Development of live bacterial delivery systems for presentation of dengue EDIII to the mucosal immune system 2

A live bacterial vector that expresses and delivers EDIII to the host immune system represents a potential means of augmenting the immune response.. Table 1.1: Dengue vaccines in develop

FOR PRESENTATION OF DENGUE EDIII TO THE

MUCOSAL IMMUNE SYSTEM

LAM JIAN HANG

NATIONAL UNIVERSITY OF SINGAPORE

2014

FOR PRESENTATION OF DENGUE EDIII TO THE

MUCOSAL IMMUNE SYSTEM

LAM JIAN HANG

I hereby declare that this thesis is my original work and it has been written by

me in its entirety I have duly acknowledged all the sources of information

which have been used in the thesis

This thesis has also not been submitted for any degree in any university

previously

_

Lam Jian Hang

20 August 2014

Hoo, R., Lam, J.H., Huot, L., Pant, A., Li, R., Hot, D., Alonso, S (2014)

Evidence for a role of the polysaccharide capsule transport proteins in pertussis pathogenesis PLoS One 9(12): e115243

Ng, J.K., Zhang, S.L., Tan, H.C., Yan, B., Maria Martinez Gomez, J., Tan,

W.Y., Lam, J.H., Tan, G.K., Ooi, E.E., Alonso, S (2014) First experimental

in vivo model of enhanced dengue disease severity through maternally

acquired heterotypic dengue antibodies PLoS Pathog 10(4): e1004031

PRESENTATION AT INTERNATIONAL CONFERENCES

Lam, J.H., Alonso, S (2012) Expression and delivery of dengue EDIII by

Bordetella pertussis BPZE1 via the BrkA autotransporter Oral presentation

prize In: 4th Australasian Vaccines and Immunotherapeutics Development Meeting, Brisbane, Australia

I would like express to my heartfelt thanks to my supervisor Assoc Prof Sylvie Alonso for first giving me the chance to join her lab in 2008, and subsequently offering invaluable advice, guidance and opportunities to develop my mind and skills as a researcher Over the years, I have known her

to be the most patient and encouraging supervisor I could ask for and, once again, I offer my sincere appreciation for all she has taught me

To my lab mates, past and present, I thank you all for making the working environment lively, entertaining, exciting and, frequently, edible I really treasure the friendships that we’ve built and the time we spent as a lab A big thank you to all the names listed in Table A below :)

Regina Hoo May Ling Julia Maria Martinez Gomez

Annabelle Lim Rui Fen Michelle Ang

Grace Tan Kai Xin Emily Ang

Anna Ker Yeo Huimin Eshele

Table A: List of awesome SA lab mates encountered over the course of

my PhD candidature

I would also like to express my appreciation to my TAC Dr Ooi Eng Eong and Dr Ratha Mahendran for offering useful comments and advice during our meeting

Lastly, I would like to express my deepest gratitude to my parents and my sister who have been hugely supportive and understanding when I had to shift

my residential address to CeLS during my most intensive months None of this

1.2 THE ADAPATIVE IMMUNE RESPONSE

1.2.1 The anti-DENV immune response in

1.2.1.1.1 Antibody-dependent enhancement

1.4.1 Structural and serological characteristics 19 1.4.2 EDIII as a vaccine candidate: some considerations 20

1.4.2.3 Quality of the anti-EDIII antibody response 21

1.5 LIVE BACTERIAL VECTORS FOR

1.5.1 Rationale for using a live bacterial vector 22

1.5.2.1 B pertussis as a live vector for

nasal delivery of heterologous antigens 25 1.5.2.1.1 Resurgence of pertussis despite vaccination 25 1.5.2.1.2 Developing a live attenuated B pertussis vaccine 26

1.5.3.1 L lactis as a live vector for delivery of

1.6 HETEROLOGOUS PRODUCTION OF EDIII

IN B PERTUSSIS AND L LACTIS – SOME

1.7 OBJECTIVES OF THIS PROJECT 51

2.1 ASSESSING THE SUITABILITY OF EDIII

2.1.1.1 E coli strains, plasmids and culture conditions 53

2.1.1.2.3 Restriction enzyme digest, agarose gel

electrophoresis, gel extraction and generation

2.1.1.3.1 Expression of ediii by IPTG induction 55

2.1.1.3.3 Solubilisation of inclusion body and refolding of rEDIII 56 2.1.1.3.4 Purification of rEDIII using Ni-NTA chromatography 56

2.1.3.1.2 Growth conditions 59

2.1.3.1.4 Plaque reduction neutralisation test (PRNT) 60

2.2 BORDETELLA PERTUSSIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

2.2.1.1 E coli strains, plasmids and culture conditions 61

2.2.1.2.2.1 DNA amplification for cloning work 64

2.2.2.1 B pertussis strains and culture conditions 67

culture supernatant 70 2.2.2.3.1.2 Sodium dodecyl sulphate-polyacrylamide gel

enzyme-linked immunosorbent assay (ELISA) 74

2.2.4.1 Virus strain, cell lines, growth conditions,

2.3 LACTOCOCCUS LACTIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

2.3.1.1 E coli strains, plasmids and culture conditions 76

2.3.1.2.4 E coli heat shock transformation, plasmid extraction

2.3.2.1 L lactis strains and culture conditions 79

2.3.2.2.3 Selection of antibiotic resistant transformants 81

2.3.2.3.3.1 Preparation of gfp-expressing L lactis strains for analysis 84

2.3.2.3.3.2 Preparation of LL-EDIII(CWA) strain for analysis 84

2.3.2.4.2.2 Immunisation and blood collection schedule 85 2.3.2.4.2.3 Assessment of serum IgG response by ELISA 86

2.3.2.4.2.4.1 Preparation of single cell suspension 87

CHAPTER 3: RESULTS 88 3.1 ASSESSING THE SUITABILITY OF EDIII

3.1.1 Production of rEDIII and immunisation of BALB/c mice 88 3.1.2 Assessment of serum antibody responses towards

3.2 BORDETELLA PERTUSSIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

3.2.1 Recombinant BPZE1 producing FHA-EDIII

3.2.1.1 Construction of BP-FHA-EDIII, BP-FHA-EDIII(1Cys)

3.2.1.2 Production and secretion of the FHA-EDIII chimera 92 3.2.2 Recombinant BPZE1 producing BrkA-EDIII chimeric protein 94

3.2.2.3 Surface exposure of EDIII was confirmed by

3.2.2.4 The in vitro and in vivo fitness of BP-BrkA-EDIII

3.2.2.5 Serum IgG response against DENV2 was not detected

after nasal immunisation with BP-BrkA-EDIII 99 3.2.3 Recombinant BPZE1 producing BrkA-EDIII-S1 chimeric protein 102 3.2.3.1 Construction of BP-BrkA-EDIII-S1 strain 102 3.2.3.2 Fusing EDIII and S1 to BrkA resulted in severe

degradation of the chimeric protein and the secretion

3.2.3.3 SecP33 appeared to compete with native S1 for

3.2.3.5 Fitness of BP-BrkA-EDIII-S1 strain was significantly

3.2.3.6 Serum IgG response against DENV2 was not detected

after nasal immunisation with BP-BrkA-EDIII-S1 110 3.2.4 Recombinant BPZE1 producing S1-EDIII chimeric protein 112

3.2.4.2 Production of S1-EDIII chimeric protein by BPSE3 strain

3.2.4.3 Growth impairment was not observed in BPSE3 115 3.2.4.4 Serum IgG response against DENV2 was not detected

3.3 LACTOCOCCUS LACTIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

3.3.1 Recombinant L lactis producing cytoplasmic EDIII 120 3.3.1.1 Construction of gfp-expressing L lactis strains 120 3.3.1.2 Effect of nisin concentration on the yield of EDIII-GFP 122 3.3.1.3 Comparing the effects of semi-aerobic and aerobic culture

conditions on the yield and folding of cytoplasmic EDIII 123 3.3.1.3.1 Yields of cytoplasmic EDIII were comparable after

3.3.1.3.2 Folding of EDIII seemed to be improved after

3.3.2.2 Production of EDIII(Sec) and EDIII(CWA) proteins

3.3.2.3 Presence of EDIII in culture supernatant of LL-EDIII(Sec)

3.3.2.4 Bicarbonate-buffered medium improved the yields of

3.3.2.6 Subcellular location determined stability of EDIII

during erythromycin and nisin deprivation 138

3.3.2.7.1 Intratracheal immunisation of C57BL/6 with

LL-EDIII-GFP or LL-EDIII(CWA) but not LL-EDIII(Sec) generated low-level IgG responses

3.3.2.7.2 Assessing the ability of ediii-expressing L lactis to

prime the immune system to respond to DENV challenge

4.1 ASSESSING THE SUITABILITY OF EDIII AS A

4.2 BORDETELLA PERTUSSIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

4.3 LACTOCOCCUS LACTIS AS A LIVE VECTOR

FOR THE PRODUCTION AND MUCOSAL

As the target of strongly neutralising monoclonal antibodies, EDIII is a highly popular subunit vaccine candidate However, monomeric EDIII is poorly immunogenic A live bacterial vector that expresses and delivers EDIII to the host immune system represents a potential means of augmenting the immune response BPZE1, the live attenuated vaccine strain of the respiratory pathogen

Bordetella pertussis, was used to deliver DENV2 EDIII to the mucosal

immune system of the respiratory tract To facilitate interaction with B cells, EDIII was produced as a fusion to the FHA (surface exposed and secreted), BrkA (surface exposed) or PTX (secreted) virulence factor The FHA-EDIII chimera proved unsuitable as the disulphide bond within EDIII was incompatible with FHA translocation across the outer membrane, causing a severe reduction in yield of the chimeric protein Abolishing the disulphide bond via substitution of one or both Cys residues with Gly restored cellular protein levels but not secretion, and also led to the loss of a major neutralising epitope On the other hand, fusing EDIII to either BrkA or PTX resulted in successful surface exposure and secretion, respectively Immunogenicities of

all ediii-expressing recombinant BPZE1 strains generated in this study were

assessed via intranasal immunisation of BALB/c mice Strong serum antibody responses against the bacterial vector were consistently detected by ELISA However, EDIII-specific responses were weak and no significant response was detected against DENV2 Absence of viral-neutralising activity in the immune sera was confirmed by the plaque reduction neutralisation test These results therefore indicated that BPZE1 was not ideal for the presenting EDIII to the respiratory mucosal immune system As an alternative approach, we chose the

lactic acid bacterium L lactis for EDIII delivery Three recombinant strains of

L lactis were constructed, producing cytoplasmic, secreted or cell wall associated EDIII Gene expression was driven by the inducible nisA promoter

cloned into the replicative pMG36e plasmid Production of EDIII in the different cellular compartments was confirmed by western blot While increasing doses of nisin (5 to 20 ng/ml) did not lead to higher EDIII yields, addition of bicarbonate buffer to control acidity and suppress the bacteria acid

assessed by immunising C57BL/6 mice via the respiratory route Strong serum

antibody responses against L lactis were detected by ELISA However,

EDIII-specific responses were weak and only a single mouse immunised with the cytoplasmic EDIII strain exhibited significant activity against DENV2 Despite the general absence of antibody responses against the virus,

recombinant L lactis was able to efficiently prime the host immune system, as

evidenced by high levels of IFN-γ in the supernatants of rEDIII- or restimulated splenocytes isolated from mice immunised with the cytoplasmic

DENV2-EDIII L lactis strain Altogether, our work indicates that B pertussis and L lactis bacteria may not represent promising live vectors for the mucosal

delivery of EDIII They may however allow efficient priming of the host immune system, thereby providing some protection against a subsequent

DENV challenge

Table 1.1: Dengue vaccines in development 17 Table 1.2: Immune response and protection studies

Table 1.3: Cytoplasmic production of foreign antigens by L lactis 43

Table 1.4: Secretion of foreign proteins by L lactis 44

Table 1.5: Surface display of foreign proteins by L lactis 45 Table 1.6: Immune response and protection studies with

Table 2.5: Primers used for E coli or B pertussis work 64

Table 2.7: List of primary antibodies used in western blot 72

Table 2.9: Primers used for E coli or L lactis work 78

Table 2.11: List of primary antibodies used in western blot 83 Table 3.1: PRNT50 titres of rEDIII-immunised mice 90

Table 5.1: Strengths and weaknesses of BPZE1 and L lactis 162

Figure 1.1 (A): Schematic representation of the DENV genome 3 Figure 1.1 (B): Schematic representation of the mature DENV virion 3 Figure 1.1 (C): Organisation of the E glycoprotein on the surface of

Figure 1.1 (D): Schematic representation of the DENV E protein

Figure 1.2: Global distribution of DENV serotypes from 2000 to 2013 7

Figure 1.6: Schematic representation of homologous

Figure 3.1: Induction of neutralising antibody responses with the

Figure 3.3: Fusing EDIII to FHA was detrimental to production

and secretion of the chimeric protein 93

Figure 3.5: Production of BrkA-EDIII chimeric protein by recombinant

BPZE1 was confirmed by western blot 96 Figure 3.6: Surface exposure of BrkA-EDIII protein was confirmed by

Figure 3.7: In vitro growth of BP-BrkA-EDIII was not impaired 98 Figure 3.8: BP-BrkA-EDIII retained its lung colonisation ability 99 Figure 3.9: Serum IgG response against DENV2 was not detected

despite strong activity towards B pertussis 100

Figure 3.10: Schematics of brkA-ediii-ptxS1 cloning 102 Figure 3.11: Fusing EDIII and S1 to BrkA resulted in severe

degradation and the secretion of a protein fragment 104 Figure 3.12: SecP33 appeared to compete with native S1 for the same

Figure 3.13: Secretion of SecP33 appeared to be associated with the

Figure 3.14: Fitness of BP-BrkA-EDIII-S1 was severely compromised 109 Figure 3.15: Serum IgG response against DENV2 was not detected in

mice immunised with BP-BrkA-EDIII-S1 111

Figure 3.17: Production and secretion of S1-EDIII protein was detected

Figure 3.18: Growth impairment was not observed in BPSE3 116 Figure 3.19: Serum IgG response against DENV2 was not detected in

Figure 3.20: Schematic representation of gfp and ediii-gfp expressed

Figure 3.21: Effect of nisin concentration on the yield of EDIII-GFP 122 Figure 3.22: Assessment of protein production after semi-aerobic or

Figure 3.23: Significant improvement in fluorescent subpopulation of

LL-EDIII-GFP after aerobic induction 126

Figure 3.24: Schematic representations of ediii(Sec) and ediii(CWA)

expressed under the control of PnisA 129 Figure 3.25: Production of EDIII(Sec) and EDIII(CWA) proteins were

Figure 3.26: Presence of EDIII in culture supernatant of LL-EDIII(Sec)

Figure 3.27: Yields of EDIII were improved with the use of bicarbonate-

Figure 3.28: Surface exposure of EDIII was not detected by flow

Figure 3.29: Impact of subcellular location on stability of EDIII during

Figure 3.30: Immunising with LL-EDIII-GFP, LL-EDIII(Sec) or

LL-EDIII(CWA) failed to elicit a functional

to have successfully primed the immune system 145

ADCC Antibody-dependent cell-mediated

cytotoxicity

protein

molecule 1

Infectious Diseases

NS Non-structural

sulphate-polyacrylamide gel electrophoresis

CHAPTER 1 INTRODUCTION

CHAPTER 1 INTRODUCTION

1.1 DENGUE: VIROLOGY, DISEASE AND EPIDEMIOLOGY

1.1.1 Virion structure and assembly

Dengue virus (DENV) is an enveloped, positive stranded RNA virus that exists as four antigenically distinct serotypes (DENV1-4) The single-stranded viral RNA genome is 10.7 kb in length and contains a 5’ methyl guanosine cap, 5’ untranslated region (UTR), single open reading frame, and a 3’ UTR (Figure 1.1, A) [1] Viral RNA is translated to a polyprotein carrying three structural proteins (capsid, envelope and membrane) that form the virus particle and seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) required for viral replication

The structure of DENV has been solved using a combination of cryoelectron microscopy and image reconstruction techniques (Figure 1.1, B) [2, 3] In the culture supernatant of infected cells, the virus exists as a mix of mature and immature forms with diameters of 50 nm and 60 nm, respectively [4] Both particles possess a surface glycoprotein shell and, beneath that, a host-derived lipid bilayer At the centre of the virion is the nucleocapsid core consisting of genomic RNA and capsid (C) protein The glycoprotein shell is composed of envelope (E) and membrane (prM/M) proteins Immature virions are non-infectious and have spiky projections of 60 trimers of prM-E heterodimers Mature virions, on the other hand, are infectious and possess a smooth surface composed of 90 E homodimers arranged to form an icosahedral scaffold (Figure 1.1, C)

Cleavage of prM is necessary for DENV maturation and infectivity This cleavage is mediated by the host protease furin and generates the M protein, which associates with the mature virion as a transmembrane protein beneath the E protein shell [2-5] The E protein is involved in receptor binding and fusion with the host membrane Structurally, it consists of three domains (DI,

region DII carries a highly conserved, hydrophobic fusion loop at its tip, which is required for insertion into the host membrane to create a point of attachment for drawing host and viral membranes together [6] Viral entry into

a host cell is achieved by receptor-mediated endocytosis [5] Following the internalisation of the virus, the acidic pH of the endosome causes the E protein

to undergo conformational changes The fusion loop is exposed and membrane fusion is initiated [6] The nucleocapsid eventually exits the endosome and enters the cytoplasm The virus uncoats to release its genetic material for translation into the polyprotein, which is co-translationally cleaved by host and viral proteases into individual viral proteins RNA replication is initiated

by the NS proteins – synthesis begins with the negative-strand followed by the positive-strand and the new genetic material may be used for translation, or synthesis of more negative-strand RNA, or packaged into new virions [7] Among the NS proteins, only NS3 and NS5 possess enzymatic activity The N-terminal domain of NS3 is a protease and the C-terminal domain is an RNA helicase NS5 contains a methyltransferase at the N-terminus and an RNA-dependent RNA polymerase at the C-terminus [8] Together with its co-factor NS2B, NS3 is responsible for cleaving the DENV polyprotein at specific sites Its helicase domain is implicated in interacting with the NS5 polymerase [4]

New virions are assembled in the endoplasmic reticulum (ER), which results

in the formation of non-infectious, immature particles that are transported through the exocytotic pathway [5] prM likely functions as a chaperone for the folding of E during assembly and is eventually cleaved by the host protease furin in the trans-Golgi network [1] This allows the E protein to adopt the conformational state required for its entry functions [5] Cells infected with DENV secrete high levels (~30%) of immature particles containing prM, suggesting that cleavage of prM to M is not efficient

Although immature DENV lacks infectivity, Rodenhuis-Zybert et al (2010)

demonstrated the infection of Fcγ-receptor-expressing cells in the presence of prM-specific antibodies [9] Thus, immature virions have the potential to become highly infectious through enhancing antibodies and may contribute to severe disease

Helicase NTPase

RNA polymerase Methyltransferase

Figure 1.1: (A) Schematic representation of the DENV genome (B)

Schematic representation of the mature DENV virion Genomic RNA associates with C protein to form the nucleocapsid core Host-derived lipid bilayer (blue layer) surrounds the nucleocapsid and lies beneath the E glycoprotein shell The M protein is embedded within the lipid bilayer and

positioned below the E protein layer (C) Organisation of the E glycoprotein

on the surface of the mature virion The E protein assembles into anti-parallel

D

Putative receptor binding loop DII

DIII Fusion loop

E

F

BC loop

DE

loop

FG loop

A strand

homodimers (green box), which are arranged in rafts of three parallel dimers

(red box) 90 E homodimers are organised into an icosahedral scaffold (D)

Schematic representation of the DENV E protein primary structure Domains I (red) and II (yellow) are discontinuous and contain glycans Domain III (blue)

is located at the C-terminus and occupies a continuous stretch from residue

296-394 EDIII is non-glycosylated (E) The E homodimer The E protein

forms an anti-parallel homodimer on the surface of the mature virion Fusion loop of DII, required for mediating fusion between viral and host membranes,

is indicated in orange The putative receptor binding loop of DIII is located at

residues 382-385 (F) Ribbon diagram of EDIII protein The amino- and

carboxyl-termini are indicated in white The lateral ridge is composed of the N linker and BC, DE and FG loops The A strand is circled in yellow, AB loop

in green Both cysteines are marked by orange crosses

Picture credits

A: Adapted from Murphy and Whitehead, 2011, Annu Rev Immunol [10]; B:

Downloaded from ViralZone, SIB Swiss Institute of Bioinformatics

(http://viralzone.expasy.org/all_by_species/24.html); C, D: Modis et al., 2003,

PNAS [11]; E: †Heinz and Stiasny 2012, Vaccine [5]; F: Coordinates of 3D

model downloaded from Protein Data Bank (PDB ID: 3VTT)

†: This article was published in Vaccine, 30, Heinz, F.X., Stiasny, K , Flaviviruses and

flavivirus vaccines, 4301-4306, Copyright Elsevier (2012)

1.1.2 Disease and epidemiology

DENV is the aetiological agent of dengue fever (DF), the most prevalent

arthropod-borne viral illness in humans It belongs to the genus Flavivirus of the family Flaviviridae Over 70 viruses are grouped within this genus and

members that are recognised as important human pathogens with significant disease impact include yellow fever virus, Japanese encephalitis virus, West Nile virus and tick-borne encephalitis virus [12] In general, the majority of DENV infections pass with minimal or no symptoms Most symptomatic infections present as DF, a self-limiting illness that lasts 4–7 days DF is characterised by a sudden onset of fever accompanied by headache, pain behind the eyes, generalised myalgia and arthralgia, flushing of the face, rash, anorexia, abdominal pain, and nausea A small proportion of patients may develop life threatening dengue haemorrhagic fever or dengue shock syndrome (DHF/DSS), characterised by spontaneous bleeding, plasma leakage,

or both Although uncommon, DHF/DSS still represents the most important contributor to a severe clinical outcome [10, 13]

Transmission of DENV is mediated mainly by Aedes aegypti and, to a lesser extent, A albopictus mosquitoes Flaviviruses are highly dependent on

specific vectors for their transmission; therefore, the geographical distributions

of these viruses follow the distributions of their vectors [5] With the

worldwide presence of the Aedes mosquito in the tropics and sub-tropics,

DENV now has the greatest global reach among the flaviviruses (Figure 1.2) DENV is endemic in more than 100 countries, including most of Southeast Asia, South America, Central America and the Caribbean and South Pacific regions Additionally, co-circulation of multiple serotypes is common within countries most affected by DENV [10] 3.6 billion people live in endemic areas and the virus causes approximately 400 million infections and 100 million symptomatic cases annually Over 2 million cases of severe dengue disease and over 20,000 deaths are estimated to occur each year [1]

Several factors are implicated in the high prevalence of dengue Unprecedented global population growth and the associated unplanned and uncontrolled urbanisation, especially in tropical developing countries, create conditions ideal for mosquito proliferation and transmission of mosquito-

borne diseases [14] A aegypti is a highly adaptable species that exploits a

wide variety of containers found in domestic habitats as larval development sites, including containers ranging in size from bottles and cans to large water storage tanks Ample supplies of such larval development sites in urban areas

make the control of A aegypti a highly challenging task [15] Vector control

programmes have achieved limited success owing to the need for sustained engagement against a backdrop of decay in public health infrastructures in many developing countries [14] Increasing air travel has also contributed to increasing spread of DENV Airline passenger numbers have increased by 9% annually since 1960, enabling infected human hosts to move the viruses across long distances rapidly [16] Despite the high prevalence of DENV and the morbidity and mortality associated with the disease, no effective antiviral therapy or vaccine exists at present and treatment remains largely supportive

in nature Minimising the global disease burden is expected to require a combination of vector control, antiviral therapy and vaccination [8]

† Figure 1.2: Global distribution of DENV serotypes from 2000 to 2013

†: This article was published in Trends Microbiol., 22, Messina, J.P., Brady, O.J., Scott, T.W.,

Zou, C., Pigott, D.M., Duda, K.A., Bhatt, S., Katzelnick, L., Howes, R.E., Battle, K.E., Simmons, C.P., Hay, S.I., Global spread of dengue virus types: mapping the 70 year history, 138-146, Copyright Elsevier (2014)

1.2 THE ADAPATIVE IMMUNE RESPONSE FOLLOWING A

serotype leads to a robust immune response which provides lifelong protection against the disease upon reinfection with the same serotype Heterotypic protection is transient and lasts only a few months, after which the individual becomes susceptible to other serotypes [10, 13] Furthermore, individuals with pre-existing immune responses against DENV are at higher risk of developing severe disease following a secondary infection Observational studies

conducted by Halstead et al on children with dengue revealed the occurrence

of DHF/DSS in two age groups The first group was born from immune mothers and of 6 to 9 months of age, which represented the phase of decline in maternal antibodies below protective levels The second group consisted of children who had experienced an earlier infection, usually mild or subclinical, and were later infected with a different serotype [17] Development of severe disease following sequential, heterotypic infections has also been described in the adult population, notably the 1981 outbreak in Cuba and the 2006 outbreak in Singapore [18] Thus, heterologous antibodies acquired either passively or actively have the potential to exacerbate dengue disease

dengue-1.2.1.1 Antibodies

There is strong evidence for the critical role of neutralising antibodies in protection against DENV Passive transfer of neutralising antibodies is protective in murine and non-human primate (NHP) models of DENV infection [19-22] The prM, E and NS1 proteins are principal targets for the humoral immune response owing to their surface-exposed or secreted nature; antibody responses to other NS proteins are weak due to their intracellular localisation, which limits access to B cells [13]

1.2.1.1.1 Antibody-dependent enhancement (ADE) of infection

According to the “multiple-hit” model for flavivirus neutralisation, an antibody must engage the virion with a stoichiometry that exceeds a threshold for neutralisation to occur Whether an antibody can exceed this threshold depends on its affinity for viral antigens and the total number of accessible

epitopes on the average virion Engagement of the virion with a stoichiometry below this threshold may support antibody-dependent enhancement (ADE) of infection [23] Weakly neutralising antibodies that are cross-reactive for other serotypes or sub-neutralising concentrations of protective antibodies are hypothesised to cause severe dengue through the mechanism of ADE In the presence of such antibodies, viral infectivity is not abrogated Instead, the virus forms immune complexes that target cells expressing Fcγ receptors – notably monocytes, macrophages and dendritic cells Subsequently, there is increased viral entry mediated by the Fcγ receptors, which leads to greater viral production and viraemia Additionally, ADE appears to also modify intracellular antiviral mechanisms and enhance viral replication [24] A

murine model of DENV infection demonstrated in vivo enhancement of

infection and disease in the presence of heterologous immune serum as well

as sub-neutralising level of polyclonal or monoclonal antibody [25, 26] Evidence from human studies on the role of ADE in disease severity, however,

remained equivocal Vaughn et al (2000) first presented evidence supporting

a direct correlation between disease severity and peak viraemia titre in children with acute DENV infection [27] The same group then published another paper describing the correlation between DHF and higher mean plasma viraemia early in illness, in children with secondary DENV3 infection [28] In a recent publication on the role of maternally-acquired antibodies in disease protection or enhancement, the authors failed to find a significant association between DHF in infants and the DENV3 ADE activity of their sera [29] In light of the evidence presented by various authors, the ADE hypothesis appears biologically possible but is unlikely the sole factor in determining disease outcome in humans

1.2.1.2 T cells

Human T cell epitopes can be found across the entire DENV proteome Whereas CD8+ T cells preferentially target nonstructural proteins (NS3 and NS5), CD4+ T cells generally recognise epitopes on the envelope, capsid, and NS1 [30] According to the “original antigenic sin” hypothesis, weakly

in severe disease in a secondary, heterotypic infection These cells are preferentially expanded over newly activated T cells with higher affinity for the current infection [31] and exhibit altered cytokine production with enhanced secretion of TNF-α [32] Plasma levels of soluble TNF receptors were higher in patients with DHF compared to patient with DF during acute infection [28, 33] and TNF-α was more frequently detectable in DHF than in

DF plasma samples [33] Elevated level of circulating TNF-α exerts a plethora

of effects, including vascular leakage and shock [34] The mechanism behind this phenomenon appears to be a combination of TNF-α-induced apoptosis of vascular endothelial cells and increased expression of cellular adhesion molecules (including ICAM-1) that might lead to greater adhesion of inflammatory blood cells [35]

On the other hand, T cells appear to also confer protection against DENV In vitro studies showed virus-specific T cells to recognise infected cells and

respond with a variety of effector functions, including proliferation, cytokine production and induction of cellular lysis [36, 37] New insights into the protective role of T cells came from studies of DENV infection in IFN-α/βR-/-mice Depletion of CD8+ T cells from such mice before an infection led to significantly higher viral loads compared to undepleted mice Additionally, enhanced viral clearance was observed after immunising with immunodominant CD8+ epitopes [38] The same group subsequently studied the importance of CD4+ T cells in the control of DENV infection Although depletion did not affect viral loads, CD4+ cells were still able to lyse infected cells, and immunisation of mice with CD4+ epitopes resulted in improved viral clearance [39] More recently, a comprehensive study of CD8+ responses in the general population from the Sri Lankan hyperendemic area showed that a vigorous response by multifunctional CD8+ T cells was associated with protection from DENV [40] In light of the evidence available thus far, it is clear that T cells are important mediators of the outcome of a DENV infection

1.3 PROSPECTS FOR A DENGUE VACCINE

1.3.1 Challenges of vaccine development

1.3.1.1 Unbalanced immune responses

An effective dengue vaccine remains elusive despite decades of dengue research Development has been hindered by several factors, one of which is viral diversity Infection by any one serotype confers lifelong homotypic immunity but does not provide heterotypic protection In fact, epidemiological observations indicate that subsequent infection with a second type increases the likelihood of severe disease (DHF/DSS) [5, 17] Concern over ADE and its role in DHF/DSS instilled a belief in the necessity for a tetravalent vaccine stimulating balanced immune responses across DENV serotypes [41] Eliciting immunity with a live attenuated tetravalent vaccine (LATV) is, by far, the most popular strategy and such candidates, attenuated through various means, are currently the most advanced in clinical development (see Section 1.3.2.) Since they replicate within the host and are structurally and antigenically similar to WT viruses, LATVs are thought to be the best means

of inducing a robust and durable immune response similar to that of a natural infection However, these candidates face a common problem of unbalanced immune responses, which is believed to originate from viral interference – a phenomenon characterised by one (or more) vaccine serotype outcompeting the others, thus leading to a bias in the antibody response [42] Differences in the degree of attenuation and infectivity may give certain serotypes a proliferative advantage within the host Thus, the dose of each monovalent component must be carefully adjusted to minimise variations in immune responses Even so, the interactions between viruses can be complex and unpredictable, which makes avoidance of interference difficult [43, 44]

1.3.1.2 Immune correlates of protection

Another challenge in vaccine development is that immune correlates of protection are not completely known Most vaccines in use today rely on the induction of antibodies to control infection [45] Strong evidence from animal experiments supports the role of neutralising antibodies in protection against

for assessment of vaccine immunogenicity [46-48] However, in the context of human vaccination or infection, a correlation between defined levels of neutralisation and protection against infection or severe disease remains to be established [47, 49] For instance, the CYD-TDV vaccine from Sanofi Pasteur was able to elicit high neutralising titres against DENV2 (geometric mean PRNT50 titre of 310) but failed to confer significant protection against DENV2 infection in a phase 2b trial conducted in the Muang district, Ratchaburi province, Thailand [47] This lack of correlation can perhaps be explained by the inability of the plaque reduction neutralisation test (PRNT) to assess other aspects of the anti-DENV antibody response, such as ADE, antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-fixation [1] Additionally, using PRNT titres as the sole readout essentially neglects the contribution of other immune effectors in disease control or progression DENV-specific T cells, for instance, appear to be a pivotal component of the overall immunological response following infection or vaccination (see Section 1.2.1.2.) In conclusion, although PRNT50 titres are widely adopted as

an indicator of vaccine immunogenicity, they provide a narrow and overly simplistic view of the scope of immune responses and thus may be of limited value

1.3.1.3 Animal model

Preclinical testing of DENV vaccines is hampered by the lack of a suitable animal model Dengue fever is strictly a human disease – although infection experiments have been performed in mice and non-human primates (NHPs), none of these animal models are able to recapitulate the full clinical spectrum

of the human infection Due to the close genetic relationship between primates and humans, NHPs are widely regarded as valid models for studying DENV [50] The antibody response that NHPs develop is similar to that of humans and increase in viraemia has been demonstrated in NHPs with circulating antibodies against a heterotypic virus Nevertheless, viral replication in NHPs

is much lower than in humans and limited to lymphoid-rich tissues Additionally, infection in NHPs is typically asymptomatic [51]

For convenience, ease of handling and lower cost of maintenance, a small animal model is favoured over NHPs The main difficulty with developing a mouse model is the inability of human clinical isolates to replicate successfully in rodents [52] Initial models used intracranial (IC) injection of high-dose neurovirulent DENV strains into suckling mice and adult immunocompetent mice, resulting in neurological manifestations such as encephalitis and paralysis, which are not relevant to the human disease [51, 52] The AG129 mouse strain (lacking receptors to IFN-α/β and γ but possessing intact adaptive immunity) was the first to be shown susceptible to mouse-adapted New Guinea C (NGC) virus administered via the intraperitoneal (IP), rather than IC, route [53] High virus titres were detected

in serum and spleen shortly after infection but mice exhibited irrelevant neurological abnormalities a few days before death Further refinement led to the development of vascular leak syndrome, without neurological manifestations, following lethal intravenous (IV) administration

clinically-of 107 PFU mouse-adapted DENV2 strain D2S10 [54] Vascular leakage is a hallmark of severe dengue disease in humans; development of vascular leak in AG129 therefore significantly increased the relevance of this model for the study of dengue pathogenesis At a lower dose, however, D2S10 was non-lethal and induced clinically irrelevant paralysis Our laboratory established an AG129 model with a non-mouse-adapted clinical isolate D2Y98P that produced vascular leak after IP injection with a low dose (104 PFU) of virus [55] The uniqueness of this model lay in the onset of severe disease after the viraemic phase, which resembled the disease kinetic in humans Other mouse models, in contrast, exhibited severe disease during peak viraemia Using a plaque-purified clone D2Y98P-PP1 (Genbank accession #JF327392), we were able to reproduce the symptoms demonstrated earlier with the parental virus via the more clinically relevant subcutaneous route, thus making this model increasingly useful for studying disease pathogenesis [56]

The absence of functional type I & II IFN responses is a major weakness of the AG129 mouse model, since dengue patients are generally immune competent [57] A possible alternative is the A129 mouse strain, which is

responses to a DENV infection Compared to the AG129, though, the α/βR-/- strain is less permissive to DENV infection Initial tests with IP injection of DENV2 NGC showed that a combination of IFN-α, -β, and -γ abnormalities was necessary to induce a lethal phenotype [53] Recently, Sujan Shresta’s group was able demonstrate lethality in A129 using a DENV2 plaque-purified clone (S221) administered IV at a dose 200 times higher than that required for AG129 [25] Further characterisation of this mouse strain is necessary to determine its suitability as a DENV infection model

IFN-Humanised mice are also being established as DENV models Bente et al

(2005) infected non-obese diabetic/severely compromised immunodeficient (NOD/SCID) mice xenografted with human CD34+ cells with DENV via the subcutaneous (SC) route and observed clinical signs of DF (fever, rash, and thrombocytopenia) [58] The major weakness of this model is the absence of a human immune response To address this problem, Ramesh Akkina’s group reconstituted human haematopoiesis and immune cells in RAG2−/−γc −/− mice

by engrafting the animals with human CD34+ haematopoietic stem cells Following a DENV2 challenge, viral replication and DENV-specific antibodies were detected in sera of infected mice, though not all animals exhibited a class switch from IgM to IgG [59] Using cord blood

haematopoietic stem cell (HSC)-engrafted NOD-scid IL2rγ null (NSG) mice,

Jaiswal et al (2009) demonstrated DENV infection of engrafted mice and

DENV-specific HLA-A2-restricted T-cell function and modest antibody responses [60] This model was further improved upon by co-transplantation

of human foetal thymus and liver tissues into NSG mice, which led to enhanced DENV-specific antibody titres and generation of neutralising antibodies [61]

Considerable effort has gone into the development of a suitable animal model

to study DENV infection and disease While different models have reported varying levels of success, none of them are able to recapitulate the full clinical spectrum of the human infection, as well as the complete scope of the human immune response Thus, results from animal studies, while informative, must still be interpreted with caution

1.3.2 Vaccine candidates in development

1.3.2.1 LATVs

The high global prevalence of DENV infection and disease represents an unmet public health need for a vaccine that confers protection against all serotypes Despite the many difficulties of research and development, the goal

of achieving a safe and efficacious DENV vaccine is still being vigorously pursued by many groups Several candidates are in various stages of preclinical or clinical development (Table 1.1) and LATVs are currently in the lead A variety of methods have been employed to attenuate the virus sufficiently for human use The classical approach of serial passage in primary dog kidney cells was adopted by WRAIR/GSK to generate their vaccine candidate [62] Inviragen took the classically-attenuated DENV2 strain developed at Mahidol University and substituted the prM and E genes with the corresponding ones from DENV1, DENV3 or DENV4 [63] Similarly, Sanofi Pasteur took the yellow fever vaccine strain YFV 17D, already approved for human use, and replaced prM and E genes with those from DENV [64] NIAID/NIH chose to genetically attenuate the virus by introducing a 30-nucleotide deletion (Δ30) in the 3′ untranslated region (UTR) of DENV1 and DENV4 Since the same technique failed to sufficiently attenuate the remaining two serotypes, structural genes of DENV2 and DENV3 were cloned into the rDEN4Δ30 backbone [65]

1.3.2.2 Subunit vaccines as an alternative

Development of LATVs has been plagued by the problem of interference and unbalanced antibody responses Booster immunisations are necessary to improve the antibody responses against weaker serotypes However, subsequent immunisations can only be performed after sterilising immunity has waned [1] since heterotypic responses elicited by stronger serotypes will result in rapid clearance of the inoculum For instance, early results from a clinical trial of the PDK-passaged LATV candidate from WRAIR showed that

immunogenicity [66] Instead, a booster after six months showed an improvement in seroconversion rates across all serotypes [67] Prolonged schedules are difficult to implement and can result in poor patient compliance, particularly in developing countries where DENV is endemic

An alternative to the LATV is a recombinant subunit vaccine Majority of such vaccines are in early stages of development with only a few having entered Phase 1 clinical trial (Table 1.1) The non-replicative nature of a subunit vaccine allows for simple adjustments of dose to achieve balanced antibody responses In contrast, adjustment of viral doses does not always result in the desired outcome because of unexpected interactions between viruses [43] Moreover, subunit vaccines are one of the safest alternatives to LATV – since they do not replicate within the host, they are suitable for use in immunocompromised individuals [42] A major weakness of the subunit vaccine, however, is the limited number of viral antigens that it contains This

is in contrast to whole virus vaccine which presents a broad range of antigens

to the immune system and thus generates a diverse antibody repertoire The scope of immune response elicited by a subunit vaccine is expected to be narrow due to limited antigenic diversity Nevertheless, this provides an opportunity to focus the immune response on protective antigens and omit others that might enhance infection For instance, the pre-membrane (prM) protein was shown to elicit highly cross-reactive antibodies that potently promote ADE [68], making it a poor choice for a subunit vaccine component Instead, nearly all dengue subunit vaccines are based on the envelope (E) protein [46, 69-71] The features of the E protein that make it an attractive vaccine candidate will be described in detail in the following Section