Isolation of broadly cross reactive human antibodies against dengue virus using a human non immunized fab phage display library

It is believed that during the transport of the immature virion through the ER and Golgi apparatus to the cell surface, prM protects E from undergoing the low-pH induced conformational

Introduction

1.1 Epidemiology

Dengue virus (DENV) is an arthropod-borne virus belonging to the

Flaviviridae family Consisting of four related but antigenically distinct

serotypes (DENV1-4), they rely on mosquito vectors that live in close

association with human for effective transmission DENVs are of immense

global clinical importance; approximately 50-100 million cases of DENV

infections worldwide and about 500,000 cases of the life-threatening form of

severe dengue – dengue haemorrhagic fever (DHF) and dengue shock

syndrome (DSS) occur annually (WHO, 1997a) Dengue is now considered to

be the most rapidly spreading vector-borne disease, posing a serious public

health threat (Pinheiro et al., 1997)

The World Health Organization (WHO) estimates that at least 100 countries are endemic and about 40% of the world’s population is at risk of infection each year (Tan, 1997) Its global emergence and re-emergence may

be a result of multiple factors, including unplanned and uncontrolled rapid

urbanization combined with inadequate wastewater management, a lack of

effective mosquito control leading to the increased distribution and density of

vector and global climate change; all of which combined to result in increased

spread of the virus (Kyle et al., 2008; Pinheiro et al., 1997) Increased

movement of people among population centres via modern transportation may

also contribute to the resurgence of dengue (Gubler, 2002b) Additionally,

some studies have suggested that the resurgence of dengue infections may be

due to microevolution of the virus, resulting in more virulent strains replacing

the less virulent genotypes (Klungthong et al., 2008; Steel et al., 2010;

Weaver et al., 2009)

Substantial economic, political and social effects associated with major

dengue urban epidemics, such as those seen in Cuba (1977-79, 1997)

(Guzman et al., 2008), Delhi (1996) (Kabra et al., 1999), Taiwan (2002) (Teng

et al., 2007) and Brazil (2008) (Honorio et al., 2009) all add to the growing

concern of the disease Dengue is also a leading cause of morbidity in tourists

and military personnel who travel to dengue endemic areas (Gubler, 2002a;

Webster et al., 2009) With no approved vaccine or cure, dengue thus

continues to pose as a major public health problem Although significant effort

and resources had been applied towards DENV vaccine development over the

last 3 decades and several candidates are now entering late phase clinical

trials, a safe and efficacious vaccine is not likely to be ready in the near future

It is thus vital to improve our understanding of the pathogenesis and life cycle

of DENV to reveal potential targets which may be of use in the development

of a successful vaccine and be potentially of use in the development of

chemotherapy

1.2 Mosquito vectors

Transmission of dengue is dependent on mosquitoes belonging to the

genus Aedes A albopictus and A polynesiensis can sustain transmission, but

the primary and most important vector for DENVs is A aegypti, whose

behavior and bionomics contributed to the highly efficient inter-human

transmission of dengue This mosquito is highly adapted to the urban

environment and prefers artificial water containers such as flower vases or

coconut shells as larval habitats; its eggs are able to survive for a long period,

withstanding desiccation, therefore able to exploit increased breeding sites

provided by uncontrolled and unplanned urbanization (Halstead, 2008) Adult

A aegypti uses human habitations as resting and host-seeking habitat, where

the female mosquito feed on human hosts for blood as the source of protein for

oogenesis and energy for flight, maximizing human-vector contact and

minimizing exposure to insecticides sprayed outdoor (Weaver et al., 2010) In

addition, A aegypti is a nervous feeder which often feeds on multiple human

hosts in one single meal, increasing the number of hosts and the probability of

becoming infected Recently, studies have reported the increasingly similar

breeding behaviour of A albopitus to A aegypti, shifting from an outdoor to

indoor domestic environment, causing a greater risk of dengue transmission

(Dieng et al., 2010; Rao et al., 2010)

The environment also has a part to play as studies have shown that dengue

epidemics tend to coincide with rainy seasons, corresponding to observations

of significant increase in mosquito larval populations (Rappole et al., 2000)

Furthermore, the rate of viral propagation in mosquitoes is influenced by

ambient temperature and relative humidity, where a rise in environmental

temperature shortens the extrinsic incubation period, ie the time needed for a

female mosquito to becoming infective after acquiring an infectious

bloodmeal (Halstead, 2008) Various groups have reported on meteorological

influences on the abundance of the adult dengue vector, and efforts to develop

predictive models to instigate pre-emptive vector control operations have been

reported (Chadee et al., 2007; Dibo et al., 2008; Favier et al., 2006) While

positive relationships between rainfall, temperature, humidity and dengue

vector abundance have been reported by several groups, the relative

significance of each of these factors varied among the different reports These

conflicting observations may be due to the differing local ecology of the areas

under study, with these three variables impacting A aegypti populations in

slightly different ways; highlighting the need to investigate the diversity of

relationships between entomological and meteorological indices at local

ecological scales (Azil et al., 2010)

Community-based programs aimed at vector control through the

elimination of A aegypti breeding sites or by application of larvicides have

been carried out in the Americas (1940s – 1970s) (Soper, 1963; Tsikarishvili

et al., 1964), Cuba (1980s) (Kouri et al., 1989) and Singapore (1960s -1980s)

(Tan, 1997) and were found to be highly effective While these efforts in

Singapore and Cuba continue, those in the Americas were not sustainable as

various factors eroded their effectiveness including development of pesticide

resistance, an awareness of the side-effects of insecticide, decrease in

government funding for public health services and the failure of horizontal

programs integrating education and community to motivate community

participation (Knudsen et al., 1992) As a result, A aegypti rapidly

re-colonised the Americas after the cessation of the A aegypti eradication

program and dengue reappeared Peculiarly, Singapore experienced a surge of

dengue incidences despite low premises index after 15 years of low dengue

incidence This may be due to a combination of factors, including lowered

herd immunity as a result of reduced dengue transmission as well as a shift of

dengue transmission profile, whereby there was increasing virus transmission

away from home, consequently, cases in adults predominated as opposed to

children as in other parts of Southeast Asia (Ooi et al., 2006) Most dengue

infections, particularly in young children, are mild or silent, whereas adults are

more likely to be clinically overt, and may contribute to a resurgence of

dengue incidence (Ooi et al., 2009; Ooi et al., 2003; Seet et al., 2005)

1.3 Clinical manifestation

DENV infections cause a spectrum of clinical outcomes, ranging from

asymptomatic to mild flu-like illness to classical dengue fever (DF) and the

severe dengue hemorrhagic fever (DHF) DF is a self-limiting febrile illness

associated with fever, headache, myalgia, nausea and vomiting, accompanied

by joint pains, weakness, rash and thrombocytopenia The fever may last for 2

to 7 days, with a saddleback pattern, characterized by a drop in fever after a

few days only to rebound 12 to 24hr later DHF often follow secondary

dengue infections, but may sometimes occur in primary infections, especially

in infants (Dietz et al., 1996; Halstead et al., 2002) DHF is characterized by

high fever, haemorrhagic manifestations and circulatory failure as a result of

plasma leakage Increased vascular permeability may lead to shock or dengue

shock syndrome (DSS), which is associated with a very high mortality rate if

not managed properly Although less common, severe dengue infection may

give rise to complications such as encephalitis, hepatitis, myocarditis and renal

dysfunction (Pancharoen et al., 2002)

According to the WHO DHF case definition (WHO, 1997a) developed

in the 1970s, a patient with acute sudden onset of high fever for 2-7 days with

platelet count of less than 100 x 109/L and signs of plasma leakage as well as

haemorrhagic tendency is classified as having DHF (WHO, 1997b) However,

atypical clinical manifestations of dengue infection which can also lead to

poor clinical outcomes are increasingly reported, challenging the current

definitions (Deen et al., 2006) The Increasing trend of adult dengue has

caused a review of the WHO classification scheme, useful in case

management of paediatric patients, as it may underestimate severe dengue in

adults; the course of dengue infection in adults may be complicated by

differences in physiological state compared to children, as well as the greater

prevalence of morbid conditions, such as asthma, hypertension, cardiovascular

diseases and diabetes (Chen et al., 2004; Guzman et al., 1992; Limonta et al.,

2008) In the light of these observations, the WHO has recently reviewed the

dengue guidelines and published a different set classification scheme (WHO,

transcription and translation The ORF is translated as one polyprotein which

is co- or post-translationally processed by host and viral proteins in distinct

cellular compartments into three structural proteins, the capsid (C), precursor

membrane (prM) and envelope (E) as well as seven non-structural proteins,

NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5 (Zhang et al., 2003) (Fig 1)

The three structural proteins are associated with the membrane of the

rough endoplasmic reticulum (rER) by C-terminal membrane-spanning

regions While host signalase in the rER lumen is responsible in cleavages

distal to the membrane-spanning regions of C, prM and E; a cleavage

proximal to the membrane-spanning region of C is carried out by viral

protease in the cytoplasm, both of which are essential for viral particle

assembly (Mukhopadhyay et al., 2005) Upon cleavage, the newly generated

C-terminal of prM, which contains the transmembrane domain, is anchored to

the membrane and serves as the signal sequence for the translocation of E

The prM is able to fold into its native structure independently of E; however,

several studies have shown that E requires the co-expression of prM to acquire

its native conformation, suggesting a chaperone-like role of prM in the folding

of E (Chang et al., 2003; Kim et al., 2008) Partially assembled flavivirus

nucleocapsid, which consists of the RNA genome organized with 180 copies

of C protein, buds from the endoplasmic reticulum (ER), becoming enveloped

by the host-derived lipid bilayer that carries with it the two viral surface

proteins: prM and E, which interact to form prM-E heterodimers, forming

immature virus particles

Shortly before the virus is released from the cell, the 91 residue

precursor portion (pr) of prM in the N-terminal region is cleaved off, leaving

the C-terminal portion, the mature membrane protein (M) anchored in the

membrane Eventually, soluble pr peptide and mature virus are released into

the extracellular environment (Yamshchikov et al., 1993; Yu et al., 2009; Yu

B

Fig 1 Schemetic diagram of DENV genome and polyprotein (Adapted from (Perera et al.,

2008)) (A) The single-stranded positive sense RNA viral genome is ~11 kb in length with a

capped 5’ end (B) Membrane topology of the polyprotein The viral RNA is translated as a single polyprotein and processed by cellular and viral proteases (denoted by arrows) to give

3 structural proteins: capsid (C), pre-membrane/membrane protein (prM/M) and envelope (E); and 7 non-structual proteins: NS1, NS2A/B, NS3, NS4A/B and NS5 Signalase cleavage in the ER release prM and E from the polyprotein, but they remain anchored on the luminal side of the membrane C is also anchored in the ER membrane (on the cytoplasmic side) by a conserved hydrophobic signal sequence at its C-terminal end This signal sequence is cleaved

by the viral NS2B–NS3 protease prM is cleaved into pr peptide and M by host furin during virus maturation in the TGN The non-structural proteins are processed mainly by the viral protease NS2B–NS3 in the cytoplasm with the exception of NS1, which is released from NS2A

by a yet unidentified protease in the lumen of the ER NS2A/2B and NS4A/4B are anchored in the ER as transmembrane proteins

et al., 2008) The cleavage site of prM occurs immediately after the amino

acid (aa) sequence Arg-X-Arg/Lys-Agr (X is variable), which corresponds to

the consensus sequence of the protease furin and this cleavage reaction

probably takes place in the acidic compartment of the TGN; this is supported

by studies showing inhibition of cleavage by raising the pH of the acidic

intracellular compartments using acidotrophs or by furin inhibitors, suggesting

that this cleavage reaction involves furin and is pH dependent (Guirakhoo et

al., 1992; Junjhon et al., 2008; Randolph et al., 1990; Yu et al., 2008) It is

believed that during the transport of the immature virion through the ER and

Golgi apparatus to the cell surface, prM protects E from undergoing the

low-pH induced conformational changes which are required for membrane fusion

and infectivity, thus preventing premature fusion with host membrane, as

immature virions with uncleaved prM are non-infectious, being unable to fuse

with cells even at acidic conditions (Elshuber et al., 2003; Kim et al., 2008)

1.4.2 Structure and maturation

Fusion of viral and host cell membrane is an obligatory step of entry

and subsequent infection for enveloped viruses Viral surface proteins are the

critical agents involved, primed to facilitate fusion of the lipid bilayers and

they are usually produced as inactive precursors which require proteolytic

cleavage to achieve their fusogenic potential The best-studied example is the

influenza virus haemaglutinin (HA), which belongs to the class I fusion

proteins It is synthesized as a single-chain precursor and requires cleaveage

into two chains, HA1 and HA2, before the trimeric HA can become fusion

competent (Gething et al., 1986) HA1 is responsible for receptor binding

which leads to endocytic uptake of the virus; acidification of the endosomes

triggers conformational rearrangement of HA2, resulting in the exposure of

the N-terminal fusion peptide which inserts into the endocytic membrane,

further conformational change results in fusion of viral and host cell

membrane (Harrison, 2008) All class I fusion proteins are two-chain products

of a cleaved, single-chain precursor with a hydrophobic N-terminal fusion

peptide that is liberated by the cleavage Flavivirus fusion proteins on the

other hand, belong to an architecturally and evolutionarily distinct class of

fusion proteins, known as the class II fusion proteins These proteins associate with a second ‘protector’ protein, whereby cleavage primes the fusion protein

to respond to acidic pH (Elshuber et al., 2003; Kielian, 2006; Modis et al.,

2004)

For flaviviruses, the E protein binds cell surface receptors and is

involved in virus entry by envelope fusion with host cellular membrane It is a

glycoprotein of approximately 55kDa with twelve strictly conserved cysteine

residues forming six disulphide bridges (Roehrig et al., 2004; Stiasny et al.,

2006a) X-ray crystallographic analysis of the structure of E protein shows

that each polypeptide chain contains three distinct domains: the

amino-terminal domain I (DI) which is located in the centre of the structure, the

elongated domain II (DII), which contains the fusion peptide (FP – residues

98–109 in DENV2) at the tip and participates in membrane fusion and

dimerization of E proteins; and lastly, the carboxyl-terminal

immunoglobulin-like domain III (DIII) which is believed to be involved in receptor-binding

(Modis et al., 2004; Stiasny et al., 2009) The E protein is able to switch

between different oligomeric states: a trimer of prM-E heterodimers in

immature particles, a homodimer of E proteins in mature virus and as a

homotrimer in post-fusion state (Modis et al., 2004) The dissociation of the E

proteins combined with conformational change of the protein, most probably

facilitates the fusion of the viral membrane with the host endosomal

membrane, allowing the virus to gain entry into the cell (Li et al., 2008;

Mukhopadhyay et al., 2005)

Studies have shown that addition of acidotropic reagents, such as

NH4Cl raises the pH of the TGN and prevents the furin mediated cleavage of

prM, resulting in the release of immature virus particles (Randolph et al.,

1990) It is interesting to note that although the optimum enzymatic activity of

furin is at neutral pH, prM was found to be cleaved only under acidic

conditions (pH 5.0-6.0) Therefore, the pH dependency of the cleavage event

likely reflects accessibility of the cleavage sites on the virus surface,

suggesting that the conformation of immature virus is distinct from that of

mature virus Recent advances in cryoelectron microscopy (cryoEM) have

enabled a better understanding of the mechanisms involved (Yu et al., 2008)

At neutral pH, the immature virion has a diameter of approximately 600Å and

contains 60 prominent spikes, each consisting of a trimer of prM-E

heterodimers In contrast, a mature virion is ‘smooth’-surfaced and has a

diameter of about 500Å It contains a closely packed protein shell made up of

90 E dimers arranged in a herring-bone pattern with 180 copies of M protein

However at low pH, studies have shown that the immature virus undergoes a

major conformational change that involves the E proteins rearranging into a

configuration that is similar to that of the mature virus, resulting in a smoother

surface with a diameter of about 530 Å and may represent the conformation

required for efficient cleavage of prM Upon back-neutralization to pH 7.5,

the smooth particles reverted back to the spiky form (Fig.2), indicating that the

pH induced conformational change is reversible (Guirakhoo et al., 1991; Li et

al., 2008; Yu et al., 2008; Zhang et al., 2003)

Crystal structure of a recombinant prM-E heterodimer at pH 5.5 and

pH 7.0 has revealed that the hydrophobic FP in each of the three E proteins

within the three pairs of prM-E heterodimers is covered by the pr peptides at

neutral pH; in this conformation, furin would be sterically hindered from

binding to any of the three prM-E heterodimers within a spike, explaining the

inability of furin cleavage of prM at neutral pH (Heinz et al., 1994; Yu et al.,

2008) In contrast, the structure of immature virus at low pH shows strong

similarities to the mature virus, whereby prM-E heterodimers dissociate from

the trimeric spike-like organization and form arrangements that are essentially

the same as in the mature virus; with three E-homodimers nearly parallel to each other, lying flat against the virion surface, resulting in a ‘smoother’

particle (Fig.2) (Li et al., 2008; Yu et al., 2008; Zhang et al., 2003) The FP

becomes buried at the interface of the pr and E dimer with the pr part of the

prM peptides showing extensive interaction with the DII of one E monomer

and the DI of a neighbouring E monomer The pr peptide stabilizes the

E-dimer, and prevents its dissociation and the subsequent formation of E

homotrimer; providing an explanation for the stability of immature virus at

low pH, whereas the mature version undergoes membrane fusion (Yu et al.,

2009; Yu et al., 2008) In addition, the furin cleavage site on prM also

Fig.2 Structure of the dengue virion and conformations of the E protein (Taken from

(Perera et al., 2008)) (a) The cryo-EM reconstruction of the immature virion at neutral pH In

this structure, E protein interacts with prM to form heterodimer which in turn forms 60 trimeric spikes that extend away from the surface of the virus This arrangement of E gives the virus a ‘spiky’ morphology and represents the initial particle that buds into the ER The conformation of the E protein (grey) within a spike is shown below the virion The ‘pr’ peptide is shown in blue protecting the fusion peptide on E (shown as a red star) (b) The cryo-EM reconstruction of the immature virion at low pH The virus encounters low pH in the TGN during its transit through the secretory pathway Under these conditions, the prM-E heterodimers dissociate from their trimeric spike-like organization and form 90 dimers that lie flat against the viral surface This orientation of prM-E proteins gives the virion a ‘smooth’ morphology This conformational change is reversible, and upon raising the pH, this ‘smooth’ particle can revert back to its ‘spiky’ morphology (c) While in the TGN, the prM protein is cleaved into its ‘pr’ peptide and M protein by the host furin protease The cleaved ‘pr’ peptide maintains its position as a ‘cap’ on E and the E proteins remain as 90 homodimers lying parallel to the virion surface M, not shown in this figure lies embedded in the viral membrane beneath the E protein shell (d) The cryo-EM reconstruction of the mature virion Following furin cleavage, the mature virion is secreted into the extracellular milleu and the pr peptide is released from mature particle as pH is neutralized

becomes accessible at the low-pH conformation This conformational change

from neutral to low pH was found to be reversible for DENV as long as the

prM is intact However, once pr is cleaved, no further conformational change

was observed when pH was back neutralized (Li et al., 2008; Yu et al., 2008)

Upon cleavage of prM, the pr peptide remained bound to E at low pH,

preventing membrane fusion within the TGN; with release of the mature

virion particle into the neutral pH of the extracellular environment, pr

dissociates from the dimeric E structure and the resulting mature virus is then

able to undergo membrane fusion to begin the next infection cycle (Perera et

al., 2008; Yu et al., 2009; Yu et al., 2008) This is an excellent example of the

dynamic conformational changes of the viral structure that is essential for viral

entry and subsequent disassembly in the host cell

1.5 Primary infection

1.5.1 Receptor mediated cell entry

The initial steps to virus entry are virus attachment to the cell surface

via a cellular receptor followed by penetration into the cytoplasm The

success of these steps is dependent on the tissue tropism and cell type, but the

precise identities of both in DENV primary infection is still not unequivocally

elucidated owing to the great diversity of DENV cellular tropism in vitro;

DENV can replicate in a wide range of primary and continuous cell cultures

derived from many mammalian and arthropod tissues (Acosta et al., 2009;

Acosta et al., 2008b; Lozach et al., 2005) Studies of pathologic specimens

from patients with DHF have found expression of viral antigens in liver,

spleen, lymph node and bone marrow, suggesting many tissues are involved in

vivo (Huerre et al., 2001) Based on limited data from cadaveric human skin

explants, DENV infection was speculated to initiate at the site of inoculation

through Langerhans and dermal/interstitial dendritic cells (DC), spreading to

the aforementioned organs via the lymphatic and circulatory systems (Kou et

al., 2008; Wu et al., 2000) The presence of either a single ubiquitous receptor

or divergent receptors according to the type of host cell was therefore

suggested for DENV For mosquito cells, a tubulin-like protein and a

laminin-binding protein were reported as putative receptors in C6/36 cells for DENV2

and DENV4 respectively (Acosta et al., 2008a) For mammalian cells such as

monocytes, macrophages, DC, B and T leukocytes, endothelial cells and bone

marrow-, hepatoma-, neuroblastoma- and kidney-derived cells, numerous

proteins have been proposed; these include heparin sulphate (HS), a very

ubiquitous glycosaminoglycan present on the surface of cells, heat shock

protein (Hsp) 70 and Hsp90, laminin receptor, mannose receptor,

glucose-regulated protein 78 (GRP78), CD-14 associated protein and DC-specific

ICAM-3-grabbing nonintegrin (DC-SIGN), which is expressed in high levels

in immature DC (Acosta et al., 2008a; Acosta et al., 2008b; Hung et al., 1999;

Lozach et al., 2005)

Real-time fluorescence microscopy study of the early events in

DENV2 have revealed that DENV particles do not bind efficiently to cells,

possibly contributing to the observation of high unit-to-particle ratio, whereby

the concentration of virus particles measured by biochemical assays (protein

content and quantifying PCR) was substantially higher than that deduced by

infectivity assays (plaque assay and infectious centre assay) (van der Schaar et

al., 2007) Collectively, the multitude of conflicting data has led to a proposal

of a multistep process, whereby E interacts with at least two target molecules

on the cell surface: first, an abundant and low affinity attachment receptor,

which may be HS or DC-SIGN, concentrating virus particles on cell surface;

followed by interaction with a second high affinity receptor that mediates

virion internalization (Lozach et al., 2005; Martínez-Barragán et al., 2001)

1.5.2 Virus internalization

Following attachment, it is generally accepted that the virion is

internalized via receptor-mediated, clathrin-dependent endocytosis Several

studies have highlighted the requirement of exposure to low pH for membrane

fusion activity, with evidences provided by diverse experimental approaches,

such as viral entry inhibition by addition of lysosomotropic agents to infected

cells (Randolph et al., 1990), low pH-induced formation of syncytia in

DENV-infected C6/36 cells (Chen et al., 1994) and silencing of vacuolar ATPase

gene by siRNA (Duan et al., 2008) Mutational analysis has also suggested

that a conserved histidine residue in the FP may act as a pH sensor, regulating

rearrangement of the E domains in Tick-Borne Encephalitis virus (TBEV) by

its protonation state (Fritz et al., 2008) and also membrane insertion of the FP

during the dimer–trimer transition, preceding the final hairpin-like homotrimer

(Stiasny et al., 2009) In addition, single virus tracking studies have

demonstrated the strict pH-dependence of fusion that became abrogated with

the addition of weak base during cell entry (van der Schaar et al., 2007)

However, the study has also shown that internalized particles displayed

different types of transport behaviours, with the majority of them following a

long, three-staged transport pattern before fusing with the endosomal

membrane in the perinuclear region, whereas the rest were nearly stationary

and stayed at the cell periphery where fusion occurred (van der Schaar et al.,

2007)

These findings are in concordance with the structural analysis of E

conformational changes when exposed to low pH, which leads to the

reversible dissociation of the E homodimer, resulting in exposure of the FP

When the low-pH treatment is carried out in the presence of liposomes, the

interaction of the FP with lipids triggers an irreversible trimerization of E

ectodomains (termed soluble E (sE)), which remains stably associated with the

membrane (Stiasny et al., 2007) Solubilization of the protein from the

liposomes with nonionic detergents allowed the isolation, biochemical

characterization, and crystallization of this sE trimer (Bressanelli et al., 2004;

Stiasny et al., 2009) Crystal structure determination of this trimeric low-pH

form shows re-orientation of the sE from a horizontal anti-parallel dimeric

conformation to a vertical trimer in which the subunits display a parallel

arrangement, with the FP of the three subunits coming together to form a

membrane-insertable, ‘aromatic anchor’ at the tip of the trimer The aromatic

residues W101 and F108 previously buried in the DI/DIII pocket at the dimer

interface in the neutral-pH form become exposed in this trimeric form,

presumably interacting with aliphatic moieties of the membrane lipid bilayer

(Bressanelli et al., 2004; Modis et al., 2004; Sánchez-San Martín et al., 2009)

(Fig.3)

dII/dIII linker

Fig.3 Conformational rearrangement of protein E in mature DENV (Taken from (Bressanelli

et al., 2004)) Comparison of the overall organization of the protein in the neutral- and

acid-pH forms The ‘top’ and ‘side’ views are indicated in the top and bottom rows, respectively The three domains of sE are labelled dI, dII, and dIII (A) Neutral–pH, dimeric conformation of

sE in a surface representation The carbohydrate residues (labelled CHO) are indicated in pink A ribbon diagram is intercalated between the top and side views, at the same scale and orientation as the foreground subunit in the side view The last amino acid observed in the

crystal structure (K395; Rey et al., 1995) is indicated by an open blue star, labelled C-term

The lipid bilayer is diagrammed at the same scale underneath the dimer in the side view, with the aliphatic region in pale yellow and the lipid head regions in gray (B) Low-pH conformation of sE As in panel A, only one subunit is coloured and the others are shown in white and gray The arrows show the dimensions of the molecule, including all atoms with a Van der Waals radius of 2A° In the side view, the purple region indicates the dIII/stem linker, which ends at the last amino acid indicated by an open red star (labelled C-ter) Note the vertical groove that follows the C-terminus along the interface between neighbouring dIIs in the trimer The lipid bilayer is diagrammed as in (A), indicating the postulated interaction of the fusion peptide loops with the lipid heads

Despite consensus on the pH requirement for viral fusion, the precise

intracellular pathway for DENV internalization still proves to be elusive and

controversial Earlier reports using electron microscopy showed viral

penetration into cytoplasm via direct fusion with plasma membrane in

mosquito and BHK cells (Lim et al., 1999; Se-Thoe et al., 2000) However,

more recent studies highlighted the requirement of receptor mediated, low

pH-dependent endocytosis in HeLa, C6/36, BHK and BSC-1 cells (Acosta et al.,

2008b; Krishnan et al., 2007) This initial endocytic uptake was shown to be

mediated with the use of pharmacological inhibitors of the

clathrin-mediated pathway, chlorpromazine and dansylcadaverine; and with

overexpression of a dominant-negative mutant of the clathrin coat-associated

protein Esp15, which specifically interferes with clathrin-coated pit assembly

without affecting clathrin-independent endocytic pathways (Acosta et al.,

2008b; Krishnan et al., 2007; Suksanpaisan et al., 2009) With the use of

immunocryoelectron microscopy, WNV particles infecting Vero cells were

seen within clathrin-coated pits 2 minutes post-infection and these internalized

particles were shown to be trafficked from early endosomes to late

endosomes/lysosomes by double-labelling immunofluorescence assays and

immunoelectron microscopy performed with anti-WNV envelope or capsid

proteins and endosomal markers (EEA1 and LAMP1) (Chu et al., 2004)

Reports have also indicated a requirement of lipid rafts in successful infection

of human macrophages and mouse neuroblastoma cells (Acosta et al., 2009)

However, conflicting results demonstrating efficient membrane fusion of

DENV2 in cholesterol-depleted and therefore likely lipid raft-depleted insect

and Vero cells respectively were also reported (Acosta et al., 2009;

Umashankar et al., 2008); and an alternative non-classical pathway,

independent of clathrin, caveolae and lipid rafts but dependent on dynamin in

Vero and A549 (human alveolar adenocarcinoma) cells was proposed for

DENV2 (Umashankar et al., 2008) Elsewhere, differences in receptor usage

in human liver HepG-2 cells by different DENV serotypes was also

demonstrated, whereby laminin receptor and GRP78 were identified as

DENV1 and DENV2 receptors, respectively (Martínez-Barragán et al., 2001;

Thepparit et al., 2004; Upanan et al., 2008) Furthermore, the glycosylation

state of the two potential N-linked glycosylation sites on E were also

speculated to affect viral entry by numerous studies, in which DENV with

varying glycosylation states were reported to display diverse phenotypic

changes, such as fusion activity, tropism, virulence and morphogenesis

(Bryant et al., 2007; Guirakhoo et al., 1993; Johnson et al., 1994; Mondotte et

al., 2007).The variations in receptors used by different DENV serotypes may

be ascribed to differences in the E glycoprotein responsible for interaction

1.6 Secondary infection

1.6.1 Antibody-dependent enhancement (ADE)

Epidemiological studies have shown that infection with one serotype

of DENV results in lifelong immunity to homotypic infection, serotype

cross-reactive protection exists early after primary infection but wanes about 6

months post-infection, leaving the patient susceptible to the infection by the

remaining 3 serotypes (Burke et al., 1988; Guzman et al., 1990; Halstead et

al., 1969; Kliks et al., 1989; Sabin, 1952; Sangkawibha et al., 1984; Thein et

al., 1997) In fact, the life threatening DHF and DSS, often accompanied by a

high level of circulating viruses, is epidemiologically associated with

heterotypic sequential infection (Goncalvez et al., 2007; Vaughn et al., 2000;

Webster et al., 2009) Roughly over 80% of DHF cases occur in secondary

heterologous DENV infections, as demonstrated by several cohort studies,

associating viral burden, presence of dengue antibodies in pre-illness serum

and disease severity A hallmark prospective study by Kliks et al suggested

that the ability of dengue antibodies in pre-illness serum to enhance DENV

infection of monocytes correlated with disease severity (Kliks et al., 1989);

several dengue outbreaks in Cuba also demonstrated an epidemiological

connection between secondary infections with large outbreaks of DHF/DSS

(Guzman et al., 2008; San Martín et al., 2010) The remaining 10% of DHF

cases occur with primary infection, usually in infants more than 6 months old,

and is postulated to be caused by circulating non-neutralizing

maternally-derived dengue antibodies (Chau et al., 2009; Halstead, 1970; Halstead et al.,

2002; Kliks et al., 1989), although a recent study raises some doubt on

whether non-neutralizing antibodies alone can mediate severe illness in infants

(Libraty et al., 2009) The antibody-dependent enhancement (ADE)

hypothesis was proposed in the 1970s to provide a possible explanation for

these observations (Halstead, 1988) In this hypothesis, interaction between a

pre-existing sub-neutralizing level of dengue antibodies and a heterologous

DENV results in enhanced viral entry into FcR-bearing cells, such as monocytes, via the Fcγ receptor (FcγR); leading to increased viral load, profound immune activation and greater disease severity ADE was found to

be mediated by the Fc region of IgG antibodies as Fab fragments or IgG

mutants which abrogated the interaction with FcγR failed to cause

enhancement (Goncalvez et al., 2007)

FcγRs comprise a multigene family of integral membrane glycoproteins that upon binding to the Fc portion of antigen-complexed

immunoglobulins (Ig) G, exhibit complex activation or inhibitory effects on cell functions Two classes of human FcγRs predominate the surface of

DENV permissive cells: FcγRI and FcγRIIA (Anderson et al., 1986; Boonnak

et al., 2008; Kou et al., 2008; Mady et al., 1991; Rodrigo et al., 2006) FcγRI

is a 72-kDa protein found exclusively on antigen-presenting cells (APCs) and

exhibits high affinity for monomeric IgG-1; while FcγRIIA is a 40-kDa

protein which is more broadly distributed and preferentially binds multivalent IgG Each FcγR is composed of 3 domains: an extracellular specific ligand-binding domain, a short transmembrane domain and a cytoplasmic tail The

FcγRs modulate cell metabolism and physical behaviours when triggered by

receptor cross-linking via the immunoreceptor tyrosine-based activation motif

(ITAM)-containing γ-chain, which links the FcγR to tyrosine kinase-activated

signaling pathways and internalization of the ligand-complexed receptor either

by phagocytosis or endocytosis (Fanger et al., 1997; Libraty et al., 2001)

Anti-DENV antibodies at subneutralizing levels have been shown to augment DENV infection in FcγR-bearing cells mediated via both FcγRI and FcγRIIA where virus replication was measured by immunofluorescence and

plaque assays (Kontny et al., 1988; Libraty et al., 2001) To examine the

relative efficiency of each FcγRs individual enhancement on DENV complex infectivity and to inquire whether FcγR activation is required for

immune-were expressed in non FcγR bearing COS-7 cells and DENV infectivity was

measured using plaque assay and flow cytometry (Guzman et al, 1990) It

was shown that enhanced immune-complex infectivity mediated by FcγRI was greatest when associated with the native γ-chain, but instead of total elimination, abrogation of γ-chain ITAM signaling capacity only reduced the enhancement FcγRIIA was shown to be much more efficient at enhancing DENV immune-complex infectivity, despite ITAM-signaling incompetency

which led to impaired phagocytosis, immune enhancement appeared to be

unaffected It seemed that 2 internalization mechanisms were responsible for the above observations: the first being a γ-chain signaling-dependent pathway where DENV immune-complex aggregated to sufficient size and triggered a classical phagocytosis entry pathway; the second is less efficient and γ-chain-independent, which relied simply on the receptors to concentrate partially

neutralized virions on the cell surface for entry by a parallel endocytosis pathway Based on these observations, the authors speculate FcγRIIA to be better equipped than FcγRI in utilizing alternative signaling pathways and entry mechanisms, since it preferentially binds immune-complexes with a high dissociation rate constant, while FcγRI has high affinity for monomeric IgG

(Rodrigo et al., 2006)

Recently the ability of anti-prM antibodies to augment infectivity of poorly

infectious immature DENV to the same level as wild type virus particles in

FcR-bearing cells in a furin-dependent manner has been highlighted

(Rodenhuis-Zybert et al., 2010) It was proposed that following FcR-mediated

trafficking of the immature virus to endosomes, exposure to the lower pH of

the endosome triggered conformational rearrangement of the viral particle,

exposing the furin cleavage site of prM; this would enable cleavage by furin

shuttling between the TGN and endosomes, allowing maturation of the virus

and fusion with the endomembrane (Molloy et al., 1999; Sariola et al., 1995;

Zhang et al., 2003) However, this observation does not reconcile with the

recent study on proteolytic maturation of DENV, whereby pr peptides were

reported to remain associated with the virion upon cleavage and only released

after back-neutralization to pH 8.0 (Yu et al., 2008) Two mechanisms to

account for the removal of the pr peptide for maturation of the virus

intracellularly have been postulated; in the first pr peptides stabilize E proteins

to an extent that they survived the mildly acidic lumen of the early endosome

and that the pr peptides were released at the more acidic pH of the late

endosomes (~pH5.) without the requirement of back-neutralization; in the

second cleavage, the pr peptides associate with the anti-prM antibodies instead

of the E protein, freeing the E proteins to adopt the fusion-active conformation

(Rodenhuis-Zybert et al., 2010) While the exact mechanism remains to be

elucidated, this study highlights the potential of immature DENV to be highly

infectious and its possible contribution to disease pathogenesis

In another study, a panel of mouse-derived monoclonal antibodies was

generated from DENV2-infected mice, and both anti-E and anti-prM

antibodies were found to enhance DENV infection in a

concentration-dependent manner mediated by the FcγRIIA pathway (Goncalvez et al., 2007)

This study also showed that anti-prM enhanced DENV infection in cells

lacking FcRs, such as BHK-21 or A-549 cells, in this case binding to the cell

was mediated by the ability of the anti-prM antibodies to cross-react with host

hsp 60 (Huang et al., 2006) It appears that in this case, ADE of DENV

infection may be mediated by the antibody’s dual specific binding on DENV and target cells, bridging the virion to its receptor, suggesting an additional

pathway for DENV binding and infection via enhancing antibodies

Apart from increased viral uptake, ADE has also been shown to

enhance infectivity through suppression of the intracellular production of

antiviral mediators, as well as increased production of vasoactive cytokines

and chemokines When mouse macrophages were infected with Ross River

virus complexed with enhancing antibodies, interferon (IFN) transcription

factors, STAT-1 and NF-κB complexes were suppressed, as opposed to the

up-regulation observed in cells infected with virus only (Mahalingam et al.,

2002) Similarly, an inverse correlation between plasma nitric oxide, a potent

inhibitor of viral replication, and DENV RNA copies in secondary DHF

patients was demonstrated in a study which focused on post-entry events in

ADE DENV infection (Charnsilpa et al., 2005) The suppression of key

antiviral cytokines, such as interleukin (IL)-12, IFN-γ and tumour necrosis

factor (TNF)-α and an up-regulation of anti-inflammatory cytokines

expression like IL-6 and IL-10 may be the cause of increased viral production

observed The modulation of cytokine levels may in turn modify innate and

adaptive intracellular antiviral mechanism, such as the decreased production of

antiviral nitric oxide (NO) free radicals and preferentially induction of a

Th2-type response in monocytic cells Similar findings have been observed in

human macrophages (Ubol et al., 2008; Ubol et al., 2010)

1.6.2 Cytokine-mediated immunopathology

The enhancement of disease severity in secondary DENV infection

may also result from T cell immunopathology, whereby increased dengue viral

antigen presentation to T cells leads to rapid activation and proliferation of

CD4+ and CD8+ T cells and release of pro-inflammatory cytokines

(Chaturvedi et al., 2007) This cytokine-mediated immunopathology is

postulated to be driven by cross-reactive memory T cells according to the

‘original antigenic sin’ model, in which low-affinity memory T cells produced

in the primary infection were proposed to have selectively expanded during

the secondary heterologous infection before the nạve T cells of higher affinity

specific to the heterologous secondary DENV serotype could be activated

These cross-reactive T cells seem to show suboptimal degranulation but high

cytokine production and are apoptotic, which may contribute to the

development of the vascular leakage characteristics of DHF (Appanna et al.,

2007; Beaumier et al., 2008; Imrie et al., 2007) A high level of T cell

activation and rapid cell death, coupled with a dominating cellular immune

response by T cells with low affinity to the infecting virus may lead to

suppressed or delayed viral elimination, resulting in high viral load and

increased immunopathology

1.6.3 Autoimmune disorder

In addition to the mechanisms mentioned above, autoimmune

responses induced by the cross-reactivity of antibodies generated against viral

components, such as non-structural 1 (NS1) protein, may also contribute to

dengue pathogenesis Anti-NS1 antibodies have affinity for human

fibrinogen, thrombocytes and endothelial cells, leading to platelet lysis and

nitrogen oxide (NO)-mediated apoptosis of endothelial cells, contributing to

thrombocytopenia and vascular damage (Falconar, 1997; Lin et al., 2008; Lin

et al., 2006)

1.6.4 Role of Complement components

The complement pathway can limit Flavivirus infection by stimulating

adaptive immune responses and augments immune serum-mediated

neutralization of YFV, DENV, and Kunjin virus (Della-Porta et al., 1977;

Sabin, 1950; Spector et al., 1969) Similarly, the protective efficacy of

Flavivirus neutralizing antibodies in vivo correlates with IgG subclasses that

efficiently fix complement (Schlesinger et al., 1995) Furthermore, recent

studies indicate that complement can restrict ADE of WNV and DENV

infection in FcR-expressing cell lines and primary macrophages

However, a pathological role for complement activation has also been

suggested As key modulators of the immune system, complement-derived

proteins are capable of affecting a huge range of cell types and are implicated

to play a role in dengue infection pathogenesis For example, C3a, an

anaphylatoxin generated by the cleavage of complement component 3 (C3),

serves to recruit DENV-susceptible immune cells, such as monocytes and

DCs; it regulates vasodilation and increases permeability of blood vessels; it

also acts on specific receptors to produce local inflammatory responses110

The detection of large amounts of C3a, in patients with severe dengue,

revealed a role of complement in dengue pathogenesis (Malasit, 1987;

Nascimento et al., 2009) Furthermore, co-localization studies have shown

that membrane-bound NS1 and anti-NS1 antibodies were involved in the

activation and formation of complement attack complex (Avirutnan et al.,

2006) Soluble NS1 were also reported to differentially bind to cultured

endothelial and mesothelial cells (Avirutnan et al., 2007) and high levels of

intravascular soluble NS1, as observed in DENV-infected patients, may bind

to selective cells and contribute to tissue-specific vascular leakage that occurs

during severe secondary DENV infection due to recognition by anti-NS1

antibodies and immune complex formation (Avirutnan et al., 1998; Avirutnan

et al., 2006)

1.7 Antibody-mediated neutralization

1.7.1 Epitope localization of neutralizing antibodies

Humoral immunity is an essential part of host protection against

flavivirus infection, with studies demonstrating the protective effect of

passive-transferred immune sera or prophylactic/therapeutic monoclonal

antibodies in infected B cell-deficient mice and encephalitis mouse models

respectively (Diamond et al., 2003; Roehrig et al., 2001) In general,

antibody-mediated control of flaviviral infection in vivo has been correlated

with in vitro neutralization The majority of flavivirus-neutralizing antibodies

recognize the E protein, with the most potent ones being serotype-specific,

mapped to the upper lateral ridge of the putative receptor-binding DIII

(DIII-lr), and shown to be effective as passive prophylaxis or therapy in rodent

models (Beasley et al., 2002; Goncalvez et al., 2008; Oliphant et al., 2006)

The less neutralizing antibodies tend to be cross serotype-reactive and

predominantly localized to DII, near or within the highly conserved FP (Crill

et al., 2004; Goncalvez et al., 2004; Gromowski et al., 2007; Stiasny et al.,

2006b), and the hinge region between DI and DII (Oliphant et al., 2006);

although investigators have also mapped sub-complex cross-reactive

antibodies to DIII, which unlike the type-specific antibodies, are only weakly

to moderately neutralizing (Sukupolvi-Petty et al., 2007)

On the other hand, antibodies against NS1 were reported to protect

against infection in vivo through FcγR-dependent and –independent

mechanisms (Chung et al., 2006) Antibodies to prM are generally poorly

neutralizing (Puttikhunt et al., 2008), but have been reported to be protective

in some studies through mechanisms that remain unclear (Bray et al., 1991;

Falconar, 1999; Vázquez et al., 2002)

1.7.2 Blocking at pre- and post-attachment step

Functional characterization of anti-E antibodies on cells lacking FcRs

revealed that the majority neutralize viral infection by blocking binding of the

virus to its cellular receptors, such as highly-sulphated HS (Della-Porta et al.,

1977; Schlesinger et al., 1995; Mehlhop et al., 2007; Yamanaka et al., 2008)

The complement system increases this mode of neutralization by modulating

the occupancy requirements, such as increasing antibody avidity or the steric

effects of bound antibody, resulting in more efficient blockade of virus

attachment (Pierson et al., 2008) It has also been shown that antibodies can

neutralize infectivity post-attachment (Duan et al., 2009; Vogt et al., 2009),

for example the WNV DIII-specific E16, which through structural and

cryo-electron reconstruction analysis, was suggested to prevent fusion by imposing

steric constraints to the low-pH mediated rearrangements of E (Kaufmann et

al., 2006)

1.7.3 Antibody occupancy and affinity

Flavivirus neutralization is a ‘multiple’-hit phenomenon requiring engagement by more than a single antibody at one time Neutralization occurs

when the number of antibodies bound to an individual virion exceeds a required ‘threshold’ In this regard, the neutralization potential of an antibody

is determined by its affinity and the abundance/accessibility of its epitope on

the virion (Diamond et al., 2008; Pierson et al., 2008; Pierson et al., 2007; van

der Schaar et al., 2009) In addition, the pseudo-icosahedral arrangement of E

proteins on the virion means they are displayed in three distinct chemical

environments defined by proximity to the two-, three- and five-fold axes of

symmetry (Kuhn et al., 2002), possibly causing epitopes in each of these

environments to be differentially accessible for antibody binding due to steric

constraints imposed by the adjacent E Thus, the number of sites available for

binding may differ among distinct epitopes on the virion, as such antibodies

binding to the highly accessible DIII-lr can exceed the required threshold by

binding to a fraction of available targets; whereas antibodies against the

poorly-exposed epitopes requires >99% occupancy for complete neutralization

(Diamond et al., 2008; Nelson et al., 2008)

Some epitopes may not be accessible to antibody binding with a

stoichiometry that exceeds the neutralization threshold, therefore, even at an

antibody concentration that permit saturation, viral neutralization cannot be

achieved because too few antibodies can dock on the virus simultaneously

(Nelson et al., 2008; Pierson et al., 2007) Paradoxically, many antibodies

that recognize poorly accessible epitopes on a mature virion can still show

neutralizing activity in vitro and in vivo (Crill et al., 2004; Pierson et al., 2007;

Stiasny et al., 2006b) To reconcile this, virion maturation was demonstrated

to significantly reduce the neutralization potency of anti-WNV against

poorly-exposed epitopes on mature virus, although these antibodies retained their

affinity and capability to bind the virion (Roehrig et al., 2001) It was

proposed that maturation reduced the number of antibodies that may

simultaneously bind the virion, so that even at full occupancy, the required

neutralization threshold still cannot be fulfilled Therefore, depending on the

proportion of the virion at differing maturity states in a population, the same

antibody may have little to significant neutralizing activity

Since the E arrangement on a ‘partially mature’ virion is not yet resolved, it remains unclear how the presence of prM increases the

accessibility of epitopes that are poorly accessible on a mature virion;

although it has been suggested that transiently exposed determinants that result

from dynamic and/or lateral movement among E on the virion that is not

evident from existing structural studies which are static models of E

arrangement may be responsible (Nelson et al., 2008) A model was proposed

in another study, whereby icosahedral viruses are assumed to be able to engage in a ‘breathing’ motion and that such dynamic structural changes may

occur at elevated temperatures to an extent where previously hidden epitopes

become exposed and accessible for antibody binding (Oliphant et al., 2006)

It is presumed that during the ‘breathing’ process, the virion will tend to conserve its icosahedral symmetry and the E proteins will return to their

original positions in the mature virus, but once one or a few antibodies has

bound, the virus would probably not be able to revert to its previous range of

dynamic motions due to the steric hindrance provide by the bound antibody,

causing the rest of the virus to alter its structure to the antibody-bound

conformation Therefore, binding of a few antibody molecules can cause the

rest of E to become more accessible for further binding in a cascade motion

(Lok et al., 2008)

The balance between ADE and neutralization is therefore determined

by the number of antibodies bound to a single virion to achieve the

neutralization threshold Enhancement of infection may occur in vitro in

FcR-expressing cells once the concentrations of antibody fall below the threshold

or by antibodies that can never reach this threshold, even at maximal binding

However, depending on the mechanism of antibody neutralization, the

occupancy requirement may vary; for example, an antibody that inhibits

infection solely by sterically blocking receptor attachment may require a

higher threshold than one that promotes a cascade of E rearrangement; this

may influence the range of concentrations that promote ADE (Pierson et al.,

2008)

1.7.4 IgG antibody subclass and FcγR subtype

DENV neutralization has also been reported to be modulated by the Fc

region in an IgG subclass-dependent manner, where neutralization potency is

proportional to FcγR-DENV immune complex binding avidity (Goncalvez et

al., 2008) In this study, IgG-1, -3 and -4 antibody neutralization was

enhanced in CV-1 (African green monkey kidney) cells transfected with affinity FcγRIA and significantly reduced in low-affinity FcγRIIA transfected CV-1 cells; neutralization by IgG-2 antibody, which has low affinity for both FcγRs was diminished Neutralization in cells without FcγRs was found to be greatest by IgG-3 antibodies, presumably due to the antibody’s inherently

high-greater flexibility and allosteric effect of the IgG heavy-chain constant region

on Fab affinity and specificity (Torres et al., 2008) It was speculated that the

relatively weak binding and fast off-rate of DENV-immune complex to FcγRIIA may be responsible in leading to an antibody-mediated entry pathway that favours DENV infection, whereas DENV-immune complex comprised of

relatively tight binding IgG subclasses was co-internalized with FcγRIA upon

cross-linking and were directed to a virus-destructive intracellular pathway

(Rodrigo et al., 2009)

These studies emphasized the conditional nature of viral neutralization

or enhancement by antibodies and the gap in the understanding of how

antibodies control viral infection, since both homotypic and heterotypic

antibodies will target DENV to FcR expressing cells They also point to a

lack of knowledge on the effect of epitope specificity and antibody affinity of

the IgG in complex with the virus, and the apparent divergence in immune enhancement by different FcγRs

1.8 Human humoral response to Dengue

Our current knowledge of the antibody response to dengue and the

characterization of neutralization epitopes are mostly based on murine

monoclonal antibodies which were generated from animals immunized with E

protein or live virus (Crill et al., 2001; Halstead, 2003) Although invaluable

insights were gained with the help of these antibodies, the human antibody

responses elicited by DENV infections and the target epitopes involved are

incompletely understood

Indeed, serological studies of dengue patients have found that binding

of most primary polyclonal anti-E antibodies was cross-reactive and was

almost completely abolished by single point mutations 101WA and 108FA at

the fusion loop of DII (Lai et al., 2008); whereas murine anti-DII antibodies

generally recognized a wider range of epitopes in and out of the FP (Crill et

al., 2004; Stiasny et al., 2006a), raising the possibility that the spectrum of

human anti-E antibodies generated in primary infection maybe more restricted

than anticipated Polyclonal patient sera from secondary infection, however,

were less affected by mutations at the FP, suggesting a broadening of anti-E

antibody spectrum during secondary infection (Lai et al., 2008)

More recently, studies coupling memory B cell immortalization with a

broad screening approach to isolate large panels of DENV-reactive antibodies

from human donors also indicated that, although some observations with the

human humoral immune response agreed with those seen in studies using

mouse antibodies, deviation in target epitopes and antibodies binding

properties were observed (Beltramello et al., 2010; Lai et al., 2008; Vogt et

al., 2009) The majority of human anti-E antibodies isolated were

cross-reactive and moderately neutralizing, with epitopes mapping to DI/DII of E

protein In addition, DI/DII-reactive antibodies isolated from primary

infections were mostly serotype-specific, whereas those isolated from

secondary infections were all cross-reactive (Beltramello et al., 2010)

Consistently, a study attempting to characterize the specificity and

functionality of antibodies in DENV-immune human sera also demonstrated

that anti-DIII antibodies mainly recognized type-specific epitopes after

primary infection and cross-reactive epitopes after secondary infection

(Falconar et al., 1999)

However, only a minor proportion of human antibodies potently

neutralized DENV-infection of Vero cells and was DIII-specific (Beltramello

et al., 2010) This is consistent with serological studies following either

DENV or WNV infection, which reported an overall prevalence of

DI/DII-specific antibodies and a much lower abundance of DIII-DI/DII-specific antibodies

within the E-specific response (Crill et al., 2009; Lai et al., 2008; Oliphant et

al., 2007; Vogt et al., 2009; Wahala et al., 2009), in accordance with the

reported immunodominance of DI/DII-specific antibodies in recent studies

that generated recombinant antibodies against WNV E using phage display

libraries (Gould et al., 1985; Throsby et al., 2006)

In contrast to DIII-antibodies isolated from immunized mice, which

were usually neutralizing only against a few genotypes within a given serotype

(Shrestha et al., 2010), these human DIII-antibodies were usually sub-complex

reactive and able to neutralize three or even all four DENV serotypes

(Beltramello et al., 2010; Crill et al., 2009; Lai et al., 2008; Throsby et al.,

2006) Remarkably, a study has also shown that depletion of DIII-binding

antibodies only made a minor impact to the total neutralizing capacity of

human immune sera, coinciding with findings using WNV–immune human

sera (Oliphant et al., 2006) This raised the possibility that DI/II- and maybe

M-specific antibodies are mainly contributing to the neutralization activity of

human immune sera and DIII-antibodies may not play as big a part in

neutralization as previously thought

Human antibodies to NS1 had limited cross-reactivity and were mostly

conformational sensitive (Beltramello et al., 2010; Dejnirattisai et al., 2010),

like those against prM and E (Falconar et al., 1991; Lai et al., 2008; Roehrig

et al., 1998) A strong antibody response to prM in both primary and

secondary infection patients was also reported These antibodies were highly

cross-reactive and non-neutralizing even at high concentrations, but potently

enhancing over a broad range of concentrations and are capable of promoting

infectivity of non-infectious immature DENV (Beltramello et al., 2010;

Dejnirattisai et al., 2010); similar to observations reported for mouse anti-prM

antibodies (Huang et al., 2006; Rodenhuis-Zybert et al., 2010)

1.9 Objectives

Despite the fact that DENV vaccines are entering large scale clinical

testing, we still know remarkably little about the complex relationship and

mechanistic details between DENV antibodies in human immune sera and

other immunological components as well as the functional outcome of these

interactions Human monoclonal antibodies are therefore an immensely useful

and versatile tool in the study to unravel these mysteries

Hybridoma technology has revolutionized biochemical research and

diagnosis; many non-human monoclonal antibodies against DENVs obtained

through this conventional technique have proven to be valuable for serotype

determination (Henchal et al., 1982; Henchal et al., 1985) and early diagnosis

(Henchal et al., 1983; Xu et al., 2006) However, non-human antibodies such

as murine antibodies may not be ideal in studies that require efficient human effector function mediation Attempts to ‘humanize’ non-human antibodies by genetic engineering and complementary-determining region (CDR) grafting,

resulted in chimeric antibodies that have human constant region and mouse

variable regions or antibodies that contain antigen-recognising mouse CDR

loop and human variable-domain framework (Brekke et al., 2003b; Kim et al.,

2005; Rajamanonmani et al., 2009) However, given that the human antibody

response elicited by DENV infection appears to be somewhat different to that

generated in mice, this approach is still not ideal

Efforts in isolating human antibodies include immunization of

transgenic mice, whose Ig-heavy chain and Igκ-light chain loci are disrupted

and have transgenes encoding genes for human Ig (Green et al., 1994;

Lonberg et al., 1994) But the process of generating transgenic mice is costly

and laborious; additionally, immune response in transgenic mice is sometimes

less robust and an increased number of immunizations or antibody screens is

known to be required (Yamashita et al., 2007) Attempts in generating human

monoclonal antibodies by hybridoma techniques were unsuccessful due to

poor fusion of human myelomas and instability of human B-cell lines (Stahli

et al., 1980) The overall transformation efficiencies of EBV-immortalization

of B cell lines by traditional methods are extremely low (about 1–3%), and

largely depend on the maturation status of the cell and are frequently

performed after stimulation (Aman et al., 1984; Borrebaeck et al., 1988; Crain

et al., 1989; James et al., 1987; Redmond et al., 1986) But recently, efficient

EBV-mediated transformation of human memory B cells (Beltramello et al.,

2010; Dejnirattisai et al., 2010; Schieffelin et al., 2010; Traggiai et al., 2004)

and single B cell-based human antibody gene cloning using RT-PCR to isolate

monoclonal antibodies derived from single B cells at any maturation stage

(Iizuka et al., 2010) have been developed with success However, all these

methods isolate only antibodies generated by natural human clonal selection,

therefore, the target epitopes of these antibodies are ones that are under natural

immune selection Furthermore, anti-DENV antibodies isolated using these

methods are usually ones generated against one single serotype in the case of

primary infection or 2-3 serotypes from patients with secondary infections

However, with the development of phage-displayed libraries and the

successful expression of antibody fragments in E coli (Jostock et al., 2004),

an additional alternative for the isolation and further genetic modification of

antibody fragments against important pathogens is available The power of

phage display lies in its ability to maintain a physical link between phenotype

and genotype and the flexibility to select for antibodies with a desired

specificity or cross-specificity by sequentially screening against multiple

antigens Successful isolation of anti-DENV antibody fragments from

chimpanzee immune-libraries (Goncalvez et al., 2004) and human

single-chain Fv (scFv) nạve library (Cabezas et al., 2008; Cabezas et al., 2009) have

been reported However, chimpanzee-derived Fabs may present the same

problems as murine antibodies since dengue virus infection does not occur

naturally in non-human primates either; and IgG chains amplified from

immunized-animals can be biased by host immune response (Griffiths et al.,

1998) Whereas soluble scFv may appear to have higher apparent affinities

for they have a tendency to form mulitmers in a clone-dependent manner,

making affinity determination difficult (Holliger et al., 1993; Marks et al.,

1993; Weidner et al., 1992) This tendency to multimerize may also lead to

selection of phages based on avidity instead of affinity during the screening

process, resulting in isolation of clones with very low actual affinity (de Haard

et al., 1999; Griffiths et al., 1998)

The natural antibody repertoire within B cells contains a large array of

antibodies and non-immunized human phage library such as the Humanyx Fab

phage display library was constructed by amplifying the V-genes from the

IgM mRNA of B cells of non-immunized individuals This procedure

provides antibodies that have not yet encountered antigen, which allows

access to a more diverse immune repertoire (de Haard et al., 1999; Klein et al.,

1997) Biopanning with phage display libraries may also allow for selection

of antibodies against unique epitopes that are not under the natural selection

pressure of the human immune system, since the antigen presentation is in

vitro and can be manipulated to isolate antibodies that may not be possible to

raise in vivo (Brekke et al., 2003a) In this case, complex-reactive anti-DENV

antibodies against surface determinants maybe isolated by screening all 4

DENV serotypes sequentially The stringency of the screening procedure,

such as number of washes and amount of antigens maybe adjusted to obtain

high affinity binders Since epitopes that are involved in neutralization are

constantly under selection pressure, they are more prone to mutation, hence, it

is not surprising that many viruses are highly adept at keeping their most

critical (and conserved) determinants of pathogenesis cryptic With the

flexibility to manipulate antigen presentation and diversity of antibodies

repertoire of phage display, we may be able to isolate antibodies with

unprecedented cross-serotype binding and/or neutralization spectrum against

conserved regions on DENVs

Therefore, the objectives of this study are to isolate unique human

anti-DENV antibodies by screening all 4 serotypes of DENV sequentially

using a human non-dengue immunized phage display library (Humanxy

library) and to convert the Fab fragments isolated into full-length human IgG1

in a mammalian system via transient transfection for efficient effector

functions mediation This system would be a useful and rapid alternative

which can complement the existing method of generating human anti-dengue

antibodies by B cell immortalization, with the potential of isolating unique

antibodies against conserved and/or cryptic epitopes that are not selected for in

natural immune clonal selection These anti-DENV antibodies will be fully

characterized and their usefulness will be tested for their potential to be used

in the investigation of DENV pathogenesis