepatitis B virus-like mouse model. Gene Ther, 2010. particle. Hepatology, 2005. 42(6): p. 17(8): p. 1033-41. 1300-9. 683. Fang, C.Y., et al., Analysis of the size of 673. Padgett, B.L., et al., Cultivation of DNA packaged by the human JC virus- papova-like virus from human brain like particle. J Virol Methods, 2012. with progressive multifocal 182(1-2): p. 87-92. leucoencephalopathy. Lancet, 1971. 684. Stubenrauch, K., et al., Conjugation of 1(7712): p. 1257-60. an antibody Fv fragment to a virus 674. White, M.K., et al., Human coat protein: cell-specific targeting of polyomaviruses and brain tumors. recombinant polyoma-virus-like Brain Res Brain Res Rev, 2005. 50(1): particles. Biochem J, 2001. 356(Pt 3): p. 69-85. p. 867-73. 675. Laghi, L., et al., JC virus DNA is present 685. Ashley, C.E., et al., Cell-specific in the mucosa of the human colon and delivery of diverse cargos by in colorectal cancers. Proc Natl Acad bacteriophage MS2 virus-like Sci U S A, 1999. 96(13): p. 7484-9. particles. ACS Nano, 2011. 5(7): p. 676. Zheng, H., et al., Oncogenic role of JC 5729-45. virus in lung cancer. J Pathol, 2007. 686. Kawano, M., et al., Chimeric SV40 212(3): p. 306-15. virus-like particles induce specific 677. Shin, S.K., et al., Oncogenic T-antigen cytotoxicity and protective immunity of JC virus is present frequently in against influenza A virus without the human gastric cancers. Cancer, 2006. need of adjuvants. Virology, 2014. 107(3): p. 481-8. 448: p. 159-67.

228

687. Rivera-Hernandez, T., et al., Self- cell responses in the absence of adjuvanting modular virus-like systemic side effects. J Immunol, particles for mucosal vaccination 2004. 172(3): p. 1777-85. against group A streptococcus (GAS). 697. Gerber, S., et al., Human Vaccine, 2013. 31(15): p. 1950-5. papillomavirus virus-like particles are 688. Wang, B.Z., et al., Incorporation of efficient oral immunogens when membrane-anchored flagellin into coadministered with Escherichia coli influenza virus-like particles enhances heat-labile enterotoxin mutant the breadth of immune responses. J R192G or CpG DNA. J Virol, 2001. Virol, 2008. 82(23): p. 11813-23. 75(10): p. 4752-60. 689. Skountzou, I., et al., Incorporation of 698. Young, S.L., et al., Transcutaneous glycosylphosphatidylinositol- vaccination with virus-like particles. anchored granulocyte- macrophage Vaccine, 2006. 24(26): p. 5406-12. colony-stimulating factor or CD40 699. Klimek, L., et al., Assessment of ligand enhances immunogenicity of clinical efficacy of CYT003-QbG10 in chimeric simian immunodeficiency patients with allergic virus-like particles. J Virol, 2007. rhinoconjunctivitis: a phase IIb study. 81(3): p. 1083-94. Clin Exp Allergy, 2011. 41(9): p. 1305- 690. Derdak, S.V., et al., Direct stimulation 12. of T lymphocytes by immunosomes: 700. Li, J., et al., Messenger RNA vaccine virus-like particles decorated with T based on recombinant MS2 virus-like cell receptor/CD3 ligands plus particles against prostate cancer. Int J costimulatory molecules. Proc Natl Cancer, 2014. 134(7): p. 1683-94. Acad Sci U S A, 2006. 103(35): p. 701. Sun, Y., et al., Intracellular delivery of 13144-9. messenger RNA by recombinant PP7 691. Krug, A., et al., Identification of CpG virus-like particles carrying low oligonucleotide sequences with high molecular weight protamine. BMC induction of IFN-alpha/beta in Biotechnol, 2016. 16(1): p. 46. plasmacytoid dendritic cells. Eur J 702. Pan, Y., et al., Development of a Immunol, 2001. 31(7): p. 2154-63. microRNA delivery system based on 692. Boggs, R.T., et al., Characterization bacteriophage MS2 virus-like and modulation of immune particles. FEBS J, 2012. 279(7): p. stimulation by modified 1198-208. oligonucleotides. Antisense Nucleic 703. Lin, M.C., et al., Inhibition of BK virus Acid Drug Dev, 1997. 7(5): p. 461-71. replication in human kidney cells by 693. Krieg, A.M., Now I know my CpGs. BK virus large tumor antigen-specific Trends Microbiol, 2001. 9(6): p. 249- shRNA delivered by JC virus-like 52. particles. Antiviral Res, 2014. 103: p. 694. Hartmann, G., et al., Rational design 25-31. of new CpG oligonucleotides that 704. Hoffmann, D.B., et al., In Vivo siRNA combine B cell activation with high Delivery Using JC Virus-like Particles IFN-alpha induction in plasmacytoid Decreases the Expression of RANKL in dendritic cells. Eur J Immunol, 2003. Rats. Mol Ther Nucleic Acids, 2016. 5: 33(6): p. 1633-41. p. e298. 695. Marshall, J.D., et al., Identification of 705. Rokavec, M., et al., IL-6R/STAT3/miR- a novel CpG DNA class and motif that 34a feedback loop promotes EMT- optimally stimulate B cell and mediated colorectal cancer invasion plasmacytoid dendritic cell functions. and metastasis. J Clin Invest, 2014. J Leukoc Biol, 2003. 73(6): p. 781-92. 124(4): p. 1853-67. 696. Storni, T., et al., Nonmethylated CG 706. Fritzenwanger, M., et al., motifs packaged into virus-like Cardiotrophin-1 induces interleukin-6 particles induce protective cytotoxic T synthesis in human umbilical vein

229

endothelial cells. Cytokine, 2006. family. J Virol, 2000. 74(24): p. 11950- 36(3-4): p. 101-6. 4. 707. Kim, J.Y., et al., Visfatin through 718. Stuart, A.D. and T.D. Brown, STAT3 activation enhances IL-6 Alpha2,6-linked sialic acid acts as a expression that promotes endothelial receptor for Feline calicivirus. J Gen angiogenesis. Biochim Biophys Acta, Virol, 2007. 88(Pt 1): p. 177-86. 2009. 1793(11): p. 1759-67. 719. Tamura, M., et al., Genogroup II 708. Lee, C., et al., Janus kinase-signal noroviruses efficiently bind to transducer and activator of heparan sulfate proteoglycan transcription mediates phosphatidic associated with the cellular acid-induced interleukin (IL)-1beta membrane. J Virol, 2004. 78(8): p. and IL-6 production. Mol Pharmacol, 3817-26. 2006. 69(3): p. 1041-7. 720. Nystrom, K., et al., Histo-blood group 709. Peacey, M., Creation and antigens act as attachment factors of Investigation of a Versatile Rabbit rabbit hemorrhagic disease virus Haemorrhagic Disease Virus-like infection in a virus strain-dependent Particle Vaccine. 2007, University of manner. PLoS Pathog, 2011. 7(8): p. Otago. e1002188. 710. Bencze, W.L. and K. Schmid, 721. Pei, D.S., et al., Oncolytic-adenovirus- Determination of Tyrosine and expressed RNA interference for cancer Tryptophan in Proteins. Anal Chem, therapy. Expert Opin Biol Ther, 2010. 1957. 29(8): p. 1193-1196. 10(9): p. 1331-41. 711. Lovrien, R. and D. Matulis, Assays for 722. Anesti, A.M., et al., Expression of RNA total protein. Curr Protoc Microbiol, interference triggers from an 2005. Appendix 3: p. Appendix 3A. oncolytic herpes simplex virus results 712. Touze, A., et al., Gene transfer using in specific silencing in tumour cells in human polyomavirus BK virus-like vitro and tumours in vivo. BMC particles expressed in insect cells. J Cancer, 2010. 10: p. 486. Gen Virol, 2001. 82(Pt 12): p. 3005-9. 723. Bennink, J.R., et al., Anti-influenza 713. Choi, J.M., et al., Atomic resolution virus cytotoxic T lymphocytes structural characterization of recognize the three viral polymerases recognition of histo-blood group and a nonstructural protein: antigens by Norwalk virus. Proc Natl responsiveness to individual viral Acad Sci U S A, 2008. 105(27): p. 9175- antigens is major histocompatibility 80. complex controlled. J Virol, 1987. 714. Cao, S., et al., Structural basis for the 61(4): p. 1098-102. recognition of blood group 724. Gotch, F., et al., Identification of viral trisaccharides by norovirus. J Virol, molecules recognized by influenza- 2007. 81(11): p. 5949-57. specific human cytotoxic T 715. Bu, W., et al., Structural basis for the lymphocytes. J Exp Med, 1987. receptor binding specificity of 165(2): p. 408-16. Norwalk virus. J Virol, 2008. 82(11): p. 725. Falk, K., et al., Identification of 5340-7. naturally processed viral 716. Farkas, T., et al., Genetic diversity and nonapeptides allows their histo-blood group antigen quantification in infected cells and interactions of rhesus enteric suggests an allele-specific T cell caliciviruses. J Virol, 2010. 84(17): p. epitope forecast. J Exp Med, 1991. 8617-25. 174(2): p. 425-34. 717. Ruvoen-Clouet, N., et al., Binding of 726. Townsend, A.R., F.M. Gotch, and J. rabbit hemorrhagic disease virus to Davey, Cytotoxic T cells recognize antigens of the ABH histo-blood group fragments of the influenza

230

nucleoprotein. Cell, 1985. 42(2): p. 736. Kono, K., et al., Multicenter, phase II 457-67. clinical trial of cancer vaccination for 727. Townsend, A.R., et al., The epitopes of advanced esophageal cancer with influenza nucleoprotein recognized by three peptides derived from novel cytotoxic T lymphocytes can be cancer-testis antigens. J Transl Med, defined with short synthetic peptides. 2012. 10: p. 141. Cell, 1986. 44(6): p. 959-68. 737. Yoshitake, Y., et al., Phase II clinical 728. Sercarz, E.E., et al., Dominance and trial of multiple peptide vaccination crypticity of T cell antigenic for advanced head and neck cancer determinants. Annu Rev Immunol, patients revealed induction of 1993. 11: p. 729-66. immune responses and improved OS. 729. Brehm, M.A., et al., T cell Clin Cancer Res, 2015. 21(2): p. 312- immunodominance and maintenance 21. of memory regulated by unexpectedly 738. Slingluff, C.L., Jr., et al., A randomized cross-reactive pathogens. Nat phase II trial of multiepitope Immunol, 2002. 3(7): p. 627-34. vaccination with melanoma peptides 730. Slifka, M.K., et al., Preferential escape for cytotoxic T cells and helper T cells of subdominant CD8+ T cells during for patients with metastatic negative selection results in an melanoma (E1602). Clin Cancer Res, altered antiviral T cell hierarchy. J 2013. 19(15): p. 4228-38. Immunol, 2003. 170(3): p. 1231-9. 739. Ruedl, C., et al., Virus-like particles as 731. Rodriguez, F., et al., carriers for T-cell epitopes: limited Immunodominance in virus-induced inhibition of T-cell priming by carrier- CD8(+) T-cell responses is specific antibodies. J Virol, 2005. dramatically modified by DNA 79(2): p. 717-24. immunization and is regulated by 740. Schwarz, K., et al., Efficient gamma interferon. J Virol, 2002. homologous prime-boost strategies 76(9): p. 4251-9. for T cell vaccination based on virus- 732. Tarhini, A.A., et al., Safety and like particles. Eur J Immunol, 2005. immunogenicity of vaccination with 35(3): p. 816-21. MART-1 (26-35, 27L), gp100 (209- 741. Ramshaw, I.A. and A.J. Ramsay, The 217, 210M), and tyrosinase (368-376, prime-boost strategy: exciting 370D) in adjuvant with PF-3512676 prospects for improved vaccination. and GM-CSF in metastatic melanoma. Immunol Today, 2000. 21(4): p. 163- J Immunother, 2012. 35(4): p. 359-66. 5. 733. Cho, H.I. and E. Celis, Design of 742. Estcourt, M.J., et al., Prime-boost immunogenic and effective multi- immunization generates a high epitope DNA vaccines for melanoma. frequency, high-avidity CD8(+) Cancer Immunol Immunother, 2012. cytotoxic T lymphocyte population. 61(3): p. 343-51. Int Immunol, 2002. 14(1): p. 31-7. 734. Derouazi, M., et al., Novel Cell- 743. McShane, H., Prime-boost Penetrating Peptide-Based Vaccine immunization strategies for infectious Induces Robust CD4+ and CD8+ T Cell- diseases. Curr Opin Mol Ther, 2002. Mediated Antitumor Immunity. 4(1): p. 23-7. Cancer Res, 2015. 75(15): p. 3020-31. 744. Tanghe, A., et al., Improved 735. Phuphanich, S., et al., Phase I trial of immunogenicity and protective a multi-epitope-pulsed dendritic cell efficacy of a tuberculosis DNA vaccine vaccine for patients with newly encoding Ag85 by protein boosting. diagnosed glioblastoma. Cancer Infect Immun, 2001. 69(5): p. 3041-7. Immunol Immunother, 2013. 62(1): p. 745. McShane, H., et al., Enhanced 125-35. immunogenicity of CD4(+) t-cell responses and protective efficacy of a

231

DNA-modified vaccinia virus Ankara antigen. Proc Natl Acad Sci U S A, prime-boost vaccination regimen for 2001. 98(20): p. 11491-6. murine tuberculosis. Infect Immun, 754. Irvine, K.R., et al., Enhancing efficacy 2001. 69(2): p. 681-6. of recombinant anticancer vaccines 746. Feng, C.G., et al., Priming by DNA with prime/boost regimens that use immunization augments protective two different vectors. J Natl Cancer efficacy of Mycobacterium bovis Inst, 1997. 89(21): p. 1595-601. Bacille Calmette-Guerin against 755. Ilett, E.J., et al., Dendritic cells and T tuberculosis. Infect Immun, 2001. cells deliver oncolytic reovirus for 69(6): p. 4174-6. tumour killing despite pre-existing 747. Brooks, J.V., et al., Boosting vaccine anti-viral immunity. Gene Ther, 2009. for tuberculosis. Infect Immun, 2001. 16(5): p. 689-99. 69(4): p. 2714-7. 756. Prestwich, R.J., et al., Immune- 748. Robinson, H.L., et al., Neutralizing mediated antitumor activity of antibody-independent containment reovirus is required for therapy and is of immunodeficiency virus challenges independent of direct viral oncolysis by DNA priming and recombinant pox and replication. Clin Cancer Res, virus booster immunizations. Nat 2009. 15(13): p. 4374-81. Med, 1999. 5(5): p. 526-34. 757. Qiao, J., et al., Purging metastases in 749. Haglund, K., et al., Robust recall and lymphoid organs using a combination long-term memory T-cell responses of antigen-nonspecific adoptive T cell induced by prime-boost regimens therapy, oncolytic virotherapy and with heterologous live viral vectors immunotherapy. Nat Med, 2008. expressing human immunodeficiency 14(1): p. 37-44. virus type 1 Gag and Env proteins. J 758. Smyth, M.J., G.P. Dunn, and R.D. Virol, 2002. 76(15): p. 7506-17. Schreiber, Cancer 750. Amara, R.R., et al., Control of a immunosurveillance and mucosal challenge and prevention of immunoediting: the roles of immunity AIDS by a multiprotein DNA/MVA in suppressing tumor development vaccine. Science, 2001. 292(5514): p. and shaping tumor immunogenicity. 69-74. Adv Immunol, 2006. 90: p. 1-50. 751. Schneider, J., et al., A prime-boost 759. Le, D.T., D.M. Pardoll, and E.M. Jaffee, immunisation regimen using DNA Cellular vaccine approaches. Cancer J, followed by recombinant modified 2010. 16(4): p. 304-10. vaccinia virus Ankara induces strong 760. Kottke, T., et al., Broad antigenic cellular immune responses against coverage induced by vaccination with the Plasmodium falciparum TRAP virus-based cDNA libraries cures antigen in chimpanzees. Vaccine, established tumors. Nat Med, 2011. 2001. 19(32): p. 4595-602. 17(7): p. 854-9. 752. Gilbert, S.C., et al., Enhanced CD8 T 761. Drake, C.G., E. Jaffee, and D.M. cell immunogenicity and protective Pardoll, Mechanisms of immune efficacy in a mouse malaria model evasion by tumors. Adv Immunol, using a recombinant adenoviral 2006. 90: p. 51-81. vaccine in heterologous prime-boost 762. Santisteban, M., et al., Immune- immunisation regimes. Vaccine, induced epithelial to mesenchymal 2002. 20(7-8): p. 1039-45. transition in vivo generates breast 753. Bruna-Romero, O., et al., Complete, cancer stem cells. Cancer Res, 2009. long-lasting protection against 69(7): p. 2887-95. malaria of mice primed and boosted 763. Moody, S.E., et al., The transcriptional with two distinct viral vectors repressor Snail promotes mammary expressing the same plasmodial tumor recurrence. Cancer Cell, 2005. 8(3): p. 197-209.

232

764. Kalluri, R. and R.A. Weinberg, The transgenic mice immunized with basics of epithelial-mesenchymal various human MUC1 vaccines. transition. J Clin Invest, 2009. 119(6): Cancer Immunol Immunother, 2000. p. 1420-8. 48(10): p. 588-94. 765. Taheri, M., et al., Self recognition in 774. Gong, J., et al., Reversal of tolerance the Ig superfamily. Identification of to human MUC1 antigen in MUC1 precise subdomains in transgenic mice immunized with carcinoembryonic antigen required fusions of dendritic and carcinoma for intercellular adhesion. J Biol cells. Proc Natl Acad Sci U S A, 1998. Chem, 2000. 275(35): p. 26935-43. 95(11): p. 6279-83. 766. Screaton, R.A., L.Z. Penn, and C.P. 775. Kass, E., et al., Induction of protective Stanners, Carcinoembryonic antigen, host immunity to carcinoembryonic a human tumor marker, cooperates antigen (CEA), a self-antigen in CEA with Myc and Bcl-2 in cellular transgenic mice, by immunizing with transformation. J Cell Biol, 1997. a recombinant vaccinia-CEA virus. 137(4): p. 939-52. Cancer Res, 1999. 59(3): p. 676-83. 767. Screaton, R.A., et al., The specificity 776. Rowse, G.J., et al., Tolerance and for the differentiation blocking immunity to MUC1 in a human MUC1 activity of carcinoembryonic antigen transgenic murine model. Cancer Res, resides in its glycophosphatidyl- 1998. 58(2): p. 315-21. inositol anchor. J Cell Biol, 2000. 777. Thompson, J.A., et al., Expression of 150(3): p. 613-26. transgenic carcinoembryonic antigen 768. Ordonez, C., et al., Human (CEA) in tumor-prone mice: an animal carcinoembryonic antigen functions model for CEA-directed tumor as a general inhibitor of anoikis. immunotherapy. Int J Cancer, 1997. Cancer Res, 2000. 60(13): p. 3419-24. 72(1): p. 197-202. 769. Benchimol, S., et al., 778. Andersen, M.H., et al., The universal Carcinoembryonic antigen, a human character of the tumor-associated tumor marker, functions as an antigen survivin. Clin Cancer Res, intercellular adhesion molecule. Cell, 2007. 13(20): p. 5991-4. 1989. 57(2): p. 327-34. 779. Mita, A.C., et al., Survivin: key 770. Huang, J.Q., et al., Spatiotemporal regulator of mitosis and apoptosis expression of murine and novel target for cancer carcinoembryonic antigen (CEA) gene therapeutics. Clin Cancer Res, 2008. family members during mouse 14(16): p. 5000-5. embryogenesis. Development, 1990. 780. Altieri, D.C., Validating survivin as a 110(2): p. 573-88. cancer therapeutic target. Nat Rev 771. Spicer, A.P., et al., Molecular cloning Cancer, 2003. 3(1): p. 46-54. and analysis of the mouse homologue 781. Harfouche, R., et al., Mechanisms of the tumor-associated mucin, which mediate the antiapoptotic MUC1, reveals conservation of effects of angiopoietin-1 on potential O-glycosylation sites, endothelial cells. Microvasc Res, transmembrane, and cytoplasmic 2002. 64(1): p. 135-47. domains and a loss of minisatellite- 782. Tran, J., et al., Marked induction of like polymorphism. J Biol Chem, 1991. the IAP family antiapoptotic proteins 266(23): p. 15099-109. survivin and XIAP by VEGF in vascular 772. Ho, S.B., et al., Heterogeneity of endothelial cells. Biochem Biophys mucin gene expression in normal and Res Commun, 1999. 264(3): p. 781-8. neoplastic tissues. Cancer Res, 1993. 783. Mesri, M., et al., Suppression of 53(3): p. 641-51. vascular endothelial growth factor- 773. Acres, B., et al., MUC1-specific mediated endothelial cell protection immune responses in human MUC1

233

by survivin targeting. Am J Pathol, 793. Zhao, H., et al., DNA topoisomerase II- 2001. 158(5): p. 1757-65. alpha as a proliferation marker in 784. Idenoue, S., et al., A potent human gliomas: correlation with immunogenic general cancer vaccine PCNA expression and patient survival. that targets survivin, an inhibitor of Clin Neuropathol, 2008. 27(2): p. 83- apoptosis proteins. Clin Cancer Res, 90. 2005. 11(4): p. 1474-82. 794. Schrader, C., et al., Topoisomerase 785. Tsuruma, T., et al., Phase I clinical IIalpha expression in mantle cell study of anti-apoptosis protein, lymphoma: a marker of cell survivin-derived peptide vaccine proliferation and a prognostic factor therapy for patients with advanced or for clinical outcome. Leukemia, 2004. recurrent colorectal cancer. J Transl 18(7): p. 1200-6. Med, 2004. 2(1): p. 19. 795. Schoeffler, A.J. and J.M. Berger, DNA 786. Lennerz, V., et al., Immunologic topoisomerases: harnessing and response to the survivin-derived constraining energy to govern multi-epitope vaccine EMD640744 in chromosome topology. Q Rev patients with advanced solid tumors. Biophys, 2008. 41(1): p. 41-101. Cancer Immunol Immunother, 2014. 796. Niimi, A., et al., Co-localization of 63(4): p. 381-94. chicken DNA topoisomerase IIalpha, 787. Aucouturier, J., et al., Montanide ISA but not beta, with sites of DNA 720 and 51: a new generation of replication and possible involvement water in oil emulsions as adjuvants for of a C-terminal region of alpha human vaccines. Expert Rev Vaccines, through its binding to PCNA. 2002. 1(1): p. 111-8. Chromosoma, 2001. 110(2): p. 102- 788. Hofmann, U.B., et al., Identification 14. and characterization of survivin- 797. Grue, P., et al., Essential mitotic derived H-2Kb-restricted CTL functions of DNA topoisomerase epitopes. Eur J Immunol, 2009. 39(5): IIalpha are not adopted by p. 1419-24. topoisomerase IIbeta in human H69 789. Woessner, R.D., et al., Proliferation- cells. J Biol Chem, 1998. 273(50): p. and cell cycle-dependent differences 33660-6. in expression of the 170 kilodalton 798. Kosari, F., et al., Identification of and 180 kilodalton forms of prognostic biomarkers for prostate topoisomerase II in NIH-3T3 cells. Cell cancer. Clin Cancer Res, 2008. 14(6): Growth Differ, 1991. 2(4): p. 209-14. p. 1734-43. 790. Adachi, N., et al., Cell-cycle regulation 799. Wong, N., et al., TOP2A of the DNA topoisomerase IIalpha overexpression in hepatocellular promoter is mediated by proximal carcinoma correlates with early age CCAAT boxes: possible involvement of onset, shorter patients survival and acetylation. Gene, 2000. 245(1): p. chemoresistance. Int J Cancer, 2009. 49-57. 124(3): p. 644-52. 791. Al-Kuraya, K., et al., HER2, TOP2A, 800. Hande, K.R., Clinical applications of CCND1, EGFR and C-MYC oncogene anticancer drugs targeted to amplification in colorectal cancer. J topoisomerase II. Biochim Biophys Clin Pathol, 2007. 60(7): p. 768-72. Acta, 1998. 1400(1-3): p. 173-84. 792. Willman, J.H. and J.A. Holden, 801. Park, J.S., et al., Topoisomerase II Immunohistochemical staining for alpha as a universal tumor antigen: DNA topoisomerase II-alpha in antitumor immunity in murine tumor benign, premalignant, and malignant models and H-2K(b)-restricted T cell lesions of the prostate. Prostate, epitope. Cancer Immunol 2000. 42(4): p. 280-6. Immunother, 2010. 59(5): p. 747-57.

234

802. Osuna, J., A. Perez-Blancas, and X. 812. Pouyanfard, S., et al., Induction of Soberon, Improving a circularly protective anti-CTL epitope responses permuted TEM-1 beta-lactamase by against HER-2-positive breast cancer directed evolution. Protein Eng, 2002. based on multivalent T7 phage 15(6): p. 463-70. nanoparticles. PLoS One, 2012. 7(11): 803. Venken, K.J., et al., MiMIC: a highly p. e49539. versatile transposon insertion 813. Nakamura, Y., T. Gojobori, and T. resource for engineering Drosophila Ikemura, Codon usage tabulated from melanogaster genes. Nat Methods, international DNA sequence 2011. 8(9): p. 737-43. databases: status for the year 2000. 804. Yant, S.R., et al., Site-directed Nucleic Acids Res, 2000. 28(1): p. 292. transposon integration in human 814. Nagesha, H.S., L.F. Wang, and A.D. cells. Nucleic Acids Res, 2007. 35(7): Hyatt, Virus-like particles of calicivirus p. e50. as epitope carriers. Arch Virol, 1999. 805. Luque, D., et al., Epitope insertion at 144(12): p. 2429-39. the N-terminal molecular switch of 815. Matthaei, M., et al., Comparative the rabbit hemorrhagic disease virus T quantitative monitoring of rabbit = 3 capsid protein leads to larger T = 4 haemorrhagic disease viruses in capsids. J Virol, 2012. 86(12): p. 6470- rabbit kittens. Virol J, 2014. 11: p. 109. 80. 816. DeMaula, C.D., et al., Infection 806. Levy, A., et al., A melanoma kinetics, prostacyclin release and multiepitope polypeptide induces cytokine-mediated modulation of the specific CD8+ T-cell response. Cell mechanism of cell death during Immunol, 2007. 250(1-2): p. 24-30. bluetongue virus infection of cultured 807. Kim, Y., et al., Immune epitope ovine and bovine pulmonary artery database analysis resource. Nucleic and lung microvascular endothelial Acids Res, 2012. 40(Web Server cells. J Gen Virol, 2001. 82(Pt 4): p. issue): p. W525-30. 787-94. 808. Diez-Rivero, C.M., E.M. Lafuente, and 817. Baccam, P., et al., Kinetics of influenza P.A. Reche, Computational analysis A virus infection in humans. J Virol, and modeling of cleavage by the 2006. 80(15): p. 7590-9. immunoproteasome and the 818. Rosenberg, S.A., P. Spiess, and R. constitutive proteasome. BMC Lafreniere, A new approach to the Bioinformatics, 2010. 11: p. 479. adoptive immunotherapy of cancer 809. Nielsen, M., et al., The role of the with tumor-infiltrating lymphocytes. proteasome in generating cytotoxic T- Science, 1986. 233(4770): p. 1318-21. cell epitopes: insights obtained from 819. Pajtasz-Piasecka, E., et al., Loss of improved predictions of proteasomal tumorigenicity of murine colon cleavage. Immunogenetics, 2005. carcinoma MC38/0 cell line after 57(1-2): p. 33-41. transduction with a retroviral vector 810. Velders, M.P., et al., Defined flanking carrying murine IL-12 genes. Folia Biol spacers and enhanced proteolysis is (Praha), 2004. 50(1): p. 7-14. essential for eradication of 820. Lute, K.D., et al., Human CTLA4 knock- established tumors by an epitope in mice unravel the quantitative link string DNA vaccine. J Immunol, 2001. between tumor immunity and 166(9): p. 5366-73. autoimmunity induced by anti-CTLA-4 811. Rueda, P., et al., Influence of flanking antibodies. Blood, 2005. 106(9): p. sequences on presentation efficiency 3127-33. of a CD8+ cytotoxic T-cell epitope 821. Esche, C., et al., CD154 inhibits tumor- delivered by parvovirus-like particles. induced apoptosis in dendritic cells J Gen Virol, 2004. 85(Pt 3): p. 563-72. and tumor growth. Eur J Immunol, 1999. 29(7): p. 2148-55.

235

822. Gilfillan, S., et al., DNAM-1 promotes and human compatible adjuvants. activation of cytotoxic lymphocytes by Vaccine, 2001. 19(30): p. 4121-9. nonprofessional antigen-presenting 834. Field, C., Attenuation of Immune cells and tumors. J Exp Med, 2008. Suppression to Complement Glioma 205(13): p. 2965-73. Immunotherapy. 2016. 823. Ball, L.A., et al., Phenotypic 835. Ali, O.A., et al., Vaccines Combined consequences of rearranging the P, with Immune Checkpoint Antibodies M, and G genes of vesicular stomatitis Promote Cytotoxic T-cell Activity and virus. J Virol, 1999. 73(6): p. 4705-12. Tumor Eradication. Cancer Immunol 824. Dietzschold, B., et al., Rhabdoviruses, Res, 2016. 4(2): p. 95-100. in Fields Virology. 1996, Lippincott- 836. Avogadri, F., et al., Combination of Raven. Alphavirus Replicon Particle–Based 825. Carneiro, F.A., et al., Membrane Vaccination with Immunomodulatory recognition by vesicular stomatitis Antibodies: Therapeutic Activity in the virus involves enthalpy-driven B16 Melanoma Mouse Model and protein-lipid interactions. J Virol, Immune Correlates. Cancer Immunol 2002. 76(8): p. 3756-64. Res, 2014. 2(5): p. 448-458. 826. Ochsenbein, A.F., et al., Control of 837. Lemke, C., et al., Combination cancer early viral and bacterial distribution immunotherapy using checkpoint and disease by natural antibodies. blockade and intratumoral virus-like Science, 1999. 286(5447): p. 2156-9. particles containing CpG ODN. Cancer 827. Thomsen, A.R., et al., Cooperation of Res, 2016. 76(14 Supplement): p. B cells and T cells is required for 1417-1417. survival of mice infected with 838. Li, K., Adoptive Cell Therapy using CD4 vesicular stomatitis virus. Int T Helper 1-like and CD8 Cytotoxic T Immunol, 1997. 9(11): p. 1757-66. Lymphocytes in a Mouse Model of 828. Scullion, S.L., Investigating cytotoxic T Melanoma. 2015, University of cell responses to RHDV virus-like Otago. particles. 2012, University of Otago. 839. Geller, D.S., et al., Targeted therapy of 829. Shtivelman, E., et al., Fused transcript osteosarcoma with radiolabeled of abl and bcr genes in chronic monoclonal antibody to an insulin-like myelogenous leukaemia. Nature, growth factor-2 receptor (IGF2R). 1985. 315(6020): p. 550-4. Nucl Med Biol, 2016. 43(12): p. 812- 830. Clark, S.S., et al., Expression of a 817. distinctive BCR-ABL oncogene in Ph1- 840. Ma, Y. and J. Li, Vesicular stomatitis positive acute lymphocytic leukemia virus as a vector to deliver virus-like (ALL). Science, 1988. 239(4841 Pt 1): particles of human norovirus: a new p. 775-7. vaccine candidate against an 831. Shepard, H.M., et al., Monoclonal important noncultivable virus. J Virol, antibody therapy of human cancer: 2011. 85(6): p. 2942-2952. taking the HER2 protooncogene to 841. Knudsen, N.P., et al., Different human the clinic. J Clin Immunol, 1991. 11(3): vaccine adjuvants promote distinct p. 117-27. antigen-independent immunological 832. Pietras, R.J., et al., Antibody to HER- signatures tailored to different 2/neu receptor blocks DNA repair pathogens. Sci Rep, 2016. 6: p. 19570. after cisplatin in human breast and 842. Dolen, Y., et al., Co-delivery of PLGA ovarian cancer cells. Oncogene, 1994. encapsulated invariant NKT cell 9(7): p. 1829-38. agonist with antigenic protein induce 833. Peter, K., et al., Induction of a strong T cell-mediated antitumor cytotoxic T-cell response to HIV-1 immune responses. proteins with short synthetic peptides Oncoimmunology, 2016. 5(1): p. e1068493.

236

843. Dow, L.E., et al., Inducible in vivo immune response: combined delivery genome editing with CRISPR-Cas9. of tumor antigen and small Nat Biotechnol, 2015. 33(4): p. 390-4. interference RNA for 844. Ran, F.A., et al., In vivo genome immunosuppressive gene to dendritic editing using Staphylococcus aureus cells. Acta Biomater, 2014. 10(5): p. Cas9. Nature, 2015. 520(7546): p. 2169-76. 186-91. 855. Warashina, S., et al., A lipid 845. Swiech, L., et al., In vivo interrogation nanoparticle for the efficient delivery of gene function in the mammalian of siRNA to dendritic cells. J Control brain using CRISPR-Cas9. Nat Release, 2016. 225: p. 183-91. Biotechnol, 2015. 33(1): p. 102-6. 856. Dejean, A.S., S.M. Hedrick, and Y.M. 846. Tabebordbar, M., et al., In vivo gene Kerdiles, Highly specialized role of editing in dystrophic mouse muscle Forkhead box O transcription factors and muscle stem cells. Science, 2016. in the immune system. Antioxid 351(6271): p. 407-11. Redox Signal, 2011. 14(4): p. 663-74. 847. Zhang, Y., I-TASSER server for protein 857. Dejean, A.S., et al., Transcription 3D structure prediction. BMC factor Foxo3 controls the magnitude Bioinformatics, 2008. 9: p. 40. of T cell immune responses by 848. Hecht, N., Regulation of'haploid modulating the function of dendritic expressed genes' in male germ cells. J cells. Nat Immunol, 2009. 10(5): p. Reprod Fertil, 1990. 88(2): p. 679-693. 504-13. 849. Zhao, G., et al., Iron and hydrogen 858. Rao, R.R., et al., Transcription factor peroxide detoxification properties of Foxo1 represses T-bet-mediated DNA-binding protein from starved effector functions and promotes cells A ferritin-like DNA-binding memory CD8(+) T cell differentiation. protein of Escherichia coli. Journal of Immunity, 2012. 36(3): p. 374-87. Biological Chemistry, 2002. 277(31): 859. Tejera, M.M., et al., FoxO1 controls p. 27689-27696. effector-to-memory transition and 850. Kershaw, N.J., et al., Regulation of maintenance of functional CD8 T cell Janus kinases by SOCS proteins. memory. J Immunol, 2013. 191(1): p. Biochem Soc Trans, 2013. 41(4): p. 187-99. 1042-7. 860. Togher, S., et al., FoxO3 is a negative 851. Tamiya, T., et al., Suppressors of regulator of primary CD8+ T-cell cytokine signaling (SOCS) proteins expansion but not of memory and JAK/STAT pathways: regulation formation. Immunol Cell Biol, 2015. of T-cell inflammation by SOCS1 and 93(2): p. 120-5. SOCS3. Arterioscler Thromb Vasc Biol, 861. Wang, H., et al., Biocompatible, 2011. 31(5): p. 980-5. chimeric peptide-condensed 852. Metcalf, D., et al., Polycystic kidneys supramolecular nanoparticles for and chronic inflammatory lesions are tumor cell-specific siRNA delivery and the delayed consequences of loss of gene silencing. Chem Commun the suppressor of cytokine signaling-1 (Camb), 2014. 50(58): p. 7806-9. (SOCS-1). Proc Natl Acad Sci U S A, 862. McCulloch, T. Recombinant virus-like 2002. 99(2): p. 943-8. particles presenting epitopes for 853. Davey, G.M., et al., SOCS-1 regulates antibody generation : a thesis IL-15-driven homeostatic submitted for the degree of Bachelor proliferation of antigen-naive CD8 T of Biomedical Science with Honours, cells, limiting their autoimmune Department of Microbiology and potential. J Exp Med, 2005. 202(8): p. Immunology, University of Otago, 1099-108. Dunedin, New Zealand. 2013. 854. Heo, M.B., M.Y. Cho, and Y.T. Lim, Polymer nanoparticles for enhanced

237

238

7 Appendix I: Supplementary Figures

Figure 7.1 Chimaeric VLP Sequencing Data N-terminal sequencing data provided for (A) T.VP60, (B) S.VP60 and (C) TS.VP60, sequenced from pAcUW51(GUS) expression plasmids by Massey Genome Service (Massey University, NZ) . (D) Identification of peptide sequences confirmed through mass spectrometry by the Centre for Protein Research (University of Otago, NZ).

239

Figure 7.2 Mass Spectrometry Data Mass spectrometry data for (A) T.VP60, (B) S.VP60 and (C) TS.VP60, by the Centre for Protein Research (University of Otago, NZ).

240

8 Appendix I: Antibody Compendium

Antibody* Fluorophore Clone Supplier Anti-CD11c APC N418 BioLegend Anti-CD40 PE-Cy7 3123 BioLegend Anti-CD80 Pacific Blue 16-10A1 BioLegend Anti-CD86 PE GL-1 BioLegend Anti-I-A/I-E FITC M5/114.15.12 BioLegend Rabbit Anti-Stat3 None 79D7 Cell Signalling Technology Rabbit Anti-TopIIα None 4733S Cell Signalling Technology Rabbit Anti-Survivin None 7164B7 Cell Signalling Technology Goat Anti-Actinβ None I-F1 Santa Cruz Biotechnology Donkey anti-Rabbit DyLight 800 SA5-10044 Thermo Scientific Donkey anti-Goat DyLight 680 SA5-10090 Thermo Scientific Anti-IFN-γ None R4-6A2 BD Pharmingen Anti-IFN-γ Biotinylated XMG1.2 BD Pharmingen * The target species is mouse unless otherwise stated.

241

9 Appendix II: Recipes

9.1 Assorted PBS Solutions

9.1.1 10X PBS 10X PBS was prepared with 160 g sodium chloride (NaCl), 4 g potassium chloride (KCl),

22.7 g disodium hydrogen phosphate (Na2HPO4) and 4 g potassium dihydrogen phosphate

(KH2PO4), with sufficient dH2O as to bring the total volume up to 2000 mL.

9.1.3 iPBS iPBS was prepared with 0.14 g Na2HPO4, 1.43 g KH2PO4, 8.18 g NaCl and 2.98 g KCl, with sufficient dH2O as to bring the total volume up to 1000 mL. iPBS was autoclaved, and stored at room temperature.

9.1.4 cPBS cPBS was prepared with 5.2 g sodium dihydrogen phosphate dehydrate (NaH2PO4.2H2O),

23.66 g Na2HPO4 and 17.54 g NaCl, with sufficient dH2O as to bring the total volume up to 1000 mL. cPBS was autoclaved, and stored at 4 °C.

9.2 Transformation Reagents

9.2.1 LB Broth and Agar LB broth was prepared with 10 g bactotryptone (BD Biosciences, California, USA), 5 g bacto yeast extract (BD Biosciences, California, USA) and 5 g NaCl in 1000 mL of distilled H2O

(dH2O). The broth was distributed into 3 mL aliquots in glass bijous, autoclaved, and stored at 4 °C. LB agar was prepared using the recipe for LB broth, with the addition of 15 g bacteriological agar (Scharlau, Spain). The LB agar solution was autoclaved, and then cooled to 50 °C before adding 12.5 μg/mL tetracycline or 50 μg/mL ampicillin. LB agar containing the appropriate antibiotic was poured into petri dishes, left to set and dry overnight at room temperature, and stored at 4 °C.

242

9.2.2 Transformation Buffer I

Transformation buffer I was prepared with 0.3 g potassium acetate (KCH3COO), 0.15 g

CaCl2, 1 g magnesium chloride (MgCl2), 1.21 g rubidium chloride (RbCl) and 15 mL

Glycerol, made up to 100 mL with Milli-Q H2O, adjusted to pH 5.6 with acetic acid and sterilised by passing through a 0.45 μm filter.

9.2.3 Transformation Buffer II Transformation buffer II was prepared with 1.1 g CaCl2, 0.12 g RbCl, 0.2 g 3-(N- Mopholino)propanesulfonic (MOPS) acid and 15 mL Glycerol, made up to 100 mL with Milli-Q H2O, adjusted to pH 6.5 with 1 M potassium hydroxide (KOH) and sterilised by passing through a 0.45 μm filter.

9.3 Electrophoresis Reagents

9.3.1 50X TAE Buffer 50X TAE buffer was prepared with 242 g Tris base, 57 mL acetic acid and 100 mL 0.5 M

EDTA pH 8, made up to 1000 mL with dH2O.

9.3.2 Agarose Gels Agarose gels were prepared by adding 0.6 g (1%) or 1.2 g (2%) DNA grade agarose (HydraGene, China) to 60 mL 1X TAE, microwaving for 2 minutes or until completely dissolved and melted, and cooled slightly before pouring into gel moulds. Gels were allowed to set for 30 minutes at room temperature.

9.3.3 2X Sample Buffer

2X sample buffer was prepared with 500 μL 10% SDS in Milli-Q H2O, 200 μL glycerol, 120

μL 1M Tris-HCl at pH 6.8, 10 μL 1% bromophenol blue and 70 μL Milli-Q H2O. 5 μL of 2- mercaptoethanol was added to 45 μL of 2X sample buffer just prior to use.

243

9.3.4 SDS-PAGE Stacking Gel Buffer

Stacking gel buffer was prepared with 0.4 g SDS and 6.04 g Tris in 70 mL Milli-Q H2O. The solution was adjusted to pH 6.8 with HCl, and then topped up with Milli-Q H2O to a total volume of 100 mL.

9.3.5 SDS-PAGE Resolving Gel Buffer

Resolving gel buffer was prepared with 0.4 g SDS and 18.2 g Tris in 70 mL Milli-Q H2O.

The solution was adjusted to pH 8.8 with HCl, and then topped up with Milli-Q H2O to a total volume of 100 mL.

9.3.6 SDS-PAGE Gel Production A 10% resolving gel solution was prepared with 1.9 mL 40% acrylamide, 1.9 mL resolving gel buffer, 3.8 mL Milli-Q H2O, 37.5 μL 10% ammonium persulphate (APS) and 7.5 μL tetramethylethylenediamine (TEMED). This solution was added to the interlocked space between glass SDS-PAGE gel plates (Bio-Rad, California, USA) until around half-full. A thin layer of water-saturated butanol was layered on top, and the cassette was agitated slightly to produce an even horizontal interface. The gel was then allowed to set for 45 minutes at room temperature. A 4% stacking gel was prepared with 0.25 mL 40% acrylamide, 1.25 mL stacking gel buffer, 1 mL Milli-Q H2O, 17.5 μL 10% APS and 3.5 μL TEMED. The water- saturated butanol was poured of the set resolving gel, and excess was blotted with filter paper. 4% stacking gel solution was layered over the resolving gel, filling the remaining of the cassette up to and around the inserted comb. The gel was then allowed to set for 45 minutes at room temperature. SDS-PAGE gels were stored at 4 °C for up to 7 days prior to use, sealed in a plastic bag containing tissues moistened with dH2O to prevent dessication.

9.3.7 10X SDS-PAGE Electrophoresis Buffer 10X SDS-PAGE electrophoresis buffer was prepared with 144 g glycine, 30 g Tris and 10 g

SDS with sufficient dH2O to bring the total volume to 1000 mL.

244

9.3.8 Coomassie Blue Stain Coomassie blue stain solution was prepared with 1.25 g Coomassie brilliant blue G-250 dissolved in 225 mL methanol, 45 mL acetic acid and 230 mL Milli-Q H2O. The stain was filtered through Whatman filter paper to remove insoluble precipitates.

9.3.9 Destain Solution

Destain solution was prepared with 100 mL methanol, 100 mL acetic acid, and 800 mL dH2O.

9.4 Western Blot Reagents

9.4.1 Anode Buffer I

Anode buffer I was prepared with 36 g Tris and 100 mL methanol, with sufficient dH2O as to bring the total volume up to 1000 mL, at pH 10.4.

9.4.2 Anode Buffer II

Anode buffer II was prepared with 3.03 g Tris and 100 mL methanol, with sufficient dH2O as to bring the total volume up to 1000 mL.

9.4.3 Cathode Buffer Cathode buffer was prepared with 3.03 g Tris, 3 g glycine and 100 mL methanol, with sufficient dH2O as to bring the total volume up to 1000 mL, at pH 9.4.

9.4.4 PBS-T Wash Buffer PBS-T wash buffer was prepared with 20 mL 10X PBS and 1 mL Tween 20 (Sigma-Aldrich,

Montana, USA), with sufficient dH2O as to bring the total volume up to 2000 mL.

245

9.5 Cell Culture Reagents

9.5.1 Trypan Blue Trypan blue cell viability stain was prepared by added 0.5 g trypan blue to sufficient Milli-

Q H2O as to produce a total volume of 10 mL. The solution was sterilised by passing it through a 0.45 μm filter.

9.5.2 SeaPlaque Agarose SeaPlaque agarose was prepared by dissolving 1.5 g of SeaPlaque (Lonza, Switzerland) in 50 mL of Milli-Q H2O. The solution was autoclaved, and stored at 4 °C.

9.5.3 cIMDM cIMDM was prepared by adding 500 μL of sterile 5.5x10-2 M 2-mercaptoethanol to 500 mL of IMDM.

9.6 VLP-specific Reagents

9.6.1 CsCl Solutions CsCl solutions were prepared with 2.86 g CsCl for 1.2 g/cm3, and 6.15 g for 1.4 g/cm3, dissolved in 10 mL Milli-Q H2O and sterilised by passing through a 0.45 μm filter.

9.6.2 Disassembly Buffer Disassembly buffer was prepared with 12.1 g Tris and 17.5 g NaCl dissolved in 1800 mL dH2O. The solution was adjusted to pH 7.5 with HCl, and sufficient dH2O was added as to bring the total volume up to 2000 mL. Disassembly buffer was autoclaved and stored at 4 °C.

246

9.7 Flow Cytometry Reagents

9.7.1 Red Cell Lysis Buffer

Red cell lysis buffer was prepared with 4.15 g ammonium chloride (NH4Cl), 0.5 g potassium hydrogen carbonate (KHCO3) and 0.0186 g EDTA, with sufficient Milli-Q H2O as to bring the total volume up to 500 mL. Red cell lysis buffer was filter sterilised, and stored at 4 °C.

9.7.2 FACS Buffer FACS buffer was prepared with 0.1 g sodium azide and 1 g BSA, with sufficient 1X PBS as to bring the total volume up to 1000 mL.

9.7.3 AutoMACS Buffer AutoMACS buffer was prepared with 29.2 g EDTA and 0.25 g BSA, with sufficient DPBS as to bring the total volume up to 50 mL. AutoMACS buffer was filter sterilised, and stored at 4 °C.

9.7.4 AutoMACS Running Buffer AutoMACS running buffer was prepared with 29.2 g EDTA, with sufficient DPBS as to bring the total volume up to 50 mL. AutoMACS running buffer was filter sterilised, and stored at 4 °C.

247

10 Appendix III: Presentations and Publications

10.1 Presentations Australasian Society for Immunology Meeting 2012; Dunedin, New Zealand

Australasian Society for Immunology Meeting 2012; Melbourne, Australia

Otago School of Medical Sciences Postgraduate Symposium 2013; Dunedin New Zealand Poster: RHDV VLP Modification for Encapsulation of Nucleic Acids

Microbiology and Immunology Department Retreat 2013; Dunedin, New Zealand Poster: Investigating Encapsulation of Nucleic Acids in RHDV VLPs for Immunomodulation and Gene Delivery

Australasian Society for Immunology Meeting 2013; Wellington, New Zealand Poster: Engineering Multi-target Chimaeric VLPs to Enhance Efficacy of Immunotherapeutic Cancer Vaccines Recipient: Australian Society for Immunology Student Travel Grant

Australasian Virology Society Meeting 2013; Queenstown, New Zealand Poster: Engineering Multi-target Chimaeric VLPs to Enhance Efficacy of Immunotherapeutic Cancer Vaccines Recipient: Australasian Virology Society Poster Presentation Award

Keystone Symposia: Immune Evolution in Cancer 2014; Whistler, Canada Poster: Engineering Multi-target Chimaeric VLPs to Enhance Efficacy of Immunotherapeutic Cancer Vaccines Recipient: Maurice and Phyllis Paykel Trust Travel Grant

Fulbright New Zealand 2014; Wellington, New Zealand

Recipient: Fulbright Science and Innovation Graduate Award

Otago School of Medical Sciences Postgraduate Symposium 2015; Dunedin, New Zealand Presentation: Combination Immunotherapy for Colorectal Cancer with Oncolytic VSV and Chimaeric RHDV VLP Vaccines Recipient: Otago School of Medical Sciences Presentation Award

International Congress of Immunology 2016; Melbourne, Australia Presentation: Multi-Epitope Chimaeric VLP as a Therapeutic Vaccine for Colorectal Cancer Recipient: Australian Society for Immunology Student Travel Grant

248

10.2 Publications Donaldson, B., Al-Barwani, F., Young, V., Scullion, S., Ward, V. K., & Young, S. L. (2014). Virus-like particles, a versatile subunit vaccine platform. C. Foged, T. Rades, Y. Perrie, & S. Hook (Eds.). Springer. Al-Barwani, F., Donaldson, B., Pelham, S. J., Young, S. L., & Ward, V. K. (2014). Antigen delivery by virus-like particles for immunotherapeutic vaccination. Therapeutic delivery, 5(11), 1223-1240.

Donaldson, B., Walker, G., Young, S., Ward V. K. (2016), Rapid Detection of Virus-like Particle Disassembly and Reassembly. Journal of Virological Methods (In Review)

Donaldson, B., et al (2016), Multi-Target Chimaeric VLP as a Therapeutic Vaccine in a Model of Colorectal Cancer. (In Draft)

249