murine polyomavirus vlps as a platform for cytotoxic t ...for bioengineering and nanotechnology, the...

Murine Polyomavirus VLPs as a Platform for Cytotoxic T Cell Epitope-Based Influenza

Vaccine Candidates

Rufika Shari Abidin

S.Si., M.Biotech., M.Kes.

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland in 2016

Australian Institute for Bioengineering and Nanotechnology

i

Abstract

This research examined the ability of virus-like particles (VLP) assembled from the coat proteins of

Murine polyomavirus (MPyV) to carry cytotoxic T cell (Tc) epitopes. MPyV VLPs can be assembled

in vitro from purified subunits composed of pentamers of the major coat protein VP1 called

capsomeres; and this approach provides fine control over VLP composition and structure. Tc-epitopes

are often highly hydrophobic, which poses a significant bioengineering challenge and two distinct

approaches were pursued to determine the optimal delivery strategy of Influenza Tc-epitopes by

MPyV VLP. Firstly, identified epitopes were externally presented using established sites for peptide

insertion on the MPyV major coat protein, VP1. Secondly, polypeptides possessing multiple epitopes

were encapsidated within VP1 VLP using precise elements of the minor coat protein VP2/3 that

interact with the interior cavity of capsomeres. Hydrophobicity-related capsomere aggregation arising

from single Tc epitope presentation was successfully circumvented by engineering charged amino

acids (DD) flanking the epitope displayed on a surface-exposed loop of VP1 and by use of an

optimised buffer condition. A multiparametric, high-throughput buffer screen was implemented to

optimise pH and buffer additives during affinity tag removal. Stable modified capsomere recovered

from this reaction were assembly competent in the presence of L-Arginine, and VLP yield was high

when modular capsomeres were co-assembled with unmodified VP1 capsomeres forming stable

mosaic VLPs. The ability of the MPyV VLP to encapsidate foreign protein was first evaluated and

optimised with green fluorescent protein (GFP) as a model protein to enable monitoring and analysis.

A minimal VP1-interacting sequence was defined from the carboxy-terminus of VP2/3 (VP2C) and

fused to the amino-terminus of GFP, ensuring internalisation of the reporter protein. VP1:VP2C-GFP

capsomere complexes were successfully recovered when VP1 was co-expressed with VP2C-GFP,

whereas complex formation was not observed when VP1 and VP2C-GFP were mixed in vitro.

Recovered capsomere complexes were assembly competent and amount of encapsidated GFP was

controllable when VP1:VP2C-GFP capsomere complexes were co-assembled with unmodified VP1

capsomeres in a defined molar ratio. The encapsidation approach was then applied to a subdomain of

the Influenza matrix protein M1 by co-expressing VP1 with VP2C-M1-I. M1-I capsomere complexes

were assembly competent as indicated by gel electrophoresis, dynamic light scattering, and

transmission electron microscopy. To enable recovery of VLPs for analysis and characterisation as

well as to confirm encapsidation and the formation of mosaic VLPs, a semi-preparative size exclusion

chromatography method was developed. This research was able to address bioengineering challenges

posed by hydrophobic epitope presentation and developed MPyV VLPs into effective carriers of

hydrophobic peptides such as influenza Tc-epitopes which may have use as a broadly protective

influenza vaccine.

ii

Declaration by author

This thesis is composed of my original work, and contains no material previously published

or written by another person except where due reference has been made in the text. I have clearly

stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance,

survey design, data analysis, significant technical procedures, professional editorial advice, and any

other original research work used or reported in my thesis. The content of my thesis is the result of

work I have carried out since the commencement of my research higher degree candidature and does

not include a substantial part of work that has been submitted to qualify for the award of any

other degree or diploma in any university or other tertiary institution. I have clearly stated which parts

of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and,

subject to the policy and procedures of The University of Queensland, the thesis be made available

for research and study in accordance with the Copyright Act 1968 unless a period of embargo has

been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright

holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright

holder to reproduce material in this thesis

iii

Publications during candidature

Conference abstracts

Abidin RS, Middelberg APJ, Sainsbury F. Multiparametric buffer screening to prevent aggregation

of engineered MuPyV capsomeres during processing. NanoBio Australia 2014; 2014 7-10 July. The

International Conference on BioNano Innovation Australia. Brisbane, Australia. Australian Institute

for Bioengineering and Nanotechnology, The University of Queensland. Poster presentation.

Abidin RS. Insert design and high-throughput screening to improve recovery of virus-like particle

subunits for cytotoxic T cell epitope-based influenza vaccines. Inspired by Biology: Exploring

Nature's Toolbox; 2014 23-25 October. 16th EMBL PhD Symposium. Heidelberg, Germany.

European Molecular Biology Laboratory. Oral and poster presentation.

Abidin RS, Middelberg APJ, Sainsbury F. Encapsulation of EGFP via fusion to VP2 minor coat

protein inside murine polyoma virus-like particles. BioNano Innovations; 2015 2 October. ASA

Student Conference 2015. Brisbane, Australia. Australian Institute for Bioengineering and

Nanotechnology, The University of Queensland. Oral presentation.

Peer-reviewed papers

Abidin RS, Lua LH, Middelberg AP, Sainsbury F. 2015. Insert engineering and solubility screening

improves recovery of virus-like particle subunits displaying hydrophobic epitopes. Protein Science

24(11):1820-8.

iv

Publications included in this thesis



24(11):1820-8 – incorporated as Chapter 3.

Contributors Statement of contribution

Rufika Shari Abidin (Candidate) Designed experiments (30%)

Conduct experiments (100%)

Trouble-shoot experiments (70%)

Analyse results (60%)

Wrote the paper (80%)

Linda H. Lua Designed experiments (20%)

Revised the paper (20%)

Anton P. Middelberg Designed experiments (20%)


Frank Sainsbury Designed experiments (30%)

Trouble-shoot experiments (30%)

Analysed results (40%)

Wrote the paper (20%)


v

Contributions by others to the thesis

This thesis was drafted and written by the Candidate under the supervision of Dr Frank Sainsbury

and Professor Anton Middelberg.

Sainsbury, F. contributed in conception and design of the project, non-routine technical work, analysis

and interpretation of research data, and critically revising drafts so as to contribute to the

interpretation. Middelberg, A. contributed to conception and design of the project, and critically

revising drafts.

A Master student, Sri Kartika Wijaya, conducted research and analysis under Candidate’s supervision

of data presented in this thesis on Section 4.3.1 (including Table 4.1), and Section 4.4.1 (including

Figure 4.2), submitted for MMolBiol(adv)., The University of Queensland, 2014. Degree awarded

19 December 2014.

Professional editor, Robyn Kent, provided proof reading services for Chapter 5, according to the

guidelines laid out in the university-endorsed national ‘Guidelines for editing research theses’.

All of the data and results presented in this thesis are solely the work of the Candidate with the

exception of the following figures:

Figure Contributors

Figure 3.6 B Transmission electron micrographs (TEM) were taken by Frank Sainsbury

Figure 4.2 Data was collected and analysed by Sri Kartika Wijaya under Candidate’s

supervision.

Figure 4.3 A & B Construct design and cloning was performed with Frank Sainsbury.

Figure 4.6 C Western blot was performed with Frank Sainsbury.

Figure 4.7 A & B Asymmetric flow field flow fractionation (AF4) was performed by Frank

Sainsbury. TEM was taken by Frank Sainsbury.

Figure 4.8 Data co-collected and co-analysed by Hui Jean Liem under Candidate’s

supervision. TEMs were taken by Frank Sainsbury.

Figure 5.2 C TEM was taken by Frank Sainsbury.

Figure 5.3 Cloning was performed by Frank Sainsbury.

Figure 5.5 C & D Western blot was performed with the help of Philippe Varennes-Jutras.



vi

Statement of parts of the thesis submitted to qualify for the award of another degree

None.

vii

Acknowledgements

All praise is due to The Almighty, Allah SWT, Lord of the worlds. My Protector. My Sustainer. The All-Knowing. The All-Wise.

Firstly, I would like to cordially express my sincere gratitude to my Principal Advisor Dr. Frank Sainsbury for the continuous support, patience, and mentorship during my PhD studies and related research. His valuable guidance and encouragements have helped me throughout the challenging phases of my PhD research. I would like to cordially express my gratitude to my Advisor Professor Anton Middelberg, for insightful comments, and helpful advice. Moreover, I am grateful to both of my Advisors for the opportunity to conduct cutting-edge research and to gain advanced knowledge and experience in the exciting field of Vaccine and Protein Engineering under their advisorship.

My PhD studies was financially supported by the University of Queensland International Scholarship, and Australian Institute for Bioengineering and Nanotechnology Group Leader Top-Up Scholarship. Therefore, I would like to thank these organisations for financially supporting my studies in Australia. I also would like to thank the European Molecular Biology Laboratory Australia for the Travel Grant to present at 16th EMBL PhD Symposium in Heidelberg, Germany in 2014.

For insightful discussions I would like to thank Associate Professor Linda Lua, Dr. Nani Wibowo, Dr. Christopher de Bakker, and Alemu Tekewe Mogus. In addition, I would like to thank Hui Jean Liem for laboratory assistance during her summer research. I would also like to thank the Protein Expression Facility for technical support. For excellent administrative support I would like to thank the AIBN RHD Program Team.

For friendship, warm collegiality, and encouragements, I would like to thank the CBE Team, especially Sri Kartika Wijaya, Hossam Tayeb, Phillippe Varennes-Jutras, Noor Dashti, Emeline Catrice, and Mark Clayton. To my brothers and sisters in faith at the MSA-UQ I would like to express my gratitude for the mutual support, especially to Khairatun Najwa for friendship and collaboration. I also would like to thank the Indonesian community at the AIBN for the camaraderie, especially Swasmi Sutanto and Amanda Septevani for selfless support. Thank you for my housemates Zakiyah and Anita Ratna for togetherness. For lifelong friendship and constant support, I am cordially grateful to Suryani Aris, Yusmaeni Dahlan, and Ulya Zainal. Cordial appreciation to the ladies at Chai Wives for supportive chats and discussions as well. I would like to extend my gratitude to my healthcare team: Dr. Andrea Panitz, Dr. Robert Moyle, Dr. Maree Mungomery, Tuyet Pham, and Sybilla Wilson whose professional care, support, and advice have kept me going.

I would not have made it this far without the never-ending support and prayers of my family and thus, I would like to cordially extend my gratitude to my parents, Dr. Ir. H. Baharuddin Abidin and Hj. Hasnawati Parenrengi; my mother in-law, Nazir Begum and her family; and my siblings, Thauhuriah Adliah, Rizkie Aliah Muhammad, Hikmah Mujtahidah, Amaliah Imaniah, Ikhlasul Islam Sabil Isni Ghani and their respective families. Last but not least, I would like to cordially express my gratitude to my dearest husband, Muhammad Usman Asghar Bajwa, for the meaningful and invaluable support, care, motivation, prayers, love, and companionship. It is still great to be a part of all of you.

viii

Keywords

virus-like particles, vaccine platform, influenza, protein engineering, bioprocess development,

encapsidation, Tc-epitope, high-throughput screen, aggregation, controlled self-assembly

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 090301, Biomaterials, 40 %

ANZSRC code: 100306, Industrial Molecular Engineering of Nucleic Acids and Proteins, 40%

ANZSRC code: 030403, Characterisation of Biological Macromolecules, 20%

Fields of Research (FoR) Classification

FoR code: 0903, Biomedical Engineering, 40%

FoR code: 1003, Industrial Biotechnology, 40%

FoR code: 0304, Medicinal and Biomolecular Chemistry, 20%

ix

Table of Contents Abstract ................................................................................................................................................. i

Declaration by author ........................................................................................................................... ii

Publications during candidature .......................................................................................................... iii

Publications included in this thesis ..................................................................................................... iv

Contributions by others to the thesis .................................................................................................... v

Statement of parts of the thesis submitted to qualify for the award of another degree ....................... vi

Acknowledgements ............................................................................................................................ vii

Keywords .......................................................................................................................................... viii

Australian and New Zealand Standard Research Classifications (ANZSRC) .................................. viii

Fields of Research (FoR) Classification ........................................................................................... viii

Table of Contents ................................................................................................................................ ix

List of Figures ................................................................................................................................... xiv

List of Tables .................................................................................................................................... xvi

List of Abbreviations ....................................................................................................................... xvii

1 Project Overview.......................................................................................................................... 1

1.1 Influenza Tc-epitope-based vaccines .................................................................................... 4

1.2 Murine polyoma viral coat proteins as a vaccine delivery system ........................................ 8

1.3 Research Aim and Objectives ............................................................................................. 11

1.3.1 Cytotoxic T cell epitope delivery by external presentation ......................................... 11

1.3.2 Optimization of fusion protein encapsidation within MPyV VLP .............................. 12

1.3.3 Characterisation and analysis of MPyV VLP based influenza vaccines via external and

internal presentation ................................................................................................................... 13

1.4 Thesis organisation .............................................................................................................. 13

2 Literature Review ....................................................................................................................... 14

2.1 Overview ............................................................................................................................. 14

2.2 The murine polyomavirus-like particle vaccine delivery technology ................................. 14

2.2.1 Established MPyV Vaccine Platform .......................................................................... 20

x

2.2.1.1 Capsomere platform ............................................................................................................ 21

2.2.1.2 VLP platform ....................................................................................................................... 21

2.2.2 Further development for hydrophobic Tc-epitope presentation .................................. 22

2.2.3 Problems related to hydrophobic Tc-epitope delivery ................................................. 24

2.2.3.1 Mosaic particles ................................................................................................................... 25

2.2.3.2 Protein Encapsidation by MPyV ......................................................................................... 25

2.3 MPyV VLP-based influenza Tc-epitope vaccines .............................................................. 28

2.3.1 Influenza....................................................................................................................... 28

2.3.2 Current influenza vaccines and their limitations.......................................................... 31

2.3.3 The mechanism of cytotoxic T cell activity ................................................................. 34

2.3.4 Advantages of Tc-epitope delivery with MPyV VLP platform ................................... 37

2.4 Size exclusion chromatography ........................................................................................... 40

2.4.1 Principles of size exclusion chromatography ............................................................... 40

2.4.2 Column efficiency and influencing factors .................................................................. 42

3 Insert engineering and solubility screening improves recovery of virus-like particle subunits

displaying hydrophobic epitopes ....................................................................................................... 46

3.1 Abstract ............................................................................................................................... 47

3.2 Introduction ......................................................................................................................... 47

3.3 Results ................................................................................................................................. 49

3.3.1 MPyV VP1 displaying Influenza cytotoxic T cell epitopes ......................................... 49

3.3.2 Aggregation of modified VP1 during affinity tag removal.......................................... 49

3.3.3 High-throughput buffer screen ..................................................................................... 49

3.3.4 SEC using selected buffer conditions for capsomere recovery .................................... 57

3.4 Discussion ........................................................................................................................... 57

3.5 Materials and Methods ........................................................................................................ 61

3.5.1 Insertion of cytotoxic T cell epitopes into MPyV VP1 ................................................ 61

3.5.2 Expression and Purification of MPyV VP1 capsomeres ............................................. 61

3.5.3 Buffer Condition Screening ......................................................................................... 62

3.5.4 Absorbance and turbidity ............................................................................................. 62

xi

3.5.5 Dynamic Light Scattering (DLS) ................................................................................. 62

3.5.6 Size exclusion chromatography ................................................................................... 63

3.5.7 In vitro VLP Assembly ................................................................................................ 63

3.5.8 Electron Microscopy .................................................................................................... 63

3.6 Supplementary Material ...................................................................................................... 63

3.7 Acknowledgements ............................................................................................................. 64

4 Encapsidation of GFP inside MPyV VLP via fusion with VP2C .............................................. 70

4.1 Abstract ............................................................................................................................... 70

4.2 Introduction ......................................................................................................................... 70

4.3 Methods ............................................................................................................................... 71

4.3.1 Cloning of free VP2C-GFP .......................................................................................... 71

4.3.2 Cloning of VP1:VP2C-GFP co-expression ................................................................. 74

4.3.3 Expression .................................................................................................................... 74

4.3.4 Affinity chromatography purification .......................................................................... 74


4.3.5.1 Capsomere SEC (S200) ....................................................................................................... 76

4.3.5.2 VLP SEC (S500) ................................................................................................................. 76

4.3.6 N-terminal sequencing ................................................................................................. 76

4.3.7 Dot blot ........................................................................................................................ 76

4.3.8 Assembly ...................................................................................................................... 77

4.3.9 Dynamic Light Scattering ............................................................................................ 77

4.3.10 Electron Microscopy .................................................................................................... 77

4.3.11 Asymmetrical Flow Field Flow Fractionation ............................................................. 77

4.4 Results ................................................................................................................................. 78

4.4.1 VP2C-GFP failed to bind to VP1 when mixed in vitro ............................................... 78

4.4.2 Design of VP1 and VP2C-GFP co-expression ............................................................ 78

4.4.3 Expression and solubility test ...................................................................................... 81

4.4.4 IMAC indicated successful binding between VP2C-GFP and VP1 ............................ 81

xii

4.4.5 Isolation of VP1:VP2C-GFP interaction from free VP2C-GFP by size exclusion

chromatography ......................................................................................................................... 84

4.4.6 Western blot to analyse purified SEC fractions ........................................................... 84

4.4.7 VP1:VP2C-GFP was able to assemble into VLP ........................................................ 86

4.4.8 Controlled GFP encapsidation was achievable by co-assembly with wt capsomeres . 86

4.5 Discussion ........................................................................................................................... 89

4.6 Conclusion ........................................................................................................................... 91

5 Cytotoxic T Cell Epitope-Based MPyV VLP Influenza Vaccine Development, Analysis, and

Characterisation ................................................................................................................................. 92

5.1 Abstract ............................................................................................................................... 92

5.2 Introduction ......................................................................................................................... 92

5.3 Materials and methods......................................................................................................... 94

5.3.1 Column packing ........................................................................................................... 94

5.3.2 Column efficiency testing ............................................................................................ 94

5.3.3 Cloning of VP1:VP2C-M1-I and VP1:VP2C-M1-II ................................................... 94

5.3.4 Expression .................................................................................................................... 95

5.3.5 Affinity chromatography purification .......................................................................... 95


5.3.6.1 Capsomere SEC (S200) ....................................................................................................... 95

5.3.6.2 VLP SEC (S500) ................................................................................................................. 98

5.3.7 Western blot ................................................................................................................. 98

5.3.8 Assembly ...................................................................................................................... 98

5.3.8.1 Standard wtVP1 assembly ................................................................................................... 98

5.3.8.2 VP1:VP2C-M1-I assembly .................................................................................................. 98

5.3.8.3 Mosaic DDIT1 assembly ..................................................................................................... 98

5.3.9 Dynamic light scattering .............................................................................................. 99

5.3.10 Electron microscopy .................................................................................................... 99

5.4 Results ................................................................................................................................. 99

5.4.1 Optimisation of a new S500 column ............................................................................ 99

5.4.2 Column validation with wt MPyV VLPs ................................................................... 102

xiii

5.4.3 Design of encapsidated influenza M1 protein construct ............................................ 102

5.4.4 IMAC purification of VP1:VP2C-M1-I indicated successful binding with VP1 ...... 102

5.4.5 VP2C-M1-II oligomerisation complicates complex purification .............................. 105

5.4.6 Development, analysis, and characterisation of M1-I-based VLPs as vaccines ........ 105

5.4.7 Development, analysis, and characterisation of DDIT1-based VLPs as vaccines ..... 109

5.5 Discussion ......................................................................................................................... 109

5.6 Conclusion ......................................................................................................................... 113

6 Conclusion and Future Direction ............................................................................................. 114

6.1 Summary of research findings ........................................................................................... 114

6.1.1 External Tc-epitope presentation ............................................................................... 115

6.1.2 Optimisation of encapsidation ................................................................................... 116

6.1.3 Tc-epitope-based MPyV VLP candidate vaccines..................................................... 118

6.1.4 SEC-S500 to analyse and characterise VLPs ............................................................. 121

6.2 Future Direction ................................................................................................................ 122

6.3 Concluding Remarks ......................................................................................................... 125

List of References ............................................................................................................................ 127

xiv

List of Figures Figure 1.1. Lessons from recent 2009 H1N1 pandemic. ..................................................................... 2

Figure 1.2 M1 protein soluble domain and epitope distribution. ......................................................... 6

Figure 1.3 The MPyV vaccine delivery platform. ............................................................................. 10

Figure 2.1 Structure of MPyV VP1, capsomere, and VLP. ............................................................... 17

Figure 2.2 MPyV capsomere and VLP microbial platform established at AIBN-CBE .................... 19

Figure 2.3 VP1:VP2 interaction. ........................................................................................................ 26

Figure 2.4 Schematic diagram of the influenza virus. ....................................................................... 29

Figure 2.5 The mechanism of Tc activity. ......................................................................................... 36

Figure 2.6 Principles of size exclusion chromatography. .................................................................. 41

Figure 2.7 Zone broadening variables................................................................................................ 43

Figure 3.1 Schematic representation of designed and cloned MPyV VP1 constructs. ...................... 50

Figure 3.2 Modified capsomere aggregation during GST removal. .................................................. 51

Figure 3.3 Spectrophotometric analysis of the GST removal reaction. ............................................. 52

Figure 3.4 Validation of buffer screening with wtVP1. ..................................................................... 55

Figure 3.5 Heat maps identifying optimal buffer conditions for the processing of modified

capsomeres. ........................................................................................................................................ 56

Figure 3.6 Isolation of modified capsomere, DDIT1 using optimised buffer conditions. ................. 58

Supplementary Figure 3.1 Confirmation of thrombin activity under the buffer conditions used in this

study. .................................................................................................................................................. 66

Supplementary Figure 3.2 Aggregation status of each construct subject to the buffer screen. ......... 67

Supplementary Figure 3.3 Normalized aggregation index at the end of the tag removal reaction. ... 68

Supplementary Figure 3.4 Assessment of screen hits by SEC. ......................................................... 69

Figure 4.1 Design of encapsidation within the MPyV VLP via VP2C fusion................................... 72

Figure 4.2 VP2C-eGFP expressed separately from VP1 failed to bind when mixed in vitro. .......... 79

Figure 4.3 Design of VP2C-eGFP and VP1 co-expression. .............................................................. 80

Figure 4.4 Purification of VP1:VP2C-eGFP 31 aa. ........................................................................... 82

Figure 4.5 Purification of VP1:VP2C-eGFP 23 aa. ........................................................................... 83

Figure 4.6 Purification of VP1:VP2C-GFP complex......................................................................... 85

Figure 4.7 VLP formation after assembly as confirmed by AF4 and TEM. ...................................... 87

Figure 4.8 Controlled encapsidation. ................................................................................................. 88

Figure 5.1 VP2C-M1 constructs. ....................................................................................................... 97

Figure 5.2 Column efficiency test. ................................................................................................... 101

Figure 5.3 wtVP1 VLP was tested with S500 5/200 column. ......................................................... 103

xv

Figure 5.4 IMAC of VP1:VP2C-M1-I. ............................................................................................ 104

Figure 5.5 S200 of VP1:VP2C-M1-I. .............................................................................................. 106

Figure 5.6 S200 of VP1:VP2C-M1-II. ............................................................................................. 107

Figure 5.7 S500 of VP1:VP2C-M1-I. .............................................................................................. 108

Figure 5.8 S500 of mosaic VP1 S1S4 DD IT1. ............................................................................... 110

xvi

List of Tables Table 1.1 Current developments in influenza Tc-epitope delivery system.......................................... 5

Table 2.1 MPyV-based candidate vaccines. ...................................................................................... 23

Table 2.2 Currently licensed influenza vaccine. ................................................................................ 33

Table 2.3 Advantages of MPyV VLP platform for Tc-epitope-based vaccines. ............................... 38

Table 3.1 Influenza T cell epitopes used in this study. ...................................................................... 50

Table 3.2 Additives used in buffer screening. ................................................................................... 54

Supplementary Table 3.1 Oligonucleotides used in epitope insertion. .............................................. 65

Table 4.1 Oligonucleotides used in the cloning of free VP2C-GFP. ................................................ 73

Table 4.2 Oligonucleotides used in the cloning of VP1:VP2C-GFP co-expression vectors. ............ 75

Table 5.1 Oligonucleotides used in the cloning of VP1:VP2C-M1-I and VP1:VP2C-M1-II. .......... 96

Table 5.2 Calculation of variables for column efficiency testing. ................................................... 100

xvii

List of Abbreviations

6His 6x Histidine

AdDd Adenoviral dodecahedron

AF4 Asymmetric flow-field flow fractionation

AI Aggregation Index

AIBN Australian Institute for Bioengineering and Nanotechnology

AIDS Acquired immunodeficiency syndrome

APC Antigen presenting cell

BSL Biosafety level

CapM2e Capsomeres carrying the M2e epitope

CBE Center for Biomolecular Engineering

CCMV Cowpea chlorotic mottle virus

CFE Chicken embryo fibroblast

CIP Calf intestinal phosphatase

DLS Dynamic light scattering

DTT Dithiothreitol

EDTA Ethylenediaminetetraacetic acid

ER Endoplasmic reticulum

ERAAP ER aminopeptidase associated with antigen processing

FDA Food and Drug Administration

FRET Fluorescence resonance energy transfer

GAS Group A Staphylococcus

GDP Gross domestic product

GFP Green fluorescence protein

H190 Helix 190

HA Hemagglutinin

HB Hepatitis B

HbcAg Hepatitis B core antigen

HbsAg Hepatitis B surface antigen

HE Hepatitis E

HINI-SOIV Strain H1N1 of swine-origin influenza virus strain

HIV Human immunodeficiency virus

xviii

HLA Human leukocyte antigen

HPV Human papilloma virus

HPyV Hamster polyomavirus

IEX Ion exchange chromatography

IIV Inactivated influenza vaccines

Ig Immunoglobulin

IL Interleukin

IMAC Immobilised metal affinity chromatography

IPTG Isopropyl-β-D-thiogalactopyranoside

iSE-UHPLC Interlaced size exclusion ultra-high performance chromatography

Kd Dissociation constant

L Length of the column

LAIV Live-attenuated influenza vaccines

L-Arg L-Arginine

M1 Matrix 1

M2 Matrix 2

M2e M2 ectodomain

MCS Multiple cloning site

MHC Major histocompatibility complex

MPyV Murine polyomavirus

MVA Modified virus Ankara

MW Molecular weight

N Number of theoretical plates

NA Neuraminidase

NP Nucleoprotein

NS2 Non-structural protein 2

NV Nipah virus

PA Polymerase acidic protein

PapMV Papaya mosaic virus

PAMP Pathogen associated molecular patterns

PCA Polymerase cyclic assembly

PB1 Polymerase basic protein 1

PB2 Polymerase basic protein 2

PI Protease inhibitor

xix

RNP Ribonucleoprotein

RT Room temperature

SEC Size exclusion chromatography

S200 Size exclusion chromatography in Superdex 200 for capsomeres

S500 Size exclusion chromatography in Sephacryl S500 for VLPs

SOE Splice overlap extension

SV40 Simian virus 40

Tc Cytotoxic T cell

TCR T cell receptor

TEM Transmission electron microscopy

TFH T follicular helper cell

Th1 T helper cell type 1



Treg Regulatory T cell

VLP Virus-like particle

VP2C A variety of minimal sequences omitting some part of N-terminal region but

still retaining the hydrophobic binding site of VP2.

WHO World Health Organisation

wt Wild type

1

1 Project Overview

The prevention of an influenza pandemic still faces many challenges (WHO 2011). Firstly,

conventional antibody based influenza vaccines have to be updated frequently to match mutating

influenza strains (Subbarao and Joseph 2007). Secondly, conventional influenza vaccines can only

become available approximately six months after identification of the pandemic strain, while global

spread can occur within two months (Broadbent and Subbarao 2011; WHO 2011), as observed in

recent 2009 H1N1 pandemic (Figure 1.1). Thirdly, current vaccine production capacity can only

cover doses for 12.8% of the global population (Broadbent and Subbarao 2011; PRB 2009). And last,

but not least, vaccine production is still costly and requires technology transfer for developing

countries where pandemic strains are more likely to emerge (Fedson 2009; Pagliusi et al. 2013a;

Pagliusi et al. 2013b).

Influenza outbreaks can either be epidemic or pandemic. Epidemic influenza is seasonal, occurring

during winter in temperate regions. While the majority of normal people can recover from it, it still

poses a major health risk to the very young, the elderly, and the chronically ill. Epidemic influenza is

estimated to cause 250 000 – 500 000 deaths per year (WHO 2009a). Pandemic influenza is caused

by sometimes highly lethal, but mainly highly transmissible strain that can accelerate the global

spread of the virus globally within a short time. A pandemic strain usually arises from zoonotic strains

that jump into the human host by genetic drift, genetic shift, and genetic reassortment (Herfst et al.

2014; Sriwilaijaroen and Suzuki 2014). As these strains are not usually circulating in humans,

immunity against these strains is generally weak, which results in accelerated spread and higher

pathogenicity of the virus (McAuley et al. 2015). The estimated death toll of the recent H1N1 strain

of swine-origin influenza virus (H1N1-SOIV) 2009 pandemic is estimated to be nearly 300 000

during the first year of virus circulation (Dawood et al. 2012).

Current influenza vaccines are mostly made of whole-virus inactivated vaccines grown in

embryonated chicken eggs (Lambert and Fauci 2010). Production of this type of vaccine is limited

by chicken egg availability, and this becomes a major limiting factor in pandemics by strains

originating from chicken (Milian and Kamen 2015; Singh et al. 2010). A further advancement of

currently licensed influenza vaccine includes live-attenuated influenza vaccines also grown in eggs

(Jin and Subbarao 2015), insect cell derived subunit vaccines (FDA 2013), and cell derived

inactivated vaccines (Dormitzer 2015; Kistner et al. 1998). The latter two do not rely on chicken

2

Figure 1.1. Lessons from recent 2009 H1N1 pandemic. Pandemic influenza was declared by the WHO two months after the first H1N1 case, but vaccination only started six months later when most countries globally have been affected and cases have risen to almost half a million. Based on WHO 2009 Influenza Pandemic data (WHO 2009b).

3

egg production and may reduce production time (Vlecken et al. 2013). However, production cost is

high due to employment of cell cultures with strict biosafety requirements (Pattenden et al. 2005;

Perdue et al. 2011). Administered influenza vaccines usually contain antigens from two influenza A

strains and one influenza B strain matching identified circulating strains in the environment (Lambert

and Fauci 2010). Recently a quadrivalent vaccine consisting of an additional influenza B strain has

been available in the market (Gorse et al. 2015; Pepin et al. 2013). As influenza virus mutates

frequently, strain identification has to be done periodically to update the vaccines for each season.

However, traditional vaccine production takes 4 to 6 months to complete and might not match the

newly mutated circulating strain (Milian and Kamen 2015; Soema et al. 2015).

In order to improve influenza pandemic and epidemic preparedness, as well as to reduce economic

burden, new strategies have to be developed. A broadly cross-protective influenza vaccine could

reduce updating frequency or provide some level of protection to slow initial stages of a pandemic

(Brown and Kelso 2009). This may be achieved by targeting the more conserved internal proteins to

stimulate cellular immunity (Houser and Subbarao 2015; Milian and Kamen 2015; Soema et al.

2015), by delivering conserved Tc-epitopes into dendritic cells to be presented via the Major

Histocompatibility Complex (MHC) Class I pathway (Brown and Kelso 2009; Sette et al. 2002;

Woolard and Kumaraguru 2010). A versatile delivery system capable of performing this function

needs to be developed while maintaining short production time, low production cost, as well as high

production capacity.

My research project used MPyV viral coat proteins VP1 and VP2 to deliver identified conserved

influenza Tc-epitopes. Delivery of these epitopes was done by either external modular insertion onto

surface-exposed loop, or internal encapsidation of M1 protein containing these epitopes. The aim is

to develop a MPyV-based vaccine to stimulate cellular immunity against influenza. This type of

vaccine would be broadly cross-protective, and potentially be used as prepandemic vaccine before

antibody based conventional vaccine technology becomes available. Utilizing the MPyV VP1

delivery platform is advantageous because the proteins can be grown in Escherichia coli which will

reduce production cost, shorten production time, and optimise yield. This may, in turn, increase

affordability in developing countries where pandemic strains are more likely to emerge, accelerate

preventive response against seasonal and pandemic influenza outbreaks, and augment vaccine

accessibility and coverage for the global population. If the delivery of Tc-epitopes with MPyV coat

proteins to stimulate cellular immunity in influenza model is successful, it will open up the possibility

to examine Tc-epitope delivery with this platform in other disease models that require cellular

immunity for protection or therapy, such as human immunodeficiency virus (HIV) and cancer.

4

1.1 Influenza Tc-epitope-based vaccines

A Tc-epitope-based vaccines usually consist of 8 to 10-mer peptides with sequence identical to

known, conserved Tc-epitopes of a pathogen, conferring cross-reactive cellular immunity against, for

example, different influenza strains (Brown and Kelso 2009). In cellular immunity, Tc are stimulated

to kill cells infected with a pathogen via Tc-epitope presentation. Tc recognize infected cells by the

presentation of 8-10 mer pathogenic peptide (Tc-epitope) by the infected cells via the MHC Class I

presentation pathway. These Tc can only recognize the pathogenic antigen if they have been primed

before with the epitope via dendritic cells in peripheral lymphoid organs (Murphy et al. 2008). A Tc-

epitope-based vaccine aims to prime naïve Tc with the target pathogenic antigen artificially before

the host encounters the real disease-causing pathogen (Woolard and Kumaraguru 2010). This

mechanism and activity is explained in more detail in Section 2.3.2. Even though Tc-epitope-based

vaccine cannot prevent infection due to a prerequisite of pathogen entry into the cell, this type of

vaccine can reduce the severity of the disease and prevent further spread of the pathogen in the host

(Brown and Kelso 2009). Moreover, Tc-epitope presentation via the MHC Class I and MHC Class II

pathways allows targeting of conserved internal proteins of influenza virus, and thus may provide

broad protection against heterologous strains, possibly reducing vaccine updating frequency (Milian

and Kamen 2015; Soema et al. 2015). Thus, it could be used as prepandemic vaccine before

conventional vaccine specific to the pandemic strain becomes available (Woolard and Kumaraguru

2010), or between influenza seasons by boosting pre-existing T cell immunity (Gilbert 2012).

A number of bioinformatic studies have identified conserved Tc-epitopes derived from past pandemic

and seasonal strains. Some of these have further been shown to be positive Human Leukocyte

Antigen (HLA, human MHC) binders and elicit Tc immune responses (Alexander et al. 2010;

Assarsson et al. 2008; Bui et al. 2007; Heiny et al. 2007; Parida et al. 2007; Somvanshi et al. 2008;

Tan et al. 2010; Wang et al. 2007; Wang et al. 2010a). The Tc-epitope commonly used to test a

vaccine candidate is the matrix 1 (M1) protein M158-66 epitope (Table 1.1). This epitope is conserved

and is a high HLA*0201 binder (Abidin 2010). The M1 protein consists of a soluble N-terminal

domain (1-164 aa) and a C-terminal domain (165-252 aa) linked by a proteolytic loop (Zhang et al.

2012). The N-terminal domain consists of two 4-helix bundles, which in this research are termed M1-

I (2-67 aa), and M1-II (1-164). M1-II consists of M1-I in addition to the rest of the downstream amino

acids of the N-terminal domain. Beside M158-66 epitope (further termed IT1 in this thesis), other

epitopes covering a number of HLAs (Alexander et al. 2010; Assarsson et al. 2008; Babon et al. 2009;

Lee et al. 2008; Liu et al. 2013; Tan et al. 2010; Wang et al. 2007) have been identified in the M1 N-

terminal domain as depicted on Figure 1.2.

5

Table 1.1 Current developments in influenza Tc-epitope delivery system.

Delivery system Antigen type Antigen source

Expression system Development Phase

Large-scale platform

Reference

Lipopeptide Peptides M1, PA, NS1 Peptide synthesis Proof of concept None (Day et al. 2007; Deliyannis et al. 2006; Tan et al. 2013)

Liposomes Peptides PA, PB1, PB2, M2, M1

Peptide synthesis Proof of concept None (Ichihashi et al. 2011; Matsui et al. 2010)

Virosomes Peptide NP Peptide synthesis Proof of concept None (Arkema et al. 2000) Subunit protein Flagellin Recombinant

peptide HA, NP, M1 Salmonella Proof of concept Developing (Adar et al. 2009)

VLP PapMV Recombinant

peptide M1 E.coli Proof of concept None (Babin et al. 2013)

AdDd Recombinant peptide

M1 Insect cells Proof of concept Developing (Szurgot et al. 2013)

HbsAg Recombinant peptide

M1 HEK cells Proof of concept Developed (Cheong et al. 2009)

SV40 Recombinant peptide

M1 Insect cells Proof of concept Developing (Kawano et al. 2014)

Viral vector MVA Peptide or subunit

protein M1, NP CEF cells Clinical trial Developed (Berthoud et al. 2011)

6

Figure 1.2 M1 protein soluble domain and epitope distribution. 3D structure of influenza M1 soluble domain (PDB 1EA3SID) (A). In this thesis they are used in two forms, which are the M1-I subdomain (light blue) and the M1-II subdomain. M1-II consists of M1-I and additional M1-II domain (dark blue) downstream (Zhang et al. 2012). Distribution of identified Tc-epitopes based on IEDB database (www.iedb.org) along the soluble domain is depicted with epitopes shown as black and green lines (B). IT1 and IT2 epitopes are used for research in Chapter 3, and the M1 soluble domain for research in Chapter 5.

M1-I domain M1-II additional domain

Self-interacting region

M1-I domain M1-II additional domain

N-terminus

C-terminus

A

B

IT1

IT2

7

A number of delivery systems have been developed to deliver Tc-epitopes, either as a peptide or as a

protein subunit to stimulate cellular immunity against heterologous strains (Table 1.1). Lipopetides

(Day et al. 2007; Deliyannis et al. 2006; Tan et al. 2013), liposomes (Matsui et al. 2010; Uchida

2011), and virosomes (Arkema et al. 2000; Moser et al. 2011) are amongst the systems that deliver

Tc-epitope in the form of synthetic peptides, and have shown immunogenicity as a proof of concept.

Although peptide synthesis is scalable (Bray 2003), the necessary coupling with delivery platform

post-synthesis complicates the production and may add to production cost (Soema et al. 2015)

Another strategy is genetically inserting these peptide sequences into a recombinant carrier protein

such as capsid proteins in the form of virus-like particles (VLP), or subunit proteins such as flagellin.

A VLP is generally an assembly of viral coat proteins, which may be partially embedded within cell-

derived membranes in the case of VLPs derived from enveloped viruses, taking the shape of the virus

but lacking all other viral components such as DNAs and RNAs (Lua et al. 2014). The VLPs that

have been tested with inserted influenza Tc-epitopes for immunogenicity include VLPs of Papaya

Mosaic Virus (PapMV) (Babin et al. 2013; Denis et al. 2008), adenoviral dodecahedron (AdDd)

(Coughlan et al. 2015; Lambe et al. 2013; Szurgot et al. 2013), Hepatitis B surface antingen (HbsAg)

(Cheong et al. 2009), and Simian Virus 40 (SV40) (Kawano et al. 2014) (Table 1.1). They all have

shown some degree of immunogenicity and were broadly-crossprotective.

Of these, HbsAg has a well-established scale-up platform in different hosts (Hardy et al. 2000; James

et al. 2007; Shah et al. 2015; Valenzuela et al. 1982), while large scale platforms for AdDd in insect

cells (Szurgot et al. 2015), and SV40 in insect cells (Kosukegawa et al. 1996) are still in development.

The use of flagellin as a carrier protein has also shown some degree of protection (Adar et al. 2009),

and large scale production is also in development (Mardanova et al. 2015; Simon et al. 2014). The

use of cultured eukaryotic cells as host for expression however, may increase production cost of these

candidate vaccines.

A Tc-epitope delivery platform to induce broadly-crossprotective immunity against influenza that has

reached clinical trial phase to date is the modified Vaccinia Ankara (MVA) based vaccine (Krammer

and Palese 2015). Using an HIV model protein subunit fusion has shown to give better protection in

comparison to multiple Tc-epitope presentation on the MVA platform (Gasteiger et al. 2007).

Influenza M1 and NP subunit proteins have been inserted into the MVA viral protein and the resulting

vaccine candidate was shown to be immunogenic (Berthoud et al. 2011) and elicit T cell specific

response (Powell et al. 2013). Clinical trials have shown that the vaccine worked in the elderly

(Antrobus et al. 2012), and gave better protection if combined as prime-boost with DNA vaccines

(Degano et al. 1999), or co-administered with conventional seasonal vaccines (Antrobus et al. 2014).

8

A large scale production platform in chicken embryo fibroblast (CFE) cells has also been established,

with an estimated production capacity of 60 million doses a year (Rimmelzwaan and Sutter 2009).

However, MVA is a replicating attenuated live virus, which still poses some health risk. The use of

mammalian cells as expression host means production cost is high, and even though production

capacity is relatively high, it is still not enough to provide global supply especially during a pandemic.

1.2 Murine polyoma viral coat proteins as a vaccine delivery system

As described above, the development of next generation vaccines does not only have to take into

account the immunogenicity of vaccine candidates, but also aspects of production such as technology,

cost, and time.

Virus-like particles are generally derived from the coat proteins of a virus, which self-assemble into

a spherical structure mimicking the original virus (hence its name) yet lacking DNA or RNA,

rendering them non-infectious (Rodriguez-Limas et al. 2013). They have been developed for various

uses such as antigen carriers (Babin et al. 2013; Kawano et al. 2014; Wibowo et al. 2012) or as carriers

of other cargos such as drugs (Glasgow and Tullman-Ercek 2014), proteins (Abbing et al. 2004), and

both oligonucleotides and plasmid DNA (Braun et al. 1999). Depending on the VLP, they can be

produced in different hosts such as mammalian cells, insect cells, plants, yeasts, and bacteria (Kushnir

et al. 2012).

The MPyV capsid consists of three structural proteins, which are VP1, VP2, and VP3. The VP1

subunit forms a pentamer known as a capsomere, and 72 of these capsomeres form an icosahedral

VLP (Schmidt et al. 2000). The VP1 subunit alone is sufficient to form a spherical VLP (Salunke et

al. 1986), and several surface-exposed loops of VP1 have been identified as suitable for foreign

sequence insertion (Gedvilaite et al. 2000). However, as has been demonstrated before in Hepatitis B

core antigen (HbcAg) particles, insertion of a foreign sequence in these loops may influence capsid

protein stability, therefore, their ability to assemble into VLPs. . Depending on the nature of the

inserted foreign sequence, they could either maintain the stability of VLP subunits and their assembly

competency (Billaud et al. 2005), or destabilize them rendering them assembly incompetent

(Jegerlehner et al. 2002), or prone to severe aggregation (Tekewe et al. 2015). More details are

explained in Section 2.2.3.

VP2 and VP3, though not essential for VLP formation, have been shown to enable encapsidation of

fusion proteins inside the VLP (Abbing et al. 2004; Boura et al. 2005). The C-terminus of the

VP2/VP3 protein interacts with capsomeric VP1 to encapsidate protein fused to this domain. Fusion

9

proteins, when delivered by MPyV VLPs, have been shown to enter dendritic cells and to be presented

by the cellular antigen presentation system (Boura et al. 2005). More details are explained in Section

2.2.3.2.

At the Center for Biomolecular Engineering (CBE) of the Australian Institute for Bioengineering and

Nanotechonology (AIBN), a VLP platform consisting of MPyV VP1 subunits has been developed to

deliver modular antigens to elicit a humoral immune response (Middelberg et al. 2011) (Figure 1.3).

The VP1 subunit, derived from MPyV (Middelberg et al. 2011; Wibowo et al. 2013) forms

capsomeres in vivo, inside the E.coli expression host cells. The purified capsomeres may then be

directed to self-assemble under defined conditions into VLPs and either platform (capsomeres and

VLPs) may be used to develop vaccine designs (Chuan et al. 2010). Multiple matrix 2 ectodomain

(M2e) epitopes have been inserted to the VP1 protein to improve immunogenicity, and adjuvanted

capsomeres presenting the M2e epitopes at the external loop of the protein were able to stimulate

humoral immunity in mice (Wibowo et al. 2013) . The helical subunit helix 190 (H190) from

influenza protein hemagglutinin has also been presented on MPyV VP1 and assembled VLPs were

able to elicit a strong antibody response dependent on the epitope design (Anggraeni et al. 2013).

Both capsomere production in E. coli cells with a 4.38 g/L volumetric yield and in vitro assembly

of modular capsomeres into VLP have been shown to be scalable (Liew et al. 2010; Liew et al. 2012a).

Process simulation has estimated production cost to be less than 1 cent (USD) per dose if capsomeres

and VLPs were grown in 500 L and 1500 L fed-batch cultures, respectively (Chuan et al. 2014). An

estimation of 320 million doses of vaccine (at 50 µg per dose) can be produced with 10 kL batch in

2.3 days or 4.7 days for capsomeres or VLPs, respectively (Chuan et al. 2014). Robustness in

production and humoral immunogenicity of modular MPyV capsomeres renders the platform

attractive to further develop into a broadly-crossprotective influenza vaccine for cellular immunity.

The potential of the MPyV VLP platform to deliver Tc-epitopes and stimulate cellular immunity

against influenza has not been tested. Virus-like particles are a versatile delivery system for Tc-

epitopes because they can enter dendritic cells and present antigen via the MHC Class I and MHC

Class II pathways (Boura et al. 2005; Eriksson et al. 2011); they can act as adjuvants by presenting

pathogen-associated molecular patterns (PAMP) (Raghunandan 2011) and thus may initiate

costimulatory signal required for Tc expansion (Murphy et al. 2008); and they do not contain nucleic

acid, thus minimizing the risk of contamination with infectious particles (Josefsberg and Buckland

2012). It would be useful to explore the potential of the MPyV VLP platform to deliver Tc-epitopes

and stimulate broadly cross-protective cellular immunity against influenza. If this approach is

10

Figure 1.3 The MPyV vaccine delivery platform. The MPyV vaccine delivery platform utilizes either the VLP (3D structure from VIPERdb) (a) or the pentameric capsomere (PDB 1SID) (b) form to present epitopes genetically inserted into the surface-exposed loop of the VP1 subunit. Homology model created by the SWISS model portal (www.swissmodel.expasy.org/ workspace), inserted epitope in red (c).

a b c

11

successful, it will aid in better prevention and management of epidemic and pandemic influenza

complementary to antibody based vaccine

1.3 Research Aim and Objectives

The aim of my research in this thesis is to explore the potential of the MPyV VLP system to deliver

identified influenza Tc-epitopes both externally on surface-exposed loop, and internally by

encapsidation. To achieve this aim, the research is guided by the following research objectives:

1) To evaluate the ability of MPyV VLP platform to carry identified influenza Tc-epitope via

external presentation;

2) To develop and optimise encapsidation of foreign protein using the prokaryotic production of

the in vitro assembled MPyV VLP platform;

3) To demonstrate the assembly of MPyV VLPs presenting external and internal influenza Tc-

epitopes, and to analyse and characterise developed VLP vaccines.

These research objectives are supported by rationale explained in the following sections.

1.3.1 Cytotoxic T cell epitope delivery by external presentation

Tc-epitopes can be delivered either as a single epitope, a string of multiple epitopes, or as a subunit

protein (Rimmelzwaan and Sutter 2009; Sette et al. 2002). There are two delivery strategies that can

be approached with the MPyV VLP platform, which are either delivering these epitopes as an

insertion at the surface-exposed loop of VP1 subunit, or encapsidating the epitopes inside the VLP

by fusion with VP2/VP3 C-terminus.

Surface-exposed loop insertion of B cell epitopes has been established and has been commonly used

at the CBE (Middelberg et al. 2011) (Figure 1.2). Insertion sites identified in Hamster polyoma virus

(HPyV) (Gedvilaite et al. 2000) have been adapted to the MPyV capsomere platform by insertion of

B cell epitopes, which in turn have shown immunogenicity when tested in influenza models

(Anggraeni 2014; Wibowo et al. 2013). Insertion of Tc-epitopes in this platform has not been tested.

Further use of this loop to insert Tc-epitopes could be practical because loop insertion has been shown

to be amenable to modification without VP1 structure disruption, so that capsomeres could be formed

and VLPs could be assembled.

A number of epitopes have been identified immuno-informatically and are accessible in the literature

as well as on online databases. However, Tc-epitopes are usually hydrophobic (Mitic et al. 2014) and

this poses some challenge as their exposure to the surface of the protein may lead to aggregation due

12

to hydrophobic protein-protein interactions (Chong and Ham 2014b; Cromwell et al. 2006), which in

turn may interfere with the assembly competency of VLP subunits displaying them. As insertion of

hydrophobic entities into the existing surface-exposed loop had not been tested before, the extent to

which aggregation would occur due to hydrophobicity was not known. Moreover, hydrophobicity-

related aggregation would depend on individual physico-chemical characteristics of each inserted Tc-

epitope.

The possibility of aggregation has to be resolved accordingly, both by rational protein design and

engineering, as well as providing conditions that enable recovery of soluble and assembly competent

capsomeres. Rather than simply inserting an unmodified Tc-epitope on the surface-exposed loop, the

inserted epitope may be modified by adding charged amino acid to reduce hydrophobicity (Perchiacca

et al. 2012). The use of a linker sequence which have been shown to aid proteasomal degradation

may also be applied (Livingston et al. 2001). Lastly, the assembly of mosaic particles may overcome

hydrophobicity-related capsomere aggregation during VLP assembly (Brown et al. 2009).

1.3.2 Optimization of fusion protein encapsidation within MPyV VLP

Even though the literature provides evidence of successful encapsidation of proteins such as green

fluorescent protein (GFP) when fused with VP2/VP3 (Abbing et al. 2004; Boura et al. 2005), there

has been to my knowledge no report to apply this to deliver subunit proteins of pathogens to elicit

cellular immunity. In addition, there has only been one report that utilizes in vitro assembly for GFP

and methotrexate (Abbing et al. 2004), and the approach they took does not ensure encapsidation of

the fusion protein. More details in Section 2.2.3.2.

Epitope delivery by encapsidation may be done by fusing either a single epitope, a string of multiple

epitopes, and a subunit protein with the VP2/VP3 protein. Encapsidation of subunit protein may be

preferable due to several reasons. Firstly, information on the behaviour and solubility of subunit

proteins of influenza are available in the literature, while the behaviour of a newly tested string of

multiple epitopes is unpredictable depending on the physico-chemical properties of the peptide and

may interfere with the engineering of the VLP. Secondly, a subunit protein contains multiple epitopes

of different HLAs and thus enables the vaccine to cover a more diverse human population.

The engineering of VP1-VP2/3 protein interaction to encapsidate fused protein is not well explored

yet, and thus application to the existing MPyV VLP platform as well as troubleshooting should be

carried out with a model protein such as GFP to facilitate monitoring and analysis.

13

1.3.3 Characterisation and analysis of MPyV VLP based influenza

vaccines via external and internal presentation

Hydrophobic epitopes presented on surface-exposed loop of capsomeres may interfere with efficient

VLP assembly due to aggregation. This may be resolved by mixing the modular VP1 with wild type

(wt)VP1 in a pre-determined ratio to form a mosaic VLP vaccine. This is aimed to reduce global

surface hydrophobicity and enable more efficient VLP formation under optimised buffer conditions.

Foreign protein encapsidation optimised with the existing MPyV VLP platform using the model

protein GFP to be applied to the encapsidation of influenza subunit protein containing identified Tc-

epitopes. Encapsidated influenza protein is expected to still be able to stimulate cellular immunity as

this kind of immunity does not depend on external presentation and conformation (Murphy et al.

2008).

The vaccine particles developed should be analysed and characterised. Also isolation of the particles

from other components such as aggregates and capsomeres may be needed. A preparative size

exclusion chromatography (SEC) purification method could be used to address both of these needs.

At the time this research was conducted, such a method had not been reported elsewhere.

1.4 Thesis organisation

In addition to this introductory chapter, this PhD thesis consists of 5 other chapters.

Chapter 2 provides a literature review of key topics in this research. It includes a review of the MPyV

VLP vaccine technology and the basis of further development; a review on how the cellular immunity

works; and a review about the fundamentals of size exclusion chromatography.

Chapter 3 highlights limitations posed by external Tc-epitope presentation in harvesting soluble

capsomeres and high-throughput buffer screen to resolve this problem.

Chapter 4 investigates ways to encapsidate foreign antigen inside the VLP by application to and

optimisation within the already established MPyV VLP platform.

Chapter 5 explores the assembly of Tc-epitope-based MPyV VLP influenza vaccines, employing

both external and internal presentation; and assesses methods to enable their analysis and

characterisation.

Chapter 6 summarises the findings of all the work above and suggests possible future research.

14

2 Literature Review

2.1 Overview

This chapter will provide basic concept and extended knowledge on topics pertinent to the scope of

my research. It contains three main sections. In the first main section (Section 2.2), the development

of murine polyomavirus-like particles (MPyV VLP) as vaccine delivery technology will be explained.

In the second main section (Section 2.3), the topics pertinent to the specific development of the

platform into Tc-epitope-based influenza vaccine will be described. In the third main section (Section

2.4), the principles and evaluation of SEC will be briefly explained.

2.2 The murine polyomavirus-like particle vaccine delivery technology

A VLP is generally an assembly of coat proteins, forming the capsid while maintaining protein shape

and structure of the virus. It has an empty hollow, and does not contain other viral components such

as pathogenic proteins and genetic materials, and is thus non-pathogenic and non-infectious (Lua et

al. 2014; Rodriguez-Limas et al. 2013; Teunissen et al. 2013). VLPs can either be enveloped or non-

enveloped. The envelope consists of a lipid bilayer with protruding glycoproteins outside the viral

coat proteins (Kushnir et al. 2012).

The existence of naturally occurring VLPs was first reported in 1965 (Blumberg et al. 1965) and their

ability to induce cellular and humoral immunity was subsequently elucidated in 1968 (Bayer et al.

1968). The advancement in genetic engineering further allowed the expression and purification of

synthetic VLPs (Hagensee et al. 1993; Kirnbauer et al. 1992; Li et al. 1997), which enabled further

characterisation and enhanced our understanding of their assembly (Li et al. 1997; Salunke et al.

1986). Over the course of more than four decades, 110 VLPs from 75 different viral families have

been constructed and evaluated (Yan et al. 2015; Zeltins 2013). VLPs are deemed an ideal biological

vehicle due to their biocompatibility, solubility, uptake efficiency, and their capability for targeted

delivery and drug loading (Yan et al. 2015). Thus, they have been developed for vaccines, drug

carrying, gene delivery, and in vivo imaging (Teunissen et al. 2013; Yan et al. 2015).

In vaccine development, VLPs can be used in two ways. Firstly, native viral VLPs from pathogenic

strains may be utilised by itself as vaccines. Secondly, chimeric VLPs carrying foreign antigens either

by genetic engineering or chemical conjugation may be used as vaccines (Glasgow and Tullman-

15

Ercek 2014; Kushnir et al. 2012; Lua et al. 2014; Rodriguez-Limas et al. 2013; Yan et al. 2015). VLPs

are attractive for vaccine development not only because they are immunogenic, but also because they

can be rapidly and efficiently produced (Yan et al. 2015). Different expression systems have been

established for VLP production such as bacteria, yeast, insect cells, plant cells, mammalian cells,

plant cells, and even cell-free systems (Kushnir et al. 2012).

Currently different VLP-based candidate vaccines have been developed not only for diseases caused

by infectious agents such as hepatitis, HIV, human papillomavirus (HPV), influenza, malaria, Nipah

virus (NV), and respiratory syncytial virus (RSV); but also non-pathogen-associated cancers such as

breast cancer and melanoma; as well as other non-pathogen-associated conditions such as allergy,

asthma, Alzheimer's, type II diabetes mellitus, hypertension, and nicotine addiction (Kushnir et al.

2012). Many of these VLP-based vaccine candidates have entered the clinical trial phase and to date,

there are three VLP-based vaccines that have been officially approved for clinical use (Kushnir et al.

2012). The first to be approved was the recombinant Hepatitis B (HB) VLP expressed in yeast

(Recombivax HB, Merck), approved by Food and Drug Administration (FDA) for use as a vaccine

in 1986. The second was recombinant HPV VLP vaccine produced either in yeast (Gardasil, Merck),

approved by the FDA in 2006; or produced in insect cells (Cervarix, GlaxoSmithKline), approved by

the FDA in 2009. The third was a recombinant hepatitis E (HE) VLP vaccine produced in E.coli

(Hecolin, Xiamen Innovax Biotech), approved by Chinese FDA in 2011 (Rodriguez-Limas et al.

2013). However, all of the approved VLP-based vaccines use native viral coat protein as vaccines

and none of them are in the form of chimeric VLPs carrying foreign antigens. Recently, the WHO

recommended the pilot implementation of a recombinant HbsAg VLP based candidate vaccine

carrying foreign circumsporozoite protein fusion produced in yeast (RTS,S/AS01 Mosquirix, funded

by GlaxoSmith Kline Biological SA and PATH Malaria Vaccine Initiative) (Gordon et al. 1995;

Otieno et al. 2016). More research in the application of recombinant VLPs carrying foreign antigen

could be beneficial for the prevention and treatment of pathogen and non-pathogen-associated

diseases.

My current research aims to explore the ability of a chimeric VLP to carry foreign antigen to elicit

cellular immunity using the non-enveloped MPyV coat proteins. MPyV is a virus oncogenic in

infected mice cells (Benjamin 2001). The virus infects multiple cell types due to its promiscuous

surfaced-exposed loops able to bind to a variety of sialic acid receptors in the host (Buch et al. 2015).

The virus is a double-stranded DNA virus belonging to Polyomaviridae family (ICTV 2012). The

coat protein of MPyV consists of the major protein VP1 and the minor proteins VP2 and VP3

(Liddington et al. 1991; Rayment et al. 1982). The icosahedral (T=7d) MPyV VLP and has a diameter

of 45 nm and consists of 360 VP1 subunit proteins, arranged in 72 pentamers (these pentamers are

16

called capsomeres). As for other members of the polyomaviridae, sixty capsomeres are hexavalent

and the other twelve are pentavalent (Liddington et al. 1991; Stehle and Harrison 1997). The VP1

subunit (42.5 kDa) consists of three major parts which are: 1) N-terminal part, which extends to

clockwise neighbouring VP1 subunit to form a capsomere; 2) the core anti-parallel β-sandwich part

with jelly-roll topology, which is uniform across all VP1 subunits; and 3) the C-terminal extension

into neighbouring capsomeres, the conformational variability of which plays a major role in shaping

the pentavalent and hexavalent interaction between capsomeres to form the icosahedral VLP

(Liddington et al. 1991; Rayment et al. 1982) (Figure 2.1). The VP1subunit self-assembles into

capsomeres when expressed in E.coli (Leavitt et al. 1985; Salunke et al. 1986). The capsomere is

cylindrical with a hollow conical interior, approximately 80 Å in diameter and 70 Å tall. The hollow

conical interior has a diameter of 50 Å at its base and 12 Å at its upper neck and the top of the

capsomere contains a dimple 20 Å wide and 12 Å deep (Liddington et al. 1991). The VP1 subunit

alone can form VLPs both in vivo when expressed in insect cells (Montross et al. 1991), yeast

(Salunke et al. 1986), or plants (Catrice and Sainsbury 2015), and in vitro when capsomeres are

dialysed in optimised assembly buffer conditions (Salunke et al. 1986; Schmidt et al. 2000). In vitro

dialysis experiments demonstrated that capsomere assembly into VLPs depends on calcium and

disulphide bridges, and thus, capsomere form can be maintained in buffers containing

ethylenediaminetetraacetic acid (EDTA) to chelate calcium, and dithiothreitol (DTT) to prevent the

formation of disulphide bridges (Salunke et al. 1986; Schmidt et al. 2000). Salt concentration and pH

play a role in assembly and disassembly of capsomeres into VLP and vice versa (Chuan et al. 2010;

Salunke et al. 1986). The ability to manipulate assembly in vitro eliminates the need for disassembly

and reassembly process usually applied to remove genetic material inside in vivo assembled VLPs

(Pattenden et al. 2005).

Besides the three major domains of the VP1 subunit, Liddington et al. (1991) also identified surface-

exposed loops that protrude into the outer part of the capsomere, connecting the antiparallel β-sheets

with each other. These loops are named DE, EF, BC, and HI based on the β-sheets they are connecting

(Figure 2.1A). Subsequently, Gedvilaite et al. (2000) identified insertion sites at these surface-

exposed loops of the VP1 subunit at HPyV and named them site 1 (80-89 aa), site 2 (221-224 aa),

site 3 (243-247 aa), and site 4 (288-295 aa). Insertion at these sites, based on theoretical and structural

considerations, would not interfere with the correct folding of the protein and would expose the

inserted sequences, such as antigenic determinants, into the surface of the protein, a prerequisite to

stimulate humoral immunity. This information is crucial in further attempts to insert foreign peptides

into the VP1 subunit without interfering with the structure of both the carrier and the foreign entities.

Even though the loops have been identified in Simian virus 40 (SV40) VP1 (Liddington et al. 1991)

17

Figure 2.1 Structure of MPyV VP1, capsomere, and VLP. A. Structure of VP1 subunit including intruding arms from neighbouring subunits. Loops and insertion sites identified in the literature are indicated (Adapted from Liddington et al. (1991)). B. VP1 monomer (without intruding arms from neighbouring subunits) forms a pentamer called a capsomere, and 72 capsomeres form the VLP (Adapted from Pattenden et al. (2005)).

18

and the insertion sites in HPyV VP1 (Gedvilaite et al. 2000), they can be adapted to MPyV as they

are structurally similar. The adaptation will be explained further in the next subsections.

MPyV VLPs can enter the MHC Class I and MHC Class II pathways as they can be processed by the

lysosome and the proteasome, demonstrated by Abbing et al. (2004) and Boura et al. (2005). MPyV

VLPs enter the cell by monopinocytosis (Richterova et al. 2001) followed by fusion of the

monopinocytic vesicle with the endosome (Liebl et al. 2006) as can be observed 30 min after entry

into the cell (Boura et al. 2005). After 90 minutes the MPyV VLP has migrated to the perinuclear

space and fused to the lysosome, where it gets degraded (Boura et al. 2005). After 6 hours the

degraded MPyV VLP translocate into the cell periphery and enter the cytoplasm, to be further

degraded by the proteasome 2 hours later (Boura et al. 2005). This intracellular migratory pattern was

also observed in antigen presenting cells (APC), and MPyV VLP cellular uptake by dendritic cells

has been shown to stimulate interleukin (IL)-12 production (Boura et al. 2005). The MHC Class I

pathway will be explained in more detail in Section 2.3.2 and the MHC Class II will also be explained

briefly. The MPyV VLP can be taken up by antigen presenting cells of diverse host species via a

groove on the VP1 protein promiscuous to different ganglioside and sialic acid receptors, especially

those with [α-2,3], [α-2,6], and [α-2,8]-linked sialic acid (Buch et al. 2015). When injected in mice,

MPyV VLPs could elicit a dose-dependent high titer humoral response and a T helper cell 1 (Th1)

type cellular response (Caparros-Wanderley et al. 2004). The ability of MPyV VLPs to enter the

MHC Class I and II pathways and their promiscuity to different types of eukaryotic cells are some of

the advantages of the MPyV VLP platform that makes it attractive to develop into Tc-epitope-based

vaccine.

The AIBN-CBE Group, led by Professor Anton Middelberg, has developed the MPyV VLP platform

as a commercially viable platform using E.coli as the expression host (Chuan et al. 2008b; Liew et

al. 2010) (Figure 2.2). Briefly, at laboratory scale, capsomeres are produced within the bacterial cells

with a single-batch culture under optimised conditions and isopropyl-β-D-thiogalactopyranoside

(IPTG) induction (Chuan et al. 2008b; Middelberg et al. 2011). The cells are then harvested and lysed

by sonication and the capsomeres purified by glutathione-S-transferase (GST) affinity

chromatography. The GST-tag is removed by enzymatic cleavage and the free capsomeres are then

isolated from aggregates and free GST by SEC (Lipin et al. 2009). The capsomeres can be maintained

in capsomere form or assembled into VLP in vitro in optimised buffer conditions (Chuan et al. 2010).

The laboratory scale production has subsequently been translated into a large-scale process, involving

fed-batch E.coli expression with a volumetric protein yield of 4.38 g/l which is 15- fold laboratory

scale production (Liew et al. 2010). Capsomere purification can also be achieved by ion exchange

chromatography (IEX) (Chuan et al. 2014), and assembly into VLP can be performed by diafiltration

19

Figure 2.2 MPyV capsomere and VLP microbial platform established at AIBN-CBE. The established platform includes genetic engineering, microbial production, affinity chromatography, tag removal, SEC, and capsomere assembly into VLPs. The platform is scalable (Modified from unpublished collection of AIBN-CBE).

20

or dilution method (Liew et al. 2012a; (Liew et al. 2012a; Liew et al. 2012b). Optimised assembly

buffer for the assembly via diafiltration increased yield by 42-56% in comparison to laboratory scale

assembly (Liew et al. 2012b). Computational simulation with SuperPro Designer has been used to

estimate production costs (Chuan et al. 2014). Optimised large scale production of capsomeres and

VLPs at 500-L and 1500-L volumes respectively were calculated to have an 80% probability of

manufacturing vaccine at less than 1 cent (USD) per dose (50 µg per dose). Moreover, the simulation

also predicted that at a 10k-L production scale, 320 million capsomeres or VLP vaccine doses could

be produced within 2.3 or 4.7 days, respectively.

The wealth of information available as well as having established a microbial production platform for

the MPyV capsomere and VLP platforms enabled the group to develop these platforms further into

vaccine candidates. The initial approach was to insert influenza antibody epitopes into identified

insertion sites, develop them into either capsomere or VLP vaccines, and test their immunogenicity

and ability to protect against influenza strains in mice.

2.2.1 Established MPyV Vaccine Platform

Vaccine candidate development using the MPyV coat proteins as an antigen carrier has been

established at the AIBN-CBE. The vaccines can come in either of two forms, capsomeres or VLPs.

Epitopes which stimulate antibody production from a variety of pathogens, including influenza and

group A staphylococcus (GAS) (Chuan et al. 2013; Middelberg et al. 2011; Rivera-Hernandez et al.

2013; Wibowo et al. 2012), have been inserted into engineered sites of the VP1 subunit protein.

Insertion of antigens of other pathogens, such as the rotavirus, are also in the pipeline (Tekewe et al.

2015).

Positions corresponding to site 1 and site 4 identified by Gedvilaite (2000) in HPyV are used in MPyV

VP1 by inserting distinct restriction enzyme sequences at amino acids 85 and 293, respectively, to

enable epitope insertion via homologous recombination (Middelberg et al. 2011; Wibowo et al. 2012).

Capsomeres with successful foreign peptide insertion are then assembled into VLPs and their

immunogenicity tested in mice (Anggraeni et al. 2013; Chuan et al. 2013; Middelberg et al. 2011;

Rivera-Hernandez et al. 2013). In cases where epitope insertion interfered with the ability of the

modular capsomeres to form VLPs, leading to incomplete VLPs or aggregation, the C-terminus can

be truncated (Wibowo et al. 2012). The VP1 C-terminus plays a significant role in VLP formation by

forming interpentamer interaction, and its truncation results in stable and assembly incompetent

capsomeres. As it has previously been reported that capsomeres alone were able to stimulate humoral

immunity (Fligge et al. 2001), and that stimulation of humoral immunity occurs extracellularly

(Murphy et al. 2008), it was reasonable to engineer the capsomeres into stable and assembly

21

incompetent epitope presentation platforms. However, this strategy may not be applicable for cellular

immunity stimulation as it requires Tc-epitopes to enter the cytoplasm of APCs and capsomeres alone

are not able to enter cells. The following section will describe the current state of MPyV vaccine

development, both in capsomere (Section 2.2.1.1) and VLP (Section 2.2.1.2) forms.

2.2.1.1 Capsomere platform

Stable and assembly incompetent capsomeres have been engineered to carry the M2e epitope

(CapM2e) of influenza virus. Beside site 1 and site 4, restriction enzyme sequences were also inserted

at truncated N-terminus (28 aa) and truncated C-terminus (63 aa) to enable epitope insertion

(Middelberg et al. 2011; Wibowo et al. 2012). It was demonstrated that the number of epitopes

inserted as well as insert position, influenced capsomere expression levels (Wibowo et al. 2012). Co-

formulation with adjuvants in combination with a higher number of epitopes have been shown to

increase immunogenicity (Wibowo et al. 2012). Homologous influenza challenge in mice vaccinated

with CapM2e showed significant protection, reflected by high antibody titer (>105), reduction in lung

viral load (90.4%), and prevention of body weight loss in comparison to non-vaccinated mice

(Wibowo et al. 2014). Non-chromatographic purification economically optimised production process,

and purified and adjuvanted modular capsomeres using this process resulted in high immunogenicity

when injected in chicken (Wibowo et al. 2015).

2.2.1.2 VLP platform

The ability to assemble MPyV capsomeres into VLPs in vitro is a major advantage in VLP-based

engineering as it removes the disassembly and reassembly process that can be necessary to formulate

in vivo assembled VLPs and results in loss of protein due to chemical denaturation. The in vitro

assembly buffer is a solution of fine-tuned function and of simple components: desalting effect of

ammonium sulphate; facilitation of disulphide and calcium bridge formation by the removal of DTT

and EDTA, and the addition of CaCl2; and the manipulation of VLP-compatible pH and salt

concentration (Chuan et al. 2010).

Insertion of H190, an antigen from influenza hemagglutinin (HA), into site 4 of VP1 has been reported

not to interfere with assembly (Anggraeni 2014). Two insertion approaches were applied; the use of

linker sequences flanking the antigen, or a dual tandem repeat of the antigen. Modular capsomeres

were assembly competent and results showed that the tandem repeat modularisation was better

regarding structural preservation of the antigen and VLPs were immunogenic when injected into mice

(Anggraeni 2014). A different antigen, the J8i peptide from GAS was also inserted into site 4

(Middelberg et al. 2011; Rivera-Hernandez et al. 2013). In this case, and unlike the aforementioned

22

capsomeres, there was no difference in immunogenicity observed between adjuvanted and non-

adjuvanted VLPs (Middelberg et al. 2011). There was also no significant difference in

immunogenicity when VLPs carry single or dual tandem epitopes, and they were both similarly

protective for mice against challenge with GAS (Rivera-Hernandez et al. 2013). Chuan et al. (2013)

also showed that pre-existing immunity against the carrier, a major concern for heterologous

presentation platforms, did not hinder the production of antigen-specific antibodies, which is good in

terms of vaccine development.

Even though a variety of MPyV-based vaccine candidates in the form of capsomeres and VLPs have

been developed (Table 2.1), they all are targeting antibody production to raise humoral immunity.

The ability of the MPyV VLP to stimulate cellular immunity by carrying Tc-epitopes has not been

tested. Cellular immunity stimulation may provide an alternative solution in influenza pre-pandemic

and seasonal epidemic preparation as eliciting cellular immunity can exploit the more conserved

internal influenza proteins, which are less prone to mutation in comparison to antibody targeted

external proteins (Subbarao and Joseph 2007). Other diseases such as cancer and acquired

immunodeficiency syndrome (AIDS) also may benefit from stimulation of cellular immunity as

humoral immunity does not work in these conditions in general. Therefore, the ability of the MPyV

platform to carry Tc-epitopes needs to be explored and challenges particular to the nature of Tc-

epitopes in combination with the nature of MPyV coat proteins need to be addressed.

2.2.2 Further development for hydrophobic Tc-epitope presentation

Influenza vaccine development to stimulate cellular immunity by delivering Tc-epitopes has several

advantages over antibody-based vaccines. Firstly, unlike humoral immunity, stimulation of cellular

immunity does not depend on the protein conformation of the antigen (Murphy et al. 2008). Secondly,

stimulation of cellular immunity occurs inside the cell, where the antigenic protein is degraded into

nonameric Tc-epitopes (Section 2.3.3), which does not require the antigen to be presented on the

surface of the carrier. Thus, thirdly, the Tc-epitope-based vaccine can make use of the internal

proteins of influenza virus which are more conserved and are less prone to mutation (Brown and

Kelso 2009; Krammer and Palese 2015). However, cellular immunity stimulation presents several

challenges which are: 1) the antigen is required to enter the APC's cytoplasm (Yewdell et al. 2003);

2) the antigen needs to be processed into smaller peptides by the proteasome, which then enter the

endoplasmic reticulum (ER) to couple with MHC molecules (Yewdell et al. 2003); and 3) the majority

of Tc-epitopes are hydrophobic in nature, which may lead to aggregation (Mitic et al. 2014) (More

on these mechanisms in Section 2.3.3). An MPyV VLP based system may address the first two

challenges as the VLP has been shown to enter dendritic cells, and be processed by both the

23

Table 2.1 MPyV-based candidate vaccines.

MPyV platforms Antigen Pathogen References

Capsomere

M2e Influenza (Wibowo et al. 2012)

VLP

J8i GAS (Rivera-Hernandez et al. 2013)

H190 Influenza (Anggraeni 2014)

VP8 Rotavirus (Tekewe et al. 2015)

24

proteasomal and lysosomal pathways (Abbing et al. 2004; Boura et al. 2005). The hydrophobicity of

Tc-epitopes and its potential to cause aggregation, however, remains to be addressed and has been a

major goal of my thesis research. The next subsections will outline problems associated with

hydrophobic Tc-epitope presentation and ways to resolve aggregation related to hydrophobicity, and

also discuss other potential approaches in Tc-epitope delivery with the MPyV VLP platform.

2.2.3 Problems related to hydrophobic Tc-epitope delivery

Insertion of foreign peptides into a viral coat protein can either maintain protein stability (Anggraeni

2014; Billaud et al. 2005; Middelberg et al. 2011; Rivera-Hernandez et al. 2013); or interfere with

protein stability, rendering the VLP components assembly incompetent as has been experienced with

M2e insertion into MPyV VP1 (Wibowo et al. 2012), or prone to aggregation as has been

demonstrated with hydrophobic rotavirus epitope insertion within MPyV VP1 (Tekewe et al. 2015).

Normal protein aggregation is usually initiated by nucleation of partially unfolded protein

intermediates which expose hydrophobic patches as well as extended unfolding arms (Wang et al.

2010b). This phenomenon then leads to disordered self-assembly via hydrophobic and electrostatic

protein-protein interaction (Fink 1998; Wang et al. 2010b). Initial disruption of conformational

stability leads to the disruption of colloidal stability (Chi et al. 2003) as proteins interact with each

other to form soluble aggregates. Gradual increase in aggregate size may extend beyond their

solubility limit, leading to precipitated insoluble aggregates (Wang et al. 2010b). Simulation has

shown aggregation of proteins with hydrophobic entities on the surface before (Zhang et al. 2008).

Therefore, insertion of hydrophobic Tc-epitopes on the surface-exposed loop of VP1 protein may

emulate hydrophobicity related protein aggregation. Aggregation of therapeutic proteins usually

reduces or eliminates their biological activity or immunogenicity (Wang et al. 2010b), and this may

also apply to an MPyV-based vaccine presenting hydrophobic Tc-epitope on the surface if they

aggregate. In addition, aggregation in therapeutic protein formulation is undesirable as the presence

of particles >10 µm renders the preparation unsuitable for administration (Cromwell et al. 2006).

Aggregation is influenced by several environmental factors such as pH, temperature, ionic strength,

sequence hydrophobicity, local surface charges, protein concentration, buffering agent, and the

presence of chaperones and additives (Chong and Ham 2014b; Fink 1998; Wang et al. 2010b). Ionic

strength and pH play a major role in influencing intramolecular and intermolecular interactions, both

hydrophobic and electrostatic. To prevent protein aggregation, pH and ionic strength of the solvent

is usually optimised, and a selection of additive, depending on the nature of the protein, is often added

to the solvent (Wang 2005). For example, additives commonly used to prevent aggregation include

25

detergents, chaperones, osmolytes, and charged amino acids such as L-Arginine (L-Arg) (Arakawa

et al. 2007; Lanckriet and Middelberg 2004; Leibly et al. 2012).

2.2.3.1 Mosaic particles

One way to reduce hydrophobicity on the surface of a VLP to prevent aggregation and to maintain

VLP structure is by combining hydrophobic modified capsomeres with unmodified capsomeres,

forming a mosaic particle. The bioengineering of mosaic particles has been demonstrated before with

the Qβ bacteriophage and the plant virus cowpea chlorotic mottle virus (CCMV). The Qβ virus-like

particle can self-assemble inside E.coli and co-expression of modified Qβ monomer subunit

displaying foreign protein on the surface with unmodified Qβ monomer can result in stable VLP

mosaic particles. This approach permitted insertion of either the Z protein (Brown et al. 2009) or the

epidermal growth factor (EGF) protein (Pokorski et al. 2011) on the surface of Qβ. Expression of

modified Qβ alone interfered with the assembly competency of the particles and co-expression with

unmodified Qβ restored the assembly competency of the particles in vivo.

Mosaic particle formation has also aided in the encapsidation of GFP inside the CCMV VLP (Minten

et al. 2010). Modified subunit was co-assembled with unmodified subunit in vitro to control the

amount of encapsidated GFP. Co-assembly resulted in stable mosaic particles encapsidating GFP in

a controlled manner. Similar to MPyV VLP, the CCMV VLP can be assembled in vitro. Co-assembly

of modified MPyV VP1 carrying antigens on the surface or in the internal cavity of capsomeres with

unmodified wtVP1 capsomeres in vitro has not been reported before. Mosaic particle formation could

be a good strategy in reducing hydrophobic effects on the surface of modified MPyV VLPs, and in

controlling the number of encapsidated proteins. The protein encapsidation approach is explained in

the next subsection.

2.2.3.2 Protein Encapsidation by MPyV

Another way to overcome hydrophobicity-related aggregation is by encapsidating the hydrophobic

antigen inside the cavity of MPyV VLP rather than presenting it on the surface. The minor coat

proteins VP2 or VP3 are not required for self-assembly, but they have been shown to enable foreign

protein encapsidation inside the MPyV VLP (Abbing et al. 2004; Boura et al. 2005; Inoue et al. 2008).

The VP2 differs from VP3 only by additional 115 aa at the N-terminus of VP2 (Barouch and Harrison

1994). One VP2 (and VP3) protein binds to the hydrophobic inner cavity of one VP1-capsomere via

a conserved hydrophobic region close to the C-terminal of VP2, forming a VP1:VP2 capsomere

complex (Barouch and Harrison 1994; Chen et al. 1998) (Figure 2.3). While the hydrophobic binding

26

Figure 2.3 VP1:VP2 interaction. The diagram shows three VP1 subunits as members of a pentameric capsomere, with one VP2 inside the cavity of the capsomere. VP2 binds to the inner cavity of VP1 via a hydrophobic binding region at the C-terminus of VP2 (Adapted from Chen et al. (1998)).

27

region of VP2 is very conserved, amino acid sequences at the N-terminal and C-terminal arms of this

binding region are not as conserved, underlining the important function of the binding region

(Barouch and Harrison 1994; Sapp and Day 2009). The binding affinity of the capsomere:VP2

interaction is quite high, with reported dissociation constant (Kd) of 5 µM (Barouch and Harrison

1994) or 190 nM (Abbing et al. 2004).

Foreign protein encapsidation with VP2 has been demonstrated by fusion of the foreign protein to

either the N-terminus or C-terminus of VP2. VP2 and a variety of minimal sequences omitting some

part of the N-terminal region but still retaining the hydrophobic binding site of VP2 (VP2C) have

been engineered for this purpose (Abbing et al. 2004; Boura et al. 2005; Inoue et al. 2008).

Encapsidation has been achieved by either co-expression of VP1 and foreign protein fused VP2/VP2C

within insect cells (Boura et al. 2005; Inoue et al. 2008), or by separate expression in E.coli and

separate chromatographic purification before co-assembly in vitro (Abbing et al. 2004). The GFP has

been fused to VP2/VP2C in most cases to enable chromatographic detection at A488 nm (Abbing et

al. 2004; Boura et al. 2005; Inoue et al. 2008) to validate the approach. Abbing et al. (2004) reported

an encapsidation capacity of 64 GFP molecules when unmodified VP1 was co-assembled in vitro

with VP2/VP2C-GFP. It was also demonstrated that GFP encapsidation did not interfere with the

ability of MPyV VLPs to enter cells (Abbing et al. 2004; Boura et al. 2005). While Abbing et al.

(2004) and Boura et al. (2005) fused GFP at the N-terminus arms of MPyV VP2, Inoue et al. (2008)

fused GFP at either N-terminus or C-terminus of SV40-VP2C. Inoue et al. (2008) were able to show

that C-terminal fusion was preferable as it resulted in regularly shaped VLPs while N-terminal fusion

resulted in irregularly shaped VLPs. Furthermore, Pleckaityte et al. (2015) demonstrated that fusion

of foreign proteins into truncated N-terminus of VP2 resulted in the foreign protein being exposed to

the surface of the VLP and in antibody recognition of the foreign protein. Their results suggested that

N-terminal fusion resulted in the protrusion of the fusion proteins through the upper opening of the

capsomere into the external space of the VLPs. Thus, foreign protein fusion to the C-terminus of

VP2C was more preferable for encapsidation. The encapsidation strategy enables carriage of a longer

stretch of amino acids, even in the form of whole protein or soluble subdomains, without interfering

with assembly competency and cell entry. Regarding MPyV VLP development as a Tc-epitope

carrier, a longer stretch of amino acid would entail the ability to carry multiple numbers of epitopes

with different HLA specificities. Encapsidation of soluble subdomains of conserved internal proteins

of influenza could be an immunological advantage because proteins and longer peptides have been

shown to be better processed by the MHC Class I pathway and are more likely to be presented by

APCs than short peptides (Berthoud et al. 2011).

28

This section started by describing the existing MPyV vaccine platform and pointed out limitations

associated with hydrophobic Tc-epitope carriage. Further, approaches to circumvent bioengineering

challenges were explained, such as buffer optimisation, the engineering of mosaic particles, and

antigen encapsidation strategies. The next section will further discuss the advantages and

disadvantages of the MPyV platform as influenza Tc-epitope carrier, by first explaining the influenza

phenomenon, as well as limitations in current influenza vaccines.

2.3 MPyV VLP-based influenza Tc-epitope vaccines

2.3.1 Influenza

The influenza virus belongs to the Orthomyxoviridae family, characterised by an enveloped virus

containing negative single-stranded ribonucleic acid (RNA). The family includes six different genera

which are Influenza A, Influenza B, Influenza C, Thogovirus, Isavirus, and Quaranjavirus. Of these

genera, Influenza A is the most relevant to humans (Subbarao and Joseph 2007; Szewczyk et al.

2014). Different HA and neuraminidase (NA) types define serological features of an influenza strain.

Currently, there are 18 HA and 11 NA serological types (Szewczyk et al. 2014).

The HA and NA proteins are the external proteins of influenza, which mutate frequently (Subbarao

and Joseph 2007). The HA protein comprises 80% of the external proteins protruding like a spike.

The NA comprises the other group of major external proteins protruding like a mushroom (Norkin

2010; Szewczyk et al. 2014) (Figure 2.4). HA plays a role in viral entry into the cell, and NA plays

a role in viral exit from the cell (Norkin 2010; Szewczyk et al. 2014). Another minor external protein

is the M2, which is embedded in the viral membrane and acts as an ion channel. The internal proteins

are more conserved (Subbarao and Joseph 2007) and include the M1, nucleoprotein (NP), polymerase

basic protein 1 (PB1), polymerase basic protein 2 (PB2), polymerase acidic protein (PA), and non-

structural protein 2 (NS2) (Norkin 2010; Szewczyk et al. 2014). The most abundant protein is the M1

capsid protein, partly embedded along the inner part of the membrane, and binds to the

ribonucleoproteins (RNP) in the cavity of the virus. The RNP complex consists of NP; and the

ensemble of RNA-dependent RNA polymerases PB1, PB2, and PA. The NP protein encases each of

the eight RNA segments separately (Norkin 2010; Szewczyk et al. 2014).

Influenza pathogenicity is associated with mutations more frequently in the external and less

frequently in the internal proteins, causing an elevation in RNA replication, cytokine and chemokine

dysregulation, antiviral cytokine resistance, and interference with human signalling pathway

29

Figure 2.4 Schematic diagram of the influenza virus. A. The influenza virus consists of major external protein HA and NA and minor external protein M2. Internal proteins include M1, and NEP/NS1 and the RNP. B. RNP consists of viral RNA wrapped in NP protein which bind to polymerase proteins PA, PB1, and PB2 (Adapted from Norkin (2010)).

30

(Korteweg and Gu 2008; Neumann et al. 2010). In particular, HA cleavage plays an important role in

influenza virus replication and pathogenic protein translation inside the host cell, including

subsequent pathogenic viral protein translation and virus replication (Korteweg and Gu 2008;

Neumann et al. 2010; Peiris et al. 2007).

The influenza viral proteins, especially the external proteins, are prone to mutation by three

mechanisms: 1) antigenic drift; 2) antigenic shift; and 3) recombination (Szewczyk et al. 2014).

Antigenic drift is caused by the lack of proofreading ability of influenza RNA polymerases,

introducing single or more amino acid mutations (Holland et al. 1982; Szewczyk et al. 2014).

Mutation occurs continuously, accumulatively leading to new strains unrecognisable by humoral

immunity. This, allows transmissibility and maintenance of the virus in the host population and results

in seasonal influenza epidemic (Ryan et al. 2014). Antigenic shift can happen in two ways which are:

reassortment; and viral jump into a different host. Reassortment occurs when two different strains

infect the same host cells and exchange genetic material (Ryan et al. 2014; Stevens et al. 2006). An

example of this is when influenza virus of avian and of human origin infect the same swine cell.

Avian influenza viruses have different receptor specificities [α-2,3 sialic acid] than human influenza

viruses [α- 2,6 sialic acid] and they usually do not cross-infect. However, the swine cell contains both

avian and human receptors (Liu et al. 2009). Other intermediates such as ferrets, horses, and terrestrial

avian species, and even human, have also been identified (Herfst et al. 2014; Matrosovich et al. 2004).

When reassortment becomes phenotypic, it is called antigenic shift. An example of this is when host

cell specificity changes from avian to human due to possible reassortment in swine or even human

cells (Matrosovich et al. 2004; Wang and Jiang 2009). The H1N1-SOIV strain responsible for the

2009 pandemic was a reassortant strain containing six RNA segments from North-American swine-

origin strains and 2 RNA segments originating from Eurasian strains (Mamun and Huda 2011).

Influenza virus can also jump directly to a different host species with or without intermediates

(Schnitzler and Schnitzler 2009; Wang and Jiang 2009). Both forms of antigenic shift, reassortment

and host jump, can result in a pandemic as the new host does not have any herd immunity against the

new strain (McAuley et al. 2015; Schnitzler and Schnitzler 2009; Wang and Jiang 2009). Especially

when the new strain becomes highly transmissible among the new host population as experienced in

the H1N1-S-OIV pandemic (Herfst et al. 2014). Homologous and non-homologous RNA

recombination, may also occur at a rarer frequency, but may transform a low pathogenic strain into a

high pathogenic strain (Suarez et al. 2004).

An influenza epidemic is an increase, often suddenly, in the number of cases of influenza above what

is usually expected in that population in that area (Dicker et al. 2012). Epidemic influenza usually

affects a particular region and occurs seasonally, especially during winter. The estimated death rate

31

of epidemic influenza is about 250 000-500 000 per year and especially prevalent in the very young,

elderly, and chronically ill (WHO 2009a).

An influenza pandemic occurs when a new influenza subtype emerges against which there is minimal

immunity in the population, resulting in concurrent epidemics with a high number of cases and

fatalities worldwide (WHO 2005). The 1981 H1N1 Spanish flu and the 2009 H1N1 S-OIV pandemic

were some of the examples in human history. The estimated fatal case of the recent H1N1 S-OIV

2009 pandemic was estimated to be 300 000 (Dawood et al. 2012). International airline travel

accelerated its global spread as the rate of transmission increased proportionally to the number of

people travelling from affected countries (Davis et al. 2013; Neatherlin et al. 2013). The outbreak

started in Mexico in April 2009 and spread to North America and Europe in May 2009, and World

Health Organisation (WHO) declared a full-blown pandemic in June 2009 (WHO 2013). Vaccines

were only available six months (October 2009) after the onset of the pandemic, when it had spread

across the globe and reached Australia, with delays in certain countries where antivirals were

distributed before the pandemic (WHO 2013).

During the 2009 pandemic, school closures occured across affected countries and there were many

absentia from the workplace due to treatment or avoidance of possible transmission (Keogh-Brown

et al. 2010). These factors, along with the pathological effects of influenza were estimated to place a

heavy economic burden on the individual level (Sander et al. 2009), and loss in gross domestic

product (GDP) on the national level (Keogh-Brown et al. 2010). The use of antiviral drugs and

vaccination were estimated to reduce these economic burdens significantly (Keogh-Brown et al.

2010; Sander et al. 2009), which underlines the importance of vaccination in mitigating influenza

pandemic and epidemic, both health wise and economy wise.

2.3.2 Current influenza vaccines and their limitations

The aim of vaccination is to activate the adaptive immunity of a patient against a target pathogen.

Although the adaptive immunity lags in response during primary pathogen exposure, it acquires

memory of the pathogenic antigen. When the same pathogen attacks a second time, the adaptive

immunity via memorised antigen recognition, clears away the pathogen before the disease can

progress (Murphy et al. 2008). There are two types of adaptive immunity, humoral immunity and

cellular immunity. Humoral immunity plays a role in clearing extracellular pathogens by the

production of antibody, leading to neutralisation, complement activation, and opsonisation. Cellular

immunity plays a role in clearing intracellular pathogens by killing infected cells displaying pathogen-

derived antigens on their surface (Murphy et al. 2008).

32

Vaccination by administration of harmless antigen into the patient aims to mimic the first round of

infection to activate the adaptive immunity of the patient. Administered influenza antigens may be in

the form of killed or attenuated-live pathogens or their subunit proteins. Administration of harmless

antigen would mimic the first infection without causing severe disease, to enable the adaptive immune

system to acquire memory about the antigen. In turn, the next time the real pathogen attacks, the

adaptive immune system would be able to clear it immediately without the pathogen causing the

disease to the vaccinated individual (Murphy et al. 2008; Nolz and Harty 2011a). Section 2.3.3 will

explain the mechanism of memory acquisition and subsequent selective cytotoxicity of the cellular

immunity in more detail.

Current influenza vaccines are usually trivalent or quadrivalent. A trivalent vaccine contains two

Influenza A strains and one Influenza B strain according to WHO recommendation for that season.

A quadrivalent influenza vaccine contains and additional Influenza B strain (Krammer and Palese

2015). Licensed influenza vaccines to date come in the form of inactivated influenza vaccines (IIV),

live-attenuated influenza vaccines (LAIV), and recombinant influenza protein vaccine (Cox and

Hollister 2009; Del Giudice and Rappuoli 2015; Jin and Subbarao 2015; Krammer and Palese 2015)

(Table 2.2). An IIV consists of killed whole virus that could further be disrupted into split vaccines

and protein subunit vaccines. This method was first developed in 1937 and has been further optimised

ever since and recent development includes the addition of adjuvants and high-dose administration

(Del Giudice and Rappuoli 2015; Krammer and Palese 2015). An LAIV is an artificially reassortant

virus with six backbone gene segments that introduce attenuation of viral replication in tissues >25°C.

Two other gene segments are HA and NA from circulating strains. LAIV vaccines were the second

influenza vaccine technology licensed after IIV and gave better protection than IIV (Jin and Subbarao

2015). A new generation of licensed influenza vaccine, recently approved by FDA in 2013, is the

trivalent Flublok vaccine containing recombinant HA proteins produced in insect cells, without the

whole virus (Cox and Hollister 2009). Other vaccines based on more advanced technologies such as

reverse genetic viruses are currently in the pipeline (Gomila et al. 2013; Suphaphiphat et al. 2010).

There are several impediments in currently licensed vaccines, one of which is egg-dependent

production, and in the case of an avian influenza outbreak, highly virulent avian influenza virus may

not be grown in high titres in embryonated eggs, and therefore, conventional vaccine production

would be delayed (Buckland 2015). Recent advancement in technology allows the use of mammalian

or insect cells for production (Table 2.2), which, however, requires higher biosafety measures, adding

significantly to production cost, rendering the vaccines less affordable in developing countries

(Buckland 2015). Furthermore, global vaccine production capacity would not be able to satisfy global

demand in the event of a pandemic. As an example, seasonal influenza vaccine production capacity

33

Table 2.2 Currently licensed influenza vaccine.

Vaccine technology

Expression system

Biosafety Level

Product Immunity

Type Breadth

IIV (Del Giudice and Rappuoli 2015)

Eggs. BSL-2 Inactivated virus, split virus, subunit vaccines.

Humoral. Strain specific

Mammalian cells.

BLS-3 Inactivated virus, split virus, subunit vaccines.

Humoral. Strain specific

LAIV (Jin and Subbarao 2015)

Eggs. BSL-2 Attenuated virus.

Humoral, cellular.

Strain specific but broader than IIV

Mammalian cells.

BSL-3 Attenuated virus.

Humoral, cellular.

Strain specific, but broader than IIV

Recombinant protein (Cox and Hollister 2009)

Insect cells. BSL-2 HA proteins. Humoral. Strain specific

34

was 876 million doses during the 2009 H1N1 S-OIV pandemic according to WHO report (Girard et

al. 2010). Monovalent pandemic vaccine production six months after strain identification was

estimated to be 4.9 billion doses according to a joint report by the International Federation of

Pharmaceutical Manufacturers and Associations – Influenza Vaccine Supply task force (IFPMA IVS)

and the European Vaccine Manufacturers (EVM) (Abelin et al. 2011; Collin et al. 2009). However,

global population was estimated to be 6.8 billion people (PRB 2009) at that time. Global supply is

still not able to satisfy global demand, exacerbated by logistic problems such as vaccine distribution

and vaccine shelf-life. These problems necessitate new approaches in vaccine manufacture

technology which preferably involves simpler and cost efficient procedures, shorter production time,

high yield, and longer vaccine shelf-life.

Moreover, all of the currently licensed vaccines aim to stimulate the humoral immunity for antibody

production and are thus strain specific as the influenza virus external proteins frequently mutate due

to antigenic drift, antigenic shift, and recombination (Broadbent and Subbarao 2011; Krammer and

Palese 2015; Subbarao and Joseph 2007). Due to the highly mutative behaviour of the external

proteins, influenza vaccines have to be updated from season to season and in the occasional pandemic

outbreak to match mutated circulating strain (Broadbent and Subbarao 2011; Subbarao and Joseph

2007). Current vaccine manufacturing technology requires six months, including lengthy regulatory

process, to produce distributable vaccines (Buckland 2015). As circulating strains keep mutating,

vaccine strains may not match the circulating strain resulting in reduced immunogenicity (Buckland

2015; Singh et al. 2010). These drawbacks of the dominant existing technology necessitate the

development of broadly cross-protective influenza vaccine able to raise immunity against

heterologous strains to reduce vaccine update frequency and to facilitate better preventive measures

for an emerging pandemic (Brown and Kelso 2009; Subbarao and Joseph 2007). There are several

options to achieve this, for example by utilising conserved B cell epitopes of HA and M2e identified

recently to increase heterologous humoral immunity (Krammer and Palese 2015). Moreover, a

strategy to raise the cellular immunity may also be explored, and may have several advantages

including the utilisation of more conserved internal proteins of influenza to enable the stimulation of

broadly cross-protective immunity. To develop an MPyV VLP-based Tc-epitope carrier to enable

cellular immunity stimulation against influenza is the main aim of my research.

2.3.3 The mechanism of cytotoxic T cell activity

The aim of Tc-epitope-based vaccination is to protect from severe infection by the generation of

memory Tc specific to the pathogen in question. Memory Tc generated via Tc-epitope vaccination

35

are expected to recognise the antigen and exert its cytotoxic activity to eliminate the pathogen faster

than the time it takes for the pathogen-related disease to progress (Nolz and Harty 2011b).

APCs, especially dendritic cells, prime the Tc to initiate the acquisition of memory and subsequent

selective cytotoxicity (Murphy et al. 2008; Yewdell et al. 2003) (Figure 2.5A). Dendritic cells have

to internally process and externally present the antigens to Tc to prime them with the foreign

pathogenic antigens. Dendritic cells process antigens by degrading intracellular viral or bacterial

proteins into smaller peptides via cytoplasmic proteasomal activity. The degraded peptides are then

transported into the ER via the transporter associated with antigen processing (TAP). In the ER, the

peptides are further degraded into smaller 8-10 amino acid, usually 9-mer peptides by the ER

aminopeptidase associated with antigen processing (ERAAP). These 8-10 amino acid long peptides

are called Tc-epitopes. One Tc-epitope then bind to the Tc-epitope binding cleft of one MHC Class I

and the combination of the MHC Class I binding cleft and the Tc-epitope is called the antigenic

determinant. MHC Class I is partly embedded into the inner membrane layer of the ER cavity,

anchored there by several other proteins, including tapasin, which binds to both MHC Class I and

TAP, thus bringing the Tc-epitope and MHC Class I in proximity to each other. Once the Tc-epitope

binds to MHC Class I, the MHC Class I becomes stabilised and travels from the ER to the surface of

the dendritic cells (Murphy et al. 2008; Yewdell et al. 2003).

Dendritic cells usually take up foreign pathogens at the periphery by several mechanisms, including

macropinocytosis and phagocytosis. Proteins degraded in the phagolysosome can enter the cytoplasm

and be further degraded by the proteasome and enter the MHC Class I pathway, and this event is

called cross-presentation (Murphy et al. 2008). Upon foreign antigen uptake in the periphery,

dendritic cells become activated and travel to peripheral lymphoid organs such as the lymphatic

nodules or the spleen. Surface antigen presentation by dendritic cells activates the naïve Tc, which is

a Tc that has not been primed with foreign antigen before (Murphy et al. 2008; Yewdell et al. 2003).

The Tc recognises the surface presented peptide:MHC Class I complex (antigen determinant) by its

T cell receptor (TCR). Activated Tc (effector Tc) then exit the peripheral lymphoid organs and travel

across the lymphatic system and circulate in the blood to scan for other host cells displaying the same

peptide:MHC Class I complex that they have been primed before by the dendritic cells in the

lymphoid organs. Host cells infected by the same pathogen as the dendritic cells display the same

antigenic determinant. Thus, the Tc only select these cells to exert its cytotoxic activity. Upon

antigenic determinant recognition via the TCR, effector Tc release their granules into the cytoplasm

of the infected cells. These granules contain perforin, granzymes A and B, and granulysin. These

cytokines stimulate apoptosis in the target cells by initiating the caspase pathway, leading to DNA

36

Figure 2.5 The mechanism of Tc activity. A. The MHC Class I pathway: 1) viral entry into host cell; 2) viral protein synthesis; 3) proteasomal degradation; 4) entry into ER via TAP; 5) Tc-epitope binding to MHC Class I; 6) peptide:MHC complex migrate to the surface of the cell; 7) peptide:MHC complex recognition by TCR. B. Activation of Tc: 1) dendritic cell pathogen uptake; 2) dendritic cell migration to lymphoid organs; 3) Tc-epitope presentation by the MHC Class I pathway; 4) peptide:MHC complex recognition by TCR; 5) naïve Tc activation and proliferation; 6) Tc differentiation into effector and memory cells; 7) activated Tc migrate to periphery; 8) Tc selectively kills cells which present the same antigen they have been exposed to in the lymphoid organs (Adapted from Abidin (2010)).

37

damage of both pathogen and host cell and eventual cell death. By signalling the death of pathogen-

infected host cells, the Tc stops pathogen replication and also hinders pathogen spread by killing the

cells the pathogens reside in (Murphy et al. 2008; Yewdell et al. 2003) (Figure 2.5B). After 12 days,

90-95% Tc undergo apoptosis while 40 days after, the rest differentiate into central memory T cells

(Tcm). The number of Tcm is about 500-1000 fold of its naïve precursor and they live for extended

period of time. They can home to peripheral lymphoid organs and exert robust recall response upon

secondary antigen exposure (Harty and Badovinac 2008).

The MHC Class II pathway occurs with a slightly different mechanism where protein degradation

and MHC binding takes place in the acidic endolysosomal compartment (Murphy et al. 2008). T cells

corresponding to MHC Class II differentiate into a number of cell types: 1) T helper cells type 1 (Th1)

which activate macrophages to kill intracellular pathogens; 2) T helper cells type 2 (Th2) which

mature B cells into plasma cells to produce antibodies of immunoglobulin (Ig)E types, as well

activating mast cells and eosinophils; 3) T follicular helper cells (TFH) which also matures B cells into

plasma cells to produce antibody of IgG types; 4)T helper cells type 17 (Th17) which helps in

recruiting neutrophils into site of infection; and 5) regulatory T cells (Treg) which prevent cytotoxic

activity against self-antigens (Murphy et al. 2008).

A Tc-epitope-based vaccine would target cellular immunity by delivering Tc-epitopes into dendritic

cells to prime naïve Tc cells. However, this type of vaccine would have to take into account the

extensive polymorphism of HLA (human version of MHC) molecules. As of April 2016, there are

10 730 recorded HLA Class I and 3 743 HLA Class II alleles (Robinson et al. 2015). The analysis of

nonameric Tc-epitopes revealed that amino acid 2 and amino acid 9 bind specifically to pocket B and

F of HLA Class I binding pockets respectively, defining their specificity to different HLA Class I

types (Sette and Sidney 1999; Sidney et al. 2008). HLAs bind specifically to Tc-epitopes, however,

some HLAs may bind to the same Tc-epitopes due to similarity in the B and F pockets, and based on

this most HLAs can be grouped into supertypes (Lund et al. 2004; Sette and Sidney 1999; Sidney et

al. 2008). According to Sette and Sidney (1999), Tc-epitope-based vaccines targeting 9 supertypes

(A2, A3, B7, B44, A1, B27, B62, and B58) may cover 99.8% of the global population.

2.3.4 Advantages of Tc-epitope delivery with MPyV VLP platform

The potential advantages of using the MPyV VLP platform as a vaccine carrier for influenza Tc-

epitope had been pointed out throughout this literature review and will be summarised again in this

subsection. These advantages may be categorised based on the following aspects: 1) immunogenicity;

2) process development; 3) production; 4) safety; and 5) stability (Table 2.3).

38

Table 2.3 Advantages of MPyV VLP platform for Tc-epitope-based vaccines.

Advantage Implication for

influenza vaccine production

Disadvantage Implication for

influenza vaccine production

Immunogenicity Enter different cell

types. Enter dendritic cells

and the MHC Class I and II pathways.

Immunogenic repeated molecular patterns (PAMP).

Pre-existing immunity does not interfere with protection.

Can vaccinate hosts of different species.

Cellular immunity stimulation for broader cross-protection.

Less updating frequency.

Adjuvant not needed. Allows booster

vaccination.

Process development Encapsidation

possible. In vitro assembly. Well established

knowledge-base.

Multiple epitope delivery for broader population coverage.

Circumvent hydrophobicity related aggregation.

Easier manipulation.

Downstream processing may be less applicable in developing countries.

Need to develop simpler purification methods.

Production Bacterial production

system. Safe, BSL-1. Scalable.

Flexible vaccine production.

Simple manufacturing process.

Cheap, fast, and high yield production.

Genetically modified product.

Stricter regulation.

Safety No genetic material. Safe.

Stability Stable at RT for 9

weeks. Stable at -80°C for a

couple of years.

Stockpiling and remote distribution possible.

39

Viral capsids have a distinct characteristic of repetitive molecular patterns commonly known in

immunology as PAMP, recognised by the innate immunity and dendritic cells as a hallmark of a

foreign entity. Upon recognition of the foreign entity by special receptors, the innate immune system

and dendritic cells produce different signals which lead to dendritic cell maturation and activation to

prime T cells (Jennings and Bachmann 2008; Melchjorsen 2013). The MPyV VLP has been shown

to activate and maturate dendritic cells and is one of only two polyomaviral-like particles of the

polyomaviridae family that do so effectively (Gedvilaite et al. 2006). MPyV VLPs are also able to

enter dendritic cells and be processed by both the MHC Class I and MHC Class II pathways as has

been demonstrated by Abbing et al. (2004) and Boura et al. (2005). Moreover, they are able to enter

different type of cells via different sialic acid receptors (Section 2.2) and thus can be used as Tc-

epitope-based vaccine for a different host species. These are the immunological advantages of the

MPyV VLP as a carrier for influenza Tc-epitopes.

In addition to external epitope presentation by insertion at the surface-exposed loop, the MPyV VLP

may also encapsidate foreign protein inside its cavity (Abbing et al. 2004; Boura et al. 2005), allowing

the carriage of soluble domain proteins with multiple epitopes of different HLA Class I specificities.

This strategy may circumvent hydrophobicity-related aggregation and possibly strengthen T cell

response. Assembly of MPyV VLP could also be done in vitro, which eliminates disassembly and

reassembly process usually needed to remove genetic material (Chuan et al. 2010). The process may

potentially be manipulated for capsomere co-assembly to form either mosaic particles or to control

the amount of encapsidated foreign protein. Recombinant technology and process development of the

MPyV VLP platform has been well established at AIBN-CBE and thus may facilitate process

development. These are the process developmental advantages of the MPyV VLP as a carrier for

influenza Tc-epitopes.

A scalable bacterial expression platform in a biosafety lever (BSL)-1 environment has been developed

at the AIBN-CBE for the MPyV VLP, which allows simple manufacturing process, lower production

cost, shorter production time, and higher production yield (Chuan et al. 2014). These are the

production advantages of the MPyV VLP as a carrier for influenza Tc-epitopes. However,

downstream processing involves relatively advanced purification apparatus, and thus adjustment may

be needed if manufacturing is set in developing countries. Nevertheless, low production cost and high

yield, as simulated for large scale production (Chuan et al. 2014), may minimise vaccine

manufacturing plant localisation. Vaccine price would then only be limited by distribution cost and

vaccine shelf-life. In addition, a precipitation method to harvest MPyV capsomeres is being

developed at the AIBN (Wibowo et al. 2015). MPyV VLP for up to a couple of years of storage is

stable at -80°C, and at room temperature (RT) for up to 9 weeks as previously mentioned (Caparros-

40

Wanderley et al. 2004). The MPyV VLP is also safe as it does not contain genetic material (Chuan et

al. 2010). However, MPyV VLP is a genetically modified product and could meet stricter regulation

procedures (Kushnir et al. 2012).

This main section has summed up the advantages and disadvantages of the MPyV VLP platform to

develop into a Tc-epitope-based influenza vaccine. Even though the MPyV VLP platform is easily

adjustable, a method to analyse and characterise modified VLPs to confirm successful VLP formation

and modification was not available at the time I pursued my research. Thus, a new SEC method was

developed alongside to analyse and characterise VLPs modified as Tc-epitope-based influenza

vaccine. The next main section will describe briefly the basic principles of SEC and factors

influencing an efficient SEC method.

2.4 Size exclusion chromatography

Limited availability of methods to isolate the VLPs for analysis and characterisation hinders the

development of VLP-based vaccines. A few of these methods have been recently reported (Ladd Effio

et al. 2016; Lagoutte et al. 2016), but were not readily available at the time I did my research. Thus,

a new preparative SEC method to enable isolation and characterisation of VLPs was developed in my

research. This section briefly describes the principles of SEC and gel filtration in process

development.

2.4.1 Principles of size exclusion chromatography

SEC or gel filtration chromatography is a method to separate molecules based on their size. The

sample is run through a liquid mobile phase and a solid stationary phase. The stationary phase consists

of a porous matrix with a defined porosity, typically in the form of inert beads, made from a variety

of materials such as agarose, dextran, or cross-linked polyacrylamide or the combination thereof

(O'Fagain et al. 2011). The liquid mobile phase consists of external volume and internal volume. The

external volume is the volume outside the pores, and is also called the void volume. The internal

volume is the volume inside the pores. Total liquid volume comprises both the external and the

internal volumes (O'Fagain et al. 2011).

Due to their sizes, particles have different abilities to travel inside the pores. Particles that are too

large to travel through the pores will elute earlier, and molecules of smaller sizes travel through more

pores and thus elute in decreasing particle size (Hagel 2011; Hagel and Janson 1992; O'Fagain et al.

2011) (Figure 2.6). The volume of the peak when a particle of a certain size elutes from the column

41

Figure 2.6 Principles of size exclusion chromatography. A. Magnified diagram of a media bead and its pores. B. Illustration of selective pore entry based on sample particle size. C. SEC run: 1) sample application; 2) small molecules (yellow) travel slower through the column while large molecules (red) travel faster; 3) large molecules are eluted earlier from the colum. D. SEC chromatogram. Vo is void volume, Ve is elution volume, and Vt is total liquid volume (Adapted from GE-Healthcare (2010b).

42

is called retention volume, or elution volume. Besides particle size, elution volume is also dependent

on the hydrodynamic diameter of the particles. Particles of asymmetrical shapes, such as fibrous

proteins, travel faster than globular particles of the same molecular weight, appearing as if they have

a larger particle size (Hagel 2011; Hagel and Janson 1992; O'Fagain et al. 2011). The size of the pores

and the beads in the matrix depends on the composition of the media, and media with different particle

size separation range are available commercially. Examples of commercially available media include:

1) Superdex, composed of dextran and agarose, with 0.1 -102 kDa separation range; 2) Superose,

composed of highly cross-linked porous agarose particles, with a

1 - 103 kDa separation range; and 3) Sephacyrl, composed of covalently cross-linking allyl dextran

with N,N’-methylene bisacrylamide, with a 1-105 kDa separation range. Each of these media are

divided into types with a narrower separation range, for example, Sephacryl S100 HR has a 1-102

kDa separation range while Sephacryl S500 HR has a 10-105 kDa separation range (GE Healthcare

Lifesciences). SEC has several applications such as high-resolution protein or peptide fractionation,

desalting, solvent removal, removal of inhibitors from enzymes, separation of cells and virus

particles, molecular mass estimation, and removal of byproducts of chemical reactions such as

PEGylation (GE-Healthcare 2010b; O'Fagain et al. 2011). A good media should have a reasonable

pore volume to provide maximum separating volume, and a reasonable particle diameter to give

optimum column efficiency (Hagel and Janson 1992).

2.4.2 Column efficiency and influencing factors

A good SEC separation is marked by a good resolution, which is the degree of separation between

peaks in an elution profile chromatogram. Resolution is a function of selectivity and column

efficiency. Selectivity relates to separation range of the pore sizes of the media beads discussed in the

previous subsection. Column efficiency is the efficiency of the column to produce narrow peaks and

to reduce zone broadening (GE-Healthcare 2010b; Hagel 2011). Column efficiency can be measured

by determining the number of theoretical plates, height equivalent of a theoretical plate, reduced plate

height, and asymmetry. The value of the number of theoretical plates, height equivalent of a

theoretical plate, and reduced plate height are derived from relative peak width, while asymmetry is

a measurement of deviation from ideal Gaussian peak shape (GE-Healthcare 2010a; GE-Healthcare

2010b; Hagel 2011).

A theoretical plate is a theoretical zone required for the mobile phase and the stationary phase to reach

an equilibrium with each other (Braithwaite and Smith 1999; Hamilton and Sewell 1982). The number

of theoretical plates is conversely related to elution volume, but inversely related to peak width at

half peak height which correlates to zone broadening (Figure 2.7). The higher the number of

43

Figure 2.7 Zone broadening variables. Wh is peak width at half peak height. a is forward peak and b is rear peak length at 10% peak height. Ve is elution volume of the particle (Adapted from Hagel (2011)).

44

theoretical plates, the less the zone broadening, the better the column efficiency. This relationship is

formulated in the following equation:

5.54 2.1

where N is the number of theoretical plates, Ve is the elution volume, and Wh is peak width at half

peak height.

The theoretical plate can also be translated into its height, defined as height equivalent of a theoretical

plate. Shorter plate height reflects better column efficiency (Braithwaite and Smith 1999; Hamilton

and Sewell 1982). The height equivalent of a theoretical plate can be derived by dividing the length

of the column with the calculated N of the column, formulated in the following equation:

2.2

where HETP is height equivalent of a theoretical plate and L is the length of the column.

Further, the HETP can be translated into reduced plate height that denotes the number of vertical

media beads within a theoretical zone by dividing HETP with the bead diameter. An excellent column

efficiency would have a reduced plate height <3 (Hamilton and Sewell 1982), formulated in the

following equation:

2.3

where h is reduced plate height, and dp is bead diameter.

Zone broadening and column efficiency may also be measured by calculating the asymmetry of the

peak. Ideal solute distribution through the column should follow a Gaussian distribution, and

deviation is visible as peak tailing. Peak tailing may indicate low column efficiency or solute

interaction with the media (Braithwaite and Smith 1999). Asymmetry is the ratio between the rear

peak part and the forward peak part at 10% peak height, both measurable as volume or time (Figure

2.7). Good asymmetry may fall between 0.8-1.3 depending on the media, and formulated with the

following equation:

2.4

where As is asymmetry, b is rear peak part width at 10% peak height, and a is forward peak part width

at 10% peak height.

45

Total diffusion time and total diffusion distance influence column efficiency. Flow rate, column

length, and solute viscosity affect total diffusion time while particle size and pore size distribution

affect total diffusion distance (Hagel and Janson 1992). In addition, mixing chambers and external

dead volumes in connectors, adaptors, and other parts may contribute to zone broadening and reduce

column efficiency (Hagel 2011). Sample volume may also cause zone broadening, and a sample

volume of 1-2% of total column volume is usually recommended (Hagel 2011).

The information available in the literature as described in this main section would be helpful in the

development of an SEC method to analyse and characterise modified VLPs developed in my research.

The information will aid in media selection and overall evaluation of column performance and

efficiency, which is required to confirm the validity of the column to analyse and characterise

modified VLPs. The variables described in Section 2.4 will only be used in the context of semi-

preparative SEC column efficiency testing in this thesis.

.

46

3 Insert engineering and solubility

screening improves recovery of virus-like

particle subunits displaying hydrophobic

epitopes

The entire Chapter 3 consists of the journal article submitted as:



24(11):1820-8.

The following modifications were made to the original submitted article:

- Page numbers of the original submitted article were changed into the numbers consistent with

those on the remainder of the thesis page numbers.

- Figure numbers were changed into the numbers consistent with those on the remainder of

figure in the thesis.

- Section and sub-section numbers were changed into the numbers consistent with those on the

remainder of sections and sub-section numbers in all chapters of the thesis.

- The reference style of the original submitted article was changed into the style consistently

used in all chapters of the thesis.

47

3.1 Abstract

The Polyomavirus coat protein, VP1 has been developed as an epitope presentation system able to

provoke humoral immunity against a variety of pathogens such as Influenza and GAS. The ability of

the system to carry Tc-epitopes on a surface-exposed loop and the impact on protein solubility has

not been examined. Four variations of three selected epitopes were cloned into surface-exposed loops

of VP1, and expressed in E. coli. VP1 pentamers, also known as capsomeres, were purified via a

GST-tag. SEC indicated severe aggregation of the recombinant VP1 during enzymatic tag removal

resulting from the introduction of the hydrophobic epitopes. Inserts were modified to possess double

aspartic acid residues at each end of the hydrophobic epitopes and a high-throughput buffer condition

screen was implemented with protein aggregation monitored during tag removal by

spectrophotometry and dynamic light scattering (DLS). These analyses showed that the insertion of

charged residues at the extremities of epitopes could improve solubility of capsomeres and revealed

multiple windows of opportunity for further condition optimisation. A combination of epitope design,

pH optimisation and the additive L-Arg permitted the recovery of soluble VP1 pentamers presenting

hydrophobic epitopes and their subsequent assembly into virus-like particles.

Keywords: aggregation, cytotoxic T cell epitope, high-throughput screening, hydrophobicity,

hydrophobic epitopes, protein aggregation, protein engineering, virus-like particles.

3.2 Introduction

The coat protein subunits (VP1) of MPyV form a pentameric structure, called a capsomere. Purified

capsomeres self-assemble to VLPs under defined conditions (Chuan et al. 2010; Garcea et al. 1987;

Salunke et al. 1986) and both capsomeres and VLPs have been developed into vaccine platforms

(Middelberg et al. 2011; Wibowo et al. 2013). For example, MPyV VLPs presenting a helical epitope

from influenza hemagglutinin via genetic insertion into surface-exposed loops of VP1 elicited a

strong antibody response (Anggraeni et al. 2013). The capsomere platform allows for the presentation

of antigens at the VP1 termini as well as internal loops and multiple influenza M2e epitopes presented

in this way are able to stimulate protective humoral immunity in mice (Wibowo et al. 2013). As

capsomeres can be produced in bacterial cultures, both platforms have the potential to reduce vaccine

production cost in comparison to that of conventional vaccines (Chuan et al. 2014).

In addition to stimulating humoral responses, MPyV VLPs are known to stimulate strong cellular

immune responses (Caparros-Wanderley et al. 2004). MPyV VLPs have great potential as a delivery

system for Tc-epitopes because they enter the cytoplasm during infection, thus escaping the

48

endosome (Sapp and Day 2009) and this ability has been observed using isolated dendritic cells in

vitro (Boura et al. 2005). Processing by the immune-proteasome leads to MHC Class I presentation

of epitopes and a cellular immune response (Schirmbeck et al. 1995). Furthermore, MPyV VLPs are

able to activate dendritic cells (Boura et al. 2005) leading to antigen specific T cell proliferation and

likely immune-proteasome processing (Fric et al. 2008). Therefore, MPyV VLPs displaying Tc-

epitopes could provide effective vaccines for the stimulation of cellular immune responses that are

both relatively cheap and quick to manufacture.

However, selection bias for MHC Class I presentation dictates that Tc-epitopes are commonly

hydrophobic (Mitic et al. 2014). Aggregation is a limiting factor in every step of protein bioprocessing

(Cromwell et al. 2006) and a major role of hydrophobicity in protein aggregation has been

demonstrated in recent studies (Chong and Ham 2014a; Chong and Ham 2014b). The stability of a

protein in aqueous solution is heavily dependent on the solvent environment and unfavourable

interactions between the solvent environment and protein surfaces can de-stabilise colloidal and

conformational stability (Chi et al. 2003). Protein colloidal stability is influenced by intermolecular

protein-protein interaction, which depends on overall electrostatic and hydrophobicity of the protein

(Chi et al. 2003; Chong and Ham 2014b)(15; 17). Conformational stability can be lost by unfolding,

leading to intermediate and fully denatured species. Partial unfolding, either by natural or chemical

denaturation, or by incomplete folding, usually exposes the hydrophobic region of the protein and

thus increases attractive intermolecular interactions and initiates aggregation (Wang et al. 2010b).

Insertion of hydrophobic entities into surface-exposed loop of a protein may emulate this, also leading

to aggregation, due to artificial hydrophobic interactions not seen in the native protein. A promising

strategy to counter this effect is to co-engineer charged amino acid residues alongside the inserted

hydrophobic entities (Perchiacca et al. 2012).

Our current research further examines the ability of MPyV VLPs to carry Tc-epitopes to potentially

elicit cellular immune responses against influenza. To this end, we designed four variants of MPyV

VP1 with single or multiple Tc influenza epitopes, with or without charged amino acid inserts,

inserted into a surface-exposed loop. Severe aggregation in the standard buffer condition chosen for

unmodified VP1 was observed upon enzymatic removal of the affinity purification tag. To help

overcome this aggregation a buffer condition screen comprising six different additives across five

different pH values was implemented and analysed by different methods. Our results showed that

such high-throughput buffer screening for solubility identified windows of opportunity to improve

capsomere recovery. Furthermore, in the presence of additives, insert engineering made a significant

difference to modified capsomere solubility, resulting in improved recovery and permitting the

assembly of VLP-like structures.

49

3.3 Results

3.3.1 MPyV VP1 displaying Influenza cytotoxic T cell epitopes

Four insertions were designed from three Tc-epitopes (Table 3.1), which were inserted into the HI

loop of MPyV VP1 (Middelberg et al. 2011). For each of the four constructs the spacer amino acid K

was added at the C-terminus of the epitopes for better cellular processing (Livingston et al. 2001).

IT1 was inserted alone and a longer insert possessing all three epitopes was created (IT123).

Additional designs for both the IT1 and triple epitopes were generated using double Aspartic acid

(DD) inserts at the N- and C-terminus of IT1 (DDIT1) and at the N-terminus of each epitope in IT123

(DDIT123). The resulting constructs encoded glutathione S-transferase (GST)-VP1 fusion proteins

linked by thrombin cleavage sites (Figure 3.1).

3.3.2 Aggregation of modified VP1 during affinity tag removal

Wild type VP1 and the modified variant IT1 were expressed in E. coli, purified by GST affinity

chromatography and treated with thrombin for two hours to remove the GST tag. SEC of the digestion

reaction showed that in contrast to wtVP1, no soluble capsomere could be detected for IT1 using

standard buffer conditions, previously optimised for unmodified VP1 (Figure 3.2). Moreover, the

apparent reduction in absorbance at A280 nm indicates that a significant portion of IT1 is lost as

insoluble aggregates following centrifugal clarification prior to SEC. Spectrophotometric monitoring

of the enzymatic digest showed that there was an increase in aggregates as the GST tag was released

from IT1 (Figure 3.3). Initial values suggested that even before the tag removal there was a difference

in the aggregation status of IT1 compared to wtVP1 capsomere. After an initial period of slight

decrease, the Aggregation Index (AI) of the IT1 reaction increased steadily from 20 minutes

indicating the formation of IT1 aggregates as the GST tag is removed. In contrast, wtVP1 also showed

a slight decrease in AI during the early stages of the reaction, but this was not followed by an increase

in AI, indicating no induction of aggregation. No change in solubility for both wtVP1 and IT1 was

observed in the absence of thrombin (data not shown). Therefore, in these reaction conditions,

optimised for wtVP1 capsomeres (Chuan et al. 2008b; Lipin et al. 2008b), IT1 tended to aggregate as

the proportion of tag-free capsomere accumulated.

3.3.3 High-throughput buffer screen

A 40 buffer condition library was developed on 96 well plates by varying pH and additives during

enzymatic GST removal. The additives used comprised detergents (Leibly et al. 2012), charged amino

50

Table 3.1 Influenza T cell epitopes used in this study.

Epitope1 Name Protein Origin References

GILGFVFTL IT1 Matrix 1 (Alexander et al. 2010)

LLTEVETYV IT2 Matrix 1 (Adar et al. 2009)

RLIQNSLTI IT3 Nucleoprotein (Ishizuka et al. 2009)

1 The Tc-epitopes are presented by HLA A*0201.

Figure 3.1 Schematic representation of designed and cloned MPyV VP1 constructs. Cytotoxic Tc-epitopes IT1, IT2, and IT3 were inserted into GST-tagged VP1 in 4 different configurations. K, lysine spacer; DD, double aspartic acid insert. The arrow indicates the location of the thrombin cleavage site.

51

Figure 3.2 Modified capsomere aggregation during GST removal. Following enzymatic tag removal wtVP1 can be separated by SEC into an aggregate peak, soluble capsomere peak and GST peak. The small peak eluting as a shoulder to the capsomere peak has been previously identified as a chaperone-VP1 complex (Liew et al. 2010). Tag removal from IT1 VP1, containing a hydrophobic epitope, results in only free GST and aggregated IT1 VP1.

52

Figure 3.3 Spectrophotometric analysis of the GST removal reaction. (A) SDS-PAGE analysis of wtVP1 digestion reaction with samples taken during the reaction. The position of the GST fusion protein and thrombin digestion products, VP1 and GST, are shown. (B) Change in the aggregation index of wtVP1 and IT1 VP1 digestion mixtures over the course of the reaction.

53

acids (Ho et al. 2003; Leibly et al. 2012; Tsumoto et al. 2004), osmolytes (Leibly et al. 2012), and

molecular chaperones (Lanckriet and Middelberg 2004; Prashar et al. 2011; Serno et al. 2011) as

detailed in Table 3.2. Thrombin activity in the different buffer conditions was assessed and confirmed

that complete digestion of the GST-VP1 fusion was achieved between pH 7.0 and pH 9.0, irrespective

of the additive present. Below pH 7.0, there was reduced digestion efficiency in the presence of some

additives and, therefore, we chose to proceed with pH values at or above neutral.

The AI measurement proved to be a good means by which to detect aggregation in a buffer screening

system. As samples showed considerable differences in their starting values, we compared conditions

by calculating the difference in these values between the start and end of the thrombin digest. Plotted

as heat maps, this parameter showed some differences in the aggregation status of wtVP1 across the

buffer conditions tested (Figure 3.4). In general, VP1 capsomeres were more stable above neutral pH

and additives had mixed effects on solubility. In order to compare the relative changes in AI to an

absolute parameter in the measurement of aggregation status that would enable comparison between

constructs and not just between conditions, we turned to DLS. Plotting a heat map of the endpoint Z-

average size revealed a similar distribution of aggregation status as those observed by

spectrophotometric measurement (Figure 3.4). Thus, taken together, Z-average size and ΔAI can

provide complementary parameters that can be measured in a high-throughput mode to assess the

solubility of MPyV capsomeres in various buffer conditions.

The high-throughput screen was then applied to the four modified capsomere constructs and heat

maps were used to assess the impact of buffer conditions for each construct. The screening showed

that optimal conditions varied considerably for each construct and that complementarity was again

observed between the assessment parameters (Supplementary Figure 3.2). As was observed for

wtVP1, alkaline conditions (pH 8.5 to 9.0) were better for solubility of the modified constructs. The

Z-average size was then used to normalise the heat maps across constructs (Figure 3.5) to determine

the effect of construct design on capsomere solubility. This analysis showed that the inclusion of

charged residues (DD) could, to some extent, overcome the hydrophobicity of the IT1 epitope,

resulting in a solubility profile similar to wtVP1. This was not the case for the larger, though less

hydrophobic IT123 epitope. This result could also be confirmed by normalising final AI

measurements across the constructs (Supplementary Figure 3.3). However, the resolution of this

parameter does not appear to be as good as Z-average size, which may be explained by the fact that

the coefficient of variation across all constructs and conditions is 1.7 times higher for the final AI

than for Z-average measurements.

54

Table 3.2 Additives used in buffer screening.

Class Additive Conc. References

Detergent

CHAPS

10 mM

(Leibly et al. 2012)

Triton X-100 0.05% (v/v) (Leibly et al. 2012)

Amino acid

L-Arginine

50 mM

(Ho et al. 2003;

Tsumoto et al. 2004)

Osmolyte

Trehalose

150 mM

(Leibly et al. 2012)

Sorbitol 200 mM (Leibly et al. 2012)

Molecular chaperone

α-cyclodextrin

10 mM

(Lanckriet and

Middelberg 2004;

Prashar et al. 2011;

Serno et al. 2011)

55

Figure 3.4 Validation of buffer screening with wtVP1. Analysed with two parameters, Z-average size (DLS) and the change in aggregation index (ΔAI), heat maps representing aggregation status give a snapshot of 40 buffer conditions employing various additives across a range of pH values as indicated in 0.5 pH unit increments.

56

Figure 3.5 Heat maps identifying optimal buffer conditions for the processing of modified capsomeres. Z-average size was normalised across all constructs to allow the comparison of constructs as well as the identification of optimised buffer conditions.

57

3.3.4 SEC using selected buffer conditions for capsomere recovery

Several buffer conditions were analysed further with high-resolution SEC to determine whether they

enable IT1 or DDIT1 capsomere recovery. Confirming the results of the screen, high resolution SEC

on scaled down reaction volumes showed that promising conditions such as including the additives

L-Arg and CHAPS at pH 8.5 could help in the recovery of DDIT1, whereas little to no soluble

capsomere could be obtained from IT1 (Supplementary Figure 3.4). The most promising condition,

50 mM L-Arg at pH 8.5 showed a major peak for DDIT1 between the aggregate and the GST peaks

(Figure 3.6A) and SEC on this peak was scaled up as per the materials and methods section. The

appearance of this peak indicated reduced aggregation and enabled recovery of soluble capsomeres.

However, the capsomere peak was shifted to the right in comparison to wt capsomeres (Figure 3.2).

This shift may be due to interaction of the charged amino acid inserts with the stationary phase or

may be an indication of dissociation of the capsomeres into VP1 dimers. Sodium dedocylsulphate

polyacrylamide gel electrophoresis (SDS-PAGE) analysis (Figure 3.6B) showed that the peak

consisted of isolated DDIT1 and to test this the fraction was examined by transmission electron

microscopy (TEM), which showed the presence of pentameric capsomeres (Figure 3.6C). Given the

purity of the capsomere peak, it is likely that the majority of species visible by TEM are capsomeres

in varying orientations, with some in the ideal orientation for identification (arrowheads in Figure

3.6C). However, some species may be capsomeres that are distorted or degraded by TEM grid

preparation. Attempts to assemble the recovered DDIT1 capsomere in the standard assembly buffer,

optimised for wtVP1, resulted in complete precipitation and loss of the protein. However, in the

presence of L-Arg some spherical particles could be observed under TEM, indicating the formation

of VLP-like structure (Figure 3.6D). These results show that the high-throughput solubility screen

can successfully identify conditions for the recovery of MPyV capsomeres modified to present

hydrophobic epitopes and initial assembly trials also indicated that these capsomeres are assembly

competent.

3.4 Discussion

VLPs offer an interesting platform for Tc-epitope presentation because they can act as adjuvants by

presenting PAMP and thus may initiate co-stimulatory signal required for Tc expansion (Murphy et

al. 2008; Raghunandan 2011). In addition, VLPs present a manufacturing advantage over traditional

live-attenuated or killed vaccines through the use of recombinant expression technologies that can

significantly reduce cost (Chuan et al. 2014). In addition, they do not contain nucleic acid, thus

reducing the risk of contamination with infectious agents (Josefsberg and Buckland 2012). Here we

58

Figure 3.6 Isolation of modified capsomere, DDIT1 using optimised buffer conditions. (A) SEC showed a capsomere peak (b) for DDIT1 following GST removal in the presence of 50 mM L-Arginine, distinct from an aggregation peak (a) and GST peak (c). (B) SDS-PAGE gel analysis of SEC fractions, the products of thrombin digest and size of molecular weight standards are shown. Lanes are numbered by the corresponding SEC fraction volume (mL); M, molecular weight standards; a, b, and c correspond to peak fractions as indicated on the chromatogram (A). (C) The DDIT1 capsomere fraction was analysed by TEM showing the presence of capsomeres (black arrowheads point to examples). (D) The DDIT1 capsomere fraction was assembled in the presence of 50 mM L-Arginine and spherical entities resembling VLPs were observed.

59

have shown that capsomeres, the building block of MPyV VLPs, can accommodate hydrophobic Tc-

epitopes. We devised a high-throughput screen to identify buffer conditions for their soluble recovery

and show that such modified capsomeres are assembly competent.

Examination of the thrombin digest to remove GST tag from the IT1 construct indicated a slight

increase in solubility during the initial stage of GST removal, followed by steadily increasing

aggregation as the reaction progressed. There was also a slight reduction in the aggregation of wtVP1

as GST was removed, which continues until 60 minutes into the reaction, after which the aggregation

status stabilised. Increased protein concentration correlates proportionally to aggregation (Treuheit et

al. 2002), as increased concentration reduces proximity of proteins in solution enhancing their

interaction. In our system initial aggregation of affinity-purified capsomeres appears to involve the

GST-tag itself, which can occasionally induce aggregation, often when fusion partners possess

hydrophobic regions (Terpe 2003).

IT1 is a well characterised Tc-epitope corresponding to amino acids 58-66 of the influenza M1 protein

and its hydrophobicity stems from its localisation, buried within the globular M1 protein. The use of

charged residue inserts, such as double aspartic acid, have been shown to help overcome aggregation

due to hydrophobicity in vaccine protein engineering (Anggraeni 2014) and, indeed, we observed a

considerable effect of the charged inserts in increasing the solubility of MPyV capsomeres displaying

IT1. However, the addition of DD residues alone was not able to overcome this problem in standard

buffer conditions.

Optimisation of the buffer condition was attempted by adding a number of additives shown to prevent

aggregation in previous studies. Buffer conditions were screened in a high-throughput manner and

multiparametric analysis was used to assess the solubility of capsomeres as they were released from

GST. The parameters used to assess solubility were validated with wtVP1 whose capsomere recovery

pattern has been empirically established. The AI has been used to detect the formation of aggregates

with respect to the total protein content (Esfandiary and Middaugh 2012), while the Z-average size

measured by DLS has recently been shown to be inversely related to capsomere formation (Tekewe

et al. 2015). When the buffer screen was applied to the four modified VP1 constructs, heat maps of

the two parameters were in general agreement with each other and suggested that higher pH was more

preferable for capsomere recovery. Differences between the two parameters for certain buffer

conditions likely reflect the fact that AI detects both soluble and insoluble aggregates, whereas DLS

measurements are made only after sample clarification. Therefore, the use of AI can provide a

valuable check on the interpretation of favourable DLS results in cases where there are discrepancies

60

between the two. The detection of insoluble aggregates also explains the higher co-efficient of

variation across the AI measurements compared to DLS measurements.

The suitability of pH 9.0 for IT1 and IT123 and pH 8.5 for DDIT1 may be related to their iso-electric

points (6.6, 7.1 and 5.6 respectively), as buffer conditions further away from the iso-electric point

enhances protein solubility (Cohn 1920). However, a higher pH than 9.0 was not feasible for

downstream processing due to chromatography media incompatibility. Among the additives tested in

this study, the detergents CHAPS and triton X-100 as well as the positively charged amino acid L-

Arg, consistently enhanced the solubility of modified capsomeres. Non-ionic detergents have been

shown to promote solubility of modified capsomeres (Tekewe et al. 2015), however, they can be

difficult to remove completely. The commonly used zwitterionic detergent CHAPS is a relatively

expensive additive although it proved to be very promising in the present study.

Overall, the high-throughput screen successfully identified windows opportunity for the further

processing of MPyV capsomeres as discussed below. The screen was able to provide information on

the effect of construct design on capsomere solubility as well as the selection of buffer conditions

suitable to prevent aggregation of modified capsomeres. Of the 400 data points from 200 reactions

monitored, four DLS measurements failed and one of these corresponded to a failed

spectrophotometric measurement. Therefore, due to the multiparametric analysis, no data was

obtained for just one of the 200 samples, an error rate of 0.5%. The results of the screen were

confirmed by the high resolution, low-throughput SEC, which showed that the use of L-Arg, and to

some extent CHAPS, at pH 8.5 could be used to obtain soluble DDIT1 capsomeres.

L-Arg is a promising additive for preparative processes because it is relatively inexpensive and easier

to remove from the solution compared to other effective additives such as the detergents used in this

study and is also safe for human use (Polycarpou et al. 2013; Yeo et al. 2013). L-Arg is known to be

able to prevent aggregation caused by hydrophobic interactions (Tsumoto et al. 2004), by masking

hydrophobic regions and reducing attractive intermolecular interactions (Ho et al. 2003). In silico

modelling has shown that the guanidium group of L-Arg interacts with hydrophobic patches of

protein while the polar carboxyl part of L-Arg forms a hydrophilic layer (Li et al. 2010). The

combination of this surfactant-like behaviour and self-interactions leading to clusters of L-Arg around

hydrophobic protein domains can provide colloidal stability to protein solutions. Maintaining the

presence of L-Arg was also required to prevent aggregation and precipitation of DDIT1 capsomeres

during in vitro assembly.

Here we have developed an engineering approach and solubility screen to enable the recovery of

recombinant VLP subunits carrying hydrophobic Tc-epitopes. Using a combination of light scattering

61

and spectrophotometry in a high-throughput manner, we were able to identify optimal buffer

conditions for the processing of wt and modified MPyV capsomeres. In addition, we show that

flanking charged amino acids could mitigate the negative impact on capsomere solubility imparted

by some epitope inserts. Conditions identified by the screen could be confirmed to improve

capsomere isolation for VP1 presenting engineered epitopes by SEC and electron microscopy showed

that capsomeres were indeed purified and that they could be assembled into VLP structures. The

colloidal stability of VLPs displaying hydrophobic epitopes is the subject of ongoing work and it will

be interesting to compare the ability of L-Arg and CHAPS to permit effective assembly and maintain

solubility of VLPs. We are also exploring the possibility of mixing modified and wt capsomeres to

reduce the negative influence of hydrophobic inserts, although this must be balanced with the

resulting reduced valency of the epitope. In future work we intend to test MPyV VLPs displaying Tc-

epitopes for their immunogenicity and their ability to stimulate cellular immunity. If this is successful,

it will offer the possibility of developing this platform for other disease models that require cellular

immunity for protection or for therapeutic vaccines against chronic diseases such as cancer. The work

presented here provides a valuable approach in the development of processes to manufacture such

vaccine candidates and those of other epitope presentation systems.

3.5 Materials and Methods

3.5.1 Insertion of cytotoxic T cell epitopes into MPyV VP1

The coding sequence for MPyV VP1 (M34958) in pGEX-4T1 was generously provided by Professor

Robert Garcea (University of Colorado, Colorado, USA) and Tc-epitopes were inserted into a surface-

exposed loop of VP1. Four constructs, containing either single (IT1) or multiple Tc-epitopes (IT1,

IT2, IT3; See Table 3.1 for epitope sequences), with or without double aspartic acid inserts were

generated by in vitro gene assembly and in vivo homologous recombination. See supplementary

information for oligonucleotide sequences and detailed methods. The resulting constructs (Figure

3.1) encoding the VP1 variants IT1, DDIT1, IT123 and DDIT123 fused to GST at the N-terminus

were all confirmed by Sanger sequencing.

3.5.2 Expression and Purification of MPyV VP1 capsomeres

Plasmid of constructs were transformed into competent E. coli Rosetta (DE3) pLysS cells (Novagen,

www.merckmillipore.com), and expressed as previously described (Chuan et al. 2008a) with slight

modification. Briefly, following cell lysis and clarification, GST-tagged protein was purified with

affinity chromatography, the GST tag removed with thrombin (GE Healthcare,

62

www.gehealthcare.com), followed by further purification of the cleaved product with y SEC to

recover purified capsomeres (Chuan et al. 2008a; Lipin et al. 2008b). Running buffer for affinity

chromatography (L-buffer) was composed of 200 mM NaCl, 40 mM Tris base, 1 mM EDTA

disodium, 5% Glycerol, and 5 mM DTT at pH 8. Running buffer for SEC was either L-buffer, or

buffer specific to the condition screened for digestion reaction. Standard SDS-PAGE was performed

to analyse the quality of the protein.

3.5.3 Buffer Condition Screening

Concentration of GST-tagged protein fractions of each construct were measured with Bradford

reagent (Thermoscientific, www.thermofisher.com), and adjusted to 4 mg/ml. This was then mixed

with buffer solution in 1:1 ratio to a final protein concentration of 2 mg/ml, to which thrombin was

added at a ratio of 1:25. Base components of the reaction buffer were 70 mM Tris-base, 150 mM

NaCl, 0.6 mM EDTA, 5% Glycerol (v/v), and 5 mM DTT. In addition, the buffer solution also

contained additives as listed in Table 3.2 and the pH of the buffer solutions were adjusted to give a

range of final pH values from 7.0 to 9.0 in 0.5 pH unit increments such that each additive was tested

at each pH. In addition, a control with no additive and a single combination of additives (L-Arg and

CHAPS) were included resulting in 40 conditions. Buffer screening during the thrombin digest was

carried out on UV-transparent 96-well plates (Corning, www.corning.com) in 100 µL reaction

volume.

3.5.4 Absorbance and turbidity

To assess aggregation during GST tag removal, absorbance at 280 nm and turbidity at 350 nm were

measured every 20 minutes over 2 h with 5 sec of shaking before each measurement using an

Infinite200 Pro Plate Reader (Tecan, www.tecan.com). For solubility screening, absorbance and

turbidity measurements were taken before and after the 2 h digest. These measurements were then

translated into initial and final AI according to the following formula (Esfandiary and Middaugh

2012):

100350

280 3503.1

3.5.5 Dynamic Light Scattering (DLS)

Samples were clarified by centrifugation at 5000 g for 5 min and 25 μL was transferred in duplicate

to 384-well polystyrene plates (Corning, www.corning.com). Scheduled analysis was performed

using a DynaPro Plate Reader (Wyatt Technologies, www.wyatt.com) and DLS results were analysed

with Dynamics software (Wyatt Technologies, www.wyatt.com). DLS was validated by measuring

63

mixtures of capsomeres and aggregates in different ratios to reflect possible particle distribution and

concentration after cleavage. Using Z-average size, the primary statistic given by cumulants analysis,

an inverse correlation was shown with increasing percentage of capsomeres (Tekewe et al. 2015).

3.5.6 Size exclusion chromatography

SEC was performed to detect and recover capsomeres after enzymatic GST tag removal. Digestion

was carried out with selected buffer conditions in a total of 1 ml. SEC was performed on the reaction

using a Superdex 200 10/300 GL column (GE Healthcare, www.gehealthcare.com) on an AKTA®

Explorer 100 system (GE Healthcare, www.gehealthcare.com). The running buffer was matched to

the digestion reaction and run at 0.5 ml/min. Where fractions were collected, the designated aggregate

peak was collected in 1 ml fractions, putative capsomere peak and the GST peak were collected in

250 μl fractions.

3.5.7 In vitro VLP Assembly

Modified DD IT1 capsomere recovered via SEC in buffer containing 50 mM L-Arginine (L-Arg)

(>98%, Sigma-Aldrich, www.sigmaaldrich.com) at pH 8.5 was assembled by dialysis in standard

assembly buffer containing 0.5 M (NH4)2SO4, 20 mM Tris-base, 5% glycerol (v/v), 1 mM CaCl2 at

pH 7.4, with or without L-Arg at 50 mM.

3.5.8 Electron Microscopy

Samples were placed on 200-mesh carbon-coated copper grids and left to settle for 90 seconds.

Following two washes in water, grids were negative stained with 2% Uranyl acetate for 60 seconds.

TEM was carried out on a JEOL 1010 at 80 kV and images were cropped and adjusted for brightness

using Adobe Photoshop (CS6; www.adobe.com).

3.6 Supplementary Material

The supplementary material contains Supplementary Figure 3.1 (SuppFigure1.pdf) showing the

activity of thrombin in all buffer conditions, Supplementary Figure 3.2 (SuppFigure2.pdf) showing

the aggregation status for the four different constructs, Supplementary Figure 3.3

(SuppFigure3.pdf) showing the normalized final AI, and Supplementary Figure 3.4

(SuppFigure4.pdf) showing high resolution SEC assessment of buffer screen hits for IT1 and DDIT1.

In addition, there is also a Supplementary Methods Section (SuppMethods.pdf) detailing the

cloning approach and containing a table of the oligonucleotides used.

64

Supplementary methods: Assembly and cloning of epitopes into MPyV VP1

Four constructs containing either single (IT1) or three Tc-epitopes (IT1, IT2, IT3) together, with or

without DD linker were inserted into the HI loop of MPyV VP1 by in vivo homologous

recombination. Epitope sequences were assembled by annealing or polymerase cyclic assembly

(PCA), optionally followed by amplification by PCR using end primers with the appropriate

extensions for recombination. Oligonucleotides designed with optimal Escherichia coli codon usage

(Suppelementary Table 3.1) using DNAWorks (http:// helixweb.nih.gov/dnaworks/).

The constructs IT1 0010, DD IT1, and IT123 were assembled by annealing complementary

oligonucleotides encoding the entire sequence and, in the case of IT1 and DD IT1, extensions for

homologous recombination. Oligonucleotides were mixed at a final concentration of 8 µM each in

annealing buffer (10 mM Tris pH 7.5, 50 mM NaCl, 1 mM EDTA), incubated at 95°C for 5 minutes,

at 72°C for 30 minutes, and slowly brought to room temperature at 0.1° sec-1. DD IT123 was

assembled by PCA, each oligonucleotide was mixed at a final concentration of 0.2 µM and the

mixture was used as a template for Phusion polymerase (Finnzymes, Thermofisher Scientific,

www.thermofisher.com) according to the manufacturer’s instructions. IT123 and DD IT123

assembled by annealing and PCA, respectively, were PCR-amplified using end primers that

introduced homologous recombination sequences at both ends (Hr primers; Supplementary Table

3.1). Phusion polymerase was again used according to the manufacturer’s instructions with 1 µl of

the assembled epitopes as the template. Fragments were gel-purified using Pure-LinkTM Quick Gel

Extraction Kit (Life Technologies, www.lifetechnologies.com) according to manufacturer’s

instructions.

The pGEX-4T1 VP1 S1S4 (Middelberg et al., 2011) was linearized with AfeI (New England Biolabs,

MA, USA) and treated with Calf intestinal Phosphatase (CIP, Thermofisher Scientific) according to

manufacturer’s instructions. Epitope fragments were mixed with the linearized expression vector and

used to transform OmniMAX 2T-1 (Life Technologies, www.lifetechnologies.com) by heat shock.

Colonies were analysed by colony PCR and positive clones were confirmed by sanger sequencing.

3.7 Acknowledgements

This work was supported by an Australian Research Council (ARC) Discovery Early Career Research

Award (DE140101553) to FS and a Queensland Government 2010 Smart Futures Premier’s

Fellowship to APJM. The Authors would like to thank Dr Nani Wibowo for fruitful discussions at

the onset of this project.

65

Supplementary Table 3.1 Oligonucleotides used in epitope insertion.

Construct Method Oligonucleotide sequences

IT1

Annealing

Sense: 5’tggagagttacaagaagcggcattctgggctttgtgtttaccctgaaagcttatgatgtccatcac3 Antisense:’ 5’gtgatggacatcataagctttcagggtaaacacaaagcccagaatgccgcttcttgtaactctcca3’

DD IT1

Annealing

Sense: 5’tggagagttacaagaagcgatgatggcattctgggctttgtgtttaccctgaaagatgatgcttatgatgtccatcac3’ Antisense: 5’gtgatggacatcataagcatcatctttcagggtaaacacaaagcccagaatgccatcatcgcttcttgtaactctcca3’

IT123

Annealing and PCR

Sense: 5’ggcattctgggctttgtgtttaccctgaaactgctgaccgaagtggaaacctatgtgaaacgtctgattcagaacagcattaccattaaa3’ Antisense: 5’tttaatggtaatgctgttctgaatcagacgtttcacataggtttccacttcggtcagcagtttcagggtaaacacaaagcccagaatgcc3’ Hr Forward: 5’tggagagttacaagaagcggcattctgggctttgtgtt3’ Hr Reverse: 5’gtgatggacatcataagctttaatggtaatgctgttctgaatc3’

DD IT123

PCA and PCR

PCA Forward 1: 5’gatgatggcattctgggct3’ PCA Reverse 1: 5’ttccacttcggtcagcagatcatctttcagggtaaacacaaagcccagaatgccatcatc3’ PCA Forward 2: 5’tctgctgaccgaagtggaaacctatgtgaaggatgatcgtctgattcagaatagcattac3’ PCA Reverse 2: 5’tttaatggtaatgctattctgaatcagacgat3’ Hr Forward: 5’tggagagttacaagaagcgatgatggcattctgggc3’ Hr Reverse: 5’gtgatggacatcataagctttaatggtaatgctattctgaatc3’

66

Supplementary Figure 3.1 Confirmation of thrombin activity under the buffer conditions used in this study. The digestion reaction was performed on wtVP1 across a range of pH as indicated in the absence and presence of each of the additives. ND, non-digested GST-VP1 fusion.

67

Supplementary Figure 3.2 Aggregation status of each construct subject to the buffer screen. Colouring for both Z-average size (DLS) and the change in aggregation index (ΔAI) was generated for each construct individually. The range of values for each construct is given beside each heat map.

68

Supplementary Figure 3.3 Normalized aggregation index at the end of the tag removal reaction. Colouring was normalised across all constructs to allow the comparison of constructs as well as the identification of optimised buffer conditions.

69

Supplementary Figure 3.4 Assessment of screen hits by SEC. High-resolution SEC was conducted following tag removal reactions in selected buffer conditions for IT1 and DDIT1. 100 µl of reaction mixture was injected using an Agilent 1200 HPLC system and Wyatt DAWN EOS detector, aside from IT1 in Blank, which was run as described in materials and methods with a 1 ml reaction volume. Note that the accumulation of soluble aggregates of IT1 is improved relative to the standard pH 8.0 condition shown in Figure 3.2.

70

4 Encapsidation of GFP inside MPyV VLP

via fusion with VP2C

4.1 Abstract

Optimised buffer condition and insert engineering enabled recovery of soluble capsomeres displaying

a single Tc-epitope on the surface-exposed loop of VP1 protein. An alternative approach is needed to

deliver more epitopes with MPyV VLPs. Here I explored the ability of MPyV VLP to encapsidate a

foreign protein via fusion with the VP2 C-terminal region using GFP as a model protein. Preliminary

results showed that truncation in the hydrophobic binding region of VP2C-GFP hinders their binding

with wtVP1 capsomeres when mixed in vitro. To resolve this problem wtVP1 was co-expressed with

VP2C-GFP and purified with Immobilised Metal Affinity Chromatography (IMAC) via a 6x

Histidine (6His) tag at the C-terminus of VP2C-GFP. UV monitoring at A280 nm and A488 nm and

SDS-PAGE showed successful co-purification of VP2C-GFP and wtVP1, indicating successful in

vivo complex formation. Bound VP1:VP2C-GFP capsomere complex was separated from free VP2C-

GFP by SEC. The purified VP1:VP2C-GFP capsomere complex was assembly competent as

demonstrated by AF4 and TEM. I further explored if the amount of encapsidated GFP could be

controlled, and compared assembly mixtures of 100% wtVP1, 100% VP1:VP2C-GFP, and 50%

VP1:VP2C-GFP capsomere complexes mixed with 50% wtVP1 capsomeres. Absorbance analysis

showed that the amount of VP2C-GFP encapsidated at 50% concentration was half of that

encapsidated at 100% concentration suggesting that encapsidation can be controlled. This approach

is suitable, therefore, for encapsidating proteins of a foreign pathogen to deliver Tc-epitopes by

MPyV VLPs.

4.2 Introduction

Capsomere aggregation due to Tc-epitope presentation at a surface-exposed loop of MPyV VP1

necessitated the optimisation of buffer conditions (Abidin et al. 2015). Although this approach

enables recovery of soluble and assembly competent modular capsomeres, Tc-epitope delivery is

limited to a single epitope, while a Tc-based influenza vaccine conferring broad protection and

coverage may benefit from multiepitope delivery (Berthoud et al. 2011; Brown and Kelso 2009).

71

Activation of Tc-epitopes is facilitated by parent protein degradation rather than protein

conformational presentation at the surface (Yewdell et al. 2003). Thus an approach to engineer

encapsidation of foreign protein inside the MPyV VLP was explored using GFP as a model.

Encapsidation is achievable by utilising a segment of the C-terminus of VP2 protein. Research has

shown that a segment of the C-terminus of VP2 protein can bind to at least three VP1 subunits inside

a capsomere (VP1 pentamer) cavity via a hydrophobic binding site (Figure 4.1) (Barouch and

Harrison 1994; Chen et al. 1998). Fusion of GFP protein at the N-terminus of VP2 was attempted to

encapsidate GFP inside the MPyV VLP. However, this approach resulted in externalisation of the

GFP (Abbing et al. 2004; Boura et al. 2005; Pleckaityte et al. 2015).On the other hand, fusion at the

C-terminus of the C-terminal binding fragment of VP2 (VP2C) of SV40 and HPyV allowed

encapsidation of GFP inside MPyV VLPs (Inoue et al. 2008; Pleckaityte et al. 2015).

Encapsidation of foreign protein via fusion with C-terminus of MPyV VP2C has not been tested, and

I aim to apply this to the established MPyV VLP platform with E.coli as expression host at AIBN-

CBE (Chuan et al. 2008b). GFP was used as a model protein to enable monitoring and evaluation. In

vitro mixing between wtVP1 capsomeres and free VP2C-GFP failed due to truncation in the

hydrophobic binding region. To resolve this problem, wtVP1 was co-expressed with VP2C-GFP.

VP2C was varied as 23, 31, and 51 aa fragments for optimisation and VP2C-GFP 31 aa gave the

highest yield and considerable co-purification with VP1. VP1:VP2C-GFP capsomere complexes

were assembly competent, and the amount of encapsidated VP2C-eGFP were controllable by co-

assembly with wtVP1 capsomeres in a defined molar ratio.

4.3 Methods

4.3.1 Cloning of free VP2C-GFP

The pGEX-4T-1 plasmid was linearised with PCR, excluding the GST sequence, using primers in

Table 4.1. The forward primer corresponds to the downstream section of the multiple cloning site

(MCS) and the reverse primer corresponds to LacO gene site of the plasmid, immediately upstream

of the GST sequence. VP2C-GFP was amplified from a synthetic construct containing the GFP gene

with an upstream sequence encoding the 50 aa VP2C separated by a double GGGGS linker and a

downstream 6His-tag. VP2C(50)GFP-6His, and the linearized plasmid were ligated by Gibson

Assembly and the sequence was confirmed by Sanger sequencing.

72

Figure 4.1 Design of encapsidation within the MPyV VLP via VP2C fusion. A protein model of engineered VP2C fused with GFP, aiming to encapsulate GFP inside MPyV VLP. 3D structure of VLP from VIPERdb (PDB 1SID) (A). GFP was fused at the the C-terminus of VP2C (266-302 aa segment of VP2) (B). The length variation of VP2C was designed based on literature.

73

Table 4.1 Oligonucleotides used in the cloning of free VP2C-GFP.

Name Function Oligonucleotide sequence1

pGEX-empty-F Linearise pGEX

backbone 5’CTGACGATCTGCCTCGCG3’

pGEX-empty-R Linearise pGEX

backbone 5’GAATACTGTTTCCTGTGTGAAATTG3’

VP2C-(pGEX)-F Amplify VP2C-

GFP 5’ttcacacaggaaacagtattcatgAAGTTCTATCAGGCCCAGGTG3’

eGFP-(pGEX)-R Amplify VP2C-

GFP 5’cgcgaggcagatcgtcagTCAGTGATGGTGATGGTGATGC3’

1 Upper case letters correspond to sequence the homology to the target sequence and lower case letters correspond to extensions with homology to vector.

74

4.3.2 Cloning of VP1:VP2C-GFP co-expression

Fragments of VP1, the intergenic region of pETDuet-1, and VP2C-GFP-6His (three variations: 23aa,

31aa, and 50aa) were created by PCR reaction with the primers listed in Table 4.2. These three

fragments were joined to create the insert fragment via splice overlap extension (SOE). Plasmid pET-

Duet1 was cut using XhoI and NcoI to remove the entire region from the first ribosome-binding site

to the T7 terminator. The insert fragment and the linearised plasmid were joined via Gibson

Assembly, and Sanger sequencing confirmed successful cloning.

4.3.3 Expression

Plasmid of wtVP1, free VP2C-GFP, and VP1:VP2C-GFP were transformed into competent E. coli

Rosetta (DE3) pLysS cells (Novagen, www.merckmillipore.com), and expressed as previously

described (Chuan et al. 2008a) with slight modifications: 1) free VP2C-GFP was induced with 0.2

mM IPTG overnight at 20°C a; 2) VP1:VP2C-GFP was induced with 0.5 mM IPTG overnight at

26°C.

4.3.4 Affinity chromatography purification

For VP1:VP2C-GFP and free VP2C-GFP constructs, thawed cell pellets were each resuspended in

50 ml lysis buffer composed of 20 mM Tris, 500 mM NaCl, 5% glycerol, 1 mM DTT, 20 mM

imidazole and complete protease inhibitor (PI) (Roche, lifescience.roche.com) at pH 8.0 for

VP1:VP2C-GFP capsomere complex, or composed of 50 mM Na2HPO4 and 500 mM NaCl at pH

7.5 for free VP2C-GFP. Resuspended pellets were lysed by sonication and the clarified supernatant

was recovered. For IMAC purification of VP1:VP2C-GFP capsomere complex, an AKTA Pure (GE

Healthcare Lifesciences, www.gelifesciences.com) was used to equilibrate a 5mL HisTrap HP (GE

Healthcare Lifesciences) column with 1 column volume (CV, 5 ml) of MQW, 1 CV equilibrium

buffer (lysis buffer without PI). 40 ml of clarified supernatant was then passed through the column,

followed by washing with 6 CVs equilibrium buffer, and elution of bound material with 4 CVs of

elution buffer (equilibrium buffer with 500 mM imidazole). All chromatography operations were

conducted at 3 ml/min with the exception of the equilibration step, which was performed at 5 ml/min.

Solution exiting the column was monitored inline at A280 nm and A480 nm and was collected as

0.25 ml fractions using a built-in fractionator, once absorbance in the elution region showed elevation.

Optimum elution peak fractions (as analysed by SDS-PAGE) were dialysed with capsomere

compatible buffer (L-buffer) composed of 200 mM NaCl, 40 mM Tris base, 1 mM EDTA disodium,

5% Glycerol, and 5 mM DTT at pH 8 at 4°C for 16 h protected from light. If not immediately dialysed,

all protein solutions were stored at -80°C. For free VP2C-GFP, IMAC purification was performed as

75

Table 4.2 Oligonucleotides used in the cloning of VP1:VP2C-GFP co-expression vectors.

Name Amplicon Oligonucleotide sequence1

VP1-(Duet)-F VP1 5’ttaactttaagaaggagatatacATGGCCCCCAAAAGAAAAAGC3’

VP1-(adapter)-R VP1 5’atgcggccgTTAATTTCCAGGAAATACAGTCTTTGT3’

adapter-(VP1)-F

pETDuet

intergenic

region

5’ttcacacaggaaacagtattcatgAAGTTCTATCAGGCCCAGGTG3’

adapter-(VP2C23)-

R

Intergenic

region for 23aa

VP2C

5’ggtgggacatTGTATATCTCCTTCTTATACTTAACT3’

adapter-(VP2C31)-

R

Intergenic

region for 23aa

VP2C

5’tagaacttcatTGTATATCTCCTTCTTATACTTAACT3’

adapter-(VP2C50)-

R

Intergenic

region for 23aa

VP2C

5’catttcccatTGTATATCTCCTTCTTATACTTAACT3’

VP2C23-(adapter)-

F

23aa VP2C-

GFP-6His 5’gagatatacaatgTCCCACCAAAGAGTCACTCC3’

VP2C31-(adapter)-

F

31aa VP2C-

GFP-6His 5’ggagatatacaatgAAGTTCTATCAGGCCCAGGTG3’

VP2C50-(adapter)-

F

50aa VP2C-

GFP-6His 5’gagatatacaATGGGAAATGGTGGGCCTAC3’

GFPhis(Duet)-R VP2C-GFP-

6His 5’gcggtttctttaccagacTCAGTGATGGTGATGGTGATGC3’

1 Upper case letters correspond to sequence the homology to the target sequence and lower case letters correspond to extensions with homology to vector or flanking inserts.

76

reported elsewhere (Li et al. 2016) with slight modification in lysis buffer (50 mM Na2HPO4, 500

mM NaCl), equilibrium buffer (lysis buffer + 20 mM imidazole), and elution buffer (lysis buffer +

500 mM imidazole).

For wtVP1, purification was performed as previously described (Chuan et al. 2008a; Lipin et al.

2008b)


4.3.5.1 Capsomere SEC (S200)

For VP1:VP2C-GFP, 1 ml of 1-3 mg/ml dialysed fractions were then injected through a Superdex

200 GL 10/300 column (24 ml bed volume and 14.4 ml internal column volume) (GE Healthcare

Lifesciences), equilibrated with L-buffer, at 0.5 ml/min. VP1:VP2C capsomere complex eluted

between 10.8 ml to 13.8 ml, aggregates eluted between 7.5 ml to 10.3 ml, and free VP2C-GFP

between 16 ml to 20 ml.

For free VP2C-GFP, SEC was performed after a mixture experiment with VP1 capsomeres (Section

4.4.1) the same way as wtVP1 SEC as previously described (Chuan et al. 2008a; Lipin et al. 2008b).

4.3.5.2 VLP SEC (S500)

Assembled fractions (Section 4.3.7) were concentrated to >2 mg/ml and 80 µl was injected through

an in-house Sephacryl 5/200 column (4 ml bed volume and 3.6 ml internal column volum),

equilibrated with GL2 buffer (200 mM NaCl, 20 mM Tris, 5% glycerol, 1 mM CaCl2, at pH 7.4) at

0.1 ml/min with standard wt VLP peak eluting between 2.6 ml to 3.1 ml. The method will be described

in more detail in Chapter 5.

4.3.6 N-terminal sequencing

SDS-PAGE gel for an SEC run was blotted onto a PVDF membrane (Amersham GE Lifesciences)

and band to be analysed excised from the membrane and sent for N-terminal sequencing.

4.3.7 Dot blot

Protein separated by SDS-PAGE were electroblotted onto nitrocellulose (Amersham GE

Lifesciences). PBST with 5% skim milk was used as blocking solution, in-house anti-VP1 (1:4000)

was used as a primary antibody, and HRP-conjugated goat anti-mouse antibody (Sigma-Aldrich,

www.sigmaaldrich.com) as secondary antibody. Clarity Western ECL Blotting Substrate (Biorad,

www.bio-rad.com) was used to visualise the bands on the membrane, and image acquisition was

performed with Gel Doc XR System (Biorad).

77

4.3.8 Assembly

Wild-type VP1 capsomeres and VP1:VP2C-GFP capsomere complex recovered via SEC in L-buffer

at pH 8 were each assembled by dialysis with SnakeSkin Dialysis Tubing 10K MWCO

(Thermofisher, www.thermofisher.com) in standard assembly buffer (GL1) containing 0.5 M

(NH4)2SO4, 20 mM Tris-base, 5% glycerol (v/v), 1 mM CaCl2 at pH 7.4, at a concentration between

0.55 – 0.65 mg/ml for 24 hours. This was followed by dialysis in GL2 for 16 hours. A mixture

assembly experiment with both wtVP1 and VP1:VP2C-GFP capsomere complexes in 1:1 molar ratio

was also carried out in the same manner.

4.3.9 Dynamic Light Scattering

Samples were clarified by centrifugation at 5000 g for 5 min,and 25 μl was transferred in duplicate



with Dynamics software (Wyatt Technologies). DLS measures light scattering fluctuations as a result

of molecular diffusion and translates these fluctuations into diffusion coefficient via autocorrelation

analysis. The diffusion coefficient can then be converted into hydrodynamic radius via Stokes-

Einstein relationship. There are several types of autocorrelation analysis, amongst them the cumulant

and the regularisation analysis used in my research (Wyatt 2010). The cumulant analysis assumes the

estimated particle size distribution to fit the Gaussian distribution, and thus calculates the

hydrodynamic radius as the average of all particles in the solution (Koppel 1972). Regularisation

analysis, on the other hand, does not assume any fitted particle size distribution, and thus estimates

the radii and relative abundance of all species in the solution (Chu 2013; Provencher 1979).

4.3.10 Electron Microscopy





4.3.11 Asymmetrical Flow Field Flow Fractionation

Asymmetrical Flow Field Flow Fractionation (AF4) was done as previously described (Chuan et al.

2008a). Briefly, concentration of samples was adjusted to 1.2 mg/ml prior to analysis. Clarified

supernatant (0.5 ml) was injected for each fractionation with an Eclipse 2 AF4 system (Wyatt

Technology Corporation). AF4 of VLP samples was performed in GL2. The liquid flow rate out of

78

the channel was maintained at 0.75 mL min1 for all AF4 operations. All buffers used were degassed

and filtered through a 0.1 mm inline-filter (Millipore, www.merckmilipore.com).

4.4 Results

Encapsidation of GFP inside the MPyV VLP was attempted by fusing with VP2C, which naturally

binds to VP1 capsomeres. Fusion at the C-terminus of VP2C is expected to position GFP inside the

cavity of the VLP (Inoue et al. 2008; Pleckaityte et al. 2015). This attempt was done by two

approaches, the first is to bind VP2C-GFP and VP1 capsomeres in vitro, and the second is to facilitate

binding of VP2C-GFP and VP1 capsomeres in vivo.

4.4.1 VP2C-GFP failed to bind to VP1 when mixed in vitro

A preliminary experiment showed that purification of free VP2C-GFP with a 6His-tag was successful

under optimised condition (Kartikawijaya, 2014). However, SEC indicated that VP2C-GFP was not

able to bind to VP1 when it was mixed with VP1 in standard buffer condition. If VP1 and VP2C-GFP

bind, SDS-PAGE would show bands in the same lane on SDS-PAGE, and both A280 nm (protein

monitoring) and A488 nm (GFP-specific monitoring) signals on the same peak in a chromatogram.

However, both SDS-PAGE and the chromatogram showed that VP1 and VP2C-GFP appeared in

different lanes and peaks respectively (Figure 4.2). Analysis by N-terminal sequencing showed that

the truncation was located exactly at the hydrophobic site of VP2C corresponding to the binding site

to VP1 capsomeres (Barouch and Harrison 1994; Chen et al. 1998) (Figure 4.2). Truncation at this

hydrophobic site renders the recombinant VP2C-GFP unable to bind to VP1 capsomeres.

4.4.2 Design of VP1 and VP2C-GFP co-expression

Truncation of VP2C hydrophobic binding site may be caused by proteolysis during expression or cell

lysis, as proteases target protein hydrophobic regions for protein degradation (Fredrickson et al.

2013). VP1 capsomeres and VP2 naturally bind together in infected host cells during VLP formation,

and imitation of this processs in expression host by recombinant co-expression may help to avoid

proteolysis effect. High affinity between VP1 capsomeres and VP2C (Abbing et al. 2004; Barouch

and Harrison 1994) allows binding to capsomeres in vivo, before proteolysis of VP2C, as the site

would be enclosed within capsomere cavity.

Co-expression of VP1 and VP2C-GFP was designed by cloning them into a pET-Duet1 plasmid

(Figure 4.3A). Three variations of VP2C, 23 aa, 31 aa, and 50 aa, were designed based on a

comparison between VP2 amino acid sequences of various polyomaviruses from the literature

(Barouch and Harrison 1994; Chen et al. 1998). The comparison revealed an overlapping and

conserved hydrophobic region at the C-terminal region of VP2 (VP2C), all or part of which is a

79

Figure 4.2 VP2C-eGFP expressed separately from VP1 failed to bind when mixed in vitro. SEC profile (A) showed that free VP2C-eGFP was purified separately from VP1 capsomeres after being mixed in standard condition. SDS-PAGE (B) with lanes corresponding to peak regions as indicated, confirms separate elution, and indicated truncation in VP2C-GFP. M is molecular weight standard. Arrows point to bands as indicated. N-terminal sequencing (C) showed that truncation of VP2C-eGFP was located at the hydrophobic binding site. Arrow point to truncation site.

80

Figure 4.3 Design of VP2C-eGFP and VP1 co-expression. Design of pET-Duet1 plasmid (A) to facilitate co-expression of VP2C-eGFP and VP1 capsomere. Detailed construct design for co-expression (B). A 6His-tag (yellow) was placed at C-terminus of VP2C-eGFP. VP2C was engineered in three variations, 23 aa, 31 aa, and 50 aa. SDS-PAGE gel (C) of solubility testing of co-expressed constructs with arrows pointing to respective bands.

81

putative binding region to VP1 capsomeres. The constructs were designed to determine the shortest

possible adapter which still allows encapsidation of a foreign protein. The 50 aa VP2C has been

shown to permit this in plants (Catrice and Sainsbury 2015). GFP was fused at the C-terminus of

VP2C and tagged with 6His. Both VP1 and VP2C-GFP were each preceded by a T7 promoter

upstream, and a non-coding region was placed between the VP1 and VP2C-GFP coding regions. The

non-coding region was designed to be out of frame between VP1 and VP2C-GFP (Figure 4.3 B).

4.4.3 Expression and solubility test

The constructs were expressed in E.coli and lysis was performed in the presence of protease inhibitor.

SDS-PAGE showed that fused protein of 50 aa VP2C construct had the highest total and soluble

protein yield (Figure 4.3 C), which however was truncated based on the size of the bands, and thus

was not pursued further. Soluble proteins of the 31 aa VP2C construct are both truncated and

untruncated, but the latter was less soluble. This showed that the nature of untruncated protein is

mostly insoluble, though this suggests that its presence in the soluble fraction may indicate complex

formation. The 23 aa VP2C construct did not produce sufficient VP2C protein, even though VP1

production was good, similar to other constructs.

4.4.4 IMAC indicated successful binding between VP2C-GFP and VP1

A 6His-tag was designed at the C-terminus of GFP which was fused to VP2C. Therefore, purification

with an IMAC column would selectively purify VP2C-GFP which has a 6His-tag in it. The purified

VP2C-GFP would come in two forms: free VP2C-GFP; and VP2C-GFP that bound to co-expressed

VP1 capsomeres. Only VP1 that bound to VP2C-GFP would be co-purified, while free VP1 would

not be co-purified and would be washed away from the column.

The IMAC was run with two different wavelength detectors, which were A280 nm specific for

proteins, and A488 nm specific for GFP. The chromatogram (Figure 4.4) showed a peak in both

A280 nm and A488 nm in the purified eluted region and construct 31 aa VP2C had the highest yield.

Construct 23 aa VP2C had a low yield and thus was not examined further (Figure 4.5). SDS-PAGE

analysis showed that fractions of peak regions contain both bands of VP1 and VP2C-GFP and

thickness of bands corresponded to height of peak in the chromatogram. The presence of both bands

of VP1 and VP2C-GFP in the same fractions confirmed that VP2C-GFP was able to bind to VP1.

The thicker band of VP2C-GFP indicated the co-purification of free VP2C-GFP that needed to be

separated further by SEC. To confirm binding of VP1 to VP2C-GFP a dot blot was further carried

82

Figure 4.4 Purification of VP1:VP2C-eGFP 31 aa. Chromatogram of the IMAC purification (A) of the VP1:VP2C-eGFP 31 aa complex (zoomed in B) and SDS-PAGE with lanes corresponding to peak regions as indicated (C), showed that VP1 was co-purified with VP2C-GFP as confirmed with anti-VP1 dot blot. VP1:VP2C-eGFP 31 aa had the highest yield between the three tested constructs. Absorbance at 280 nm is shown in blue and absorbance at 488 nm is shown in green. M is molecular weight standard. Arrows point to bands as indicated. Elution fractions 31*aa and 31 aa as indicated were further analysed with SEC.

83

Figure 4.5 Purification of VP1:VP2C-eGFP 23 aa. Chromatogram of the IMAC purification (A) of the VP1:VP2C-eGFP 23 aa complex (zoomed in B) and SDS-PAGE with lanes corresponding to peak region as indicated (C), showed that VP1 was co-purified with VP2C-eGFP as confirmed with anti-VP1 dot blot. M is molecular weight standard. Absorbance at 280 nm is shown in blue and absorbance at 488 nm is shown in green. Arrows point to bands as indicated.

84

out with an anti-VP1 primary antibody, and confirmed that the fractions contained VP1. As only

bound VP1 could be co-purified with VP2C-GFP, all the VP1 contained in the fractions are most

likely bound to VP2C-GFP. A 70 kDa band was also present consistently in the elution fractions of

all constructs.

4.4.5 Isolation of VP1:VP2C-GFP interaction from free VP2C-GFP by

size exclusion chromatography

To separate VP2C-GFP bound to VP1 from free VP2C-GFP, SEC was performed on the collected

IMAC elution peak fractions, preceded by dialysis in capsomere compatible L-buffer. Concerns about

the 70 kDa band to interfere with assembly, led to the division of the fractions into two groups. The

first group were IMAC fractions that contained less 70 kDa bands (three before last SDS-PAGE

fractions for construct 31 aa (Figure 4.4)). The second group (denoted with a star) were IMAC

fractions that contained not only more 70 kDa bands, but also more of the desired bands of VP1 and

VP2C-GFP (third to sixth SDS-PAGE fractions for construct 31 aa (Figure 4.4)). Fractions of

respective groups were mixed, dialysed, and run on SEC together.

SEC of IMAC fractions of 31 aa (not starred) resulted in capsomere complex peak (elution: ~10 ml)

and free VP2C-GFP peak (elution: ~ 16 ml) (data not shown). Capsomere complex peak had a small

elevation in the A488 nm signal and contained VP1 as confirmed by SDS-PAGE, while a VP2C-GFP

band could not be identified. On the contrary, free VP2C-GFP peak had a high A488 nm signal, thick

VP2C-GFP bands, and no VP1 band. SEC of IMAC fractions of 31* aa resulted in capsomere

complex peak (elution: ~ 10 ml) and free VP2C-GFP peak (elution: ~16 ml) (Figure 4.6). Although

the A488 nm signal on capsomere complex peak was low in comparison to A280 nm signal, this peak

not only contained VP1, but also contained visible VP2C-GFP band as confirmed by SDS-PAGE.

Free VP2C-GFP peak for fractions 31* aa also had a high A488 nm signal, thick VP2C-GFP bands,

and no VP1 band. Two overlapping bands at the VP2C-GFP region were even observed in free VP2C-

GFP peak, indicating that some of this protein was truncated as expected of free VP2C-GFP. The

~10 ml elution peak in both groups of fractions for both constructs are located near the elution volume

known to be the capsomere region in wtVP1 (Abidin et al. 2015). Based on the chromatogram and

SDS-PAGE, this peak is the desired peak that contains a VP1 capsomere bound to VP2C-GFP.

4.4.6 Western blot to analyse purified SEC fractions

To confirm the identity of expected bands, a western blot was run on the SEC fractions of the 31 aa

construct in the capsomere complex peak region (~10 ml) (Figure 4.6C). This showed anti-VP1

stained bands in the region where VP1 was known to be located and GFP at approximately 30 kDa.

85

Figure 4.6 Purification of VP1:VP2C-GFP complex. Isolation of capsomeres by SEC (A) enabled separation of VP1:VP2C-eGFP complex (zoomed in B) from free VP2C-eGFP. SDS-PAGE with lanes corresponding to peak region, and western blot with anti-GFP and anti-VP1 antibodies on peak fraction marked with an asterisk identified VP1 and GFP bands (C). M is molecular weight standard and L is column load. Absorbance at 280 nm is shown in blue and absorbance at 488 nm is shown in green.

86

Both proteins also appeared as dimers. GFP is known to be a dimer (Malikova et al. 2011) and a

double band appears at approximately 58 – 60 kDa. The dimer could also be due to interaction via

the hydrophobic VP2C motifs as it persists under denaturing conditions. There was also an anti-

VP1stained band in the 90 kDa region, probably incompletely denatured VP1 in the form of dimers,

which sometimes occurs on SDS-PAGE (Catrice and Sainsbury, 2015). The 70 kDa band was not

identified by either antibody, indicating that this is a co-purified host protein contaminant.

4.4.7 VP1:VP2C-GFP was able to assemble into VLP

S200 fractions collected in previous step containing VP1:VP2C-GFP capsomere complex from the

31* aa construct was assembled in assembly buffer. The complete formation of VLPs from

VP1:VP2C-GFP capsomere complexes was observed by both AF4 and TEM (Figure 4.7 A and B),

which showed that particles were externally identical. This indicated that the 70 kDa band did not

interfere with assembly. AF4 chromatogram of VP1:VP2C-GFP assembled product indicated the

presence and separation of a smaller particle in the mixture, possibly unassembled or disassembled

capsomeres as previously observed elsewhere (Chuan et al. 2008a). When the assembled product

was run on S500, a 70 kDa band did not appear in SYPRO Orange-stained SDS-PAGE gel (data not

shown).

4.4.8 Controlled GFP encapsidation was achievable by co-assembly

with wt capsomeres

To assess whether the amount of encapsidated protein could be controlled, three assembly

experiments were carried out: 1) assembly of 100% wt capsomeres; 2) assembly of 100% VP1:VP2C-

GFP capsomere complex; 3) assembly of a mixture of wt capsomere and VP1:VP2C-GFP capsomere

complex in 1:1 molar ratio (Figure 4.8). The assembled product was then run through a sephacryl-

based preparative S500 to enable separation and analysis of the VLPs developed as described in

Chapter 5. The chromatogram of assembled 100% wt capsomeres showed complete VLP formation,

as confirmed by DLS, with VLP elution peak at 2.8 mL (Figure 4.8A). Chromatogram of assembled

100% capsomere complex showed an aggregate peak at 1.9 ml elution volume and VLP peak at 2.6 ml

elution volume at 280 nm absorbance. An A488 nm absorbance corresponded to this curve displaying

the same pattern (Figure 4.8B). Chromatogram of co-assembled VP1:VP2C-GFP capsomere

complex with wt capsomeres in a 1:1 molar ratio showed an aggregate peak at 1.9 ml elution volume

and VLP peak at 2.1 ml elution volume at A280 nm absorbance. Again, an A488 nm absorbance also

corresponded to this curve displaying the same pattern (Figure 4.8C). The A488 nm absorbance

was also lower in 50% VP1:VP2C-GFP co- assembly than the 100% VP1:VP2C-GFP assembly.

Absorbance analysis of collected SEC fractions showed that 100% VP1:VP2C-GFP assembly

87

Figure 4.7 VLP formation after assembly as confirmed by AF4 and TEM. Assembly of VP1:VP2C-eGFP complex SEC fractions in standard buffer conditions resulted in VLPs comparable to WT VLPs as confirmed by AF4 (A) and TEM (B). TEM scale bar is 300k.

88

Figure 4.8 Controlled encapsidation. SEC chromatogram and TEM of 100% wt capsomere assembly reaction as a standard (A) were compared to that of 100% VP1:VP2C-eGFP (B) and an assembly mixture of wt and VP1:VP2C-eGFP capsomeres in 1:1 molar ratio (C). VLP peaks were observed in (B) and (C) even though aggregate peaks were also observed. Preparative SEC VLP peak fractions were observed under TEM, and VLP formation was observed. Absorbance at 488 nm in (C) was lower than in (B). Absorbance at 280 nm is shown in blue and absorbance at 488 nm is shown in green. TEM scale bar is 200k.

89

encapsidated 40 molecules of GFP, which indicated partial VP2C-GFP dissociation during assembly

reaction. Absorbance analysis of collected SEC fractions showed that 50% VP1:VP2C-GFP co-

assembly encapsidated 20 molecules of GFP. This indicated that the amount of foreign protein

encapsidation may be controlled by co-assembly with wtVP1 capsomeres based on a molar ratio.

Resolution between aggregate peak and VLP peak is not ideal due to limitation in the S500 method.

However, the chromatogram showed that particles of different sizes eluted at different times forming

distinct, even though not fully separated, peaks. Although tailing of the aggregate peak within the

VLP region is possible, DLS regularization analysis still measured the radius of the particles as 29

nm for wt VLPs, 30 nm for 100% VP1:VP2C-GFP VLPs, and 35 nm for 50% VP1:VP2C-GFP VLPs.

The measured VLP sizes are still within the range of reported VLP particle size (Abbing et al. 2004;

Chuan et al. 2008a) and our validation of the DLS analysis.

4.5 Discussion

In this chapter, the approach to encapsidate foreign antigens inside the MPyV VLP was developed

and optimised. The development and optimisation included molecular design and bioprocess

modification to enable expression and recovery of VP1:VP2C-GFP capsomere complexes.

Furthermore, the in vitro assembly competency of the complexes and the ability to control

encapsidation of foreign molecules was evaluated.

The first approach to mix free VP2C-GFP with wtVP1 in vitro to form a complex did not work due

to truncation in the hydrophobic binding region of VP2C-GFP. Elimination of N-terminal region of

VP2C in this project may have played a role in sterically exposing the hydrophobic binding region to

proteolysis. This phenomenon was not observed when Abbing et al. (2004) mixed N-terminal region

intact VP2-GFP with wtVP1, which resulted in successful VLP formation in vitro.

Co-expression of VP2C-GFP and wtVP1 circumvented this problem probably due to the high affinity

between those molecules. Rapid capsomere complex formation in vivo may have enclosed the

hydrophobic binding site of VP2C inside the capsomere cavity and may have further protected it from

proteolysis. VP1:VP2C-GFP capsomere complex formation was confirmed by IMAC, SEC, and their

respective SDS-PAGE analysis. Construct 31 aa VP2C performed the best and was thus pursued

further. Even though the VP2C-GFP band was very thick on IMAC SDS-PAGE gel, it was thinner

than VP1 on the SDS-PAGE gel of capsomere complex peak of SEC fractions. This is because VP2C-

GFP is in a theoretical 1:5 molar ratio to VP1 in its bound form. Empirical calculation with absorbance

data from UNICORN software results in a 1:7 molar ratio which is quite close to the theoretical ratio.

90

The difference between empirical ratio and theoretical ratio may be due to some part of the VP2C-

GFP being buried inside the capsomere cavity or partial unfolding of GFP. Correctly folded, stable

GFP, is required for absorbance at A488 nm. SEC has also enabled separation of bound VP2C-GFP

from free VP2C-GFP that eluted later in free VP2C-GFP peak based on size difference. SDS-PAGE

showed that some of free VP2C-GFP was truncated. Bands of truncated VP2C-GFP were not present

in capsomere complex peak which indicated that only untruncated VP2C-GFP could bind to VP1.

SDS-PAGE analysis of SEC fractions also showed co-purification of a 70 kDa band, co-stained with

anti-GFP after subsequent western blotting, indicating association with VP2C-GFP. The complex

involving VP2C-GFP in the 70 kDa band may contain an interacting chaperone as have been indicated

in the literature (Liew et al. 2010). This species appears in two sizes in western blot, the smaller of

which may be the truncated form of VP2C-GFP complex. Even though the 70 kDa band was further

shown not to interfere with assembly, the nature of the 70 kDa protein needs to be investigated further,

for example by mass spectrometry.

VP1:VP2C-GFP capsomere complex was recoverable in the bioprocessing methodology developed

in this project. The recovered complexes were also assembly competent in vitro and yielded externally

identical VLPs. This methodology differs with what has been demonstrated in the litereature in

several ways. First is the use of a minimal 31 aa VP2C fragment with truncated N-terminal region

and fusion at its C-terminal arm. Second is the use of co-expression in a prokaryotic host, which led

to in vivo formation of VP1:VP2C-GFP capsomere complex rather than in vivo formation of VLPs as

reported by Boura et al. (2005) and Inoue et al. (2008) in eukaryotic host. Recoverable VP1:VP2C-

GFP capsomere complex is advantageous as it eliminates the risk of genetic material inclusion. Third

is the ability to assemble capsomere complexes in vitro, that not only avoids genetic material

inclusion but also allows precise control over VLP modifications as explained in the next paragraph.

This differs from in vitro assembly reported by Abbing et al. (2004) because they employed free VP2-

GFP and wtVP1 rather than capsomere complexes in their assembly reaction. In this project, the

number of encapsidated molecules is slightly less than previously reported 64 encapsidated GFP

molecules (Abbing et al. 2004; Schmidt et al. 2000). The dissociated free VP2C-GFP might play a

significant role in aggregate formation. And more likely, the aggregate peak in the SEC analysis may

be caused by sample concentration prior to chromatography application, as an AF4 analysis of 100%

VP1:VP2C-GFP assembled product without sample concentration demonstrated low aggregate

content (Figure 4.5). AF4 analysis also showed the assembly reaction of 100% VP1:VP2C-GFP did

not result in 100% percent assembly unlike wt VP1 assembly, as free capsomeres were observed in

the AF4 profile. These impurities could be removed with SEC500 VLP purification.

91

An approach to control the amount of encapsidated GFP was also pursued by mixing VP1:VP2C-

GFP capsomere complex with wtVP1 capsomere. Spectrophotometric analysis showed that the

amount of encapsidated GFP was proportional to the ratio of co-assembled VP1:VP2C-GFP

capsomere complex and wtVP1 capsomere. This does not only show that encapsidation was

controllable, but also strongly indicates the ability to engineer capsomere composition of a mosaic

VLP.

4.6 Conclusion

In this chapter, an in vivo interaction between wtVP1 capsomeres and a VP2C-GFP was established

via molecularly designed and adjustable co-expression. Further expression and purification

elucidated the optimum reduced VP2C fragment able to facilitate encapsidation of a foreign protein.

In addition, prokaryote-based upstream and chromatography-based downstream bioprocess

development enabled purification of the VP1:VP2C-GFP capsomere complex. . These VP1:VP2C-

GFP capsomere complexes were assembly competent in vitro, and the amount of encapsidated protein

inside the MPyV VLPs appears to be controllable via a mixed assembly reaction, reported here for

polyomavirus VLPs for the first time. The encapsidation methodology may then be applied to

encapsidate foreign antigenic proteins to deliver Tc-epitope for vaccine development, and opens up

many other possibilities for biomaterial advancement.

92

5 Cytotoxic T Cell Epitope-Based MPyV

VLP Influenza Vaccine Development,

Analysis, and Characterisation

5.1 Abstract

Influenza pandemic and epidemic, high mutation frequency of the virus, and problems related to

vaccine production capacity and cost necessitate the development of next-generation vaccine

candidates that are not only broadly cross-protective, but also efficient in terms of cost and production

time. One of the ways to explore the solution to this problem is to develop influenza Tc-epitope-based

vaccines with the MPyV VLP as a carrier. In this chapter, MPyV VLP vaccines carrying influenza

Tc-epitopes either externally or internally were developed. An MPyV VLP carrying Tc-epitopes at a

surface-exposed loop of VP1 subunit protein was further developed by optimising assembly by co-

assembling modified DDIT1 capsomeres with unmodified capsomeres to form stable mosaic VLPs.

Furthermore, an MPyV VLP carrying Tc-epitopes internally was developed by encapsidating soluble

subdomains of the M1 protein. To analyse and characterise the modified VLPs, a new semi-

preparative SEC method was developed and validated with wt VLPs. Capsomeres bound to M1-I

subdomains (VP1:VP2C-M1-I) were successfully expressed and purified as indicated by SDS-PAGE

and western blot. VLPs of VP1:VP2C-M1-I and mosaic DDIT were also successfully assembled and

purified with the newly developed semi-preparative SEC method. Analysis and characterisation of

elution fractions with DLS, SDS-PAGE, and TEM showed successful formation of the desired VLPs,

both DDIT1 mosaic VLPs and M1-I encapsidating VLPs.

5.2 Introduction

So far in this research, development of a Tc-epitope-based influenza vaccine with MPyV VLP has

been explored by two approaches: external presentation (Chapter 3) and internal encapsidation

(Chapter 4). External presentation was achieved by insert engineering of DDIT1 at a VP1 surface-

exposed loop and devising a buffer screen to enable soluble and assembly competent capsomere

recovery (Abidin et al. 2015). Even though the DDIT1 capsomeres were able to form VLPs, the

93

assembly reaction still required optimisation, for example by reducing the amount of hydrophobic

epitopes on the surface of the VLPs to minimise aggregation and increase assembly efficiency. This

may possibly be achieved by co-assembling DDIT1 capsomeres with unmodified VP1 to form more

stable mosaic VLPs. This approach to co-assemble modified and unmodified subunit proteins of a

viral capsid has been demonstrated before with Qβ and CCMV particles resulting in successful

formation of mosaic VLPs (Brown et al. 2009; Minten et al. 2010; Pokorski et al. 2011). Successful

GFP encapsidation by MPyV VLP that has been demonstrated and optimised in Chapter 4 could

then be applied to vaccine development by replacing GFP with an influenza protein. In this chapter

(Chapter 5), the M1 protein was selected for several reasons: 1) it contains epitopes IT1 and IT2

used for external presentation in Chapter 3, and thus immunogenicity between externally and

internally carried epitopes may be compared; 2) IT1 is also commonly used in the literature for

influenza vaccine development because of its immunodominance (Berthoud et al. 2011; Deliyannis

et al. 2006); 3) M1 is an internal influenza protein with a high degree of conservancy; and 4) the size

of the soluble domain is relatively small so it would not overpopulate the VLP cavity.

One important aspect in the development of vaccines by modifying MPyV VLPs is a method to

analyse and characterise the modular particles to confirm the successful formation of desired

particles. The current method for MPyV VLPs at AIBN-CBE employs asymmetric flow-field flow

fractionation (AF4) which serves to measure particle size and distribution (Chuan et al. 2008a).

However, this method is only analytical and does not provide preparative samples required for SDS-

PAGE and TEM. SDS-PAGE is important to show the presence of engineered subunit components

of the modified VLPs, and TEM is important to visualise the formation of intact VLPs. Therefore, a

new semi-preparative method for analysis and characterisation with the following desired traits

needed to be developed: 1) able to isolate different assembly reaction species; 2) efficient in terms of

time; 3) efficient in terms of sample consumption; and 4) allows fraction collection to enable further

analysis and characterisation (for example for SDS-PAGE and TEM); and 4) cost effective.

Preparative protein purification may be achieved with SEC. This technique has been widely used for

several purposes including buffer exchange; separation and purification based on particle size; MWD

determination; automated analysis with different detectors such as UV, light scattering, and

conductivity; manual analysis such as SDS-PAGE and TEM; and large scale production of purified

proteins (Hagel and Janson 1992) . SEC medium works as a molecular sieve to separate particles

based on their size. Larger particles are less exposed to pores in the media in comparison to smaller

particles, and thus elute earlier than smaller particles (Hagel 2011). Various media are available

commercially to separate particles of different sizes, such as Superose, Superdex, and Sephacryl. The

medium suitable to separate MPyV VLP is Sephacryl S500 HR as its fractionation range for proteins

94

is 10 – 105 kDa while MPyV VLP size is ~16 x 103 kDa (GE-Healthcare 2010b; Rayment et al. 1982).

Commercially available prepacked S500 columns are designed for production of a large volume of

samples, and not for analysis and characterisation of small sample volumes. Therefore, a new semi-

preparative SEC method to analyse and characterise VLPs needed to be developed.

In this last part of my research, an SEC-based method to analyse and characterise modified MPyV

VLPs was developed by assessing the column efficiency of an in-house packed Sephacryl S500

column, reflected in their h and As. After the optimum flow rate range was determined, the column

was validated with wt MPyV VLPs in combination with DLS, SDS-PAGE, and TEM. The column

was first used to confirm the assembly of MPyV VLPs encapsidating GFP, as described in Chapter

4. The method was then applied to analyse and characterise modified MPyV VLP based vaccines

explained as follows. MPyV VLP encapsidating the M1-I soluble subdomain of influenza was

developed according to the previously optimised and adjusted method (Chapter 4). Meanwhile,

capsomeres with externally presented DDIT1 were co-assembled in optimised assembly buffer to

form mosaic particles. The successful formation of MPyV VLPs which carry antigens either

externally or internally was confirmed using the semi-preparative SEC method.

5.3 Materials and methods

5.3.1 Column packing

Packing of a Tricorn 5/200 Empty High Performance column (GE Healthcare,

www.gelifesciences.com) was done according to manufacturer’s instructions. Sephacryl S500 High

Resolution medium (GE Healthcare Lifesciences) was poured into the empty column in one

continuous motion avoiding the formation of bubbles. The column was then packed at 2.3 ml/min for

an hour, and then packed at 1 ml/min for two hours until the bed height was constant.

5.3.2 Column efficiency testing

Column efficiency was tested by injecting 2% acetone (8 µl) into the column, run at various flow

rates in MQW (GE Healthcare Lifesciences). N (Equation 2.1), HETP (Equation 2.2), h (Equation

2.3) and As (Equation 2.4) were measured, where dp was 50 µm and L was 20 cm. Values of h and

As were plotted against different flow rates to determine the optimum flow rate for the column.

5.3.3 Cloning of VP1:VP2C-M1-I and VP1:VP2C-M1-II

95

Cloning of VP1:VP2C-M1-I and VP1:VP2C-M1-II constructs was done by replacing GFP in

VP1:VP2C-GFP (Figure 4.3) with either M1-I or M1-II (Figure 5.1) using Gibson Assembly as

described in Chapter 4. The pET-Duet construct containing VP2C(31)GFP was linearised by PCR,

excluding the GFP sequence, and the M1 domains were amplified using the primers listed in table

5.1.

5.3.4 Expression

pET-Duet 1 plasmid containing either VP1:VP2C-M1-I or VP1:VP2C-M1-II constructs were

transformed into competent E. coli Rosetta (DE3) pLysS cells (Novagen, www.merckmillipore.com),

and expressed as previously described in Chapter 4 with a 0.5 mM IPTG induction.

5.3.5 Affinity chromatography purification

For VP1:VP2C-M1-I and VP1:VP2C-M1-II, thawed cell pellets were each resuspendend in 50 ml

lysis buffer composed of 20 mM Tris, 500 mM NaCl, 5% glycerol, 1 mM DTT, 20 mM imidazole

and complete of protease inhibitors (PI) (Roche, www.lifescience.roche.com) at pH 8.0.

Resuspended pellets were lysed by sonication and clarified supernatant was recovered. For IMAC

purification, an AKTA Pure (GE Healthcare Lifesciences) was used to equilibrate a 5mL HisTrap HP

(GE Healthcare Lifesciences) column with 1 column volume (CV, 4 ml) of MQW, 1 CV equilibrium

buffer (lysis buffer without PI). 40 ml of clarified supernatant was then passed through the column,

followed by washing with 6 CVs equlibrium buffer, and elution of bound material with 4 CVs of

elution buffer (equilibrium buffer with 500 mM imidazole). All chromatography operations were

conducted at 3 ml/min with the exception of the equilibration step, which was conducted at 5 ml/min.

Solution exiting the column was monitored inline at 280 nm and was collected as 0.25 ml fractions

using a built-in fractionator, once absorbance in the elution region showed elevation. Optimum

elution peak fractions (as analysed by SDS-PAGE) were dialysed with capsomere compatible buffer

(L-buffer) composed of 200 mM NaCl, 40 mM Tris base, 1 mM EDTA disodium, 5% Glycerol, and

5 mM DTT at pH 8 at 4 °C for 16 h. If not immediately dialysed, all protein solutions were stored at

-80 °C.

For wtVP1, purification was performed as previously described (Chuan et al. 2008a; Lipin et al.

2008b). For DDIT1 capsomere, purification was performed as previously described (Abidin et al.

2015).


5.3.6.1 Capsomere SEC (S200)

96

Table 5.1 Oligonucleotides used in the cloning of VP1:VP2C-M1-I and VP1:VP2C-M1-II.

Name Function Oligonucleotide sequence1

His-Tag-F Linearise pETDuet-

VP1 backbone 5’CATCACCATCACCATCACTGAG3’

G4S-R Linearise pETDuet-

VP1 backbone 5’AGAACCGCCACCTCCAGAAC3’

M1-(VP2C)-F2 Amplify M1 5’tctggaggtggcggttctAGTCTTCTAACCGAGGTCGA3’

M1b1-(His)-R Amplify M1-I 5’gtgatggtgatggtgatgTCCTCGCTCACTGGGCA3’

M1b2-(His)-R Amplify M1-II 5’gtgatggtgatggtgatgATGCTGGGAGTCAGCAATCT3’

1 Upper case letters correspond to sequence the homology to the target sequence and lower case letters

correspond to extensions with homology to vector or flanking inserts.

97

Figure 5.1 VP2C-M1 constructs. Two constructs were designed for encapsidation of M1 soluble subdomains inside MPyV VLP. The first is VP1:VP2C-M1-I containing a shorter subdomain (light blue) (A) and VP1:VP2C-M1-II containing an additional subdomain (dark blue) (B). The influenza proteins were fused to C-terminus of VP2C 31 aa (grey and red regions) and were tagged with 6His (yellow) at their C-terminus.

98

For VP1:VP2C-M1-I and VP1:VP2C-M1-II, 1 ml of 1-3 mg/ml dialysed fractions were then injected

through a Superdex 200 GL 10/300 column (24 ml bed volume and 14.4 ml internal column volume)

(GE Healthcare Lifesciences) equilibrated with L-buffer at 0.5 ml/min with AKTA Explorer 100 (GE

Healthcare Lifesciences)

S200 for wtVP1 capsomeres was performed in L-buffer as previously described (Chuan et al. 2008a;

Lipin et al. 2008b), while S200 for DDIT1 capsomeres was run with the addition of 50 mM L-Arg in

L-buffer as previously described (Abidin et al. 2015).

5.3.6.2 VLP SEC (S500)

Assembled fractions (Section 5.3.8) in a concentration of >2 mg/ml and 80 µl were injected through

an in-house Sephacryl S500 5/200 column (4 ml bed volume and 3.6 ml internal column volume)

equilibrated with GL2 buffer (200 mM NaCl, 20 mM Tris, 5% glycerol, 1 mM CaCl2, at pH 7.4) for

VP1:VP2C-M1-I, and with GL2 with 50 mM L-Arg for DDIT1 mosaic particles, at 0.1 ml/min.

5.3.7 Western blot

Proteins separated by SDS-PAGE were electroblotted onto nitrocellulose (Amersham GE

Lifesciences). PBST with 5% skim milk was used as blocking solution, in-house anti-VP1 (1:4000)

and anti-6His (Thermofisher, www.thermofisher.com) were used as primary antibodies, and HRP-

conjugated goat anti-mouse (Sigma-Aldrich, www.sigmaaldrich.com) as secondary antibody. Clarity

Western ECL Blotting Substrate (Bio-Rad, www.bio-rad.com) was used to visualise the bands on the

membrane, and image acquisition was performed with Gel Doc XR System (Bio-Rad).

5.3.8 Assembly

5.3.8.1 Standard wtVP1 assembly

Capsomeres of wtVP1 were assembled as described elsewhere (Abidin et al. 2015).

5.3.8.2 VP1:VP2C-M1-I assembly

VP1:VP2C-M1-I capsomeres recovered via SEC in L-buffer at pH 8 were each assembled by dialysis

in standard assembly buffer (GL1) containing 0.5 M (NH4)2SO4, 20 mM Tris-base, 5% glycerol (v/v),

1 mM CaCl2 at pH 7.4 for 24 hours. This was followed by dialysis in GL2 for 16 hours.

5.3.8.3 Mosaic DDIT1 assembly

99

A mosaic assembly experiment by mixing DDIT1 and wtVP1 capsomeres, at a 1:4 molar ratio was

also carried out by dialysis as in Section 5.3.7.2 in modified buffer conditions by adding 50 mM L-

Arg in the assembly buffers (GL1 and GL2).

5.3.9 Dynamic light scattering

Samples were clarified by centrifugation at 5000 g for 5 min and 25 μl was transferred in duplicates



with Dynamics software (Wyatt Technologies). DLS measures light scattering fluctuations as a result

of molecular diffusion and translates these fluctuations into diffusion coefficient via autocorrelation

analysis. The diffusion coefficient can then be converted into hydrodynamic radius via Stokes-

Einstein relationship. There are several types of autocorrelation analysis, amongst them the cumulant

and the regularisation analysis used in my research (Wyatt 2010). The cumulant analysis assumes the

estimated particle size distribution to fit the Gaussian distribution, and thus calculates the

hydrodynamic radius as the average of all particles in the solution (Koppel 1972). Regularisation

analysis on the other hand does not assume any fitted particle size distribution, and thus estimates the

radii and relative abundance of all species in the solution (Chu 2013; Provencher 1979).

5.3.10 Electron microscopy





5.4 Results

5.4.1 Optimisation of a new S500 column

An in-house Sephacryl-based S500 5/200 column was packed and optimised to develop semi-

preparative SEC method for VLP samples to enable their further analysis and characterisation. The

column efficiency was tested based on h and As (Table 5.2), within a flow rate range from

0.05 ml/min to 0.15 ml/min in 0.025 ml/min intervals in a total of five measurements. A good column

efficiency is indicated by h<3 (particle diameter = 50 µm) and As between 0.8 and 1.3 for Sephacryl

(GE-Healthcare 2010b). As can be seen in Figure 5.2, h was above 3 for all flow rates tested. The

value of h ranged between 5.4 and 6.5 and increased along with increasing flow rate until the flow

100

Table 5.2 Calculation of variables for column efficiency testing.

Flow

rate

(ml/min)

Ve1

(ml)

Wh2

(ml) N3 Plates/m4

HETP5

(cm) h6 a7 b8 As9

0.05 4.01 0.345 745.5 3727.6 0.027 5.4 0.34 0.3 0.87

0.075 4.16 0.374 682.2 3411.1 0.029 5.9 0.38 0.34 0.89

0.1 4.18 0.383 659.7 3298.4 0.030 6.1 0.41 0.33 0.81

0.125 4.23 0.401 615.0 3075.0 0.032 6.5 0.42 0.34 0.81

0.15 4.26 0.394 648.8 3244.0 0.031 6.2 0.47 0.33 0.71

1Ve is peak elution volume.

2Wh is peak width at half peak height.

3N is number of theoretical plates in the column, calculated according to Equation 2.1.

4Plates/m is the number of theoretical plates per meter, where length of column (L) is 20 cm.

5HETP is height equivalent of a theoretical plate, calculated according to Equation 2.2.

6h is reduced plate height, calculated according to Equation 2.3.

7a is forward peak part width at 10% peak height.

8b is rear peak part width at 10% peak height.

9As is asymmetry, calculated according to Equation 2.4.

101

Figure 5.2 Column efficiency test. The efficiency of the column was tested by running acetone solution on the column in five different flow rates, one of which (0.1 ml/min) is depicted here (A). Data collected from these runs were used to measure h and As, which indicate the efficiency of the column (B). . Optimum flow rate fell between 0.075 and 0.1 ml/min.

102

rate reached 0.125 ml/min after which the value of h decreased. The ideal h value could not be

achieved due to the system volume. As of the column ranged between 0.71 and 0.89 with a decreasing

pattern along an increasing flow rate. Based on empirical h and As value and the time it took to run a

sample in a column with a total column volume of 4 ml, the optimum flow rate for the column was

between 0.075 ml/min and 0.1 ml/min. Therefore, 0.1 ml/min was used for the rest of this chapter, as

it was closest to the intersection of the h and As curves.

5.4.2 Column validation with wt MPyV VLPs

The column was validated by running wt MPyV VLPs. The chromatogram showed a single VLP peak

at 2.85 ml elution volume within an acceptable symmetry (Figure 5.3A). DLS analysis of collected

elution fractions showed 100% VLP-size particles within the peak region with radius slightly

decreasing along the elution volume, as expected. VLP particle radius at the peak region was 29.6 nm

and polydispersity was 3.8% based on regularisation analysis. SDS-PAGE gel confirmed that the

fractions contain VP1 corresponding to the peak region (Figure 5.2B), and TEM confirmed they were

in the form of VLPs (Figure 5.3C).

5.4.3 Design of encapsidated influenza M1 protein construct

A Tc-epitope-based influenza vaccine via internal encapsidation by MPyV VLPs was developed.

Previous encapsidation optimisation with GFP as the model protein in Chapter 4 was applied to

encapsidate the soluble domain of influenza M1 protein. The soluble domain of M1 protein used in

this research was the N-terminal region of M1 (2-159) (more in Section 2.3.5). The first subdomain,

M1-I (2-73), contains IT1 and IT2 that have been presented externally as described in Chapter 3.

The second subdomain, (74-159), contains a self-interacting domain, and M1-II consists of both first

subdomain (M1-I) and the second. Both M1-I and M1-II were fused to VP2C 31 aa in place of GFP

(Figure 5.1). Similar to the VP2C-GFP 31 aa construct on Chapter 4, either M-I or M-II were fused

to the C-terminus of VP2C and a 6His-tag was engineered in the C-terminal end of the inserted protein

to enable IMAC purification. The 6His-tag allows the purification of free VP2C-M1-I or VP2C-M1-

II as well as VP1:VP2C-M1-I and VP1:VP2C-M1-II complexes. Only VP1 forming a complex with

the VP2C fusion proteins can be purified with IMAC.

5.4.4 IMAC purification of VP1:VP2C-M1-I indicated successful

binding with VP1

VP1:VP2C-M1-I was purified with IMAC (Figure 5.4A & B) with conditions optimised in Chapter

4. Elution fractions were collected and analysed with SDS-PAGE. The gel showed that VP1 was co-

103

Figure 5.3 wtVP1 VLP was tested with S500 5/200 column. The S500 5/200 column was evaluated for its ability to isolate wtVP1 VLPs by semi-preparative SEC (A) for validation by DLS, SDS-PAGE (B), and TEM (C). Absorbance is at 280 nm and TEM scale bar is 100 nm. Lanes b of SDS-PAGE corresponds to VLP peak on the chromatogram as indicated. M is molecular weight standard. Lo is loading sample, Arrow points to wtVP1 subunit.

104

Figure 5.4 IMAC of VP1:VP2C-M1-I. Pellets were sonicated and VP1:VP2C-M1-I was purified with IMAC. Absorbance is at 280 nm (A). The elution peak was zoomed to indicate collected fractions (B). SDS-PAGE with lanes corresponding to peak regions as indicated, showed that the co-purification of VP1 suggests the presence of VP2C-M1 even though it is not clear on the SDS-PAGE gel (C). M is the molecular weight standard. Arrows point to bands as indicated.

105

eluted with VP2C-M1-I even though the band of the latter was not clearly visible on the gel (Figure

5.4C). Peak elution fractions of IMAC purification were further purified with SEC and resulted in

three major peaks: a) aggregate peak (8.7 ml); b) capsomere complex peak (11.7 ml); and c) free

VP2C-M1-I peak (16 ml) (Figure 5.5A). SDS-PAGE analysis of peak fractions showed that peak a

contained VP1, VP2C-M1-I, and contaminating proteins; peak b contained VP1, VP2C-M1-I, and a

70 kDa band, and peak c contained what may be dimers of free VP2C-M1-I (Figure 5.5B). The

VP2C-M1-I band is a faint band due to its small size. Co-elution of VP1 and VP2C-M1-I in peak b

indicated successful formation of VP1:VP2C-M1-I capsomere complex.

Western blot of these fractions with anti-VP1 (Figure 5.5C) also confirmed the presence of VP1

especially in peaks a and b. Other non-specific bands were also stained. Anti-6His blotting (Figure

5.5D) showed staining in the VP1 region which may indicate that the complex was not fully denatured

or that there were some cross-reactivity with VP1. The aggregate peak also contained anti-6His ~30

kDa and ~15 kDa putative VP2C-M1-I bands, the former may be a dimer of the latter. Of the two,

only the ~15 kDa band was found in peak b, while only the ~30 kDa band was found in peak c.

5.4.5 VP2C-M1-II oligomerisation complicates complex purification

The VP1:VP2C-M1-II construct was also expressed in small scale and purified in optimised

condition. IMAC elution peak (Figure 5.6A) and S200 of this peak resulted in multiple peaks (Figure

5.6B). Multiple peaks may be caused by oligomerisation due to a self-interacting region in the M1-II

domain (Zhang et al. 2012), and thus complicated the purification. Due to this, and because M1-I

contains both IT1 and IT2, this construct was not pursued further.

5.4.6 Development, analysis, and characterisation of M1-I-based VLPs

as vaccines

The VP1:VP2C-M1-I capsomere complex harvested in Section 5.4.3 was assembled in standard wt

buffer conditions into VLPs, concentrated, and then injected into S500 SEC column. The

chromatogram profile showed two peaks: a) aggregate peak at 2 ml; and b) VLP peak at 2.8 ml. Peak

fractions and other regions as indicated in Figure 5.7A were collected and analysed with SDS-PAGE

and characterised by DLS and TEM. SDS-PAGE analysis showed the presence of VP1 in peaks a and

b (Figure 5.7B). DLS characterisation was done with both cumulant and regularisation analysis.

Cumulant analysis showed a large cumulant radius in peak a, and the size of the cumulant radius of

this peak decreased with higher elution volume. Although the cumulant radius in the VLP region

(peak b) decreased significantly in comparison to the aggregate region, it did not decrease into the

reported radius for VLPs (73 nm instead of 27nm - 35 nm) indicating incomplete separation from the

106

Figure 5.5 S200 of VP1:VP2C-M1-I. IMAC purified fractions were further purified with S200 to separate modified capsomeres from aggregates and free VP2C-M1-I (A). The fractions were run on SDS-PAGE with lanes corresponding to peak regions as indicated, which showed the content of each peak (B). Western blot using anti-VP1 (C) and anti-6His (D) showed that VP2C-M1-I may appear as dimers in both aggregate and free forms, but only in monomer when binding to VP1. Absorbance is at 280 nm. Arrows point to bands as indicated.

107

Figure 5.6 S200 of VP1:VP2C-M1-II. Pellets were sonicated and VP1:VP2C-M1-II was purified with IMAC (A). IMAC fractions were further purified with S200 (B) and the chromatogram showed multiple peaks.

108

Figure 5.7 S500 of VP1:VP2C-M1-I. Assembled VP1:VP2C-M1-I was purified by semi-preparative SEC and showed aggregate and VLP peaks (A). SDS-PAGE with lanes corresponding to peak regions as indicated, showed the presence of VP1 (B), and TEM showed formation of VLPs (C). Absorbance is at 280 nm and TEM scale bar is 200 nm. M is molecular weight standard. Arrow points to bands as indicated.

109

aggregates. Regularisation analysis confirmed this by showing different particle species in peak b,

including aggregates and VLPs, with the aggregate ratio decreasing with increasing elution volume

and the VLP ratio increasing on the contrary (data not shown). According to regularisation analysis,

the VLP particle radius was 27.7 nm and the formation of VLPs in fractions derived from peak b was

confirmed by TEM (Figure 5.7C).

5.4.7 Development, analysis, and characterisation of DDIT1-based

VLPs as vaccines

To enhance VLP formation and reduce hydrophobic effects, DDIT1 capsomeres derived from

Chapter 3 were co-assembled with unmodified VP1 capsomeres at a 1:4 molar ratio. Assembly was

performed in an optimised assembly buffer containing 50 mM L-Arg to maintain the solubility of

DDIT1 during the formation of mosaic VLPs. This assembled mixture was concentrated (>2 mg/ml)

and injected into S500 column for analysis. Similar to VP1:VP2C-M1-I VLPs, the chromatogram

showed two peaks: a) an aggregate peak at 2 ml; and b) a VLP peak at 2.7 ml (Figure 5.8A). Peak

fractions were collected for further analysis with SDS-PAGE and characterisation with DLS and

TEM. SYPRO Orange-stained SDS-PAGE gels showed the presence of two adjacent bands in the

VP1 region, the upper band thinner than the lower band. The upper band corresponds to the DDIT1-

inserted modular capsomere because it was slightly larger in size than the wtVP1 capsomeres. The

thinner band also reflects its lower molar concentration in comparison to wtVP1. The lower band

corresponds to wtVP1 capsomeres (Figure 5.8B). According to densitometry analysis, the ratio

between modular and wt capsomeres on peak b is 1:2.32. The presence of these two bands in the VLP

region and the stabilising effect of co-assembled wt capsomeres strongly indicated that mosaic VLPs

had been successfully assembled. Cumulant analysis of the fractions by DLS showed a similar pattern

to VP1:VP2C-M1-I, and according to regularisation analysis the VLP particle radius is 26 nm.

Formation of intact VLP was again confirmed by TEM (Figure 5.8C).

5.5 Discussion

In this project, a semi-preparative SEC for VLP purification, analysis, and characterisation was

developed and optimised. The MPyV VLP encapsidation platform optimised in Chapter 4 was used

to encapsidate influenza M1 protein for vaccine development, and a DDIT1-based VLP from

Chapter 3 was improved. Using the developed semi-preparative SEC method these vaccines were

then analysed and characterised.

The semi-preparative SEC utilised a 5/200 Tricorn column with a 4 ml total volume using Sephacryl

110

Figure 5.8 S500 of mosaic VP1 S1S4 DD IT1. Assembled mosaic VLPs of VP1 S1S4 DDIT1 and wt was purified by semi-preparative exclusion chromatography (A) and showed aggregate (a) and VLP peaks (b). Analysis of fractions with SDS-PAGE and SYPRO Orange staining showed that VLP and aggregates contained both VP1 S1S4 DD IT1 and wtVP1 capsomeres (B). Successful mosaic particle formation was confirmed by TEM (C). Absorbance is at 280 nm and TEM scale bar is 200 nm. Arrows point to bands as indicated.

111

S500 HR as a medium. This column was selected to shorten analysis time with the least amount of

sample possible. Commercially available Sephacryl-based column (Hi-Prep 16/60 Sephacryl S500

HR) requires 4 hours for one sample run, and a higher amount of protein in a larger volume was

required. This was not practical for an expression scale that aimed to analyse and characterise formed

VLPs prior to large-scale production. A prepacked Tricorn 5/200 S500 column with Sephacryl S500

was not commercially available, and so the column was packed in-house and its column efficiency

tested before use. The external volume, defined as the tubing volume from the point of sample

application to column inlet and from the column outlet to the UV detector should theoretically be 3%

of the total column volume for optimum column efficiency (GE-Healthcare 2010a; Hagel 2011).

However, as the volume of the optimised Tricorn 5/200 S500 column in this project is only 4 ml, a

3% external column volume (120 µl) could not be achieved, even by using the AKTA Pure 25 system

with the shortest tubing possible. This was reflected in measured h plotted against increasing flow

rates (Figure 5.1) that were all above 3, while an efficient column would have h < 3. Thus, efficiency

analysis for this column was more focused on the As and the performance with known samples, such

as VLPs. Considering both h and As as depicted in Figure 5.2 and all other non-adjustable factors,

column efficiency was best at flow rates of 0.075 ml/min – 0.1 ml/min. With a 0.1 ml/min flow rate,

a sample could be run within 40 min, which was selected as optimal. Increasing the flow rate would

compromise column efficiency, while decreasing the flow rate would lengthen purification time.

Validation of the column with wtVP1 VLPs showed that the VLP peak was within acceptable

symmetry (Figure 5.3). DLS cumulant and regularisation analysis of eluted fractions also

demonstrated that particles in the VLP region were 100% VLPs (monodispersed) and eluted with

slightly decreasing size along the peak volume as expected. DLS measurement before and after the

VLP peak was not consistent, and this may be due to the nature of light scattering measurement,

which is compromised when sample concentration is too low. Column optimisation and validation

results give a good level of confidence that it would be a useful tool for a semi-preparative SEC to

analyse and characterise VLPs. The method is not time consuming and the injected sample volume

required is relatively small (80 µl), which allows concentration of protein solutions harvested from a

standard laboratory expression scale. A report on concurrently developed interlaced size exclusion

ultra-high performance liquid chromatography (iSE-UHPLC) (Ladd Effio et al. 2016) and a

Captocore-based column to purify VLPs (Lagoutte et al. 2016) have recently been reported. However,

while the iSEC-UHPLC has a faster run time, it was only analytical and not preparative. Moreover,

the ability of the Captocore-based column to separate VLPs from aggregates was not clearly

demonstrated.

112

One of the modification strategies to develop MPyVLP based vaccine is to encapsidate antigenic

protein inside the VLP. Encapsidation technology had been optimised in Chapter 4 and this was then

applied to encapsidate the influenza M1 protein. The M1 protein is a transmembrane protein and thus

contains a hydrophobic region (Zhang et al. 2012). To avoid aggregation, only the soluble domain of

this protein was fused to VP2C, which came in two variations, VP2C-M1-I and VP2C-M1-II. The

M1-I domain contains several epitopes (Figure 1.2) comprising several HLA supertypes such as A2,

A3, B44, and B12 (Adar et al. 2009; Alexander et al. 2010; Assarsson et al. 2008; Babon et al. 2009;

Liu et al. 2013; Reemers et al. 2012), which would be good for broader cross-protection (Brown and

Kelso 2009) and wider population coverage (Sette and Sidney 1998). VP1:VP2C-M1-II was not

pursued further due to complication in purification related to the presence of the previously reported

self-interacting domain (Figure 5.6)(Zhang et al. 2012). VP1 band was more clearly visible than the

VP2C-M1-I on SDS-PAGE when clarified supernatant was purified with IMAC (Figure 5.4), but the

presence of VP1 indicated co-purification of VP2C-M1-I as only VP1 bound to VP2C-M1-I could be

eluted from an IMAC column. Faint bands of VP2C-M1-I were visible on SDS-PAGE and anti-6His

stained western blot of S200 (Figure 5.5). This confirmed that the VP1:VP2C-M1-I capsomere

complex was formed. Analysis of SEC fractions of VP1:VP2C-M1-I implied that VP2C-M1-I existed

as a dimer in free form, and this may be due to the presence of the hydrophobic IT1 epitope at the C-

terminal region of M1-I which may have facilitated dimerisation by hydrophobic interaction. The

presence of the 70 kDa band in peak b, the putative capsomere peak, but not in the aggregate peak of

the SEC chromatogram may indicate that the 70 kDa protein was necessary for formation of a soluble

VP1:VP2C-M1-I capsomere complex and that without it aggregation may occur. Western blot of SEC

fractions may indicate that when bound to VP1, VP2C-M1-I is monomeric, but otherwise dimeric

when unbound. This dimerisation may also have resulted in aggregation as both forms were detected

in the aggregate peak.

Characterisation of the eluted peak fractions of assembled VP1:VP2C-M1-I was done with DLS

cumulant and regularisation analysis. Cumulant radius is the average radius of total particles in the

solution, so the measurement is influenced by the size of different species in the solution, including

aggregates. The number and size of aggregated particles in the putative VLP peak may be a lot less

than to those in the first peak, but the size of the particles may be enough to mask the presence of

correctly sized VLPs in this peak and resulted in increased overall cumulant radius. The regularisation

analysis was able to detect and measure the size of VLPs, the presence of which, was confirmed by

TEM. These results indicated initial success in the development, analysis, and characterisation of an

MPyV VLP-based carrier carrying M1-I subdomain.

113

The results presented in Chapter 4 suggested that mosaic particles could be assembled from mixed

capsomere assembly reactions. Mosaic particle formation by co-assembly of DDIT1 capsomeres with

wtVP1 capsomeres in optimised buffer condition was confirmed by TEM through observation of

clearly visible VLPs and in a higher density (Figure 5.8C) than assembly solution of DDIT1

capsomeres alone (Figure 3.6D) as has been demonstrated in Chapter 3. SDS-PAGE gel showed the

simultaneous presence of DDIT1 and wtVP1 which also strongly indicated the formation of mosaic

particles. However, we cannot be certain that the VLPs consist of mixed capsomeres, only that the

ensemble of VLPs does. DLS measurement showed a similar pattern to that of VP1:VP2C-M1-I VLPs

and the radius was also not far from normal wtVP1, which is expected with the addition of a single

epitope. These results also indicated initial success in the development, analysis, and characterisation

of an MPyV VLP-based carrier carrying IT1 epitope at a surface-exposed loop of VP1.

Aggregation in the vaccine formulation may be caused by several factors, including self-interaction

between recombinant proteins due to hydrophobic interaction, and VP2C-M1-I dissociation which

initiates aggregate nucleation. Aggregation may also occur during concentration of the VLP sample.

Tailing of the aggregate peak in the VLP region as indicated by cumulant analysis suggests further

optimisation of the semi-preparative SEC method. The currently developed method is enough for

analysis and characterisation, but would be less suitable for vaccine purification as aggregates need

to be fully separated from the vaccine. Optimisation strategies may include an increase in column

length, or lower flow rates. Increasing the column length and column volume would increase the

tolerated external volume and sample volume and thus would lead to a more accurately estimated

value of h.

5.6 Conclusion

Here I have demonstrated initial success in the development of an MPyV VLP-based influenza

vaccine, carrying Tc-epitopes either externally or internally. A semi-preparative SEC method to

analyse and characterise the VLPs was also developed and was within acceptable efficiency. Further

vaccine development may be done by combining external and internal epitope presentation in one

particle. Developed vaccines still need to be produced in sufficient quantities to test their

immunogenicity and protection against influenza virus in humanised mice. The encapsidation

platform may also be further developed into an enzyme, a drug, and an imaging component carrier.

114

6 Conclusion and Future Direction

6.1 Summary of research findings

Limitations in current influenza vaccines, including high updating frequency, slow production time,

and high production cost, necessitates the development of next-generation influenza vaccine that is

broadly cross-protective and efficient in terms of time and cost of production. A microbially produced

modular MPyV-based vaccine is a promising approach to overcome these limitations. Influenza

candidate vaccines based on this platform have been developed to raise humoral immunity against

both variable (H190) and relatively conserved (M2e) influenza antigens. The development of

influenza vaccine candidates based on this platform to raise cellular immunity has not been explored.

Targeting cellular immunity may exploit antigens of the more conserved internal influenza proteins,

and may thus confer broader cross-protection and reduce vaccine updating frequency.

My research project aims to explore the potential of the MPyV VLP platform to carry identified

influenza Tc-epitopes both externally via surface-exposed loops, and internally via encapsidation.

The MPyV VLP can enter mammalian cells, can be processed by the MHC Class I pathway, and can

activate dendritic cells. These characteristics are necessary requirements to stimulate cellular

immunity. Antigenic Tc-epitopes to stimulate cellular immunity come in the form of nonameric

peptide, and thus may be inserted into surface-exposed loop or encapsidated as soluble protein domain

within the MPyV VLP platform. However, bioengineering challenges posed by the hydrophobic

nature of Tc-epitopes need to be properly addressed.

The experimental work discussed in this thesis was designed to explore four key aspects in the

application of MPyV VLP as a carrier for influenza Tc-epitopes:

1) External Tc-epitope presentation – This preliminary study evaluated the robustness of MPyV

VLP to carry identified influenza Tc-epitopes via insertion into surface-exposed loop of the

VP1 subunit. Insert engineering and buffer optimisation through multiparametric screening to

overcome hydrophobicity-related aggregation was highlighted in Chapter 3.

2) Optimisation of encapsidation – Prior to influenza antigen encapsidation, the molecular design

and bioprocess of the established MPyV VLP platform was optimised to encapsidate the

model protein GFP. Optimisation in expression, purification, and assembly was monitored

and evaluated as highlighted in Chapter 4.

115

3) Tc-epitope-based MPyV VLP candidate vaccines – Candidate vaccines carrying influenza Tc-

epitope in the form of chimeric MPyV VLP were optimised. Capsomeres externally carrying

Tc-epitope were co-assembled with unmodified capsomeres to optimise VLP formation.

Encapsidation approach was applied to soluble domain of influenza M1 to carry the antigen

internally. Development of these candidate vaccines, and their analysis and characterisation

were highlighted in Chapter 5.

4) A method to characterise and analyse VLPs – For Mosaic VLPs to be analysed and

characterised a semi-preparative SEC method was developed for this purpose as highlighted

in Chapter 5.

The following sections summarise key findings of this project obtained from experimental work.

6.1.1 External Tc-epitope presentation

Three identified Tc-epitopes, IT1, IT2, and IT3 were inserted into VP1 surface-exposed loop in four

variations which are IT1, DDIT1, IT123, and DDIT123. Standard expression and purification

indicated that capsomeres aggregated during tag-removal and were not recoverable when purified

afterwards, most likely due to the hydrophobic nature of the Tc-epitopes. Two strategies were applied

to overcome hydrophobicity-related aggregation. Firstly, charged amino acids (DD) flanking the

epitopes were engineered to reduce surface hydrophobicity. Secondly, a high-throughput

multiparametric buffer screen was implemented employing spectrophotometric (Esfandiary and

Middaugh 2012) and light scattering (Mohr et al. 2013) parameters to measure aggregation. Forty

buffer conditions were evaluated in a 96-wells plate miniaturised tag removal reaction. The buffer

conditions tested included varying pH (7-9 in 0.5 increments) together with a selection of additives

(Arakawa et al. 2007; Lanckriet and Middelberg 2004; Leibly et al. 2012).

Results showed that in the case of a single epitope (IT1), a combination of flanking charged amino

acid (DD), the presence of L-Arg or detergents as additives, as well as a high pH, were able to improve

capsomere recovery and reduce aggregation. This, however, was not true for multiple epitope inserts.

These results underline the fact that successful buffer optimisation and insert engineering depends on

the nature of individual inserts, and optimisation should be tailored according to insert characteristics.

It also suggests that the number of Tc-epitopes that could be inserted into surface-exposed loop is

limited.

L-Arg was further used as additive to recover DDIT1 capsomeres for assembly, due to the

biocompatibility and inexpensiveness of the additive. This additive is widely used in the literature to

overcome protein aggregation (Arakawa et al. 2007; Tsumoto et al. 2004). Assembly of modified

116

capsomeres was also only possible in the presence of L-Arg and was not observed in standard

assembly buffer in the absence of L-Arg. However, VLP yield was very low and a different strategy

had to be employed to reduce hydrophobicity that might have interfered with assembly. One such

strategy was the co-assembly of modified capsomeres displaying hydrophobic epitopes with

unmodified capsomeres to reduce surface hydrophobicity. This strategy was pursued in Chapter 5.

6.1.2 Optimisation of encapsidation

Another approach to circumvent VLP surface hydrophobicity is by encapsidating influenza antigen

inside the MPyV VLP. Tc-epitope proteasomal processing is intracellular, and unlike recognition of

antibody-based antigens, Tc-epitopes may be carried inside the cavity of the VLP. This approach

allows the carriage of multiple Tc-epitopes within a protein or soluble domain. The delivery of

multiple epitopes may enhance protection against heterologous strains and extend vaccine population

coverage due to a more variable HLA specificity. Prior to encapsidation of influenza antigen,

encapsidation with the MPyV VLP platform needed to be optimised in every step of the bioprocess

including genetic manipulation, expression, purification, and assembly. To achieve this,

encapsidation of GFP as a model foreign protein to facilitate monitoring and evaluation was explored

in Chapter 4.

Encapsidation of foreign protein is possible by fusing the protein of interest with the MPyV VP2

minor coat protein. VP2 binds to the inner cavity of one VP1 capsomere via a hydrophobic binding

region close to the VP2 C-terminal end. Foreign protein fusion either at the N-terminal or C-terminal

ends of polyomaviral VP2 have been reported in the literature (Abbing et al. 2004; Boura et al. 2005;

Inoue et al. 2008). In my research, GFP was fused to the C-terminus of VP2. For efficient

encapsidation, a VP2C segment was specially designed. The VP2C segment was designed in three

variations which consisted either of 23, 31, or 50 amino acids (aa). These constructs retain the

hydrophobic binding site of VP2, but most of the N-terminal amino acids upstream of the

hydrophobic binding site were truncated, and a small part of the C-terminal amino acids were also

eliminated. GFP was fused to the C-terminus of VP2C and was tagged with 6His to enable

purification. Unlike previous reports where GFP was fused to the N-terminus of MPyV VP2 (Abbing

et al. 2004; Boura et al. 2005), in my research GFP was fused to C-terminus of VP2C based on the

findings of Inoue et al. (2008) with SV40 and Pleckaityte et al. (2015) with HPyV, that suggested a

complete encapsidation with C-terminal fusion.

When free VP2C-GFP was mixed in vitro with VP1 capsomeres, no binding between them occurred.

Unsuccessful binding was due to truncation at the hydrophobic binding site of VP2C-GFP, probably

by proteolytic activity. Truncation was not observed when Abbing et al. (2004) mixed GFP-VP2 with

117

VP1 capsomeres, where binding and subsequent encapsidation occurred successfully. Intact

hydrophobic binding site in their experiment might be due to GFP fusion to the N-terminal end of

VP2 whereby the preservation of the N-terminal sequence, as well as GFP fusion to the N-terminus,

may have sterically protected the hydrophobic binding site from proteolytic activity in Abbing et al.

(2004). The elimination of most of the N-terminal sequence and GFP fusion to the C-terminus of

VP2C in my constructs may have exposed the hydrophobic binding site, rendering it accessible to

proteases.

Taking into account the high binding affinity between VP1 capsomeres and VP2, and that their

binding occurs naturally in the natural host cell, I decided to emulate this reaction in vivo by co-

expressing VP1 and VP2C-GFP inside the host cell. It was expected that the binding would occur

rapidly in vivo, protecting the hydrophobic binding site of VP2 inside the cavity of the capsomere

from protease activity. A 6His-tag was engineered downstream of the GFP to enable IMAC

purification, and SDS-PAGE of IMAC fractions indicated co-purification of VP2C-GFP and VP1.

Co-purification of VP1 with 6His-tagged VP2C-GFP indicated successful VP1:VP2C-GFP

capsomere complex formation in vivo, as only VP1 bound to VP2C-GFP could be selectively purified

with IMAC. Further purification with SEC enabled the isolation of VP1:VP2C-GFP capsomere

complex from free VP2C-GFP and aggregates. Of the three constructs tested, VP1:VP2C-GFP 23 aa

had a low expression and solubility level while VP1:VP2C-GFP 50 aa was truncated and had a low

VP1:VP2C-GFP capsomere complex yield after purification. Therefore, only VP1:VP2C-GFP 31 aa

was pursued further as it had significant expression and solubility level and relatively high

VP1:VP2C-GFP capsomere complex yield after purification.

Purified VP1:VP2C-GFP capsomere complex were then assembled in standard assembly buffer and

subsequent AF4, as well as TEM and SEC analysis, confirmed that VLP with GFP encapsidated

inside were formed. The approach in my research differed from what was previously reported in the

literature in several ways. Firstly, VP1 and VP2C-GFP were co-expressed in E.coli and purified as a

VP1:VP2C-GFP capsomere complex. Both Boura et al. (2005) and Inoue et al. (2008) co-expressed

VP1 and VP2-GFP in insect cells and thus assembly occurred in vivo and the product was purified

VLPs instead of a VP1:VP2-GFP capsomere complex. Secondly, in my research, purified

VP1:VP2C-GFP capsomere complex were assembled in vitro in standard assembly buffer condition

established in AIBN-CBE, and MPyV VLP with GFP encapsidated were successfully formed.

Abbing et al. (2004) also performed assembly in vitro, but their assembly employed free capsomeres

and excessive free VP2-GFP, where removal of remaining free VP2-GFP was done after VLP was

formed.

118

The advantage of the approach in my research is that the amount of encapsidated GFP inside the VLP

could be controlled by in vitro co-assembly of VP1:VP2C capsomere complex with unmodified VP1

capsomeres. Controlled assembly could not have been achieved by simply assembling free

capsomeres and free VP2-GFP alone. Controlled assembly was demonstrated when an assembly of

100% VP1:VP2C capsomere complex, a co-assembly of 1:1 molar ratio of VP1:VP2C-GFP

capsomere complex and unmodified VP1 capsomeres, and an assembly of 100% unmodified VP1

capsomeres were compared. Spectrophotometric analysis revealed that 100% VP1:VP2C-GFP

capsomere complex assembly resulted in encapsidation of 40 functional GFP molecules while 50%

co-assembly resulted in encapsidation of on average approximately half the amount of GFP molecules

per VLP.

Thus, these experiments elucidated the ability to engineer controlled encapsidation of foreign protein

inside the cavity of MPyV VLPs within the established platform developed at AIBN-CBE. This

bioengineering milestone was achieved by additional adjustment, optimisation, and modification of

the genetic manipulation, expression, purification, and assembly steps of the established platform.

Having established the methodology for controlled foreign protein encapsidation, its application to

encapsidate foreign antigenic proteins to stimulate cellular immunity was subsequently approached.

Developing a MPyV VLP vaccine candidate with influenza protein soluble domains encapsidated

inside was pursued in Chapter 5.

6.1.3 Tc-epitope-based MPyV VLP candidate vaccines

Building on the previous experiments, Tc-epitope-based MPyV VLP candidate vaccines were

developed in Chapter 5. Two types of candidate vaccines were developed: 1) candidate vaccine that

display identified influenza Tc-epitopes on surface-exposed loop; 2) candidate vaccine that

encapsidate influenza antigen in the form of protein soluble domain with multiple overlapping Tc-

epitopes within.

As had been summarised in Section 6.1.1, capsomeres displaying a Tc-epitope (DDIT1) on the

surface were recoverable after insert engineering and buffer optimisation via multiparametric buffer

screen. However, when the capsomeres were assembled in optimised assembly buffer, VLP yield was

low. This low yield may be due to VLP surface hydrophobicity caused by surface-exposed

hydrophobic Tc-epitopes, which may interfere with VLP formation. To improve yield, an approach

to reduce surface hydrophobicity was applied by co-assembly of DDIT1 capsomeres with unmodified

VP1 capsomeres in optimised assembly buffer. This optimised assembly buffer retained L-Arg as

additive. SEC of the assembly reaction showed two major peaks, an aggregate peak and a VLP peak.

DLS analysis indicated the formation of VLPs in the VLP peak, even though the VLPs were not

119

completely separated from aggregates. The presence of bands of the modified and unmodified

capsomeres in the VLP peak on SDS-PAGE, as well as a more densely populated VLPs shown by

TEM analysis in comparison to VLP formation in Chapter 3, strongly indicated the formation of

mosaic particles able to circumvent surface hydrophobicity, aiding in the formation of VLPs.

However, it should be noted that this result does not conclusively confirm that the VLPs consisted of

mixed capsomeres, it only shows that the ensemble of VLPs did.

The encapsidation methodology developed in Chapter 4 and summarised in Section 6.1.2 was

applied to encapsidate influenza protein inside the MPyV VLP. The M1 protein was selected for this

purpose for several reasons. Firstly, M1 protein is a relatively conserved internal protein and thus the

Tc-epitopes it contains have a higher probability to be broadly cross-protective. Secondly, the M1

protein is the most abundant protein in the influenza virus (Lamb and Krug 2001) and thus increases

the likelihood of Tc-epitope presentation by infected cells, necessary for Tc recognition. Thirdly, the

M1 protein, especially its N-terminal soluble domain, contains IT1. IT1 is used as surface-exposed

Tc-epitope insert for the other candidate vaccine in my research project as described in previous

paragraph, and thus immunogenicity and protection of externally displayed and internally

encapsidated IT1 may be compared in the future. In addition, IT1 (amino acid sequence on Table

3.1) is an immunodominant Tc-epitope widely used in the literature to test immunogenicity and

protection of influenza vaccine candidates.

Two variants of soluble M1 domain were tested. The first was M1-I which is the first N-terminal

helical domain containing IT1 and IT2. The second was M1-II which the full soluble N-terminal

domain including M1-I and a self-interacting region. SEC purification of M1-II revealed unidentified

peaks suggestive of interactions between M1 domains so it was not pursued further. SEC purification

of M1-I revealed distinct peaks of aggregates, VP1:VP2C-M1-I capsomere complex, and free VP2C-

M1-I. Western blot analysis of these peaks with anti-VP1 and anti-6His indicated the formation of

VP1:VP2C-M1-I capsomere complex at the expected peak, and that in free form, VP2C-M1-I tended

to form dimers. Interestingly, SDS-PAGE analysis of these fractions indicated that the 70 kDa

putative chaperone might be necessary for complex formation, and its absence might have led to

aggregation. The 70 kDa chaperone DnaK is known to associate with capsomeres in vivo and except

in the presence of ATP can block assembly, which can be seen as controlled encapsidation, in vitro

(Chromy et al. 2003). These results together showed that successful capsomere complex formation,

purification, as well as subsequent VLP formation depends on the nature of the foreign protein fused

to VP2C. And thus, bioengineering of each fusion protein needs to be adjusted accordingly.

120

Purified VP1:VP2C-M1-I capsomere complex were further assembled in standard buffer condition.

SEC fraction of assembly reaction was then analysed and characterised with SDS-PAGE, DLS, and

TEM. Similar to VP1:VP2C-GFP VLPs, SEC profile revealed aggregate and VLP peaks and their

analysis with TEM confirmed VLP formation with small amount of aggregates in the VLP peak

fractions. DLS regularisation analysis measured a smaller VLP radius (26 nm) in comparison to that

of VP1:VP2C-GFP (30 nm). The size difference was not unexpected as GFP is larger in size that M1-

I, and size of encapsidated particles can affect VLP size (Lipin et al. 2008a). Interestingly, the

presence of the putative chaperone did not block assembly of this construct despite the absence of

ATP in the assembly buffer. Furthermore, SEC showed that it was not associated with the assembled

VLPs.

In this section, candidate vaccines carrying influenza Tc-epitopes either externally on surface-

exposed loop or internally via encapsidation of M1 subdomain was developed. Both approaches to

carry Tc-epitopes either externally or internally have their advantages and disadvantages. The

advantage of external Tc-epitope presentation include simple genetic manipulation which only

involves the VP1 protein. If hydrophobicity-related aggregation could be circumvented by insert

engineering and buffer optimisation, a recently developed salting-out precipitation method to recover

stable capsomeres (Wibowo et al. 2015) may be applied. The use of salting-out precipitation method

could omit previous purification methods involving sophisticated equipment that may be less

applicable in developing countries. However, surface-exposed loop display does not facilitate the

presentation of multiple Tc-epitopes which limits broader cross-protection and wider population

coverage. This problem may be circumvented by co-expressing VP1 displaying different Tc-epitopes

resulting in mosaic capsomeres (Tekewe et al. 2016), or the development of mosaic VLPs composed

of capsomeres displaying different Tc-epitopes.

The advantages of internal encapsidation of Tc-epitopes include the ability to carry multiple Tc-

epitopes and the prevention of hydrophobicity-related aggregation due to antigen encapsidation inside

the VLP. The ability to carry multiple Tc-epitopes may allow a broader crossprotection and a wider

population coverage of the vaccine candidate. However, the encapsidation strategy involves VP1

binding with fused VP2 and thus requires a more complex downstream processing, which at the

current stage may still need to rely on sophisticated protein purification equipment.

Whether the surface display of antigens is more favourable for proteasomal access in comparison to

encapsidation of antigens still needs to be evaluated further. External antigen display, as well as

engineered linker sequence to facilitate proteosomal degradation, may enhance Tc-epitope

processing. However, Tc-epitope located inside a subunit protein may have the advantage of

121

emulating natural Tc-epitope processing and protection of the antigen from premature or non-specific

degradation.

These experiments aim to develop Tc-epitope-based influenza vaccine using two approaches, either

by external or internal presentation of influenza antigens. Optimisation of those approaches included

the formation of mosaic particles and the encapsidation of M1-I soluble domain. Both methods

resulted in VLP formation in considerable yield, as shown by analysis and characterisation of SEC

fractions of the assembly reactions. A new SEC method with Sephacryl S500 HR media to analyse

and characterise the VLPs was developed and validated in Chapter 5, as summarised in Section

6.1.4.

6.1.4 SEC-S500 to analyse and characterise VLPs

A method to analyse and characterise purified VLPs developed in my research, after separation from

aggregates, was needed for several reasons. Firstly, to confirm successful encapsidation of GFP

protein inside the MPyV VLP and to quantitate GFP content during controlled assembly via

spectrophotometric measurement. Secondly, to confirm the formation of mosaic particles, which

required recovery of purified VLP fractions to run on SDS-PAGE gels. Finally, to visualise the VLPs

via TEM and characterise them based on hydrodynamic radius, where preparative samples of purified

VLPs were also needed.

The VLP purification in my research is aimed only for analysis and characterisation, instead of

production and formulation. Commercially available column for SEC commonly used to purify

VLPs, such as the Hi-Prep 16/60 Sephacryl S500 HR, was not suitable for relatively small laboratory

scale VLP preparation. In addition, one sample required 4 hours to run with this column, and fractions

were distributed in a large total peak volume. A large total volume of peak fractions reduced fraction

protein concentration and rendered them insufficient for analysis and characterisation. Therefore, a

new SEC method employing Sephacryl S500 HR media in a column with custom dimension (20 cm

in height and 0.25 cm in radius) was developed. The total volume of this column was 4 ml and thus

may allow shorter run time and smaller sample volume. Column efficiency testing was performed at

flow rates ranging between 0.05 ml/min to 0.15 ml/min. Ideal h value should be equal to or less than

3, however, this was not achievable due to unmodifiable external column volume. The As, however,

was within acceptable range and indicated good column efficiency. Based on both h and As, optimum

flow rate was between 0.075 ml/min and 0.l ml/min.

The S500 column was validated with wt VLPs. The chromatogram profile showed a major VLP peak

at 3 ml, and depending on the purity of the samples, a minor aggregate peak appeared at 2 ml. The

122

VLP peak was within acceptable symmetry, close to Gaussian distribution, which not only suggests

the suitability of the method to purify VLPs, but also confirms the efficiency of the column. Light

scattering analysis showed that hydrodynamic radius of the VLP peak was consistent within the 26

nm to 32 nm range, with decreasing radius along increasing elution volume as expected. The validated

method was then applied to analyse and characterise VLPs of VP1:VP2C-GFP, VP1:VP2C-M1-I,

and DDIT1 mosaic particles as summarised in Section 6.1.2 and Section 6.1.3.

The semi-preparative SEC-S500 method to analyse and characterise VLPs by collection of purified

fractions differed from the iSE-UHPLC developed at around the same time by Ladd Effio et al.

(2016). While the iSE-UHPLC enabled chromatographic analysis within minutes, the method was

analytical and does not provide collectable VLP fractions. Another method by Lagoutte et al. (2016)

employed the recently developed Captocore-based column to purify VLPs, however, it was not clear

if it was able to separate aggregates or not. Thus, the semi-preparative SEC-S500 method has several

advantages over those two methods. However, it needs to be optimised further to improve resolution

for the purpose of VLP formulation and administration.

6.2 Future Direction

Awareness of the importance of broadly cross-protective vaccine to limit disease spread during

influenza epidemic and pandemic, has led to the development of candidate vaccines to stimulate

cellular immunity. These candidate vaccines aim to deliver Tc-epitopes from conserved influenza

proteins, and different carriers have been explored for this purpose. These carriers include

lipopeptides, liposomes, virosomes, subunit protein carriers, VLPs, and attenuated viral vectors.

However, these carriers face significant setbacks such as the use of mammalian or insect cells as

expression host which may add to cost and time of production, the lack of industrial scale production

platform, or the administration of potentially infectious material.

My research aims to explore the ability of MPyV VLP to carry influenza Tc-epitopes and address

bioengineering challenges associated with epitope hydrophobicity. The MPyV VLP has an

established and scalable microbial platform that could reduce cost and time of production. The MPyV

VLP is also able to enter dendritic cells of different cell types and enter the MHC Class I pathway

necessary for Tc activation. The MPyV VLP employs the viral coat proteins of MPyV and is thus

non-infectious. The MPyV VLP is also relatively easy to genetically manipulate, facilitating flexible

bioengineering design. This platform has been developed to include external Tc-epitope presentation,

and internal Tc-epitope encapsidation in my research project. The research outcomes collectively

123

show the potential of the MPyV VLP to carry influenza Tc-epitopes. The realisation of the MPyV

VLP as Tc-epitope-based influenza vaccine, however, requires further study in four main aspects:

1) Further development of the Tc-epitope-based candidate influenza vaccine. In my research

project, bioengineering challenges associated with epitope hydrophobicity and encapsidation

of foreign antigenic proteins have been addressed. However, further research needs to be

carried out to evaluate the immunological potential of the candidate vaccines:

a. Immunological testing – The candidate vaccines need to be tested for immunogenicity

and protection in humanised transgenic mice. The humanised transgenic mice should

express HLAs relevant for the Tc-epitope displayed in the candidate vaccine. The

HLA specific for externally displayed Tc-epitopes in the vaccine developed in my

project is the HLA A*0201.

b. Immunological comparison between externally and internally delivered Tc-epitopes –

The immunogenicity and protection between candidate vaccine displaying Tc-

epitopes in surface-exposed loop and candidate vaccine encapsidating multiple Tc-

epitopes in the form of M1 soluble domain need to be compared. This comparison will

give insight into which approach is more effective in raising cellular immunity for

further research. The outcome of this evaluation will not only be applicable for

influenza, but also for other diseases such as cancer and HIV.

c. Mannosylation – VLPs are considered relevant for cellular immunity stimulation due

to their PAMP. Nevertheless, in case MPyV VLP alone only elicit weak cellular

immunity, VLP mannosylation (Al-Barwani et al. 2014) could be explored to enhance

dendritic cell activation.

2) Vaccine bioengineering advancement. Candidate vaccine development using the MPyV VLP

platform could further be advanced with the following approaches:

a. Further investigation of surface-exposed loop insertion with DDIT2 and DDIT3 – As

insertion of a single epitope with flanking charged amino acid in optimised buffer

condition resulted in successful capsomere recovery for IT1, this could be also applied

to other epitopes such as IT2 and IT3. The recovered recombinant capsomeres could

then be co-assembled with wt capsomeres or capsomeres containing other epitopes to

form VLPs displaying different epitopes on the surface. Another option is to co-

express VP1 proteins displaying different individual epitopes each with flanking

charged amino acid to form mosaic capsomeres.

b. Combination of externally and internally presented epitopes - Co-assembly between

capsomeres displaying antigens on the surface and VP1:VP2C capsomere complex

124

with fused foreign protein may be explored. The VLP formed may target both the

humoral and cellular immunity. The surface-exposed antigen could be designed to

target antibody production and the encapsidated antigen could be designed to target

Tc activation.

c. External antigen presentation by N-terminal fusion to VP2 – Pleckaityte et al. (2015)

have shown that foreign protein fusion to the N-terminus of VP2 resulted in external

presentation of the fusion protein due to protrusion of the fusion protein through the

upper opening of the capsomere into external space. This approach could be explored

as it may permit external presentation of larger size proteins instead of just a string of

peptides.

d. Encapsidation of other influenza internal proteins – Other internal proteins of

influenza such as NP, PB1, PB2, and PA are also conserved and contain various Tc-

epitopes of different HLA specificities. Encapsidation of these proteins or their soluble

domains may enhance immunity and broader cross-protection, and is, therefore, worth

to explore further.

3) Encapsidation bioprocess optimisation. A bioengineering milestone had been achieved in my

research by establishing the methodology for in vitro foreign protein encapsidation within

MPyV VLP. However, more improvement is needed in the process development, which

include:

a. Optimisation of expression parameters – This includes IPTG concentration, induction

time, temperature, and agitation speed, tailored specific to the construct at hand.

b. Optimisation of purification of capsomere complex and VLPs to increase yield and

reduce aggregation – For example, protein concentration prior to S500 may play a role

in the aggregate formation and thus alternative ways may be explored to overcome

this problem. Endotoxin removal and subsequent vaccine formulation and

administration would require the removal of aggregates and other contaminants to

achieve a high VLP purity.

c. Simulation and subsequent establishment of large-scale production – The laboratory

scale bioprocess of foreign protein encapsidation differs in several aspects from the

already established approach for external display of foreign antigens. Thus, the foreign

protein encapsidation approach needs to be adjusted for large-scale upstream and

downstream processing.

4) Mixture assembly optimisation. Despite the controlled average GFP content following mixed

assembly and the strong indication that mosaic particles were made with DDIT1 capsomeres

mixed with wt capsomeres, it is still not conclusive whether mosaic particles containing

125

different capsomeres were really formed. The data just goes so far as to say that the ensemble

of VLPs consisted of mixed capsomeres, but not the VLPs themselves. To confirm that mosaic

particles can be formed via a mixture assembly, the following evaluation method may be

performed:

a. Fluorescence resonance energy transfer (FRET) - FRET method utilises the ability of

a pair of fluorophores with overlapping spectra to transfer the resonance energy from

the donor to the acceptor fluorophore when they are within close proximity. FRET

occurs when the emission wavelength of the donor fluorophore sufficiently overlaps

with the excitation spectra of the acceptor fluorophore such that the acceptor is excited

by light emitted from the donor and it can be detected by either increased emission by

the receptor or reduced emission by the donor, depending on the extent of spectral

separation between the fluorophores. The energy transfer occurs over a certain

distance known as the Förster radius, usually from 1 to 10 nm, thus it is frequently

used to detect protein-protein interactions (Aoki et al. 2013). If a mixture of

VP1:VP2C-GFP and VP1:VP2C-mRuby capsomere complexes (Bajar et al. 2016) are

co-assembled, and recovered VLPs exhibit FRET, then the close proximity of the

different encapsidated fluorescent proteins, which would reflect successful mosaic

VLP formation via mixture assembly, can be confirmed.

5) Delivery vehicle beyond influenza vaccine candidate. The approaches used in my research, as

well as methods to validate them could be utilised beyond influenza vaccine development,

such as:

a. Application for cancer and HIV candidate vaccines - Tc-epitope carriage by the MPyV

VLP platform both externally and internally developed in my research may be applied

to deliver antigens for other pathogens or non-pathogenic diseases such as HIV and

cancer.

b. Delivery of other types of proteins – The encapsidation methodology may be

implemented to deliver other proteins such as enzymes, drugs, imaging tool,

intracellular analysis agents, or theranostic entities. Encapsidation may be required to

maintain intact protein before entry into the target cell. The MPyV VLP pathway

inside target cells needs to be further elucidated to provide beneficial information for

more robust engineering of the MPyV VLP platform.

6.3 Concluding Remarks

126

The advancement of immuno-informatics does not only allow the prediction of Tc-epitopes of

different proteins but also provides a database of identified Tc-epitopes along with information about

their conservancy and tested immunogenicity. Recent progress in rapid genomic and proteomic

sequencing of emergent pathogenic strains aids in reverse-genetics to apply recombinant technology

in candidate vaccine bioengineering. Both immuno-informatics and reverse-genetics support the

development of next-generation vaccines based on modular synthetic VLPs to stimulate the cellular

immunity. Extensive research on MPyV VLPs results in a wealth of information that allows the

establishment of a flexible vaccine platform. Microbial production with this platform allows faster

and cheaper production and promotes vaccine affordability and timely availability in developed and

developing countries.

In my research, flexible protein design and bioprocess modification of the established MPyV VLP

platform was implemented for internal and external carriage of hydrophobic influenza antigens. This

resulted in two candidate vaccines, one which presents influenza Tc-epitope on the surface of the

MPyV VLP, and the other which encapsidates multiple overlapping Tc-epitopes in the form of protein

soluble domain inside the VLP. However, the candidate vaccines still require immunological

evaluation. Nevertheless, the high-throughput multiparametric buffer screen developed here could be

applied to optimise buffer for other hydrophobic modularised antigens. Furthermore, the

encapsidation methodology provides an adjustable molecular design and an adjustable bioprocess to

encapsidate antigens of other pathogens, and also of other protein tools beyond vaccination purposes.

Further development in the rational design of modular synthetic VLPs and their immunological

evaluation may draw us closer to the realisation of timely, safe, affordable, and effective influenza

vaccines.

127

List of References

Abbing A, Blaschke UK, Grein S, Kretschmar M, Stark CM, Thies MJ, Walter J, Weigand M, Woith DC, Hess J and others. 2004. Efficient intracellular delivery of a protein and a low molecular weight substance via recombinant polyomavirus-like particles. J Biol Chem 279(26):27410-21.

Abelin A, Colegate T, Gardner S, Hehme N, Palache A. 2011. Lessons from pandemic influenza A(H1N1): The research-based vaccine industry's perspective. Vaccine 29(6):1135-1138.

Abidin RS. 2010. The use of immuno-informatics to identify cytotoxic T cell epitopes from influenza virus to be used as vaccines [Master]. Canberra, Australia: The Australian National University.

Abidin RS, Lua LH, Middelberg AP, Sainsbury F. 2015. Insert engineering and solubility screening improves recovery of virus-like particle subunits displaying hydrophobic epitopes. Protein Sci 24(11):1820-8.

Adar Y, Singer Y, Levi R, Tzehoval E, Perk S, Banet-Noach C, Nagar S, Arnon R, Ben-Yedidia T. 2009. A universal epitope-based influenza vaccine and its efficacy against H5N1. Vaccine 27(15):2099-107.

Al-Barwani F, Young SL, Baird MA, Larsen DS, Ward VK. 2014. Mannosylation of virus-like particles enhances internalization by antigen presenting cells. PLoS One 9(8):e104523.

Alexander J, Bilsel P, del Guercio MF, Marinkovic-Petrovic A, Southwood S, Stewart S, Ishioka G, Kotturi MF, Botten J, Sidney J and others. 2010. Identification of broad binding class I HLA supertype epitopes to provide universal coverage of influenza A virus. Hum Immunol 71(5):468-74.

Anggraeni MR. 2014. Engineering of virus-like particles for alternative vaccine candidate targeting hypervariable peptide antigen-element [PhD]. Brisbane, Australia: The University of Queensland.

Anggraeni MR, Connors NK, Wu Y, Chuan YP, Lua LH, Middelberg AP. 2013. Sensitivity of immune response quality to influenza helix 190 antigen structure displayed on a modular virus-like particle. Vaccine 31(40):4428-35.

Antrobus RD, Berthoud TK, Mullarkey CE, Hoschler K, Coughlan L, Zambon M, Hill AV, Gilbert SC. 2014. Coadministration of seasonal influenza vaccine and MVA-NP+M1 simultaneously achieves potent humoral and cell-mediated responses. Mol Ther 22(1):233-8.

Antrobus RD, Lillie PJ, Berthoud TK, Spencer AJ, McLaren JE, Ladell K, Lambe T, Milicic A, Price DA, Hill AV and others. 2012. A T cell-inducing influenza vaccine for the elderly: safety and immunogenicity of MVA-NP+M1 in adults aged over 50 years. PLoS One 7(10):e48322.

Aoki K, Kamioka Y, Matsuda M. 2013. Fluorescence resonance energy transfer imaging of cell signaling from in vitro to in vivo: basis of biosensor construction, live imaging, and image processing. Dev Growth Differ 55(4):515-22.

Arakawa T, Tsumoto K, Kita Y, Chang B, Ejima D. 2007. Biotechnology applications of amino acids in protein purification and formulations. Amino Acids 33(4):587-605.

Arkema A, Huckriede A, Schoen P, Wilschut J, Daemen T. 2000. Induction of cytotoxic T lymphocyte activity by fusion-active peptide-containing virosomes. Vaccine 18(14):1327-33.

Assarsson E, Bui HH, Sidney J, Zhang Q, Glenn J, Oseroff C, Mbawuike IN, Alexander J, Newman MJ, Grey H and others. 2008. Immunomic analysis of the repertoire of T-cell specificities for influenza A virus in humans. J Virol 82(24):12241-51.

Babin C, Majeau N, Leclerc D. 2013. Engineering of papaya mosaic virus (PapMV) nanoparticles with a CTL epitope derived from influenza NP. J Nanobiotechnology 11:10.

Babon JA, Cruz J, Orphin L, Pazoles P, Co MD, Ennis FA, Terajima M. 2009. Genome-wide screening of human T-cell epitopes in influenza A virus reveals a broad spectrum of CD4(+)

128

T-cell responses to internal proteins, hemagglutinins, and neuraminidases. Hum Immunol 70(9):711-21.

Bajar BT, Wang ES, Lam AJ, Kim BB, Jacobs CL, Howe ES, Davidson MW, Lin MZ, Chu J. 2016. Improving brightness and photostability of green and red fluorescent proteins for live cell imaging and FRET reporting. Sci Rep 6:20889.

Barouch DH, Harrison SC. 1994. Interactions among the major and minor coat proteins of polyomavirus. J Virol 68(6):3982-9.

Bayer ME, Blumberg BS, Werner B. 1968. Particles associated with Australia antigen in the sera of patients with leukaemia, Down's Syndrome and hepatitis. Nature 218(5146):1057-9.

Benjamin TL. 2001. Polyoma virus: old findings and new challenges. Virology 289(2):167-73. Berthoud TK, Hamill M, Lillie PJ, Hwenda L, Collins KA, Ewer KJ, Milicic A, Poyntz HC, Lambe

T, Fletcher HA and others. 2011. Potent CD8+ T-cell immunogenicity in humans of a novel heterosubtypic influenza A vaccine, MVA-NP+M1. Clin Infect Dis 52(1):1-7.

Billaud JN, Peterson D, Barr M, Chen A, Sallberg M, Garduno F, Goldstein P, McDowell W, Hughes J, Jones J and others. 2005. Combinatorial approach to hepadnavirus-like particle vaccine design. J Virol 79(21):13656-66.

Blumberg BS, Alter HJ, Visnich S. 1965. A "New" Antigen in Leukemia Sera. JAMA 191:541-6. Boura E, Liebl D, Spisek R, Fric J, Marek M, Stokrova J, Holan V, Forstova J. 2005. Polyomavirus

EGFP-pseudocapsids: analysis of model particles for introduction of proteins and peptides into mammalian cells. FEBS Lett 579(29):6549-58.

Braithwaite A, Smith FJ. 1999. Theoretical considerations. Chromatographic Methods. Dordrecht: Springer Netherlands. p 17-43.^

Braun H, Boller K, Lower J, Bertling WM, Zimmer A. 1999. Oligonucleotide and plasmid DNA packaging into polyoma VP1 virus-like particles expressed in Escherichia coli. Biotechnol Appl Biochem 29 ( Pt 1):31-43.

Bray BL. 2003. Large-scale manufacture of peptide therapeutics by chemical synthesis. Nat Rev Drug Discov 2(7):587-93.

Broadbent AJ, Subbarao K. 2011. Influenza virus vaccines: lessons from the 2009 H1N1 pandemic. Curr Opin Virol 1(4):254-62.

Brown LE, Kelso A. 2009. Prospects for an influenza vaccine that induces cross-protective cytotoxic T lymphocytes. Immunol Cell Biol 87(4):300-8.

Brown SD, Fiedler JD, Finn MG. 2009. Assembly of hybrid bacteriophage Qbeta virus-like particles. Biochemistry 48(47):11155-7.

Buch MH, Liaci AM, O'Hara SD, Garcea RL, Neu U, Stehle T. 2015. Structural and Functional Analysis of Murine Polyomavirus Capsid Proteins Establish the Determinants of Ligand Recognition and Pathogenicity. PLoS Pathog 11(10):e1005104.

Buckland BC. 2015. The development and manufacture of influenza vaccines. Hum Vaccin Immunother 11(6):1357-60.

Bui HH, Peters B, Assarsson E, Mbawuike I, Sette A. 2007. Ab and T cell epitopes of influenza A virus, knowledge and opportunities. Proc Natl Acad Sci U S A 104(1):246-51.

Caparros-Wanderley W, Clark B, Griffin BE. 2004. Effect of dose and long-term storage on the immunogenicity of murine polyomavirus VP1 virus-like particles. Vaccine 22(3-4):352-61.

Catrice EV, Sainsbury F. 2015. Assembly and Purification of Polyomavirus-Like Particles from Plants. Mol Biotechnol 57(10):904-13.

Chen XS, Stehle T, Harrison SC. 1998. Interaction of polyomavirus internal protein VP2 with the major capsid protein VP1 and implications for participation of VP2 in viral entry. EMBO J 17(12):3233-40.

Cheong WS, Reiseger J, Turner SJ, Boyd R, Netter HJ. 2009. Chimeric virus-like particles for the delivery of an inserted conserved influenza A-specific CTL epitope. Antiviral Res 81(2):113-22.

129

Chi EY, Krishnan S, Randolph TW, Carpenter JF. 2003. Physical stability of proteins in aqueous solution: mechanism and driving forces in nonnative protein aggregation. Pharm Res 20(9):1325-36.

Chong SH, Ham S. 2014a. Interaction with the surrounding water plays a key role in determining the aggregation propensity of proteins. Angew Chem Int Ed Engl 53(15):3961-4.

Chong SH, Ham S. 2014b. Site-directed analysis on protein hydrophobicity. J Comput Chem 35(18):1364-70.

Chromy LR, Pipas JM, Garcea RL. 2003. Chaperone-mediated in vitro assembly of Polyomavirus capsids. Proc Natl Acad Sci U S A 100(18):10477-82.

Chu B. 2013. Laser Light Scattering: Basic Principles and Practice. Mineola, NY: Dover Publication.

Chuan YP, Fan YY, Lua L, Middelberg AP. 2008a. Quantitative analysis of virus-like particle size and distribution by field-flow fractionation. Biotechnol Bioeng 99(6):1425-33.

Chuan YP, Fan YY, Lua LH, Middelberg AP. 2010. Virus assembly occurs following a pH- or Ca2+-triggered switch in the thermodynamic attraction between structural protein capsomeres. J R Soc Interface 7(44):409-21.

Chuan YP, Lua LH, Middelberg AP. 2008b. High-level expression of soluble viral structural protein in Escherichia coli. J Biotechnol 134(1-2):64-71.

Chuan YP, Rivera-Hernandez T, Wibowo N, Connors NK, Wu Y, Hughes FK, Lua LH, Middelberg AP. 2013. Effects of pre-existing anti-carrier immunity and antigenic element multiplicity on efficacy of a modular virus-like particle vaccine. Biotechnol Bioeng 110(9):2343-51.

Chuan YP, Wibowo N, Lua LHL, Middelberg APJ. 2014. The economics of virus-like particle and capsomere vaccines. Biochemical Engineering Journal 90:255.

Cohn EJ. 1920. The Relation Between the Iso-Electric Point of a Globulin and Its Solubility and Acid Combining Capacity in Salt Solution. Proc Natl Acad Sci U S A 6(5):256-63.

Collin N, de Radigues X, Task WHOHNV. 2009. Vaccine production capacity for seasonal and pandemic (H1N1) 2009 influenza. Vaccine 27(38):5184-5186.

Coughlan L, Mullarkey C, Gilbert S. 2015. Adenoviral vectors as novel vaccines for influenza. J Pharm Pharmacol 67(3):382-99.

Cox MM, Hollister JR. 2009. FluBlok, a next generation influenza vaccine manufactured in insect cells. Biologicals 37(3):182-9.

Cromwell ME, Hilario E, Jacobson F. 2006. Protein aggregation and bioprocessing. AAPS J 8(3):E572-9.

Davis XM, Hay KA, Plier DA, Chaves SS, Lim PL, Caumes E, Castelli F, Kozarsky PE, Cetron MS, Freedman DO and others. 2013. International travelers as sentinels for sustained influenza transmission during the 2009 influenza A(H1N1)pdm09 pandemic. J Travel Med 20(3):177-84.

Dawood FS, Iuliano AD, Reed C, Meltzer MI, Shay DK, Cheng PY, Bandaranayake D, Breiman RF, Brooks WA, Buchy P and others. 2012. Estimated global mortality associated with the first 12 months of 2009 pandemic influenza A H1N1 virus circulation: a modelling study. Lancet Infect Dis 12(9):687-95.

Day EB, Zeng W, Doherty PC, Jackson DC, Kedzierska K, Turner SJ. 2007. The context of epitope presentation can influence functional quality of recalled influenza A virus-specific memory CD8+ T cells. J Immunol 179(4):2187-94.

Degano P, Schneider J, Hannan CM, Gilbert SC, Hill AV. 1999. Gene gun intradermal DNA immunization followed by boosting with modified vaccinia virus Ankara: enhanced CD8+ T cell immunogenicity and protective efficacy in the influenza and malaria models. Vaccine 18(7-8):623-32.

Del Giudice G, Rappuoli R. 2015. Inactivated and adjuvanted influenza vaccines. Curr Top Microbiol Immunol 386:151-80.

130

Deliyannis G, Kedzierska K, Lau YF, Zeng W, Turner SJ, Jackson DC, Brown LE. 2006. Intranasal lipopeptide primes lung-resident memory CD8+ T cells for long-term pulmonary protection against influenza. Eur J Immunol 36(3):770-8.

Denis J, Acosta-Ramirez E, Zhao Y, Hamelin ME, Koukavica I, Baz M, Abed Y, Savard C, Pare C, Lopez Macias C and others. 2008. Development of a universal influenza A vaccine based on the M2e peptide fused to the papaya mosaic virus (PapMV) vaccine platform. Vaccine 26(27-28):3395-403.

Dicker RC, Coronado F, Koo D, Gibson Parrish R. 2012. Principles of Epidemiology in Public Health Practice: An Introduction to Applied Epidemiology and Biostatistics. Atlanta, GA: CDC.

Dormitzer PR. 2015. Rapid production of synthetic influenza vaccines. Curr Top Microbiol Immunol 386:237-73.

Eriksson M, Andreasson K, Weidmann J, Lundberg K, Tegerstedt K, Dalianis T, Ramqvist T. 2011. Murine polyomavirus virus-like particles carrying full-length human PSA protect BALB/c mice from outgrowth of a PSA expressing tumor. PLoS One 6(8):e23828.

Esfandiary R, Middaugh CR. 2012. Ultraviolet Absorption Spectroscopy. In: Jiskoot W, Mahler H, editors. Analysis of Aggregates and Particles in Protein Pharmaceuticals: John Wiley & Sons, Inc. p 169-200.^

FDA. 2013. FDA approves new seasonal influenza vaccine made using novel technology. US Food and Drug Administrator. http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm335891.htm, Last Updated: 16/01/2013, Date Accessed: 01/02/2015

Fedson DS. 2009. Meeting the challenge of influenza pandemic preparedness in developing countries. Emerg Infect Dis 15(3):365-71.

Fink AL. 1998. Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold Des 3(1):R9-23.

Fligge C, Giroglou T, Streeck RE, Sapp M. 2001. Induction of type-specific neutralizing antibodies by capsomeres of human papillomavirus type 33. Virology 283(2):353-7.

Fredrickson EK, Gallagher PS, Clowes Candadai SV, Gardner RG. 2013. Substrate recognition in nuclear protein quality control degradation is governed by exposed hydrophobicity that correlates with aggregation and insolubility. J Biol Chem 288(9):6130-9.

Fric J, Marek M, Hruskova V, Holan V, Forstova J. 2008. Cellular and humoral immune responses to chimeric EGFP-pseudocapsids derived from the mouse polyomavirus after their intranasal administration. Vaccine 26(26):3242-51.

Garcea RL, Salunke DM, Caspar DL. 1987. Site-directed mutation affecting polyomavirus capsid self-assembly in vitro. Nature 329(6134):86-7.

Gasteiger G, Kastenmuller W, Ljapoci R, Sutter G, Drexler I. 2007. Cross-priming of cytotoxic T cells dictates antigen requisites for modified vaccinia virus Ankara vector vaccines. J Virol 81(21):11925-36.

GE-Healthcare. 2010a. Column efficiency testing. Uppsala: General Electric Company. GE-Healthcare. 2010b. Gel filtration: Principles and Methods. Sweden: GE Healthcare Biosciences. Gedvilaite A, Dorn DC, Sasnauskas K, Pecher G, Bulavaite A, Lawatscheck R, Staniulis J, Dalianis

T, Ramqvist T, Schonrich G and others. 2006. Virus-like particles derived from major capsid protein VP1 of different polyomaviruses differ in their ability to induce maturation in human dendritic cells. Virology 354(2):252-60.

Gedvilaite A, Frommel C, Sasnauskas K, Micheel B, Ozel M, Behrsing O, Staniulis J, Jandrig B, Scherneck S, Ulrich R. 2000. Formation of immunogenic virus-like particles by inserting epitopes into surface-exposed regions of hamster polyomavirus major capsid protein. Virology 273(1):21-35.

Gilbert SC. 2012. T-cell-inducing vaccines - what's the future. Immunology 135(1):19-26. Girard MP, Tam JS, Assossou OM, Kieny MP. 2010. The 2009 A (H1N1) influenza virus pandemic:

A review. Vaccine 28(31):4895-4902.

131

Glasgow J, Tullman-Ercek D. 2014. Production and applications of engineered viral capsids. Appl Microbiol Biotechnol 98(13):5847-58.

Gomila RC, Suphaphiphat P, Judge C, Spencer T, Ferrari A, Wen Y, Palladino G, Dormitzer PR, Mason PW. 2013. Improving influenza virus backbones by including terminal regions of MDCK-adapted strains on hemagglutinin and neuraminidase gene segments. Vaccine 31(42):4736-43.

Gordon DM, McGovern TW, Krzych U, Cohen JC, Schneider I, LaChance R, Heppner DG, Yuan G, Hollingdale M, Slaoui M and others. 1995. Safety, immunogenicity, and efficacy of a recombinantly produced Plasmodium falciparum circumsporozoite protein-hepatitis B surface antigen subunit vaccine. J Infect Dis 171(6):1576-85.

Gorse GJ, Falsey AR, Ozol-Godfrey A, Landolfi V, Tsang PH. 2015. Safety and immunogenicity of a quadrivalent intradermal influenza vaccine in adults. Vaccine 33(9):1151-9.

Hagel L. 2011. Gel filtration: size exclusion chromatography. Methods Biochem Anal 54:51-91. Hagel L, Janson J-C. 1992. Chapter 6 Size-exclusion chromatography. In: Heftmann E, editor.

Journal of Chromatography Library: Elsevier. p A267-A307.^ Hagensee ME, Yaegashi N, Galloway DA. 1993. Self-assembly of human papillomavirus type 1

capsids by expression of the L1 protein alone or by coexpression of the L1 and L2 capsid proteins. J Virol 67(1):315-22.

Hamilton RJ, Sewell PA. 1982. Chromatographic theory. In: Hamilton RJ, Sewell PA, editors. Introduction to high performance liquid chromatography. Dordrecht: Springer Netherlands. p 13-41.^

Hardy E, Martinez E, Diago D, Diaz R, Gonzalez D, Herrera L. 2000. Large-scale production of recombinant hepatitis B surface antigen from Pichia pastoris. J Biotechnol 77(2-3):157-67.

Harty JT, Badovinac VP. 2008. Shaping and reshaping CD8+ T-cell memory. Nat Rev Immunol 8(2):107-19.

Heiny AT, Miotto O, Srinivasan KN, Khan AM, Zhang GL, Brusic V, Tan TW, August JT. 2007. Evolutionarily conserved protein sequences of influenza a viruses, avian and human, as vaccine targets. PLoS One 2(11):e1190.

Herfst S, Imai M, Kawaoka Y, Fouchier RA. 2014. Avian influenza virus transmission to mammals. Curr Top Microbiol Immunol 385:137-55.

Ho JG, Middelberg AP, Ramage P, Kocher HP. 2003. The likelihood of aggregation during protein renaturation can be assessed using the second virial coefficient. Protein Sci 12(4):708-16.

Holland J, Spindler K, Horodyski F, Grabau E, Nichol S, VandePol S. 1982. Rapid evolution of RNA genomes. Science 215(4540):1577-85.

Houser K, Subbarao K. 2015. Influenza vaccines: challenges and solutions. Cell Host Microbe 17(3):295-300.

Ichihashi T, Yoshida R, Sugimoto C, Takada A, Kajino K. 2011. Cross-protective peptide vaccine against influenza A viruses developed in HLA-A*2402 human immunity model. PLoS One 6(9):e24626.

ICTV. 2012. Family - Polyomaviridae. Virus Taxonomy: Classification and Nomenclature of Viruses, Ninth Report of the International Committee on Taxonomy of Viruses. San Diego: Elsevier. p 279-290.^

Inoue T, Kawano MA, Takahashi RU, Tsukamoto H, Enomoto T, Imai T, Kataoka K, Handa H. 2008. Engineering of SV40-based nano-capsules for delivery of heterologous proteins as fusions with the minor capsid proteins VP2/3. J Biotechnol 134(1-2):181-92.

Ishizuka J, Grebe K, Shenderov E, Peters B, Chen Q, Peng Y, Wang L, Dong T, Pasquetto V, Oseroff C and others. 2009. Quantitating T cell cross-reactivity for unrelated peptide antigens. J Immunol 183(7):4337-45.

James ER, van Zyl WH, Gorgens JF. 2007. Increased Hepatitis B surface antigen production by recombinant Aspergillus niger through the optimization of agitation and dissolved oxygen concentration. Appl Microbiol Biotechnol 75(2):279-88.

132

Jegerlehner A, Tissot A, Lechner F, Sebbel P, Erdmann I, Kundig T, Bachi T, Storni T, Jennings G, Pumpens P and others. 2002. A molecular assembly system that renders antigens of choice highly repetitive for induction of protective B cell responses. Vaccine 20(25-26):3104-12.

Jennings GT, Bachmann MF. 2008. The coming of age of virus-like particle vaccines. Biol Chem 389(5):521-36.

Jin H, Subbarao K. 2015. Live attenuated influenza vaccine. Curr Top Microbiol Immunol 386:181-204.

Josefsberg JO, Buckland B. 2012. Vaccine process technology. Biotechnol Bioeng 109(6):1443-60. Kawano M, Morikawa K, Suda T, Ohno N, Matsushita S, Akatsuka T, Handa H, Matsui M. 2014.

Chimeric SV40 virus-like particles induce specific cytotoxicity and protective immunity against influenza A virus without the need of adjuvants. Virology 448:159-67.

Keogh-Brown MR, Smith RD, Edmunds JW, Beutels P. 2010. The macroeconomic impact of pandemic influenza: estimates from models of the United Kingdom, France, Belgium and The Netherlands. Eur J Health Econ 11(6):543-54.

Kirnbauer R, Booy F, Cheng N, Lowy DR, Schiller JT. 1992. Papillomavirus L1 major capsid protein self-assembles into virus-like particles that are highly immunogenic. Proc Natl Acad Sci U S A 89(24):12180-4.

Kistner O, Barrett PN, Mundt W, Reiter M, Schober-Bendixen S, Dorner F. 1998. Development of a mammalian cell (Vero) derived candidate influenza virus vaccine. Vaccine 16(9-10):960-8.

Koppel DE. 1972. Analysis of Macromolecular Polydispersity in Intensity Correlation Spectroscopy: The Method of Cumulants. J. Chem. Phys 57(11):4814-4820.

Korteweg C, Gu J. 2008. Pathology, molecular biology, and pathogenesis of avian influenza A (H5N1) infection in humans. Am J Pathol 172(5):1155-70.

Kosukegawa A, Arisaka F, Takayama M, Yajima H, Kaidow A, Handa H. 1996. Purification and characterization of virus-like particles and pentamers produced by the expression of SV40 capsid proteins in insect cells. Biochim Biophys Acta 1290(1):37-45.

Krammer F, Palese P. 2015. Advances in the development of influenza virus vaccines. Nat Rev Drug Discov 14(3):167-82.

Kushnir N, Streatfield SJ, Yusibov V. 2012. Virus-like particles as a highly efficient vaccine platform: diversity of targets and production systems and advances in clinical development. Vaccine 31(1):58-83.

Ladd Effio C, Oelmeier SA, Hubbuch J. 2016. High-throughput characterization of virus-like particles by interlaced size-exclusion chromatography. Vaccine 34(10):1259-67.

Lagoutte P, Mignon C, Donnat S, Stadthagen G, Mast J, Sodoyer R, Lugari A, Werle B. 2016. Scalable chromatography-based purification of virus-like particle carrier for epitope based influenza A vaccine produced in Escherichia coli. J Virol Methods 232:8-11.

Lamb RE, Krug RM. 2001. Orthomyxoviridae: The viruses and their replication. In: Knipe DM, Howley PM, Griffin DE, Martin MA, Lamb RA, Roizman B, Straus SE, editors. Fields Virology. Philadephia: Lippincot William & Wilkins.^

Lambe T, Carey JB, Li Y, Spencer AJ, van Laarhoven A, Mullarkey CE, Vrdoljak A, Moore AC, Gilbert SC. 2013. Immunity against heterosubtypic influenza virus induced by adenovirus and MVA expressing nucleoprotein and matrix protein-1. Sci Rep 3:1443.

Lambert LC, Fauci AS. 2010. Influenza vaccines for the future. N Engl J Med 363(21):2036-44. Lanckriet H, Middelberg AP. 2004. Operational regimes for a simplified one-step artificial chaperone

refolding method. Biotechnol Prog 20(6):1861-7. Leavitt AD, Roberts TM, Garcea RL. 1985. Polyoma virus major capsid protein, VP1. Purification

after high level expression in Escherichia coli. J Biol Chem 260(23):12803-9. Lee LY, Ha do LA, Simmons C, de Jong MD, Chau NV, Schumacher R, Peng YC, McMichael AJ,

Farrar JJ, Smith GL and others. 2008. Memory T cells established by seasonal human influenza A infection cross-react with avian influenza A (H5N1) in healthy individuals. J Clin Invest 118(10):3478-90.

133

Leibly DJ, Nguyen TN, Kao LT, Hewitt SN, Barrett LK, Van Voorhis WC. 2012. Stabilizing additives added during cell lysis aid in the solubilization of recombinant proteins. PLoS One 7(12):e52482.

Li H, Le Brun AP, Agyei D, Shen W, Middelberg AP, He L. 2016. Stabilizing and destabilizing protein surfactant-based foams in the presence of a chemical surfactant: Effect of adsorption kinetics. J Colloid Interface Sci 462:56-63.

Li J, Garg M, Shah D, Rajagopalan R. 2010. Solubilization of aromatic and hydrophobic moieties by arginine in aqueous solutions. J Chem Phys 133(5):054902.

Li M, Cripe TP, Estes PA, Lyon MK, Rose RC, Garcea RL. 1997. Expression of the human papillomavirus type 11 L1 capsid protein in Escherichia coli: characterization of protein domains involved in DNA binding and capsid assembly. J Virol 71(4):2988-95.

Liddington RC, Yan Y, Moulai J, Sahli R, Benjamin TL, Harrison SC. 1991. Structure of simian virus 40 at 3.8-A resolution. Nature 354(6351):278-84.

Liebl D, Difato F, Hornikova L, Mannova P, Stokrova J, Forstova J. 2006. Mouse polyomavirus enters early endosomes, requires their acidic pH for productive infection, and meets transferrin cargo in Rab11-positive endosomes. J Virol 80(9):4610-22.

Liew MW, Rajendran A, Middelberg AP. 2010. Microbial production of virus-like particle vaccine protein at gram-per-litre levels. J Biotechnol 150(2):224-31.

Liew MWO, Chuan YP, Middelberg APJ. 2012a. High-yield and scalable cell-free assembly of virus-like particles by dilution. Biochemical Engineering Journal 67:88-96.

Liew MWO, Chuan YP, Middelberg APJ. 2012b. Reactive diafiltration for assembly and formulation of virus-like particles. Biochemical Engineering Journal 68:120-128.

Lipin DI, Chuan YP, Lua LH, Middelberg AP. 2008a. Encapsulation of DNA and non-viral protein changes the structure of murine polyomavirus virus-like particles. Arch Virol 153(11):2027-39.

Lipin DI, Lua LH, Middelberg AP. 2008b. Quaternary size distribution of soluble aggregates of glutathione-S-transferase-purified viral protein as determined by asymmetrical flow field flow fractionation and dynamic light scattering. J Chromatogr A 1190(1-2):204-14.

Lipin DI, Raj A, Lua LH, Middelberg AP. 2009. Affinity purification of viral protein having heterogeneous quaternary structure: modeling the impact of soluble aggregates on chromatographic performance. J Chromatogr A 1216(30):5696-708.

Liu D, Liu X, Yan J, Liu WJ, Gao GF. 2009. Interspecies transmission and host restriction of avian H5N1 influenza virus. Sci China C Life Sci 52(5):428-38.

Liu J, Wu B, Zhang S, Tan S, Sun Y, Chen Z, Qin Y, Sun M, Shi G, Wu Y and others. 2013. Conserved epitopes dominate cross-CD8+ T-cell responses against influenza A H1N1 virus among Asian populations. Eur J Immunol 43(8):2055-69.

Livingston BD, Newman M, Crimi C, McKinney D, Chesnut R, Sette A. 2001. Optimization of epitope processing enhances immunogenicity of multiepitope DNA vaccines. Vaccine 19(32):4652-60.

Lua LH, Connors NK, Sainsbury F, Chuan YP, Wibowo N, Middelberg AP. 2014. Bioengineering virus-like particles as vaccines. Biotechnol Bioeng 111(3):425-40.

Lund O, Nielsen M, Kesmir C, Petersen AG, Lundegaard C, Worning P, Sylvester-Hvid C, Lamberth K, Roder G, Justesen S and others. 2004. Definition of supertypes for HLA molecules using clustering of specificity matrices. Immunogenetics 55(12):797-810.

Malikova NP, Visser NV, van Hoek A, Skakun VV, Vysotski ES, Lee J, Visser AJ. 2011. Green-fluorescent protein from the bioluminescent jellyfish Clytia gregaria is an obligate dimer and does not form a stable complex with the Ca(2+)-discharged photoprotein clytin. Biochemistry 50(20):4232-41.

Mamun MM, Huda AK. 2011. Origins and evolutionary genomics of the novel swine-origin influenza A (H1N1) virus in humans--past and present perspectives. Yakugaku Zasshi 131(4):553-62.

Mardanova ES, Kotlyarov RY, Kuprianov VV, Stepanova LA, Tsybalova LM, Lomonosoff GP, Ravin NV. 2015. Rapid high-yield expression of a candidate influenza vaccine based on the

134

ectodomain of M2 protein linked to flagellin in plants using viral vectors. BMC Biotechnol 15:42.

Matrosovich MN, Matrosovich TY, Gray T, Roberts NA, Klenk HD. 2004. Human and avian influenza viruses target different cell types in cultures of human airway epithelium. Proc Natl Acad Sci U S A 101(13):4620-4.

Matsui M, Kohyama S, Suda T, Yokoyama S, Mori M, Kobayashi A, Taneichi M, Uchida T. 2010. A CTL-based liposomal vaccine capable of inducing protection against heterosubtypic influenza viruses in HLA-A*0201 transgenic mice. Biochem Biophys Res Commun 391(3):1494-9.

McAuley JL, Kedzierska K, Brown LE, Shanks GD. 2015. Host Immunological Factors Enhancing Mortality of Young Adults during the 1918 Influenza Pandemic. Front Immunol 6:419.

Melchjorsen J. 2013. Learning from the messengers: innate sensing of viruses and cytokine regulation of immunity - clues for treatments and vaccines. Viruses 5(2):470-527.

Middelberg AP, Rivera-Hernandez T, Wibowo N, Lua LH, Fan Y, Magor G, Chang C, Chuan YP, Good MF, Batzloff MR. 2011. A microbial platform for rapid and low-cost virus-like particle and capsomere vaccines. Vaccine 29(41):7154-62.

Milian E, Kamen AA. 2015. Current and emerging cell culture manufacturing technologies for influenza vaccines. Biomed Res Int 2015:504831.

Minten IJ, Nolte RJ, Cornelissen JJ. 2010. Complex assembly behavior during the encapsulation of green fluorescent protein analogs in virus derived protein capsules. Macromol Biosci 10(5):539-45.

Mitic NS, Pavlovic MD, Jandrlic DR. 2014. Epitope distribution in ordered and disordered protein regions - part A. T-cell epitope frequency, affinity and hydropathy. J Immunol Methods 406:83-103.

Mohr J, Chuan YP, Wu Y, Lua LH, Middelberg AP. 2013. Virus-like particle formulation optimization by miniaturized high-throughput screening. Methods 60(3):248-56.

Montross L, Watkins S, Moreland RB, Mamon H, Caspar DL, Garcea RL. 1991. Nuclear assembly of polyomavirus capsids in insect cells expressing the major capsid protein VP1. J Virol 65(9):4991-8.

Moser C, Amacker M, Zurbriggen R. 2011. Influenza virosomes as a vaccine adjuvant and carrier system. Expert Rev Vaccines 10(4):437-46.

Murphy K, Travers P, Walport M, Janeway C. 2008. Janeway's Immunobiology. New York Garland Science.

Neatherlin J, Cramer EH, Dubray C, Marienau KJ, Russell M, Sun H, Whaley M, Hancock K, Duong KK, Kirking HL and others. 2013. Influenza A(H1N1)pdm09 during air travel. Travel Med Infect Dis 11(2):110-8.

Neumann G, Chen H, Gao GF, Shu Y, Kawaoka Y. 2010. H5N1 influenza viruses: outbreaks and biological properties. Cell Res 20(1):51-61.

Nolz JC, Harty JT. 2011a. Protective capacity of memory CD8+ T cells is dictated by antigen exposure history and nature of the infection. Immunity 34(5):781-93.

Nolz JC, Harty JT. 2011b. Strategies and implications for prime-boost vaccination to generate memory CD8 T cells. Adv Exp Med Biol 780:69-83.

Norkin LC. 2010. Virology: Molecular Biology and Pathogenesis. Washington DC: ASM Press. O'Fagain C, Cummins PM, O'Connor BF. 2011. Gel-filtration chromatography. Methods Mol Biol

681:25-33. Otieno L, Oneko M, Otieno W, Abuodha J, Owino E, Odero C, Mendoza YG, Andagalu B, Awino

N, Ivinson K and others. 2016. Safety and immunogenicity of RTS,S/AS01 malaria vaccine in infants and children with WHO stage 1 or 2 HIV disease: a randomised, double-blind, controlled trial. Lancet Infect Dis 16(10):1134-44.

Pagliusi S, Leite LC, Datla M, Makhoana M, Gao Y, Suhardono M, Jadhav S, Harshavardhan GV, Homma A. 2013a. Developing Countries Vaccine Manufacturers Network: doing good by making high-quality vaccines affordable for all. Vaccine 31 Suppl 2:B176-83.

135

Pagliusi S, Makhoana M, Datla M, Leite L, Hendriks J, Gholami A, Huang W, Gao Y, Jadhav S, Harshavardhan GV and others. 2013b. Developing Countries Vaccine Manufacturers Network (DCVMN): engaging to step up for vaccine discovery and access. Meeting report 2012. Vaccine 31(31):3111-5.

Parida R, Shaila MS, Mukherjee S, Chandra NR, Nayak R. 2007. Computational analysis of proteome of H5N1 avian influenza virus to define T cell epitopes with vaccine potential. Vaccine 25(43):7530-9.

Pattenden LK, Middelberg AP, Niebert M, Lipin DI. 2005. Towards the preparative and large-scale precision manufacture of virus-like particles. Trends Biotechnol 23(10):523-9.

Peiris JS, de Jong MD, Guan Y. 2007. Avian influenza virus (H5N1): a threat to human health. Clin Microbiol Rev 20(2):243-67.

Pepin S, Donazzolo Y, Jambrecina A, Salamand C, Saville M. 2013. Safety and immunogenicity of a quadrivalent inactivated influenza vaccine in adults. Vaccine 31(47):5572-8.

Perchiacca JM, Ladiwala AR, Bhattacharya M, Tessier PM. 2012. Aggregation-resistant domain antibodies engineered with charged mutations near the edges of the complementarity-determining regions. Protein Eng Des Sel 25(10):591-601.

Perdue ML, Arnold F, Li S, Donabedian A, Cioce V, Warf T, Huebner R. 2011. The future of cell culture-based influenza vaccine production. Expert Rev Vaccines 10(8):1183-94.

Pleckaityte M, Bremer CM, Gedvilaite A, Kucinskaite-Kodze I, Glebe D, Zvirbliene A. 2015. Construction of polyomavirus-derived pseudotype virus-like particles displaying a functionally active neutralizing antibody against hepatitis B virus surface antigen. BMC Biotechnol 15:85.

Pokorski JK, Hovlid ML, Finn MG. 2011. Cell targeting with hybrid Qbeta virus-like particles displaying epidermal growth factor. Chembiochem 12(16):2441-7.

Polycarpou E, Zachaki S, Tsolia M, Papaevangelou V, Polycarpou N, Briana DD, Gavrili S, Kostalos C, Kafetzis D. 2013. Enteral L-arginine supplementation for prevention of necrotizing enterocolitis in very low birth weight neonates: a double-blind randomized pilot study of efficacy and safety. JPEN J Parenter Enteral Nutr 37(5):617-22.

Powell TJ, Peng Y, Berthoud TK, Blais ME, Lillie PJ, Hill AV, Rowland-Jones SL, McMichael AJ, Gilbert SC, Dong T. 2013. Examination of influenza specific T cell responses after influenza virus challenge in individuals vaccinated with MVA-NP+M1 vaccine. PLoS One 8(5):e62778.

Prashar D, Cui D, Bandyopadhyay D, Luk YY. 2011. Modification of proteins with cyclodextrins prevents aggregation and surface adsorption and increases thermal stability. Langmuir 27(21):13091-6.

PRB. 2009. World Population Data Sheet 2009. In: Bureau PR, editor. Washington. Provencher SW. 1979. Inverse Problems in Polymer Characterization - Direct Analysis of

Polydispersity with Photon Correlation Spectroscopy. Makromolekulare Chemie-Macromolecular Chemistry and Physics 180(1):201-209.

Raghunandan R. 2011. Virus-like particles: innate immune stimulators. Expert Rev Vaccines 10(4):409-11.

Rayment I, Baker TS, Caspar DL, Murakami WT. 1982. Polyoma virus capsid structure at 22.5 A resolution. Nature 295(5845):110-5.

Reemers SS, van Haarlem DA, Sijts AJ, Vervelde L, Jansen CA. 2012. Identification of novel avian influenza virus derived CD8+ T-cell epitopes. PLoS One 7(2):e31953.

Richterova Z, Liebl D, Horak M, Palkova Z, Stokrova J, Hozak P, Korb J, Forstova J. 2001. Caveolae are involved in the trafficking of mouse polyomavirus virions and artificial VP1 pseudocapsids toward cell nuclei. J Virol 75(22):10880-91.

Rimmelzwaan GF, Sutter G. 2009. Candidate influenza vaccines based on recombinant modified vaccinia virus Ankara. Expert Rev Vaccines 8(4):447-54.

Rivera-Hernandez T, Hartas J, Wu Y, Chuan YP, Lua LH, Good M, Batzloff MR, Middelberg AP. 2013. Self-adjuvanting modular virus-like particles for mucosal vaccination against group A streptococcus (GAS). Vaccine 31(15):1950-5.

136

Robinson J, Halliwell JA, Hayhurst JD, Flicek P, Parham P, Marsh SG. 2015. The IPD and IMGT/HLA database: allele variant databases. Nucleic Acids Res 43(Database issue):D423-31.

Rodriguez-Limas WA, Sekar K, Tyo KE. 2013. Virus-like particles: the future of microbial factories and cell-free systems as platforms for vaccine development. Curr Opin Biotechnol 24(6):1089-93.

Ryan KJ, Ray CG, Sherris JC. 2014. Sherris Medical Microbiology. New York: McGrawHill. Salunke DM, Caspar DL, Garcea RL. 1986. Self-assembly of purified polyomavirus capsid protein

VP1. Cell 46(6):895-904. Sander B, Nizam A, Garrison LP, Jr., Postma MJ, Halloran ME, Longini IM, Jr. 2009. Economic

evaluation of influenza pandemic mitigation strategies in the United States using a stochastic microsimulation transmission model. Value Health 12(2):226-33.

Sapp M, Day PM. 2009. Structure, attachment and entry of polyoma- and papillomaviruses. Virology 384(2):400-9.

Schirmbeck R, Deml L, Melber K, Wolf H, Wagner R, Reimann J. 1995. Priming of class I-restricted cytotoxic T lymphocytes by vaccination with recombinant protein antigens. Vaccine 13(9):857-65.

Schmidt U, Rudolph R, Bohm G. 2000. Mechanism of assembly of recombinant murine polyomavirus-like particles. J Virol 74(4):1658-62.

Schnitzler SU, Schnitzler P. 2009. An update on swine-origin influenza virus A/H1N1: a review. Virus Genes 39(3):279-92.

Serno T, Geidobler R, Winter G. 2011. Protein stabilization by cyclodextrins in the liquid and dried state. Adv Drug Deliv Rev 63(13):1086-106.

Sette A, Newman M, Livingston B, McKinney D, Sidney J, Ishioka G, Tangri S, Alexander J, Fikes J, Chesnut R. 2002. Optimizing vaccine design for cellular processing, MHC binding and TCR recognition. Tissue Antigens 59(6):443-51.

Sette A, Sidney J. 1998. HLA supertypes and supermotifs: a functional perspective on HLA polymorphism. Curr Opin Immunol 10(4):478-82.

Sette A, Sidney J. 1999. Nine major HLA class I supertypes account for the vast preponderance of HLA-A and -B polymorphism. Immunogenetics 50(3-4):201-12.

Shah S, Hayden CA, Fischer ME, Rao AG, Howard JA. 2015. Biochemical and biophysical characterization of maize-derived HBsAg for the development of an oral vaccine. Arch Biochem Biophys 588:41-9.

Sidney J, Peters B, Frahm N, Brander C, Sette A. 2008. HLA class I supertypes: a revised and updated classification. BMC Immunol 9:1.

Simon R, Curtis B, Deumic V, Nicki J, Tennant SM, Pasetti MF, Lees A, Wills PW, Chacon M, Levine MM. 2014. A scalable method for biochemical purification of Salmonella flagellin. Protein Expr Purif 102:1-7.

Singh N, Pandey A, Mittal SK. 2010. Avian influenza pandemic preparedness: developing prepandemic and pandemic vaccines against a moving target. Expert Rev Mol Med 12:e14.

Soema PC, Kompier R, Amorij JP, Kersten GF. 2015. Current and next generation influenza vaccines: Formulation and production strategies. Eur J Pharm Biopharm 94:251-63.

Somvanshi P, Singh V, Seth PK. 2008. Prediction of epitopes in hemagglutinin and neuraminidase proteins of influenza A virus H5N1 strain: a clue for diagnostic and vaccine development. OMICS 12(1):61-9.

Sriwilaijaroen N, Suzuki Y. 2014. Molecular basis of a pandemic of avian-type influenza virus. Methods Mol Biol 1200:447-80.

Stehle T, Harrison SC. 1997. High-resolution structure of a polyomavirus VP1-oligosaccharide complex: implications for assembly and receptor binding. EMBO J 16(16):5139-48.

Stevens J, Blixt O, Paulson JC, Wilson IA. 2006. Glycan microarray technologies: tools to survey host specificity of influenza viruses. Nat Rev Microbiol 4(11):857-64.

137

Suarez DL, Senne DA, Banks J, Brown IH, Essen SC, Lee CW, Manvell RJ, Mathieu-Benson C, Moreno V, Pedersen JC and others. 2004. Recombination resulting in virulence shift in avian influenza outbreak, Chile. Emerg Infect Dis 10(4):693-9.

Subbarao K, Joseph T. 2007. Scientific barriers to developing vaccines against avian influenza viruses. Nat Rev Immunol 7(4):267-78.

Suphaphiphat P, Keiner B, Trusheim H, Crotta S, Tuccino AB, Zhang P, Dormitzer PR, Mason PW, Franti M. 2010. Human RNA polymerase I-driven reverse genetics for influenza a virus in canine cells. J Virol 84(7):3721-5.

Szewczyk B, Bienkowska-Szewczyk K, Krol E. 2014. Introduction to molecular biology of influenza a viruses. Acta Biochim Pol 61(3):397-401.

Szurgot I, Jedynak M, Podsiadla-Bialoskorska M, Piwowarski J, Szolajska E, Chroboczek J. 2015. Adenovirus Dodecahedron, a VLP, Can be Purified by Size Exclusion Chromatography Instead of Time-Consuming Sucrose Density Gradient Centrifugation. Mol Biotechnol 57(6):565-73.

Szurgot I, Szolajska E, Laurin D, Lambrecht B, Chaperot L, Schoehn G, Chroboczek J. 2013. Self-adjuvanting influenza candidate vaccine presenting epitopes for cell-mediated immunity on a proteinaceous multivalent nanoplatform. Vaccine 31(40):4338-46.

Tan AC, Deliyannis G, Bharadwaj M, Brown LE, Zeng W, Jackson DC. 2013. The design and proof of concept for a CD8(+) T cell-based vaccine inducing cross-subtype protection against influenza A virus. Immunol Cell Biol 91(1):96-104.

Tan PT, Heiny AT, Miotto O, Salmon J, Marques ET, Lemonnier F, August JT. 2010. Conservation and diversity of influenza A H1N1 HLA-restricted T cell epitope candidates for epitope-based vaccines. PLoS One 5(1):e8754.

Tekewe A, Connors NK, Sainsbury F, Wibowo N, Lua LH, Middelberg APJ. 2015. A Rapid and Simple Screening Method to Identify Conditions for Enhanced Stability of Modular Vaccine Candidates. Biochemical Engineering Journal 100:50-58.

Tekewe A, Fan Y, Tan E, Middelberg AP, Lua LH. 2016. Integrated molecular and bioprocess engineering for bacterially produced immunogenic modular virus-like particle vaccine displaying 18 kDa rotavirus antigen. Biotechnol Bioeng.

Terpe K. 2003. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl Microbiol Biotechnol 60(5):523-33.

Teunissen EA, de Raad M, Mastrobattista E. 2013. Production and biomedical applications of virus-like particles derived from polyomaviruses. J Control Release 172(1):305-21.

Treuheit MJ, Kosky AA, Brems DN. 2002. Inverse relationship of protein concentration and aggregation. Pharm Res 19(4):511-6.

Tsumoto K, Umetsu M, Kumagai I, Ejima D, Philo JS, Arakawa T. 2004. Role of arginine in protein refolding, solubilization, and purification. Biotechnol Prog 20(5):1301-8.

Uchida T. 2011. Development of a cytotoxic T-lymphocyte-based, broadly protective influenza vaccine. Microbiol Immunol 55(1):19-27.

Valenzuela P, Medina A, Rutter WJ, Ammerer G, Hall BD. 1982. Synthesis and assembly of hepatitis B virus surface antigen particles in yeast. Nature 298(5872):347-50.

Vlecken DH, Pelgrim RP, Ruminski S, Bakker WA, van der Pol LA. 2013. Comparison of initial feasibility of host cell lines for viral vaccine production. J Virol Methods 193(1):28-41.

Wang H, Jiang C. 2009. Avian influenza H5N1: an update on molecular pathogenesis. Sci China C Life Sci 52(5):459-63.

Wang M, Lamberth K, Harndahl M, Roder G, Stryhn A, Larsen MV, Nielsen M, Lundegaard C, Tang ST, Dziegiel MH and others. 2007. CTL epitopes for influenza A including the H5N1 bird flu; genome-, pathogen-, and HLA-wide screening. Vaccine 25(15):2823-31.

Wang M, Larsen MV, Nielsen M, Harndahl M, Justesen S, Dziegiel MH, Buus S, Tang ST, Lund O, Claesson MH. 2010a. HLA class I binding 9mer peptides from influenza A virus induce CD4 T cell responses. PLoS One 5(5):e10533.

138

Wang W. 2005. Protein aggregation and its inhibition in biopharmaceutics. Int J Pharm 289(1-2):1-30.

Wang W, Nema S, Teagarden D. 2010b. Protein aggregation--pathways and influencing factors. Int J Pharm 390(2):89-99.

WHO. 2005. Pandemic Alert & Response: WHO checklist for influenza pandemic preparedness planning. Switzerland: Department of Communicable Disease Surveillance and Response Global Influenza Programme WHO.

WHO. 2009a. Influenza (Seasonal) Fact Sheet. WHO. http://www.who.int/mediacentre/factsheets/fs211/en/index.html, Last Updated: April 2009, Date Accessed: 06/12/2013

WHO. 2009b. Situation updates - Pandemic (H1N1) 2009. WHO. www.who.int/csr/disease/swineflu/updates/en/, Last Updated: 10/09/2010, Date Accessed: 16/01/2016

WHO. 2011. Strengthening Response to Pandemics and Other Public-Health Emergencies. France: WHO.

WHO. 2013. Evolution of a pandemic A(H1N1) 2009, April 2009-August 2010. In: Rashford A, editor. 2nd ed. Geneva: WHO.

Wibowo N, Chuan YP, Lua LH, Middelberg AP. 2012. Modular engineering of a microbially-produced viral capsomere vaccine. Chemical Engineering Science.

Wibowo N, Chuan YP, Lua LHL, Middelberg APJ. 2013. Modular engineering of a microbially-produced viral capsomere vaccine for influenza. Chemical Engineering Science 103:12-20.

Wibowo N, Hughes FK, Fairmaid EJ, Lua LH, Brown LE, Middelberg AP. 2014. Protective efficacy of a bacterially produced modular capsomere presenting M2e from influenza: extending the potential of broadly cross-protecting epitopes. Vaccine 32(29):3651-5.

Wibowo N, Wu Y, Fan Y, Meers J, Lua LH, Middelberg AP. 2015. Non-chromatographic preparation of a bacterially produced single-shot modular virus-like particle capsomere vaccine for avian influenza. Vaccine 33(44):5960-5.

Woolard SN, Kumaraguru U. 2010. Viral vaccines and CTL response. J Biomed Biotechnol 2010:141657.

Wyatt. 2010. Dynamics User's Guide Version 7.0 (M1400 Rev. I). Santa Barbara: Wyatt Technology Corporation.

Yan D, Wei YQ, Guo HC, Sun SQ. 2015. The application of virus-like particles as vaccines and biological vehicles. Appl Microbiol Biotechnol 99(24):10415-32.

Yeo TW, Lampah DA, Rooslamiati I, Gitawati R, Tjitra E, Kenangalem E, Price RN, Duffull SB, Anstey NM. 2013. A randomized pilot study of L-arginine infusion in severe falciparum malaria: preliminary safety, efficacy and pharmacokinetics. PLoS One 8(7):e69587.

Yewdell JW, Reits E, Neefjes J. 2003. Making sense of mass destruction: quantitating MHC class I antigen presentation. Nat Rev Immunol 3(12):952-61.

Zeltins A. 2013. Construction and characterization of virus-like particles: a review. Mol Biotechnol 53(1):92-107.

Zhang K, Wang Z, Liu X, Yin C, Basit Z, Xia B, Liu W. 2012. Dissection of influenza A virus M1 protein: pH-dependent oligomerization of N-terminal domain and dimerization of C-terminal domain. PLoS One 7(5):e37786.

Zhang L, Lu D, Liu Z. 2008. How native proteins aggregate in solution: a dynamic Monte Carlo simulation. Biophys Chem 133(1-3):71-80.

murine polyomavirus vlps as a platform for cytotoxic t ...for bioengineering and nanotechnology, the...

Documents