The family Flaviridae contains more than 70 members, many responsible for a considerable proportion of mortality and mortality worldwide, both in humans and animals. The family is divided into three genera: flavivirus, pestivirus and hepacivirus. The biological properties of each genus are distinct, although all share similar structural characteristics and genome organisation. Members of the genus flavivirus have a worldwide distribution although individual viruses are invariably restricted by climate and vector. More than half of the genus flavivirus are associated with human disease, including yellow fever virus, the prototype virus of the genus and an important cause of viral haemorrhagic fever. Other important pathogens in the genus include Japanese encephalitis virus found in Southeast Asia and tick-borne encephalitis virus present in Europe and Northern Asia.
The majority of the viruses in the genus flavivirus are arboviruses—transmitted between mammalian hosts by an invertebrate vector (mosquitoes and ticks). Virus replication takes place in both the vertebrate host and the insect vector. In the case of yellow fever and dengue viruses, these are maintained in the wild by transmission from primates to mosquitoes, and back to primates. Humans are infected as accidental event when intruding into this natural ecosystem. Dengue virus has evolved along with the expansion of major conurbations to the extent that some experts argue a sylvatic cycle through an animal reservoir is no longer necessary for its endemicity.
Flavivirus distribution is dependent on the ecology of their vertebrate and invertebrate hosts. Thus, some have a very restricted geographical distribution, e.g. Omsk haemorrhagic fever is restricted to western Siberia and Kyasanur Forest disease is found so far only in Karnataka (Mysore) state of India. The recent emergence of Alkhurma virus on the Arabian Peninsula is a puzzle, as this virus clearly shows a relationship with Kyasanur Forest virus. Dengue has expanded over the last 20 years from southeast Asia into the Pacific, the Americas, Africa and Australia, progressively establishing itself in local populations of the ubiquitous Aedes aegypti. The rapid expansion of dengue virus coupled to the increasing incidence of dengue shock syndrome and haemorrhagic fever makes this infection a global priority in terms of control and public health. Dengue is regarded as one of the top priorities for public resources by the tropical disease programme of the World Health Organisation.
The ecology of A. aegypti plays a critical part in the epidemiology of both yellow fever and dengue. A. aegypti is unusually small compared to other mosquitoes and is well adapted to virus transmission. It rarely circulates beyond human habitation and is unusually silent in flight, meaning that the female can often seek a blood meal unnoticed, aiming particularly for uncovered legs and ankles. The female can survive for many months, sometimes for a year or even longer. Transovarial passage of the virus from adult to egg ensures that the virus can survive throughout a dry season. Flaviviruses replicate in the salivary glands and are introduced into a new human host together with saliva. Again in contrast to other mosquitoes, A. aegypti lays its eggs individually, dispersed over large areas of still water: this maximises the chance of survival. Although the female avoids dirty or polluted water, eggs can survive in water that has been chlorinated.
Outbreaks of flavivirus infections often coincide with climatic events, for example, heavy rain following a period of drought. This results in numerous stagnant pools, ideal breeding grounds for both mosquitoes and ticks.
Surveillance and epidemiological monitoring are essential for the prevention and management of flavivirus infections. Continual and sustained sampling of arthropod vectors and vertebrate hosts for arbovirus activity is required to ensure such surveillance remains effective. Unfortunately, there has been all too often a heavy reliance on initiating mosquito eradication programmes without due regard for subsequent monitoring of disease and vector absence. The result is that some countries within endemic areas have all but ceased to monitor for outbreaks. Active surveillance can also provide early indications as to the emergence of arboviral and other diseases: Venezuela had ceased surveillance for some years but the emergence of Venezuelan haemorrhagic fever in the 1980s illustrated vividly the need for continual surveillance. This hitherto unrecognised disease was first attributed to dengue virus but more detailed epidemiological investigation uncovered the presence of Guanarito virus, a member of the Arenaviridae family (see Arenaviruses).
Clinical manifestations of the majority of flavivirus infections are relatively non-specific. In endemic areas, diagnosis is often based on clinical suspicion, especially if the attending physician is alerted by epidemic activity in the region. Scarce or non-existent resources in economically underdeveloped regions compound the difficulties in obtaining accurate and detailed data. For some years, the Rockefeller Foundation sponsored the surveillance for yellow fever in South America by providing equipment for collecting liver samples from deceased individuals suspected of having had yellow fever. Fixed tissue sections were then transported to regional laboratories for further analyses. This initiative was the exception, however, and sustained monitoring of yellow fever and other indigenous diseases remains as problematical as ever.
Yellow fever is the only viral haemorrhagic disease against which a vaccine is widely available for human use. Indeed, the 17D yellow fever vaccine has been extraordinarily successful. In contrast, the development of vaccines against dengue virus has been fraught with difficulties, despite the many similarities of these two viruses. The success and failures of developing flavivirus vaccines serve as useful paradigms for the development of rational vaccines against viral haemorrhagic fevers and other exotic infections.
There are many excellent reviews of flaviviruses. Particularly useful as to the scope of the problems these viruses present are the overviews on yellow fever by Monath (2001) and on dengue by Gubler (2002). The World Health Organisation and the United States Centres for Disease control maintain useful websites on dengue (http://www.who.int/health-topics/dengue and http://www.cdc.gov/ncidod/dvbid/dengue).
Yellow fever
Historical background
Yellow fever is regarded as one of the classic diseases of antiquity, instilling dread and terror in the Americas, Europe and Africa in the three centuries following the arrival of the Spanish conquistadors in the New World. The cause remained a mystery until the work of Walter Reed and his team in Cuba in 1900 proved transmission by A. aegypti.
Through the 18th and 19th centuries the colonisation of the Americas was retarded by frequent outbreaks of yellow fever. The Philadelphia outbreak of 1793 was one of many epidemics that hit the then capital of the United States in its infancy. The outbreak was particularly severe, affecting over 10% of the city's 40,000 population. The source was most likely arriving on boats from the West Indies fleeing the yellow fever outbreak of the preceding year. Rudimentary efforts at quarantine1 failed to stop the epidemic, which was widely believed to be linked to rotting coffee, waste or putrid air.
The outbreak could not have come at a worse time for the fledgling United States. Philadelphia was at this time the seat of government but the city had to be all but abandoned by George Washington and his administration. The epidemic began in August: by October of 1793 over 5000 out of a total population of 55,000 had succumbed, with 6000 inhabitants ill at any one time. In the midst of this epidemic, Benjamin Rush (Fig. 1 ) promoted a savage regimen of bleeding and purging, accompanied by massive doses of calomel. Rush, who we will meet later in the section on dengue, was both a prominent physician and one of the signatories to the Declaration of Independence. Rush subscribed heavily to the miasma theory of disease, pinpointing a spoiled load of coffee left on the city's wharves as the likely source of the outbreak. The views held by Rush eventually became entwined with his republicanism, and his counsel as to how best patients should be treated acquired increasingly partisan overtones. The political paralysis resulting from the 1793 outbreak is believed by historians to have contributed significantly to the neutralist stance of the newly emerging nation at a time when there was intense political manoeuvring to have the United States declare support for France in its war against England. Indeed, recurring epidemics of yellow fever did much over the ensuing century to shape the United States, especially given the disruptive effect of the disease on commerce and settlement of the southern states.
Fig. 1.
Benjamin Rush (1745–1813), American Founding Father and Physician.
Yellow fever is the archetypal haemorrhagic fever. It brought fear into the hearts of the early African explorers and its carriage across the Atlantic as a consequence of the slave trade did much to shape the European settlement of the New World. It emerged first in the ports of West Africa, then on slave trading vessels, to arrive finally in the Caribbean. This process took nearly a century after the Americas were first discovered, most likely because of the length of the sea voyage. An acute infection is dependent upon mosquitoes for transmission, and those persons infected before departure succumbed well before landfall. Only by maintaining the transmission cycle several times via shipboard mosquitoes could the virus survive the 10–12 week crossing, and only then if the crew and passengers survived. There is a contrary view that yellow fever moved from the Americas to Africa based on accounts of a disease resembling yellow fever before Europeans reached the New World. This is highly unlikely, especially given the high susceptibility of the indigenous population of the time and the susceptibility of New World monkeys to yellow fever virus. This susceptibility of New World primates to yellow fever virus contrasts sharply with the situation among Old World monkeys where yellow fever virus is very much in equilibrium with its animal reservoirs. There is a myth that Africans are less susceptible to yellow fever virus, but the reality is that yellow fever is part of the normal repertoire of infectious diseases that children are exposed to and from which recovery is the normal outcome. Thus, much of the indigenous population acquires immunity in the first years of life that extends into adulthood.
Yellow fever first appeared in 1647 in Barbados, quickly spreading to other parts of the Caribbean and the Yucatan Peninsula. Over the next two centuries it spread throughout Central America, south to Brazil and as far north as New York and Boston. European ports also suffered when ships berthed on return from the Americas. The effect on troops and settlers in the Caribbean was particularly devastating. In 1655 France sent a force of around 1500 men to take St Lucia from the British but only 89 survived the ravages of yellow fever and other diseases. Nearly a century later, the English Admiral Vernon in 1741 lost nearly half of a 19,000 strong force intent on taking Cartegena in present-day Columbia perished as a direct result of the disease.
In the closing years of the 19th century the United States government decided it was high time to root out the cause of yellow fever. The germ theory of disease had begun to be widely accepted and new causes of infectious disease were being described with almost monotonous regularity. The Surgeon General thus dispatched Walter Reed in 1899 to newly conquered Cuba. There he was joined among others by Carlos Finlay. Carlos Finlay, born of Scots and French parents, had experienced at first-hand the tragic effects of yellow fever when training in Philadelphia in 1853.
Reed and his colleagues first explored a bacterium proposed earlier by Giuseppe Sanarelli as the likely agent: this soon proved a false trail, however, with this bacterium subsequently being linked to hog cholera. Given the ascendancy and zeal of bacteriologists at the time, Reed and his fellow investigators were under considerable pressure to confirm a bacterium as the cause. How Reed and his team hit upon the concept of insect transmission is one of the great debates of medical endeavour. The widely quoted view is that Reed was aware of the work emanating from Ronald Ross in China whose studies on malarial transmission by mosquitoes were published a few years earlier in 1897. Two British physicians working in Havana at the same time as Reed almost certainly drew Reed's attention to the study of Ross. Reed made the connection, perhaps, that yellow fever often occurred when malaria was also prevalent. But the input of Carlos Finlay also needs acknowledging as he was aware of the investigations of Patrick Manson who described the role of mosquitoes in transmitting filariasis years earlier. Indeed mosquito transmission had been suggested as early as 1807 by John Crawford of Baltimore, a hypothesis reiterated by Joshua Nott in 1848. However the concept originated, Reed and his co-workers set up carefully controlled experiments, the outcome of which demonstrated unambiguously the role of mosquitoes in the transmission of yellow fever. Importantly for the time, they also showed that the agent was a “filterable virus” and not a bacterium.
Reed's work in Cuba led to intensive efforts to controlling urban yellow fever by reducing opportunities for individuals to be bitten by infected Aedes mosquitoes. Notwithstanding rural infections continued to occur and in response the Rockefeller Foundation's Commission for International Health embarked upon a programme of global eradication. To this end, the Commission went to Guayacil in Ecuador where yellow fever was still very much in evidence. Hideyo Noguchi soon reported he had isolated the causative organism in guinea pigs, calling it Leptospira icteroides. He raised a therapeutic antiserum which he persuaded himself and those around him to have the properties of a passive prophylactic.
Doubts as to the cause of yellow fever lingered, however, and in the early 1920s the Commission decided to re-evaluate the cause of yellow fever, but this time focusing efforts on West Africa, long regarded as the source of the disease. A permanent research station was established at Yaba, close to Lagos, and it was there that Adrian Stokes, an Irish physician appointed in 1922 to the chair in pathology at Guy's Hospital Medical School, successfully showed the viral nature of yellow fever by transmission to crown monkeys (Macacus simicus) from India using field isolates from Latch near Accra in present-day Ghana. After an incubation period of 2–6 days a period of high fever followed leading to eventually to collapse and death. Pathological changes were reproducible and re-transmission occurred after passing monkey serum from acutely ill animals through Berkefield filters. This early work in Ghana has recently been reviewed by Mortimer (2002).
This important work established the foundations whereby Sawyer in New York was able to develop what we now know as the 17D vaccine strains of yellow fever virus. Tragically Adrian Stokes succumbed to yellow fever, as had Noguchi, in Accra in 1928. These strategies prompted transferral of further work to the comparatively more controlled environment of New York.
Yellow fever continued to frustrate the aspirations of Europeans in settling and exploiting the Americas and Africa until well into the 20th century. It was common for seaman to become infected visiting the shorelines of these tropical zones but the victims inevitably perished at sea. The mosquitoes were hardier, however, and transovarial transmission ensured that infected insects could come ashore once ships had berthed in the northern hemisphere. One such episode occurred in September 1865 when the barque “Hecla” docked in Swansea when the weather was unseasonably warm (Meers, 1986). The “Hecla” had loaded a cargo of copper ore in Cuba as well as infected mosquitoes. Despite deaths amongst the crew whilst en route, the captain did not quarantine the ship before berthing. The result was a total of 27 infections and 15 deaths amongst those of the local population living or working within 200 m of the “Hecla”. Detailed weather records are available, and show that a dip in day time temperatures below 17–18°C for 3 days arrested the epidemic. Meers estimates that a maximum of 10 infected mosquitoes would have sufficed to cause this local outbreak. Although the diagnosis at this time would have been entirely on clinical observations, this outbreak almost certainly was due to yellow fever. It illustrates vividly how readily arboviruses can be transmitted across oceans, as was seen again as recently as 1999 when West Nile virus was introduced into the USA.
Epidemiology
Yellow fever is confined to the tropical regions of Africa and the Americas. Persistence of the disease is dependent upon cyclical transmission between monkeys and humans with mosquitoes as vectors. Thus, the epidemiology of the disease is driven by a series of complex interactions between the virus, its arthropod vectors and reservoir hosts.
These interactions give rise to two discrete transmission cycles, with marked variations between the cycles in Africa and South America. These interactions are summarised in Fig. 2 .
Fig. 2.
Transmission cycles of yellow fever.
Sylvatic, or “jungle” cycle
In forested areas, monkeys are the principal reservoirs, although infections in non-human primates tend to be transient and thus any viraemia is relatively short lived. The cycle differs between the Old and New Worlds both in terms of the species of monkeys infected, the outcome of this infection in primates and the mosquito vectors involved in transmission.
In Africa, yellow fever virus infects principally Cerepithecus and Colobus species, and in West Africa is transmitted by either Aedes furcifer-taylori or Aedes luteocephalus. In East and Central Africa the cycle is maintained principally by Aedes simpsoni. African non-human primates are relatively resistant to the infection and most recover. The associated viraemia is relatively short lived and therefore the chances of mosquito transmission are lessened. The distinct nature of the cycle between West and East Africa reflects the evolution of yellow fever virus in association with its host over a long period of time. These profiles are in accord with the distinctive genotypes of yellow fever virus recovered from patients in these different localities on the African continent.
In South America the sylvatic cycle is quite distinct. The virus is found in Aloutta, Ateles, Callethrix, Cebus and Saimiri monkeys and is frequently lethal. Widespread outbreaks occur centred on the river basin draining the Amazonian rain forest. Current thinking is that the virus was introduced into the rain forest ecosystem from urban outbreaks, aided by the adaptation of yellow fever virus to several species of tree-dwelling Haemagogus mosquitoes. As in the African cycle, mosquitoes remain infected for life once having bitten an infected monkey, with the virus passing transovarially to larvae. Humans only become infected by the bites of such insects on clearing trees or if the insect population spills over at the margins of forested areas into rural communities.
Human infections once initiated can instigate human-to-human transmission independent of the monkey population in the immediate environment. This is increasingly the case in Africa where the monkey populations have declined as human modification of their habitat has accelerated.
The margins of forest and savannah in Africa give rise to zones of emergence where during the rainy season the chance of human infection intensifies as vector numbers dramatically increase, only to decline once more during the dry season. It is at these margins, particularly after prolonged drought, that yellow fever re-emerges with serious consequences, especially among children with no immunity.
The urban cycle
This is maintained by peri-domestic mosquitoes such as A. aegypti. Efforts to eradicate this vector in Central and South America in the second half of the 20th century were largely successful in ensuring urban areas became free of yellow fever. Cessation of mosquito eradication programmes out of environmental concerns, however, has resulted in insect levels being restored to near pre-eradication levels in many areas. As yet, there has not been a resurgence of urban yellow fever in the Americas, mainly as a result of vigorous vaccination programmes. Humans entering forested areas infested by virus carrying infected mosquitoes become bitten by female insects. These live from 70 to 160 days and have a flight range of over 300 m. Eggs are laid in still water, and can be disseminated readily in pots, crevasses and old car tires.
In recent decades, yellow fever has been a far bigger public health problem in Africa, a continent where mosquito eradication has not been widely practiced and immunisation tends not to be widespread. The size and frequency of outbreaks in Africa varies between West Africa, and Central and East Africa. Outbreaks in East Africa are generally few and far between, although the largest outbreak ever recorded occurred in southwestern Ethiopia in 1960–1962, claiming around 30,000 lives with more than 100,000 people infected. This outbreak had a severe impact in such a sparsely populated region of less than one million. With the exception of the 1986–1988 outbreaks in Nigeria when over 44,000 cases were recorded, outbreaks in West Africa have occurred more often but tend to be limited in scope. Yellow fever has gradually spread from countries such as Côte D'Ivoire, Burkino Faso and Cameroon to Gabon, Liberia and Kenya, countries thought to be free of infection before the middle of the 20th century.
Properties of yellow fever virus
Morphology
Virions are spherical and approximately 50 nm in diameter with an outer lipid membrane enclosing an inner nucleocapsid (Fig. 3 ). Mature virions contain two membrane proteins, E and M. Detailed structure analysis of tick-borne encephalitis virus has revealed, for this flavivirus at least, that the E protein is arranged as dimers orientated parallel to the membrane surface. Thus, the flaviviruses do not show surface projections as is often the case for enveloped virus, such as influenza (family Myxoviridae) and rabies (family Rhabdoviridae). Both the arrangement of these dimers and the underlying nucleocapsid conform to the principles of icosahedral symmetry. Immature virions differ in that prM protein replaces the M protein.
Fig. 3.
Morphogenesis of yellow fever virus within infected Vero cells (Ishak et al., 1988).
Genetic organisation and gene expression
The flavivirus genome is an approximately 11 kb RNA molecule of positive sense2 with respect to protein translation. The 5′ end of the genome possess a type I cap (m-7GpppAmp) followed by the conserved dinucleotide AG. Flavivirus genomes are the only positive stranded RNA viruses of mammals that do not possess a poly(A) tract at the 3′ end: the 3′ terminus ends with the conserved dinucleotide CU. While nucleotide sequences are divergent amongst members of the flavivirus genus, the secondary structures at the 5′ and 3′ ends are conserved among mosquito-borne and tick-borne members of the genus.
In common with the picornaviruses, the viral genes are first expressed by synthesis of a large polyprotein (Fig. 4 ). This single precursor molecule then undergoes a series of cleavages thereby generating functional proteins. Cleavages are mediated either by the host signal peptidase present in the lumen of the endoplasmic reticulum or by a viral serine protease. The 5′ and 3′ ends of the genome are not translated, the secondary structure in this region having a role in mediating genome replication.
Fig. 4.
Organisation of the yellow fever genome.
Gene expression starts by ribosomes binding to a site downstream from the 5′ terminus, bypassing several AUG initiation codons before recognising a site close to the AUG located immediately upstream of the capsid, C gene. This internal ribosome entry event required for translating the viral genome is common to both flaviviruses and picornaviruses: the internal ribosome binding entry site (IRES) is formed in part by the secondary structure of the 5′ non-translated region.
A total of 10 proteins are expressed as a result of the processing of the polyprotein precursor. The three structural proteins, capsid (C), membrane (prM/M) and envelope (E) are expressed at the 5′ end, followed by the genes coding for the non-structural proteins, NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5 (Fig. 4). Polyprotein processing confers the advantage that gene expression can be controlled by the rate and extent to which these cleavage events occur. In addition, the use of alternative cleavage sites results in proteins with stretches of amino acid homology but different functions. This form of viral protein synthesis is likely inefficient, however, with some gene products being produced surplus to the requirements of virus replication.
Structural proteins
The two viral envelope proteins, E and M, are type I integral membrane proteins with C-terminal anchor sequences. By analogy with tick-borne encephalitis virus, the E protein consists predominantly of β-sheets arranged in a head-to-tail configuration with the distal ends of each monomer embedded in the lipid membrane (Post et al., 1992; Lindenbach and Rice, 1999). The E protein has both receptor binding (haemagglutin) and acid pH-dependent cell fusion activities, and composed of three structural domains. The third (domain III) contains a fold typical of an immunoglobulin constant domain and it is this domain that is thought to represent the cell receptor. There is considerable variation in amino acid sequence at the margins of this domain between tick- and mosquito-vectored flaviviruses. Some mutations in domain III equate with changes in virulence. In the case of yellow fever virus, there is one report that suggests a region of domain II may also be involved in binding virus to receptors present in monkey brains (Ni et al., 2000). Work on the structure of dengue virus E protein has progressed substantially over the past few years (see below) and comparative analyses should throw light on the detail of yellow fever receptor binding and substitutions to amino acids that equate to tissue tropism and attenuation.
Non-structural proteins
Despite the simplicity of protein expression, almost all of the non-structural proteins serve more than one function at some stage during the replication process. All seven are involved at various steps of RNA synthesis, although little is known as to how these interact with one another, and how each relates to those host proteins required for gene expression and RNA synthesis. The explosion of hepatitis C research, also a flavivirus, over the past 10 years has had a beneficial effect generally on our understanding of how flavivirus non-structural proteins function.
NS1 is an interesting protein, being glycosylated and essential for virus viability yet not found in the virus particle. It is translocated together with E protein into the lumen of the endoplasmic reticulum prior to cleavage at the E/NS1 junction by a host cell signalase. NS1 does not associate with mature virions, but locates in membrane-associated RNA complexes and thus promotes negative strand synthesis. In order to perform this function, NS1 combines with NS4A. Transcomplementation studies using NS1 deletion mutants of yellow fever virus and a replicon of Sindbis virus expressing the NS1 of dengue virus have shown that this interaction is virus specific, although variants of yellow fever virus have been detected that acquire recognition of dengue virus NS1 as a result of a single base change in the NS4A gene (Lindenbach and Rice, 1997, 1999).
NS1 can be detected both in and on the surface of infected cells, leading many workers to propose that NS1 could be the target of a CD8+ cytotoxic T cell response and a target for vaccine development. Jacobs et al. (2000) have suggested a role for NS1 as a cell activator as the addition of anti-NS1 antibody resulted in tyrosine phosphorylation of cellular proteins. NS1 is secreted by mammalian cells as a hexamer consisting of three homodimers. NS1 is not, however, secreted from infected mosquito cell lines (Smith and Wright, 1985). The extent of glycosylation and the processing of these sugar side chains contribute to the final structure of the resulting NS1 dimers. Variable glycosylation may also account for the difference in secretion properties of NS1 from mammalian and insect cells: one of the two polysaccharide side chains of the dengue virus NS1 is protected within the dimer from reaction with cellular glycosidases and remains high in mannose content, while the second remains exposed, being progressively trimmed and modified to form a multibranched endo-β-N-acetylglucosaminidase F resistant complex sugar (Flamand et al., 1999). Thus, complete processing of exposed, sugar side chains appears essential for NS1 secretion.
NS2A, NS2B, NS4A and NS4B collectively are a group of small hydrophobic proteins that facilitate the assembly of RNA replication complexes adjacent to cytoplasmic membranes. Neither of these gene products contains motifs homologous to other known mammalian or viral enzymes. Two forms of NS2A are known: one full length, the second with a truncated C-terminus. Mutations at the C-terminus of either form are lethal for yellow fever virus replication (Kummerer and Rice, 2002). The NS2B protein has two functional activities. First, NS2B combines with NS3 and acts as a co-factor for the NS3 serine protease. The second function utilises separate hydrophobic domains to facilitate the insertion of the NS2B–NS3 precursor into the endoplasmic reticulum at the time of translation. NS4A is not readily detected in a cleaved form within infected cells, being in the main part of a NS3–NS4A precursor. NS4B, in contrast, is readily seen, particularly in the perinuclear region. Late in the replication cycle NS4B can be found within the nucleus.
NS3 has two discrete functions. The first function is required for processing the polyprotein resulting from translation of the positive viral RNA strand. One-third of the N-terminal folds to form a serine protease. This part of the molecule contains motifs characteristic of the trypsin superfamily of proteases but in order to fulfil this proteolytic function prior coupling to NS2B is required, as mentioned above. Furthermore, efficient polyprotein processing requires binding to cellular membranes. The protease complex removes the anchor region from the C protein and recognises several other cleavage sites along the length of the polyprotein. All share a common motif consisting of two basic amino acids followed by an amino acid with a short side chain.
The heightened concern with regard to the increasing prevalence of persistent hepatitis C in the human population has stimulated work examining the NS3 protease of hepatitis C virus as a target for antivirals. This protein, along with other hepatitis C virus proteins, has now been analysed in considerable depth. The availability of structural coordinates allows for modelling of other NS3 proteins by threading and similar algorithms. Following this approach, a model of dengue NS3 has shown there are some differences in the substrate binding domain of the dengue virus NS3 protein (Brinkworth et al., 1999).
A fully functioning helicase is essential for flavivirus replication. The C-terminus contains regions homologous to the DEAD family of RNA helicases. The consensus is that the helicase plays a role in unwinding the secondary structure at the 3′ end of the viral positive strand template prior to the commencement of RNA synthesis, and perhaps also in releasing the nascent negative RNA strand prior to commencement of positive strand synthesis. NS3 also contains an RNA triphosphatase (NTPase) activity involved in the formation of the cap structure at the 5′ end of nascent positive RNA strands. This domain overlaps the protease activity between residues 160 and 180 of the dengue virus NS3 molecule (Li et al., 1999). NS3 interacts also with NS5, an interaction dependent on phosphorylation. The association with NS5 may regulate NS3 as NTPase activity is upregulated in its presence. The domains responsible for this interaction in dengue virus have been mapped to amino acids 320–368 on NS5, and 303–618 on NS3
NS5 is the largest and most highly conserved of the non-structural proteins and constitutes the RNA-dependent RNA polymerase. Although with an overall basic charge, NS5 has long hydrophobic stretches that are more characteristic of a membrane bound protein. The C-terminal portion contains the highly conserved GDD motif typical of viral RNA polymerases and protein modelling shows that NS5 structure resembles closely that of the poliovirus RNA polymerase. Other motifs along its length suggest that NS5 also has a role to play in the methylation of the 5′ end cap structures. Both the methyltransferase function and the NS3-encoded triphosphatase represent two of the three enzymes necessary for 5′ cap synthesis. A guanyltransferase would also be needed, but this activity has yet to be identified.
NS5 has been detected both in the cytoplasm and in the nucleus of infected cells. The nuclear form of NS5 is extensively phosphorylated. Translocation of NS5 into the nucleus is mediated by a nuclear localisation signal sited between the methyltransferase and polymerase domains. It is not known as what effect nuclear NS5 has on cellular functions, although differential phosphorylation of this protein during the replication cycle suggests nuclear-located NS5 has a role in the later stages of the replication cycle. Its presence in the nucleus may account for the basophilic staining bodies seen in infected cells. Current thinking is that cellular proteins complex with NS5 to direct and modulate RNA synthesis. Candidates for such proteins include the translation elongation factors EF-1, EF-Tu and EF-Ts. A further complexity is the reported interactions between NS5 and the nuclear transport receptor importin-β. These interactions can exclude NS3 from forming NS5–NS3 complexes, thus encouraging the migration of NS5 to the nucleus (Johansson et al., 2001).
Virus replication
The cellular receptor for yellow fever virus has yet to be defined although it is known that glycosaminoglycans play a role in virus entry. Two glycoprotein receptor molecules have been proposed for dengue 4 virus, and it may prove to be that entry requires more than one host component (Salas-Benito and del Angel, 1997). As flaviviruses also infect arthropod vectors, the cellular receptor for both mammalian cells and insect cells either has to be highly conserved or involve two separate domains.
Flavivirus particles enter the host cell by a process of receptor-mediated endocytosis followed by fusion at low pH of the viral envelope with the membrane of an endosomal vesicle. The nucleocapsid is then released into the cytoplasm. As most of the replication cycle takes place at or near the perinuclear membrane, there must be some mechanism of transporting the nucelocapsid through the cytoplasm, but at present this mechanism remains obscure.
Translation usually begins as a result of internal ribosome entry at the first AUG codon of the single open reading frame, although there is evidence of occasional initiation at the next AUG some 12–14 nucleotides downstream. After primary translation of the infecting viral genome, RNA synthesis begins by production of negative strand copies by the NS5 protein. Negative strand synthesis continues throughout the replication cycle. These negative copies are then used as templates for the generation of further positive RNA strands. RNA replication complexes are localised in the perinuclear endoplasmic reticulum and consist of both RNA double-stranded duplexes and replicative intermediates, the latter consisting of double-stranded regions and nascent single-stranded RNA molecules. These are either used as further plus strand templates or for translation. There are similarities here with poliovirus replication with the steady accumulation of structural proteins during the flavivirus replication cycle. This eventually leads to a point where nascent positive RNA strands are withdrawn from the pool of translatable plus strands to form new nucleocapsids. However, there seems to be a distinct compartmentalisation between polyprotein processing and RNA synthesis, and possibly moderated by such a topographical separation.
Hypertrophy and proliferation of cytoplasmic membranes are characteristic of flavivirus-infected cells. Nascent virus particles first assemble on the rough endoplasmic reticulum, and then these immature virions are transported progressively through the endoplasmic reticulum compartments to the cell surface where virus particles are released by exocytosis. Immediately prior to release the prM protein located within the viral envelope is cleaved by a host furin-type protease located in the trans-Golgi network. How this process occurs is difficult to analyse as visualisation of maturing yellow fever particles has proven difficult. Infected cells also release a non-infectious, subviral particle, for reasons that are unclear. These subviral particles are antigenic and represent cellular membrane fragments into which are inserted copies of the E and M proteins as well as small amounts of prM. As the E protein retains its haemagglutination properties, these particles are functionally referred to as the slowly sedimenting haemagglutinin (SHA) component.
Clinical disease
The clinical course of yellow fever develops through three distinct stages. The acute phase is characterised by a fever (39.5–40°C) of 3–4 days duration. Headache, back pain, nausea and vomiting constitute the major symptoms. At this stage, the patient is highly infectious with virus present in the blood from days 3–5. This viraemia ensures that the likelihood of human-to-human transmission by mosquitoes is high. Remission generally follows accompanied by a lowering of the fever. The headache disappears and the patient generally feels much recovered.
During the third stage, the fever returns with many if not all of the symptoms seen on presentation, but in a more severe form. The patient becomes increasingly anxious and agitated. Liver, heart and perhaps kidney failure follow rapidly accompanied by delirium. Jaundice is the inevitable result of the inflammation in the liver and death occurs 6–7 days after onset of the disease. Among those that survive, recovery can be slow. Virus cannot be recovered from the blood during this stage but anti-yellow fever virus antibodies can be detected, suggesting an immunopathological component late in the disease process.
During epidemics, the case fatality rate may exceed 50%, and for children can approach 70%, the major driver for the introduction of yellow fever vaccination into childhood immunisation programmes within endemic regions. In some outbreaks there has been a preponderance of males, for example the 1972–1973 outbreak in Brazil over 90% of cases were men. Despite suggestions to the contrary, there appears to be no link in disease severity according to ethnicity.
Other causes of viral haemorrhagic fever should always be suspected, such as Congo-Crimean haemorrhagic fever, Rift Valley fever, Marburg and Ebola viruses in Africa. Meningococcal septicaemia and leptospirosis are also infections that need to be eliminated during diagnosis.
Among other illnesses that can confound clinical diagnosis are the agents of viral hepatitis. Hepatitis A is common in endemic areas, indeed in Columbia and neighbouring countries hepatitis A is more common than other forms of viral hepatitis. Hepatitis A is rarely accompanied by a high fever, however, and serological tests for anti-hepatitis A IgM antibodies are readily available to differentiate this agent from yellow fever virus. The Rockefeller Foundation conducted surveillance measures in Brazil for many years based on the taking of liver tissue from fatal cases using a viscerotome. A re-examination by histopathology of samples taken between 1934 and 1967 has failed to show evidence of yellow fever antigens, however, although around 11% of the presumed yellow fever infections were positive for hepatitis B viral antigens (Simonetti et al., 2002), encouraging the speculation that these deaths may be more related to hepatitis B virus and its dependent agent hepatitis delta, the latter infection being particularly prevalent in the Amazonian basin.
Diagnosis
As with many virus infections, there is an emphasis on the detection of IgM antibodies during the early acute phase. The IgM capture ELISA is the test of preference, although care is needed with standardisation: cross-reactions do occur but can be minimised with care. Immunofluorescence using infected cells fixed previously in acetone is an easy method to adopt for field use and can be modified to detect either IgM or IgG antibodies, although the assay for IgG antibodies is sensitive to the presence of rheumatoid factor. A more definitive diagnosis is the presence of virus in the early viraemic period. Intracerebral or intraperitoneal inoculation of suckling mice is a sensitive method for virus detection but results may take up to 3 weeks. Intrathoracic inoculation of mosquitoes is also possible, but tissue culture isolation is perhaps preferable. Cell lines derived from either Aedes albopictus (C636 cells) or Aedes pseudoscutellaris (MOS61 cells) can readily support growth of yellow fever virus. Detection of virus replication by application of monoclonal antibodies can produce results in a few days. There is a risk that virus in samples is already complexed with antibodies thus dissociation of antigen-antibody complexes by treatment with the reducing agent dithiothreitol is recommended prior to inoculation of cell cultures.
Sequencing of yellow fever isolates has revealed that there are at least two, possibly three genotypes of yellow fever currently in circulation. Type I is found in Central and East Africa, type IIa in West Africa and type IIb in the Americas (Chang et al., 1995). These distinctions are based on phylogenetic analyses of the E protein. The relatively close association of strains from West Africa and the Americas is consistent with what we know of the historical origins of the virus in the New World, but it should be stressed that there are many phenotypic differences between isolates grouped in genotypes IIa and IIb. More needs to be done in defining to what extent such variation can result in virus adaptation to new hosts and mosquito vectors, and thus fundamentally alter the nature of the transmission cycle. Despite these differences, the presently available yellow fever vaccines protect against all three genotypes.
Pathology
The taking of a liver biopsy is not advisable given the high risk of haemorrhage. However, liver tissue taken at autopsy is useful for epidemiological purposes, as mentioned above. The hepatitis associated with yellow fever virus is manifested by the presence of extensive mid-zonal necrosis, visceral haemorrhages, sinusoidal acidophilic bodies and hypertrophy of Kupffer cells. The portal tracts become extensively infiltrated with monocytes. Histopathology shows the appearance of dark eosinophilic bodies in the cytoplasm of hepatocytes (Councilman bodies) (Fig. 5 ). These remnants of hepatocytes having undergone apopotosis were often regarded as specific to yellow fever but caution needs to be exercised as similar inclusions can be due to other causes of viral haemorrhagic disease.
Fig. 5.
Histopathological examination of liver parenchyma from a patient showing inflammation around the portal tracts (arrowed; a) and the distinctive presence of Councilman bodies (b).
Prevention and control
Yellow fever vaccines owe their origins to work carried out in the 1920s and early 1930s under the sponsorship of the Rockefeller Foundation. In 1928 it was found that monkeys were susceptible to yellow fever virus, an observation that led directly to the isolation of Asibi strain of yellow fever virus from present-day Ghana together with the Dakar strain obtained from a Senegalese patient. Shortly after, Theiler discovered that Swiss white mice could also be infected, thus opening up a method for measuring neutralising antibodies.
Two live attenuated vaccines were developed concurrently using these isolates. French workers passaged the Dakar isolate 128 times in mice brains to derive what became known as the French neurotropic vaccine. This product was used in the Francophile community for some years at the 258–260 passage level but its use was discontinued owing to systemic reactions in up to 20% of those vaccinated. The second vaccine lineage was derived from the Asibi strain. Present day vaccines all have their origin in this virus originally recovered in West Africa. It is important to note that both the French neurotropic vaccine and the 17D strains have lost both the capacity to produce viscerotropic disease and an ability to replicate in mosquitoes.
The 17D vaccine strains were originally developed by Theiler and colleagues by passage first in mouse brain and then in chick embryo tissue. Two substrains of the 17D attenuated virus form the basis of present vaccines. The first is based upon virus derived at the 204th passage (17D-204) and is used predominantly in Europe and Africa at the 234–238 passage level. The second originates from the 195th passage of the lineage, being subsequently passaged independently in embryonated eggs and used at the 286–288 passage level (17DD): this is used almost exclusively in South America. Thus, there is a dichotomy in passage history between vaccines used in the Americas and the Old World (Fig. 6 ). Although differing in lineage, both are equally effective, despite evidence of phenotypic differences between yellow fever virus circulating in the Americas and Africa.
Fig. 6.
Lineage of currently licensed yellow fever (17D) vaccine strains.
Protection against yellow fever—indeed against any human flavivirus infection—is mediated by the presence of neutralising antibodies. Seroconversion rates in healthy recipients rise to over 95% by 30 days following a single dose of live attenuated vaccine. Immunity is probably lifelong although revaccination after 10 years is required under the International Health Regulations. The vaccine is delivered subcutaneously and is well tolerated: It can be given simultaneously with other live attenuated vaccines such as measles and polio as well as together with DPT, oral cholera, typhoid, hepatitis A and hepatitis B vaccines. Mild adverse reactions are experienced by up to 25% of recipients, these generally lasting a few days and consist of pain and soreness at the site of injection, headache, malaise and myalgia.
The vaccine is contraindicated for pregnant women unless demanded by local circumstances. Inadvertent vaccination in early pregnancy can lead to anxiety, but a study by Robert et al. (1999) suggested that inadvertent exposure in the first trimester does not increase the risk of abnormalities during gestation. There is no evidence of the vaccine crossing the placenta. A study by Nasidi et al. (1993) among 40 Nigerian women and their offspring showed no evidence of congenital infection despite vaccination during pregnancy. Seroconversion was less than 40% as opposed to more than 85% among the general population, suggesting that immunosuppression during pregnancy suppressed the B cell response to the 17D virus.
The issue of immunosuppression in patients with HIV and responsiveness to yellow fever vaccine has not been sufficiently investigated. The expectation is that such individuals would not respond as well as others and could be more susceptible to the consequences of infection, but this hypothesis has not been tested. As the vaccine is produced in eggs, residual egg protein may prove difficult in those with allergies to eggs but again this has not been borne out in practice, with allergic reactions being as infrequent as one per million of individuals vaccinated.
The viraemia following vaccination is low, normally below 2 log10pfu, and is of short duration. Thus, the risk of vaccine virus transmission by insect bite is low, but in any event the 17D virus is substantially attenuated to the extent that it is unable to replicate in mosquito vectors.
Presently vaccine continues to be manufactured in embryonated chicken eggs. Attempts to replace the egg-derived products with virus grown in cell monolayers such as chick fibroblasts have been frustrated by the low yields that result from infection of cell cultures with the 17D virus. There is also the question of cost. Egg-grown virus is cheap to produce and the technology readily transposed into endemic areas where investment in expensive cell culture facilities is limited and skilled personnel rarely available. Some embryonated hens’ eggs are contaminated with avian leucosis virus; although there is no evidence to suggest the presence of this contaminant has any effect on efficacy or the health of vaccines, it is thought desirable to eliminate this agent from yellow fever vaccine products. A WHO-approved master seed lot free of ALV has been available from the Robert Koch Institute in Berlin for some years but as yet has not been universally adopted.
The long history in the use of the 17D substrains presents a unique opportunity to define the molecular basis of attenuation and obtain indicators as to how vaccines could be developed against dengue and other flaviviruses, particularly vaccines against those infections that as yet have proved too difficult to develop for various reasons.
The genome of yellow fever virus was first sequenced using 17D-204 virus held by the American Type Culture Collection (Rice et al., 1985). This landmark study has since been followed by completing the sequence of both the wild type Asibi virus as originally isolated by Stoker and the complementary 17DD virus subjected to an independent lineage beyond passage 195 in chick embryos. These studies together with related work have revealed a number of differences at the molecular level: for example, there are 32 amino acid differences between the Asibi isolate and the 17D-204 virus distributed throughout the structural and non-structural proteins with most variation in the E and NS2A proteins. Four nucleotide substitutions in the 3′ non-coding region are found consistently in all vaccine strains, and the nucleotide sequence in the 5′ non-coding region (NTR) is highly conserved. Parallel studies of the Dakar isolate and its derived French Neurotropic vaccine strain have similarly demonstrated a proportionally high mutation rate in the E and NS2A proteins (Wang et al., 1995), but with little concordance in the nature of individual substitutions. Only two substitutions are common: a leucine to phenylalanine change in M protein at position 36, and an isoleucine to methionine change at position 95 in NS4B. The fact that monoclonal antibody studies show cross-reactivity for viruses with dissimilar linear amino acid sequences indicates the importance of conformational epitopes in eliciting neutralising antibodies. Interestingly, the 17DD strain possesses an additional glycosylation site on the E protein at position 153 not present on 17D-204 virus, and the new master seed lot approved by WHO possesses a new glycosylation site close by at position 151. The significance of such differences is unclear but could affect either the conformation of a critical determinant or its immunogenicity.
An important issue is the potential of vaccine strains to revert to neurovirulence. There have been a handful of cases reported in vaccines, mainly in children less than 1 year of age. Only one fatal case has been reported: a 3-year-old girl who was immunised with the 17D-204 virus (Jennings et al., 1994). This case is notable as virus was reisolated from the child and subjected to antigenic analysis, showing that the virus had regained reactivity for a monoclonal antibody normally reactive only with wild type virus. The isolate was also neurovirulent for a single cynomolgus monkey. Sequencing of this isolate has shown a number of mutations (Barrett, 1997). Interestingly one of the changes was at position 155 on the E protein, close to the glycosylation site discussed above. Notwithstanding this and rare instances of encephalitis in vaccinees, it needs to be stressed that yellow fever vaccination remains one of the greatest medical achievements in the control of infectious diseases.
The experience of over 60 years of routine use has provided substantial evidence that the 17D virus is both safe and highly efficacious. For this reason, the 17D vaccine strain makes an ideal candidate for use as a live vector for genes of other, heterologous antigens, most notably for genes coding for the structural proteins of other flaviviruses, such as hepatitis C, dengue and West Nile viruses. Studies using chimeric yellow fever—dengue constructs are described later in this chapter. This platform technology owes its origins to the work of Pletnev et al. (1992) who showed the potential of a chimeric virus containing genes from tick-borne encephalitis virus and Japanese encephalitis virus as a human vaccine. This work established the important principle that the chimera construct had lost peripheral invasiveness as a result of the vector construction.
Dengue
Colonisation of the New World brought with it infections previously unknown to the indigenous populations of the Americas. These included smallpox, measles, scarlet fever, malaria and yellow fever. It is probable that cases of the latter could have well been mistaken for dengue fever as examination of historical records makes it impossible to distinguish the two infections apart until at least the end of the 18th century. Between 1779 and 1780 detailed descriptions were made of outbreaks resembling present-day dengue fever from the nascent United States, Africa and the Dutch East Indies. The most celebrated of these descriptions is due to Benjamin Rush, a physician of Philadelphia and a close friend and colleague of the American Founding Fathers (Fig. 1).
Rush is generally credited as being the first to provide a detailed description of what we now know as dengue fever. This was largely based on an outbreak occurred in Philadelphia in the summer of 1780 which Rush described as a bilious remittent fever. The description has remained little improved over the centuries, so fastidious was Rush in ensuring he committed to record an accurate and objective observation as to the course of the disease. His descriptions were all the more remarkable given that a number of illnesses were circulating in Philadelphia at that time. Although he did not to make the connection with a possible vector, he noted that “… mosquitoes were uncommonly numerous during the autumn” (quoted in Humphreys, 1997). That he failed to make the connection is all the more remarkable given that Rush recorded the epidemic as subsiding as soon as the temperature dropped sharply in October. The locals referred to the malady as “breakbone fever”. Coincidentally Rush was also renowned for his interest in mental disorders, and records the case of two afflicted sisters, both of whom suffered from dejection and depression, pressed Rush to change the name to “breakheart fever”.
Concurrently David Bylon, a physician in the Dutch East Indies noted a similar disease which he referred to as knokkel-koorts (“Knuckle fever”). Humphreys has pointed out that the prominence of knuckle pain means this could have been an outbreak of Chikungunya virus, not breakbone fever as was supposed by his contemporaries. The name dengue came into use after reports of an outbreak in the Spanish West Indies between 1827 and 1828 when the Swahili term Ki denga pepo was first used to describe the disease.
Dengue fever epidemics in the 18th and 19th centuries occurred in the Americas and elsewhere in regular cycles, reflecting the fact that transport by sea took many weeks, if not months, between Europe and the newly emerging empires of the Netherlands and Britain. The rapid changes of the 20th century changed dramatically the epidemiology of dengue virus, especially the upheavals accompanying the Second World War and its aftermath. It was at this time that the first extensively documented outbreak of dengue haemorrhagic fever (DHF) occurred in the Philippines in 1953. This was followed 5 years later by a much larger outbreak in Thailand. Since the 1960s dengue has spread to over 60 countries, with extensive epidemics particularly in the Caribbean and South America. Although outbreaks are known to occur in Africa, less is known about these, most likely because if and when they do occur the infrastructure for reporting such outbreaks is invariably poor.
Although dengue fever as an acute illness has a low mortality rate, its impact, like that of yellow fever, extends beyond its clinical importance. In centuries past, the introduction of infection into a community could mean economic disaster for fledgling colonies. In modern times, the impact of an infectious disease such as dengue can still result in serious economic loss in a country heavily dependent upon tourism. For example, Puerto Rico was hit by a devastating epidemic in 1979 that resulted in a loss of over $10 million, made far worse by dengue returning no less than eight times with losses amounting eventually to over $100 million. The 1981 outbreak in Cuba had a similar devastating impact, with a further economic cost of approximately $100 million. The four serotypes of dengue virus extend over regions inhabited by over 2.5 billion people, i.e. over a third of the human population are at risk. This makes dengue arguably the most important of virus diseases spread by arthropods.
Epidemiology
Dengue viruses are transmitted to humans by the bite of an infected female Aedes mosquito. The principal vector is A. aegypti, a black and white mosquito that has become highly adapted to an urban environment. The female needs regular blood meals in order to provide nutrients for its eggs, which it lays in containers of still water commonly found close to domestic dwellings, for example open water barrels, flower pots and old tins and containers. Old car tyres provide ideal vessels for still water and at least one outbreak in Taiwan has been attributed to Aedes larvae imported in a cargo of recycled tyres. The adults are difficult to detect, feeding off humans in daylight, with bites often going unnoticed. Civic programmes aimed at eliminating the vector by removing open containers of still water have done much to reduce, if not entirely eliminate, the risk of transmission in urban areas. These measures are often backed by the force of law as in the case of Singapore, for example.
A. aegypti has a worldwide distribution in the tropical zones, but for reasons unknown dengue is not always found in regions infested with this replication-competent vector. Epidemics first occurred in the 1950s in South East Asia and over the ensuring 30 years spread first through the Philippines to the South Pacific islands, and from there spread to Central and South America, parts of Africa, India and south to Queensland in Australia. Dengue viruses have a considerable propensity to spread further, particularly as A. aegypti has returned to many areas since the cessation of insecticide spraying. The disease threatens the southern United States, the populous areas of Brazil and could spread to much of Africa (Fig. 7 ).
Fig. 7.
Geographical distribution of Aedes aegypti and dengue virus. The denser the shade, the greater the incidence of disease
(courtesy of the World Health Organisation).
Once a female mosquito imbibes blood from a viraemic human, the virus replicates in the gut of the insect, moving to the salivary glands by 8–11 days after ingestion. The female mosquito remains infected for life, transmitting virus in its saliva each time the insect acquires a new blood meal. A. albopictus, the tiger mosquito—so named owing to its aggressive behaviour—is also replication competent. This species has recently invaded the southern USA, setting the stage for possible future outbreaks.
Although A. aegypti is responsible for human-to-human transmission in an urban environment, a sylvatic cycle has increasingly been recognised, at least in Asia. Forest-dwelling mosquitoes of the genus Ochlerotatus transmit the virus between monkeys—principally Macaca and Presbytis species. This cycle is likely complex as A. albopictus can also transmit the virus at the edge of forested areas. In Africa, Erythrocebus monkeys are the principal hosts in the rain forests, with other Aedine mosquitoes also contributing to maintenance of the transmission cycle, for example A. luteocephalus, A. taylor-furcifer and A. opok (Gubler, 1998; Wang et al., 2000). There is evidence of all four serotypes of dengue virus3 originating in monkeys, with adaptation to humans having occurred both relatively recently and independently for all four serotypes. Although the sylvatic cycle has been demonstrated for all four serotypes in Asia, only a dengue 2 sylvatic cycle has been demonstrated in Africa (Rodhain, 1991). Understanding more regarding the sylvatic cycle would boost our understanding of genetic variation between and within the four virus serotypes: dengue virus has evolved to the extent that it is not dependent upon the enzootic cycle for persistence in any given geographical locality.
Clinical features
Seroprevalence studies have shown that the majority of dengue virus infections are asymptomatic. Clinical disease is of two distinct types. The first is dengue fever, the “classical” disease seen in the vast majority of cases among adults who have not been previously exposed to the virus. The second is the haemorrhagic form of the disease, DHF, which may or may not progress to dengue shock syndrome (DSS).
Classical dengue fever is characterised by a sudden rise in body temperature, accompanied by nausea and a severe headache, the latter most often located towards the frontal regions and accompanied by retro-orbital pain. Other neurological signs include severe depression, apathy and complaints of disturbing dreams. However, the most distinctive feature is the severe muscle and bone pain, particularly in the lower back. Joint pains, a lymphadenopathy and vomiting all develop over the next 3–7 days. A diffuse and discrete macropapaular rash develops immediately prior to recovery. Although incapacitating, the acute illness is relatively short lived and patients eventually make a full recovery although many remain incapacitated for many weeks during convalescence.
The more severe disease of DHF is seen in children below the age of 15 and in those infected previously with dengue virus. Clinically, DHF resembles yellow fever in that the initial stages of the disease—similar to uncomplicated dengue fever—is followed by a brief respite when body temperature returns to near normal, only to be followed by a sharp onset of fever once more and a rapid deterioration in the patient's condition. The patient displays profound prostration and shows progressively all the manifestations of haemorrhage and shock that result from circulatory collapse and hypotension. The liver may become enlarged with signs of jaundice. Petechiae appear in the skin and patients give a positive tourniquet test. Ecchymoses, gastrointestinal bleeding and haemorrhagic pneumonia become evident.
The World Health Organisation has attempted to produce a case definition of DHF as a sudden febrile onset accompanied by haemorrhagic manifestations that include a positive tourniquet test and an increase in haematocrit reading to 20% or more. This case definition has been the subject of much debate, with many experts emphasising that the disease profile differs according to the age of the patient and geographical location. In order to aid the collation of clinical data, the WHO has proposed classifying DHF in four grades (Table 1 ).
Table 1.
WHO classification of dengue haemorrhagic fever
| Grade | Clinical description |
|---|---|
| I | Fever with non-specific, constitutional symptoms and the only haemorrhagic manifestations being a positive tourniquet test |
| II | As for Grade I, but accompanied by more extensive haemorrhagic manifestations |
| III | Signs of circulatory failure or hypertension |
| IV | Profound shock with pulse and blood pressure being undetectable (DSS) |
Although severe muscle and bone pain together with a sudden onset of fever are highly suggestive of dengue fever, the disease on presentation may resemble many other febrile illnesses. The more severe form, DHF, may be confused clinically with other causes of haemorrhagic disease, although haemoconcentration and indications of a coagulopathy may be useful in pointing towards dengue virus as a cause.
Although the risk of DHF and DSS increases significantly with secondary infections, the risk does not increase further on a subsequent third exposure to another serotype. If anything, the risk appears to decline, perhaps as a result of previous infection stimulating a sufficiently broad immune response that replication by a third serotype is contained.
Diagnosis
In contrast to most flaviviruses that cause human disease, virus isolation can prove difficult from cases of dengue fever. Direct intracerebral injection into suckling mice often requires several blind passages for an isolate to become evident. An alternative approach is the intracerebral inoculation of Toxorhynchites ticks, with successful isolation possible in less than 3 days. Intrathoracic inoculation of mosquitoes is another possibility, with isolation taking somewhat longer at up to a week. Virus isolation in cell culture is to be preferred, however. Insect cell lines are very sensitive to dengue virus: the most widely available is the C6-36 cell line derived from A. albopictus, but alternatives are AP-61 cells from A. pseudoscutellaris and TRA-248 cells from the tick Toxorhynchites ambionensis. Regular inspection of replicate cultures by addition of fluorescent- or peroxidase-labelled monoclonal antibody specific for dengue virus usually produces a positive response within 3 days of inoculation.
Specific serology is vital to making an accurate diagnosis. ELISA assays designed to detect IgM are increasingly replacing haemagglutinin inhibition for the detection of anti-dengue virus antibodies. IgM antibodies are present from as early as the third day of infection and may persist for as long as 3 months. Rapid tests for IgG antibodies are common, but care needs to be taken to ensure that tests are adequately controlled for unwanted cross-reactions with other flaviviruses. Immunofluorescence is useful as each visual field contains a number of uninfected cells that can serve as negative controls. The complement-fixation test has now largely fallen out of use, mainly because of its complexity, its relative insensitivity compared to other methods, and the difficulties in standardising reagents. Tests for measuring neutralising antibodies are particularly useful in confirming both the diagnosis and for determining the serotype of an isolate.
Properties of dengue virus
The dengue virus genome is a single-stranded RNA molecule of positive sense with respect to gene expression. Approximately 11 kb in length, the genome organisation is similar to that of yellow fever virus (see Fig. 4).
Electron microscopy of extracellular virus shows a particle of approximately 50 nm in diameter with clearly visible surface projections. The core protein exists in an ordered structure less well defined compared to that of the alphaviruses, where the core assumes an icosahedral symmetry dictated by the envelope glycoproteins as the virus buds from an infected cell, a process mediated by interactions with the alphavirus E2 envelope glycoprotein. Such interactions appear to be absent in dengue and other flaviviruses.
Knowledge of the three-dimensional structure is important for identifying regions that can be targeted to block virus entry and for designing new vaccines, as well as elucidating just how heterologous neutralising antibodies can cause antibody-mediated enhancement. Most flaviviruses have around 40% identical amino acids in their E proteins, and thus the overall three-dimensional structure in terms of secondary structure elements such as folds and management of functional domains is likely to be similar. The structure of the tick-borne encephalitis virus E protein was determined in 1995 (Rey et al., 1995) but despite the increasing sophistication of modelling methods standard alignment methods proved insufficient to predict the secondary structure of the remaining 60% of the dengue virus E protein.
The class II fusion (E) protein of dengue virus has a distinctly different structure to the class I fusion protein of the haemagglutinin of influenza virus. In common with the E protein of tick-borne encephalitis virus, the dengue E protein is ordered as 90 dimers flat-packed on the surface of an icosahedral-shaped virus particle. The dimers are so closely packed on the surface of dengue virus that the viral membrane is inaccessible at physiological pH and thus fusion cannot occur. Reducing the pH results in a conformational change, resulting in the dimers re-arranging to form a T = 3 icosahedral lattice (Kuhn et al., 2002). The net effect is an increase in particle diameter and the exposure of the underlying viral membrane.
The E protein is not cleaved during the maturation and assembly of new virus and is predominately non-helical, being composed largely of β-sheets. Each monomer is orientated in an opposite head-to-tail arrangement with respect to its partner, and structural studies show there are three distinct domains. Domains I and III are located at either ends of the dimer and opposite the complementary domain on the other monomer. Domain II is an elongated, finger-like structure with a hydrophobic sequence (approximating to residues 98–109 for the dengue 2 virus E protein) conserved between all of the flaviviruses. It is this region that is responsible for interacting with the host cell membranes following uptake of virus into endosomes and performs a function analogous to that of the N-terminal domain of class I fusion proteins such as the haemagglutinin molecule of influenza virus (see below).
Recently the structure of the dengue E protein has yielded to X-ray crystallography, illustrating that E protein contains a binding pocket for a hydrophobic ligand (Modis et al., 2003). Binding to a lipid on the target cell is thought to trigger a conformational change in E which presages membrane fusion. The presence of mannose appears important for receptor binding (Hung et al., 1999). There are two putative N-linked glycosylation sites: the first (ASN-153) is highly conserved among flaviviruses whereas the second (ASN-67) is specific for dengue virus. Virus replication in mosquito cells results in the addition of sugars to both asparagine residues and these play a role in identifying the C-type lectin DC-SIGN that is present on the surface of human dendritic cells (Navarro-Sanchez et al., 2003).
A number of cellular ligands have been suggested as forming the receptor for E. These include heparin sulphate (Chen et al., 1997), CD-14 (Chen et al., 1999) and a variety of glycoproteins (Marianneau et al., 1996; Hung et al., 1999). However, these are absent on the surface of dendritic cells that support virus replication. Moreover, immature dendritic cells appear more susceptible to virus than mature DCs (Wu et al., 2000). These cells express DC-SIGN, a C-type lectin that has been implicated in the susceptibility of dendritic cells to a spectrum of viruses, including Ebola (Alvarez et al., 2002). Tassaneetrithep et al. (2003) have shown that THP-1 cells transfected with DC-SIGN become susceptible to dengue and infectious particles can be recovered from culture supernatants. DC-SIGN has an homology with L-SIGN present on endothelial cells, and thus a similar interaction between virus and the epithelium may be instrumental in triggering the vascular changes that are a feature of dengue pathogenesis.
Cell fusion plays a role in initiating dengue and other flavivirus infections. Most of our knowledge as to how membrane fusion occurs has been obtained by studies in influenza virus, where fusion is initiated by a trimeric structure consisting of a coiled coil helix immediately adjacent to the fusion peptide. The latter is inserted into the host cell membrane, leading to the formation of a six-bundle helix (Skehel and Wiley, 1998). Stretches of hydrophobic amino acids make up the fusion peptide: these adapt either helices (associated with class I fusion events) and β barrels (class II fusion). Both allow charged (polar) amino acid side chains to become internalised, with hydrophobic exteriors promoting insertion amongst the aliphatic chains of a lipid bilayer.
The mechanism of these interactions has been clarified by Modis et al. (2004). After a lowering of pH within the endosome, the structure undergoes a substantial change in structure. The re-organisation of dimers into trimers results in the domain II fusion loops forming projections approximately 100 Å in length, the hydrophobic tips being inserted into the host membrane. The outer surface of the trimer contains grooves with the potential to interact with the anchor regions (domain III): these domains rotate during the fusion process, thus promoting distortion of the viral and host membranes leading to an irreversible conformational change and providing the necessary energy for fusion. The major significance of having class II structures would be one of speed (Corver et al., 2000). In the case of the class I structures of orthomyxoviruses, there is a prior requirement for the release of a fusion peptide and multiple interactions between subunits before fusion can occur. These reorganisations take longer than is probably the case for class II-mediated fusion where the envelope proteins are already ordered on the virus surface.
Phylogenetic analyses and evolution
Dengue virus is divisible into four distinct serotypes using conventional plaque-reduction neutralisation assays (Russell and Nisalak, 1967). The four serotypes may differ by as much as 40% in amino acid sequence of the envelope (E) protein which bears the ligand for neutralising antibodies. Dengue 1 and dengue 3 are the most closely related, with dengue 2 being somewhat more distant and dengue 4 virus the most divergent. Although it is possible that all four serotypes may have evolved in geographically isolated regions, today all four co-circulate in many areas of Africa, Asia and the Americas. Within each serotype, however, there does not appear to be much antigenic drift, which bodes well for vaccine development once the major obstacles of producing a polyvalent vaccine are overcome (see below).
The recent use of comparative gene analysis has stimulated work into defining the origin of dengue virus serotypes, and how the virus has evolved in association with its primate hosts. In particular, dengue virus is a paradigm of virus evolution in response to changing host numbers, immunity, behaviour and environment. Zanotto et al. (1996) compared the rate of nucleotide substitution between 123 mosquito- and tick-borne flavivirus genome sequences and concluded that those flaviviruses dependent upon mosquitoes for transmission evolve at around twice the rate of tick-borne viruses. Vector biology may account for at least some of this difference. The data discussed by Zanotto and colleagues suggests that, although dengue and yellow fever may have diverged more than 3000 years ago, the rate of evolution has not been constant, and that it is only in the last two centuries that dengue has undergone rapid divergence into four serotypes that are so distinct as no longer showing properties of serological cross-neutralisation. This accelerated divergence has paralleled the exponential increase in the global human population since the late 18th century when dengue first became apparent. The argument is that individual communities need to reach a certain size in order to sustain any new emerging variants.
Pathogenesis
The question as to why only a minority of patients develop DHF and DSS whilst the vast majority of cases do not is one of the biggest challenges of dengue research. Hypotheses abound, but the two main hypotheses are those focusing on variability of the virus and the role of the host immune response, particularly antibodies. Against this background there are clear epidemiological differences from region to region that implies host susceptibility may also be important in determining disease outcome
Strain variability
There is some evidence that virus variability may be a determining factor. Originally expounded by Rosen and colleagues over 30 years ago, this hypothesis states that virulence and disease outcome is determined by the genetic properties of the virus, the more virulent strains giving rise to the sequelae of DHF and DSS. Since this hypothesis was first proposed, there has been much debate as to whether certain strains can be linked to disease severity. The rapid advance in sequencing technology has added a further dimension to the somewhat anecdotal evidence of the past that has relied exclusively on serology. Perhaps the most convincing evidence that genotypic variation may be important stems from the observation that the so-called “American” strain of dengue 2 virus is not associated with DHF and DSS. This “American” genotype was responsible for an outbreak centred on Iquitos in Peru in 1995 (Watts et al., 1999). Over 50,000 secondary infections are known to have occurred but remarkably not a single case of DHF or DSS was seen. By comparison, in South East Asia at least 900 cases would have been expected during an outbreak of this size. In contrast, a dengue 2 virus outbreak in Cuba 14 years earlier—due to a different genotype—resulted in over 300,000 infections with 30,000 instances of illness progressing to DHF and DSS.
The relative avirulence of the “American” dengue 2 virus genotype has been attributed to an amino acid substitution at reside 390 of the envelope (E) glycoprotein, a change that certainly reduces the ability of the virus to replicate in cultures of monocyte-derived macrophages (Pryor et al., 2001). It is most unlikely that this is the sole explanation, however, as Shurtleff et al. (2001) showed that an American genotype strain was responsible for an outbreak in Venezuela characterised by a significant number of cases progressing to DHF and DSS. It is possible that the lack of DHF/DSS cases in Peru in 1991 was due to pre-existing antibodies to dengue 1 virus (Kochel et al., 2002). Antibody affinity may also be an important factor as well as antibody specificity. Also the length of time that elapsed from previous exposure to the heterologous serotype will determine the titre of residual reactive antibodies in patients suffering secondary infections and their capacity to complex with infectious virus. Nothing is known regarding immunological memory to these antigenic components, nor the extent to which a secondary antibody response can be triggered to the second virus.
Perhaps more critical in undermining this hypothesis is the failure to find significant sequence variation between dengue virus isolates taken from cases of acute dengue fever and those with the more severe DHF illness presenting at the same time and in the same locality (Rico-Hesse et al., 1997, 1998; Uzcategui et al., 2001). In addition, many of these studies have lacked data as to viral load, an important factor in determining just how readily virus can be transmitted via insect bite between humans. There is some evidence associating disease severity with viral burden (Vaughn et al., 2000) and certainly in other diseases such as Lassa fever and hepatitis C viral load can be an important marker of disease progression.
The role of antibody
Antibody-mediated enhancement has been shown for a number of virus infections, for example rabies and HIV, thus this is not a phenomenon restricted to dengue virus. A significant number of cases have been recorded in patients with pre-existing IgG antibodies to at least one of the other four serotypes of dengue virus. Halstead and O'Rourke (1977) proposed that such pre-existing antibodies complexed with infectious virus introduced during a subsequent infection with a heterologous virus: the resulting immune complexes result in a much greater uptake of infectious virus into susceptible cells expressing Fc receptors for antibody on their surface, such as macrophages. There is some experimental support for this hypothesis. Heterologous antibody can enhance the uptake of virus into cultured macrophages and the infection in vivo is more severe in rhesus monkeys injected with virus complexed with heterologous antibodies, whether these be from hyperimmune antisera or human sera. There is strong epidemiological and clinical support for this concept from work in South East Asia where DHF occurs predominantly in children. Pre-existing antibodies can be acquired from their mothers during the first year of life or in older infants and children as a result of a previous infection. In the Cuban epidemic of 1981, DHF occurred mainly in children exposed to dengue 1 virus 4 years previously when up to several million individuals were likely exposed to the virus.
There is every possibility that these findings cannot be so readily extended to understanding the pathogenesis of severe dengue infection in the Americas and elsewhere. Notwithstanding this caveat, most workers agree that antibody-mediated enhancement of virus uptake does play some role in dengue virus pathogenesis, but that it is unlikely to be the sole explanation as to why some cases progress to DHF and DSS. One of the major unknowns is just how variable an antigenic domain needs to be before a complexed antibody will fail to neutralise infectivity but still bind with sufficient affinity to allow the infectious virus–antibody complex to be taken up by susceptible cells. There have been some inroads to our understanding of the immune mechanisms that may be at work in such patients (for a review, see Kurane and Takasaki, 2001).
Host susceptibility
As with many human illnesses, there is always a suggestion that disease progression may be linked to the HLA type of the host (Chiewsilp et al., 1981; Stephens et al., 2002). This is reminiscent of attempts to link persistent hepatitis C and HIV with HLA haplotype, but studies with dengue virus often lack statistical power to offer any meaningful conclusions. T cell activation is linked to cells bearing V-β-17, but there does not appear to be any evidence of a “superantigen” effect. What is clear is that age at the time of exposure is important, together with sex and whether or not the person has had any previous exposure to dengue virus.
Treatment
Management of DHF and DSS is limited to supportive therapy. Symptoms can be alleviated by administering analgesics and aspirin to bring down the fever. Intravenous fluids counter fluid loss, especially when haemorrhage is present and the patient shows signs of shock. The administration of blood products is often considered.
Public health and the control of dengue
The impact of dengue on human health in affected countries is considerable. Apart from the obvious distress to infected persons, the economic burden is high, accounted largely by the additional medical care required, the need to implement expensive programmes of mosquito eradication and the indirect consequences on service industries increasingly dependent upon tourism. At present neither dengue nor yellow fever is a serious public health threat to countries outside of endemic areas but habitat areas supporting mosquitoes capable of transmitting dengue virus have rapidly expanded, becoming well established in the southern states of the USA and in Southern Europe.
A number of factors account for the spread and upsurge in dengue virus over the last 30–40 years. These are summarised in Table 2 . First, major demographic changes have taken place, particularly in Asia and oh the Indian subcontinent. A rapid increase in birth rate combined with a steady migration of rural populations into major conurbations has resulted in extensive proliferation of shantytowns and other areas with sub-standard sanitary conditions. These conditions are ideal for the spread of infectious diseases, including those dependent upon arthropod vectors. Modern society makes abundant use of plastic and polystyrene containers which when carelessly discarded adds to the already numerous nooks and crannies in which mosquito larvae can develop.
Table 2.
Factors accounting for the increase in dengue fever and its sequelae
|
The rise in air travel has meant that passengers incubating disease can spread infection in less than 36 h to any corner of the globe, as was so dramatically shown in 2003 when a single index case of SARS coronavirus in Hong Kong infected others staying in the same hotel. Within a week these contacts spread the disease to countries as far apart as Vietnam, Singapore and Canada. Dissemination by jet is not restricted to humans: the introduction of West Nile virus into the USA in 1999 is widely thought to be the result of an infected mosquito hitching a ride on a flight from the Middle East to New York.
A cardinal factor, however, is the progressive decline in public health infrastructure and associated surveillance programmes in countries where the virus is endemic or in areas where the vector and environmental conditions have established the potential for spread. Increasing pressure on scarce resources has resulted in the abandonment of programmes designed to prevent infections from becoming established in human populations. In their place are contingency plans focusing on crisis management and emergency measures. This trend, not confined to the developing world, has become coupled with a cessation in insecticide spraying in response to environmental concerns. Thus, A. aegypti has returned to areas previously declared free of the vector: the lack of effective surveillance in these expanding populations together with the pressure on limited funds for preventative medicine has combined to ensure the spread of dengue virus continues unabated.
Prospects for a dengue vaccine
Although a number of candidate vaccines are in various stages of clinical evaluation, currently there are no commercially available dengue vaccines. This is despite the development of an effective vaccine has been vigorously pursued now for nearly half a century. Progress has been frustratingly slow, largely because of the need to develop a tetravalent product that is equally effective against all four serotypes: the concern is that a monovalent vaccine specific for any one serotype might predispose a recipient to severe disease if that vaccine is later exposed to a serotype other than that present in the vaccine. The emphasis, therefore, has been on the development of a tetravalent vaccine consisting of attenuated strains of all four serotypes. Given the heterogeneity in properties other than just antigenicity between the serotypes, and between genotypes within each serotype, strain selection has been fraught with difficulties.
The emphasis on developing an attenuated vaccine has been driven by a consideration of manufacturing costs and also the perceived need for a vaccine that can be delivered as a single dose, particularly if realistic immunisation protocols are to be developed. Chemical inactivation of wild type virus has received considerably less attention for these very reasons, with chemical inactivation carrying an additional risk of alterations to the structural properties of the virus to such an extent that the immunogenicity of the envelope proteins is diminished. Inactivation also introduces the chance of variable damage between the different viruses going into a tetravalent product. The product profile of a putative dengue vaccine is summarised in Table 3 .
Table 3.
Product profile for a candidate dengue virus vaccine
|
Work is also hampered by the lack of a suitable animal model capable of mimicking the course of the disease in humans, particularly the progression from acute disease to DHF and DSS. Non-human primates have been used extensively but these animals, whilst supporting virus replication, do not show any manifestations of clinical disease. Thus, their use is restricted to the measurement of a reduction in viraemia that might occur in immunised animals subsequently challenged with wild type virus. The use of non-human primates also presents increasingly difficult issues of ethics and cost, the latter in particular often inhibiting the use of animals in sufficient numbers to make meaningful and statistically valid comparisons. Mice, although susceptible, present considerable limitations. Dengue is one of the least neurotropic for mice among the flaviviruses. Nevertheless, there have been some advances using mice deficient in interferon and also with mice transplanted with human hepatocellular carcinoma cells: these animals when injected with dengue virus mimic at least some of the features of the clinical disease seen in humans (Johnson and Roehrig, 1999; An et al., 1999).
Passage in cell culture is the traditional method for attenuating virus. However, dengue virus generally produces only low titres in mammalian cells. This is a major obstacle, unless large-scale tissue culture reactors are available. These systems are expensive to maintain, however, and are often plagued by contamination as well as consuming large quantities of tissue culture media. Notwithstanding these problems, the use of modern bulk cell culture techniques means that large quantities of other flaviviruses such as Japanese encephalitis virus can now be grown to titres in excess of 1012 pfu/ml (quoted in Pugachev et al., 2003).
Many attempts to produce attenuated virus in cell culture have largely proved disappointing: the virus either remains too reactogenic for human use or becomes over-attenuated and as a consequence fails to induce an immune response consistently in all recipients (Kanesa-Thasan, 1998). Early experience with this work showed the importance of carefully selecting vaccine candidate strains by comparing immunogeni-city with passage level. These efforts would have been helped by more information as to what constitutes suitable markers of attenuation. Nucleotide changes in the 5′ and 3′ NTR regions or amino acid changes in NS1 and NS3 leading to the appearance of small plaque variants, reduced replication in mosquitoes, as such are all potentially useful in vitro correlates of the extent to which virulence has been modified by cell passage.
The most promising attenuated vaccine candidates have been developed at Mahidol University in Bangkok. Candidate viruses representing each serotype were prepared by sequential passage, first in primary dog kidney cells, then in either primary African green monkey kidney cells or in foetal rhesus lung cells (Rothman et al., 2001). The complexity of these studies is illustrated by the titre of each vaccine candidate (equating to a 50% infectious dose) recovered from the blood of human volunteers. These titres were 104, 5, 3500 and 150 pfu for serotypes 1–4, respectively, i.e. representing a four log10 difference in response against the four serotypes. This makes for considerable problems in formulating a tetravalent vaccine. Phase I trials of a tetravalent vaccine have been conducted using these candidates using approximately 3–4 log10 of virus per inoculum. There were no serious adverse events associated with this vaccine, although recipients complained of headaches, a mild fever (∼38°C) and a macropapular rash that was sometimes puritic. There was evidence of a viraemia in all volunteers from 5 to 12 days after injection (Kanesa-Thasan et al., 2001). The highest anti-dengue virus antibody titres were present against dengue 3, and only dengue 3 was recovered consistently from viraemic blood samples. Despite the equivalence of dose, the data suggested that either the dengue 3 virus component was not sufficiently attenuated or that competitive interference occurred. Slightly more encouraging was the finding of proliferative and cytotoxic T-lymphocyte responses against all the serotypes except dengue 4 (Rothman et al., 2001). These results show just how much of an uphill task the process of developing a tetravalent vaccine represents.
Other more traditional approaches to developing live attenuated vaccines include exposure of wild type isolates to mutagenic chemicals followed by a process of selecting those mutated viruses with evidence of reduced virulence. This approach has not received nearly as much attention for developing a dengue vaccine, however. Blaney et al. (2001) followed this route to modify a dengue 4 candidate previously shown to give rise to fever and a rash in volunteers despite a deletion in the 3′ NTR. A number of temperature-sensitive mutants were recovered after exposure to the mutagen 5-fluorouracil: these all showed a reduced ability to grow in mice brains.
An alternative approach exploits the observation made by Bray and Lai (1991) that the structural genes of one dengue virus serotype can be exchanged for the homologous genes of another. These investigators successfully generated chimeras by taking a dengue 4 clone and replacing all of the structural genes (C, prM and E) with those of dengue 1, dengue 2 or dengue 3. The chimeras grew well in monkey kidney cells and mosquito C6–36 cells and induced high titres of neutralising antibodies in monkeys (Table 4 ). This work was taken further by attenuating the dengue 4 template by introducing non-lethal mutations at the 5′ and 3′ NTRs. This technology has been extended still further by replacing the dengue 4 template with that of the PDK strain of dengue 2, a component of the attenuated, tetravalent vaccine described above. As yet the immunogenic potential of these chimeras has only been reported using mice (Huang et al., 2000).
Table 4.
Neutralising antibody titres in rhesus monkeys injected with one or two doses of chimeric yellow fever/dengue virus
| Virus chimera | Geometric mean titre (GMT) |
||
|---|---|---|---|
| First dose | Second dose | P value | |
| Yellow fever/DEN1 | 640 | 1810 | 0.12 |
| Yellow fever/DEN2 | 1437 | 1613 | 0.80 |
| Yellow fever/DEN3 | 640 | 1016 | 0.26 |
| Yellow fever/DEN4 | 3225 | 5120 | 0.01 |
(from Guirakhoo et al., 2002)
© 2005
Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active.
Given the demonstration of chimera formation between serotypes of dengue, it is not surprising that chimera vaccines have been attempted using the yellow fever vaccine strain 17D. This methodology is being commercialised by Acambis Inc.4 as ChimeraVax™. Experience with Japanese encephalitis virus—also a mosquito-borne flavivirus—showed that the replacement of genes between different flaviviruses was restricted to the envelope proteins prM and E. The Japanese encephalitis (JE) chimera was significantly less lethal for suckling mice compared to the 17D virus. Important from the point of its potential use in humans, the 17D-JE chimera did not replicate in either Aedes or Culex mosquitoes. The development of these new immunogens is described further by Pugachev et al. (2003).
Viable yellow fever 17D/dengue chimeras have been produced for all of the dengue serotypes (Guirakhoo et al., 2001). These have been constructed using clones expressing the prM and E proteins of wild type isolates from different localities. These chimeras are significantly less virulent for mice compared to the 17D virus. All grew to a high titre in Vero cells, a significant advantage compared to wild type dengue isolates. Experiments in rhesus monkeys have shown only low viraemias but the presence of specific B cell responses. Encouragingly the induced neutralising antibodies were sufficient to protect the animals against challenge with wild type virus. Although these data are encouraging, the adjustment of the dose for each individual component is as much of a problem as when a tetravalent vaccine was used composed of attenuated virus strains. However, a uniform antibody response to all four serotypes could be achieved by giving a second dose 60 days after the first (Guirakhoo et al., 2002). Importantly, pre-existing antibodies to 17D virus did not affect the titre of anti-dengue virus antibodies generated by these chimeras (Table 5 ). At present, the ChimeraVax™ dengue 2 vaccine is the subject of phase II clinical trials (Pugachev et al., 2003).
Table 5.
Flaviviruses—key questions
|
As has been the case with many high priority vaccine development programmes, there has been intensive effort directed towards evaluating alternative approaches that avoids the need to produce large quantities of whole virus. These include sub-unit vaccines using expressed proteins from mammalian cells, insect cells or bacteria, and DNA immunogens incorporating the relevant dengue genes and recombinant vectors expressing dengue envelope genes.
Much of this effort is summarised by Kinney and Huang (2001). The key questions are, first, is one protein sufficient for the induction of protective immunity, and second, what role, if any, is played by protein conformation. Most workers agree that the envelope protein (E) is of primary importance for inducing a protective antibody response but there is a body of evidence suggesting other proteins may also be required, such as prM/M, NS1 and NS3. Certainly T cell responses can be measured against many, if not all, NS proteins in patients with flavivirus infections. It is worth remembering that the detection of antibody to the NS3 protein of hepatitis C virus gave the first clue as to the presence of this virus in cases hitherto grouped as non-A, non-B hepatitis. Recombinant hepatitis C vaccines based upon the E1 and E2 proteins have been rapidly worked up and tested as vaccine candidates. Attempts to produce a hepatitis C vaccine have shown, however, that protein conformation is vitally important if the correct B cell specificity is to be retained and presented to the immune system.
Good responses to the E protein of dengue 4 have been obtained using recombinant vaccinia virus, even though the C-terminal part of the E protein was deleted during the cloning process (Men et al., 1991). However, lower levels of protection have been seen using similar recombinants containing the truncated E protein of Japanese encephalitis virus. It is now known that one of the major roles of prM is to ensure the correct folding of E (Lorenz et al., 2002): probably for this reason the expression of subviral particles containing both prM and E also induces good titres of neutralising antibodies (Fonseca et al., 1994).
New advances in molecular virology have led to the generation of infectious clones. Attenuation of the potential virulence of such clones has focused on modifications to both the 5′ and 3′ NTR.5 For example, Men et al. (1996) took an infectious clone of dengue 4 and deleted sequences in the 3′ NTR. These clones produced smaller plaques in cell culture and a lower viraemia in monkeys. This initial finding was sufficient to encourage a small human trial in 20 volunteers using one clone with the 30 nucleotides deleted spanning bases 172–143 reading from the 3′ end. A low titre viraemia was confirmed in 14, with neutralising antibodies found in all cases (Durbin et al., 2001). This and other clones with 3′ NTR modifications do not grow well in mosquitoes and mosquito cell cultures. Care will have to be taken using single point mutations to generate this type of attenuated vaccine clone, however, as there is evidence of phenotypic reversion to wild type over prolonged periods of growth in C6-36 cells (Markoff et al., 2002).
Kyasanur Forest disease
Kyasanur Forest disease was first isolated in 1957 from a dead monkey found in Mysore (now Karnataka) in India, and is hence known locally as monkey fever. This followed a series of unexplained deaths among monkeys living in the Kyasanur Forest. In contrast to Omsk haemorrhagic fever (see below), there are a steady number of cases reported each year, predominantly among villagers living at the forest margins. A number of laboratory infections have also been reported.
Epidemiology
Kyasanur Forest disease virus is transmitted by the bite of the tick Haemaphysalis spinigera, a tick often found at the forest margins. Rodents, bats and shrews as well as monkeys are also susceptible to the virus. This disease has most likely emerged as a consequence of extensive forest clearing to make way for cattle grazing. This in turn has led to a rise in tick populations and thus an increase in risk of human exposure. Forest rodents are the most likely host reservoirs, monkeys and man serving as only coincidental hosts, with infected primates serving to amplify outbreaks.
Clinical features
A sudden fever rising rapidly to 40°C is accompanied by myalgia and headache, these early symptoms closely resembling those of dengue fever. In contrast to dengue fever, however, muscle pain is localised in the upper thorax and neck. A helpful early sign is the appearance of vesicular lesions on the upper palate. A lymphadenopathy is usually present. Haemorrhagic disease appears from day 3, consisting of gastrointestinal bleeding accompanied by bleeding of the gums and nose. Although there is a thrombocytopenia, there are neither indications of major disruption to the haematopoietic system nor signs of capillary damage. The occurrence of laboratory-acquired infections has allowed a more detailed study of the disease profile: these show a stronger involvement of the CNS than is evident from clinical cases, with signs of coarse tremors, abnormal reflexes and mental disturbances. Mortality is between 3 and 5%. It has been noted that naturally acquired infections occur predominantly among residents of villages with minimal or no infrastructure. These people have as a consequence a high parasitic burden and this in turn leads to elevated levels of interferon and IgE. Some workers have gone as far as suggesting that these immunological markers may play a role in determining the course of the clinical disease.
Diagnosis
As is the case with many flavivirus infections, direct virus isolation is possible from blood samples using either suckling mice or cell cultures. Antibodies can be detected using ELISA or other standard serological methods, including plaque reduction neutralisation.
Control
A formalin-inactivated vaccine has been prepared locally and preliminary clinical trials has shown good protection (Dandawate et al., 1994). This vaccine is now being produced in Bangalore, India, for local use.
Alkhurma virus
In 1995, a new flavivirus was recovered from patients with severe haemorrhagic disease in Saudi Arabia (Zaki, 1997). Of the six cases identified initially, four were proved fatal. All were butchers and it is probable that the infections were acquired as a result of handling infected sheep or coming into contact with ticks feeding on infected animals. All patients showed signs of fever, headache, generalised muscle pain, nausea and arthralgia. There was a marked thrombocytopenia and biochemical evidence of both liver and renal damage. A further four cases were identified serologically following the development of a test for IgM antibodies.
Charrel et al. (2001a, b) have sequenced the genome of Alkhurma virus and shown that it is closely related to Kyasanur Forest disease virus. It remains to be determined as to the nature of the common ancestor of these two agents and to how Alkhurma virus has emerged in a region some 2000 km away from the endemic region of Kyasanur Forest disease.
Omsk haemorrhagic fever
The endemic area is thought restricted to the Omsk and Novosibirsk-Oblast regions of Siberia, a landscape essentially of forest and steppe. Almost all the cases described in the literature occurred between 1945 and 1958, with a peak between 1945 and 1949. Sporadic outbreaks continue to occur, the most recent in 1991. Laboratory infections have also been reported, again indicating that this virus can be readily transmitted to humans.
Omsk haemorrhagic fever is transmitted by the ticks Dermocentor reticulates and D. marginatus. The virus can persist in ticks throughout the winter period, although there is no evidence of transovarial transmission. Interestingly, this virus cannot readily be distinguished from tick-borne encephalitis by use of hyperimmune antibodies: despite this close serological relationship, Omsk haemorrhagic fever causes severe disturbance to the haematological system whereas tick-borne encephalitis virus does not. The reason for this difference is not known, but lies at the heart of understanding why closely related viruses—at least on virological and serological grounds—can cause very different clinical diseases. The virus of Omsk haemorrhagic fever is maintained in nature by muskrats (Ondatra zibethica) and water voles (Arvicola terrestris). The muskrat is a recent introduction into this region and is highly susceptible to the virus. The infection rapidly progresses in this species to a fulminant, haemorrhagic disease that is invariably fatal. It is the hunters of the muskrat in Western Siberia and their close family members that are most at risk, with infected muskrats shedding large quantities of virus in urine and faeces. Hunters become exposed when skinning infected animals. Family members of these hunters account for approximately a quarter of human cases reported so far, indicating that human-to-human transmission can readily occur between close contacts.
Clinical features
The first signs of human infection are the sudden onset of fever accompanied by myalgia and headache. Haemorrhagic manifestations quickly follow with epistaxis and gastrointestinal bleeding. A bronchial pneumonia is frequently present, and only occasionally is there an indication of CNS involvement (Pavri, 1989). The case fatality rate is less than 3%. Diagnosis is by detection of specific antibodies using serology and virus isolation from a blood sample.
Recently the complete sequence of the Bogoluvoska strain has been reported (Lin et al., 2003). Phylogenetic analysis revealed a close relationship with other viruses of the tick-borne complex in the family Flaviviridae, but sufficiently distinct as to form a separate clade, a conclusion compatible with Omsk haemorrhagic fever virus only having minimal neurotropism. The glycoprotein (E) of Omsk haemorrhagic fever virus is also distinct from that of Kyasanur Forest disease and Alkhurma viruses, two other tick-transmitted flaviviruses with the capacity to induce haemorrhagic disease. A detailed comparison with these two and other neurotropic flaviviruses suggested only three amino acid changes correlated with the divergence in biological properties of these viruses. One of these changes, at residue 76, is close to the presumptive membrane fusion domain, the remaining two within the stem-anchor region of the protein. Further differences were noted outside of the single open reading frame, in particular the 5′ NTR secondary structure of the Omsk haemorrhagic fever viral genome is likely very different, a possible reflection on the unique virus-vector relationship of the virus with ticks of the Dermocentor genus. Although speculative, it is clear that Omsk haemorrhagic fever virus has adapted to a particular vector and ecological niche quite separate from those of other tick-borne flaviviruses co-circulating in Siberia.
Control
Prevention of disease can be accomplished readily by avoidance of ticks and taking care when handling dead muskrats. No vaccine is available, although there is an opinion that the serological relationship between Omsk haemorrhagic fever and tick-borne encephalitis virus is sufficiently close that the administration of tick-borne encephalitis vaccines to those at risk is beneficial. Although not indicated for Omsk haemorrhagic fever, vaccines against tick-borne encephalitis virus were used in Siberia to control the 1991 outbreak under a special directive of the regional authorities.
Summary
The spectrum of diseases caused by flaviviruses range from the encephalitic to the viscerotropic in terms of pathology. It is the latter that have the propensity to various haemorrhagic manifestations. Despite the availability of vaccines against yellow fever represents one of the major achievements of modern medicine, this very achievement is under threat as major yellow fever outbreaks can rapidly lead to vaccine shortages and attempts to curb childhood yellow fever by introducing the vaccine into childhood immunisation programmes has faltered in recent years. Coupled with the resurgence of A. aegypti in major conurbations—especially in the Americas—the scene is set for outbreaks on a scale not seen for over a century and there is a very real risk that these may not be fully contained if vaccine is in short supply. The threat posed to public health by dengue virus is considerable, this virus having gone a major expansion in endemicity in the last 50 years to affect over 50 countries. There has been a corresponding escalation in the number of cases of DHF and shock syndrome, especially among younger patients. Despite dengue virus now being the most widespread of all the viral haemorrhagic fevers, licensed vaccines are still not in sight, largely due to the hurdle of developing a vaccine capable of inducing protection equally against all four serotypes. Although Omsk haemorrhagic fever and Kyasanur Forest disease viruses pose problems in very restricted geographical regions, there are signs that these agents may represent the vanguard of other agents still to be discovered. The emergence of Alkhurma virus closely related to Kyasanur Forest disease virus is a worry, having been isolated in Saudi Arabia some 2000 km from the endemic zone of Kyasanur Forest disease virus. Ease of transport and further changes to the environment will only escalate the risk that flaviviruses pose to human health on a global scale. Although the short-term prospects for specific treatment with antiviral is bleak, the intensive efforts to develop new therapies for chronic hepatitis C almost certainly will generate in the longer term compounds that may prove beneficial for the treatment of a spectrum of flavivirus infections, in the same way that hepatitis B therapy has benefited enormously from the discovery of antivirals against HIV.
Footnotes
The term quarantine is derived from the Italian quaranta, meaning 40 days, a measure first introduced by the Venetian Republic in 1374.
By convention, the nucleic acid strand coding directly for protein is referred to as the positive strand (i.e. is of positive polarity). The complementary sequence of nucleotide bases arising from the replication process cannot be used directly for protein synthesis and is thus of negative sense with respect to ribosomal translation (i.e. is of negative polarity).
By convention, the four serotypes are frequently referred to as DEN-1, DEN-2, DEN-3 and DEN-4, respectively.
NTR: Non-translated regions at the end of a viral genome.