Main

Hendra virus (HeV) and Nipah virus (NiV) join a growing list of viruses for which bats have been implicated as the natural host, a list that started with rabies virus in 1934 (Ref. 1) and the most recent additions to which were severe acute respiratory syndrome (SARS)-like coronaviruses in 2005 (Refs 2, 3). Bats are classified in the order Chiroptera (from the Greek 'cheiros', hand, and 'pteros', wing), and it is within the genus Pteropus in the family Pteropodidae, or Old World fruit bats, that we find the natural hosts of HeV and NiV. Pteropid bats are commonly referred to as flying foxes4 (Fig. 1).

Figure 1: Flying foxes, their distribution and the locations of disease outbreaks caused by Hendra virus and Nipah virus.
figure 1

a | Pteropus poliocephalus is an Australian flying fox and member of the family Pteropodidae, one of 18 bat families in the order Chiroptera. There are four Pteropus species in Australia4. b | Sixty-five Pteropus species are distributed from Madagascar through the Indian subcontinent to south-eastern Asia and Australia and as far east as the Cook Islands4. Some Pteropus species are among the largest of all bats, weighing as much as 1.2 kg and displaying a wing span of up to 1.7 m. Pteropus species are unique because they lack the complex neural and behavioural mechanisms required for echolocation that characterize the vast majority of bat species. Instead, they have large eyes and they navigate visually, feeding mainly on fruit and flowers, which they locate by smell. The sites of disease outbreaks caused by henipaviruses are indicated. Map modified with permission from Ref. 4 © (2002) University of New South Wales Press.

HeV and NiV are not the only paramyxoviruses likely to have a bat origin. Menangle virus, which caused a reproductive disease in an Australian piggery in 1998, also seems to have Pteropus species as its natural hosts5,6, and Tioman virus was found in the urine of the Malaysian flying fox Pteropus hypomelanus7. Bat parainfluenza virus was isolated in India from a member of the Rousettus genus within the family Pteropodidae8. Last, Mapuera virus was isolated in Brazil from a bat of the Sturnira genus in the family Phyllostomidae (New World leaf-nosed bats)9. The genetic constitution of HeV and NiV and their zoonotic potential, high virulence and wide host range set them apart from other paramyxoviruses. This review provides an overview of the henipaviruses and summarizes recent molecular analyses of the structure and function of henipavirus proteins that have contributed greatly to our understanding of these unique pathogens.

The genus Henipavirus

Paramyxoviruses are classified in two subfamilies, Paramyxovirinae and Pneumovirinae (Box 1). The virions and nucleocapsids of HeV and NiV display morphological features that are typical of members of the subfamily Paramyxovirinae (Fig. 2a,b), and their genetic organization resembles that found in viruses in the Respirovirus and Morbillivirus genera in this subfamily10 (Fig. 2c).

Figure 2: Structure of henipaviruses and their genomes.
figure 2

a | A schematic representation of henipavirus structure. Henipaviruses, like other paramyxoviruses, contain a linear ribonucleoprotein (RNP) core consisting of a single-stranded genomic RNA molecule of negative polarity to which nucleocapsid proteins (N) are tightly bound in a ratio of one N for every six nucleotides127,139. The RNP also contains smaller numbers of the phosphoprotein (P) and the large (L) polymerase protein, both of which are required to transcribe genomic RNA into mRNA and anti-genome RNA. The RNP core is surrounded by an envelope from which two spikes protrude; one is the receptor-binding glycoprotein (G) and the other the fusion (F) protein. The G and F proteins are arranged as homotetramers and homotrimers, respectively. The matrix protein (M) which underlies the viral envelope is important in determining virion architecture and is released from the RNP core on its entry into cells. b | Electron micrograph of Hendra virus (HeV). The ultrastructural characteristics of HeV and Nipah virus have been reviewed140. c | The henipavirus genome. The negative-sense genomic RNA is presented in the 3′ to 5′ orientation. The open reading frames indicated by the yellow arrows encode the nucleocapsid protein (N), phosphoprotein (P), matrix protein (M), fusion protein (F), glycoprotein or attachment protein (G) and large protein (L) or RNA polymerase, in the order 3′-N-P-M-F-G-L-5′. The vertical lines represent gene start and stop signals. Note the long untranslated 3′ regions in all genes except the L gene. All genes except the P gene are monocistronic. The P gene of henipaviruses encodes not only the P protein, but also V, C and W proteins (Box 3). Genomic RNA in RNPs is transcribed by the viral polymerase which associates with the RNP at the 3′ terminus and sequentially generates discrete mRNAs from each of the viral genes. The mRNAs are not produced in equimolar amounts and there is a transcription gradient from the N to the L gene, with significant attenuation at the M–F and G–L gene junctions of HeV, a pattern of attenuation more closely resembling that observed in Sendai virus than in measles virus141,142,143.

Several features distinguish henipaviruses from other paramyxoviruses. Genetic attributes include the unique, genus-specific 3′ leader and 5′ trailer sequences, which function as promoters for transcription and replication of genomic RNA, respectively11,12, and the presence of the sequence GDNE in a highly conserved catalytic site in the transcriptase protein, instead of the GDNQ sequence that is found in almost all other non-segmented negative-strand RNA viruses12,13. Interestingly, one of the most prominent features that differentiates henipaviruses from other paramyxoviruses is the length of the viral genome, which at 18,234 nucleotides (nt) for HeV and 18,246 nt for NiV is approximately 2,700 nt (15%) longer than others in the family12,13. But, as far as genome length is concerned, the henipaviruses are not alone, and the genomes of the as-yet-unclassified paramyxoviruses Tupaia virus14 and J virus15 are 300 nt shorter and >700 nt longer, respectively, than those of henipaviruses. However, henipaviruses remain unique in having their 'extra' genomic nucleotides in the form of long untranslated regions, mostly at the 3′ end in five of the six transcription units, the exception being the L gene10,11,12 (Fig. 2c). Despite the extra length of the henipavirus genome, all proteins except one, the phosphoprotein (P), are approximately the same size in the Respirovirus, Morbillivirus and Henipavirus genera; the henipavirus P protein is approximately 100 to 200 amino acids longer than cognate respirovirus and rubulavirus proteins13,16.

Henipaviruses also have unique biological features. They are the only zoonotic paramyxoviruses and are highly pathogenic. Although there are no data on the total number of people infected with NiV during an outbreak in Bangladesh in 2004, 75% of patients that were identified as having NiV-associated illness on the basis of positive serology or on epidemiological grounds died17. The range of species that are susceptible to henipaviruses is also quite remarkable. In addition to at least three pteropid species, NiV infects five terrestrial species in four mammalian orders18,19,20,21,22,23. Experimental henipavirus infections extend the number of susceptible terrestrial orders to five by including the Rodentia24,25. This exceeds the extensive host range of canine distemper virus, a morbillivirus that naturally infects many species in the order Carnivora, such as dogs, ferrets, raccoons and lions, but can also experimentally infect hamsters and pigs (order Rodentia and Artiodactyla, respectively)26,27.

The susceptibility of humans, the high virulence of the viruses and the absence of therapeutic modalities and vaccines have led to the classification of HeV and NiV as Biosafety Level 4(BSL4) pathogens. Globally, only a limited number of laboratories have appropriate facilities for growth and handling of BSL4 pathogens. Among these, even fewer have facilities in which animals, particularly large animals, can be infected with HeV or NiV and in which workers can be protected in plastic suits, supplied with breathing air. This has limited the number of investigations into the interaction of henipaviruses with their natural hosts, susceptible livestock and laboratory animal species. However, issues of personal safety and biocontainment are minimized when working with cloned henipavirus genes in eukaryotic expressions systems, and it is from such studies that most of our current knowledge on the structure and function of henipavirus proteins has been obtained. Before discussing the molecular observations that shed light on the unusual biological properties of these emerging pathogens, let us first summarize the emergence of HeV and NiV in a timeline (TIMELINE) and briefly review the clinical and pathological outcomes of virus infection.

Henipavirus infections

Henipavirus infections are characterized by their systemic nature, with evidence of infection in multiple organ systems. The outcome of infection differs significantly in terrestrial and chiropteran hosts (Box 2). In terrestrial species, both HeV and NiV display a predominantly respiratory or neurological tropism, depending on the host. Infections can be associated with high morbidity and case-fatality rates, such as in HeV infection of horses and NiV and HeV infections of people, or lower morbidity and mortality rates, best represented by NiV infection of pigs20,22,28,29,30. HeV-infected horses develop acute, febrile respiratory disease that is sometimes accompanied by facial swelling, ataxia and, terminally, copious frothy nasal discharge31. Respiratory signs also predominate in NiV infection of pigs, especially in young animals that develop fever, nasal discharge, rapid and laboured respiration and a notable, harsh and non-productive cough, giving rise to the name 'barking-pig-disease'22. Nevertheless, neither HeV nor NiV cause a solely respiratory syndrome in horses and pigs, respectively. A proportion of convalescent horses have re-presented with neurological signs, and clinical signs consistent with multifocal neurological disease have also been observed in growing pigs, together with sudden death in mature animals22.

In humans, symptomatic NiV infection has mainly taken the form of severe acute encephalitis. Many NiV-infected patients have reduced levels of consciousness at presentation and signs consistent with brain-stem involvement32. However, up to 25% of cases also exhibited respiratory signs30,33. Infection with NiV can also take a more chronic course, with serious neurological disease occurring late (in excess of 4 years) following a non-encephalitic or asymptomatic infection34,35. The recurrence of neurological manifestations has also been noted in patients who had previously recovered from acute encephalitis (relapsed encephalitis)34. Cases of relapsed encephalitis presented from several months to nearly 2 years after the initial infection and, interestingly, two further cases of relapsed encephalitis were observed in the autumn of 2003, some 4 years after initial infection35. Taken together, there is nearly a 10% incidence rate of late encephalitic manifestation, with a mortality rate of 18%. So, with both NiV and HeV, a prolonged period of infection is possible before the manifestation of serious neurological disease. Viral antigen was found in neurons in patients who died of late-onset encephalitis34, raising questions about the underlying mechanisms that allow these viruses to escape immunological clearance for such an extended period.

Many fewer cases of human HeV infection have been recorded, and the associated disease syndrome is correspondingly less well defined. Affected patients have had influenza-type symptoms, and fatal HeV encephalitis has been described in one patient more than a year after a self-limiting episode of meningitis that was, in retrospect, also attributed to HeV infection29,36.

Molecular insights into henipavirus biology

The molecular basis of paramyxovirus virulence, host range and cell tropism is determined to a significant degree by the cell-attachment (G) and fusion (F) proteins, which determine host range and cell tropism by virtue of their roles in binding to cell receptors and fusing the virus and host-cell membranes; and by the products of the P gene, which modulate virulence by abrogating the cellular interferon (IFN) response. Recent studies on the structure and function of henipavirus proteins expressed from cloned genes have shown that the henipavirus G and F glycoproteins and the P protein also influence host range, cell tropism and virulence, but do so in ways that are both surprising and unique.

The henipavirus G protein. Unlike other members of the Paramyxoviridae, which have a limited host range, henipaviruses naturally infect flying foxes, horses, pigs, cats, dogs and humans, and experimental investigations have extended this host range to include guinea pigs and hamsters. The susceptibility of several cultured cell types was noted during the initial attempts to isolated HeV from clinical samples28. The broad species tropism of henipaviruses is also reflected in an in vitro fusion assay in which henipavirus F and G proteins that were expressed on the surface of effector cells by vaccinia virus facilitated fusion with adjacent target cells from a range of species, including rabbit, monkey and mouse37. The fact that an identical pattern of target-cell susceptibility was observed in fusion assays using HeV and NiV glycoproteins indicated that both viruses used the same cell receptor38. These observations indicate that henipavirus receptors are ubiquitously expressed.

Paramyxoviruses fall into two broad categories, depending on whether or not their attachment glycoprotein has domains that bind red blood cells and release terminal N-acetyl neuraminic acid residues from carbohydrate moieties. The presence of such haemagglutination and neuraminidase activities in avulaviruses and rubulaviruses correlates with their binding to N-acetyl neuraminic acid in cell-surface glycoprotein and glycolipid receptors39. However, although measles virus (MeV), a morbillivirus, displays haemagglutination activity, it binds to cells by a sialic-acid-independent mechanism40. The cell-surface-expressed proteins CD46 and SLAM (CD150) have been shown to act as MeV receptors, with SLAM now regarded as a universal morbillivirus receptor41,42,43,44. By contrast, HeV and NiV resemble members of the Pneumovirinae in possessing a third class of attachment protein, G, which displays neither haemagglutination nor neuraminidase activities. However, the G protein of HeV and NiV is structurally unrelated to the cognate pneumovirus protein45. These early observations indicated that henipaviruses, like morbilliviruses, might use cell-surface proteins as receptors in a process that does not require N-acetyl neuraminic acid11,28,45,46. Indeed, the susceptibility of target cells to HeV-mediated and NiV-mediated fusion could be destroyed by protease treatment37.

The attachment proteins of paramyxoviruses are type II membrane glycoproteins consisting of a cytoplasmic tail, a transmembrane region, which anchors the protein to the viral envelope, a stalk and a globular head, which is composed of six protein sheets organized in a propeller-shaped structure47,48. Although the HeV G attachment protein has low amino-acid-sequence homology with attachment proteins from other paramyxoviruses, its globular head retains the propeller shape predicted for members of the family, and the location of neutralizing epitopes resembles that observed for other members of the Paramyxovirinae45,49. Soluble forms of HeV and NiV G proteins (sG), generated by replacing the cytoplasmic tail and transmembrane domains with an immunoglobulin κ leader sequence, retain biological activity50,51. They also retain an oligomeric structure, and bind to cells that are susceptible to henipavirus infection but fail to attach to infection-resistant cells, and as immunogens they elicit a potent crossreactive neutralizing antibody response against infectious HeV and NiV50.

Preincubation of cells with HeV sG resulted in dose-dependent inhibition of both HeV and NiV infection, probably by blocking viral receptor engagement50, and not surprisingly, sG proteins had key roles in determining and confirming the recent identification of the henipavirus cell receptor. NiV sG fused to the Fc region of human IgG1 was used to immunoprecipitate the receptor from the plasma membrane of cells that were permissive for NiV F-mediated and G-mediated fusion. The receptor was identified as ephrin B2 by mass spectrometry51. In another approach, microarray analysis was used to identify mRNA sequences that are expressed in henipavirus-susceptible cells but not in cells refractory to henipavirus infection52. From a list of genes encoding predicted membrane-localized proteins found only in susceptible cells, only one — encoding ephrin B2 — could render resistant cells susceptible to fusion as well as infection not only by NiV but also by HeV52. Henipavirus infection of ephrin-B2-expressing susceptible cells was blocked by soluble recombinant ephrin B2 (Ref. 52), and the ability of ephrin B2 to serve as a receptor for virus was confirmed by showing that the sG protein of HeV and NiV bound to ephrin B2 in vitro with high affinity52.

Ephrin B2 is a member of a family of cell-surface glycoprotein ligands that bind to ephrin (Eph) receptors, a large family of receptor tyrosine kinases53,54. Although initially identified in vertebrates as regulators of axon path finding and neuronal cell migration, Eph receptors and ephrins have been found in arthropods, nematodes and even sponges54. This indicates a primordial function for these molecules, and they are now known to mediate cell-to-cell communication and regulate cell attachment and repulsion. Eph receptors and ephrins have key roles during development, especially in the nervous and vascular systems55,56. Ephrin B2 is found in neurons, smooth muscle, arterial endothelial cells and capillaries55,57,58,59. The structure of the Eph B2 receptor, ephrin B2 and the complex that they form has been determined by X-ray crystallography60. Ephrin B2 is glycosylated but the side chain is short, containing a single mannose residue and two N-acetyl glucosamine residues, and lacks sialic acid61. The identification of ephrin B2 as the cell receptor for both HeV and NiV and the widespread occurrence of the molecule in vertebrates, particularly in arterial, but not venous, endothelial cells, in the smooth muscle of the tunica media and in neurons, provide an explanation for the wide host range of henipaviruses and the systemic nature of the infections they cause51,52.

The henipavirus F protein. Paramyxoviruses contain two membrane glycoproteins: the G glycoprotein, which is required for cell attachment, and the F glycoprotein, which is required for the fusion of the viral and host-cell membranes (Fig. 2). Following virus attachment to a permissive host cell, fusion occurs at neutral pH, delivering the viral genetic material into the cytoplasm. In a related process, cells expressing these glycoproteins on their surfaces can fuse with other receptor-bearing cells, resulting in the formation of multinucleated giant cells (syncytia), a hallmark of the paramyxovirus cytopathic effect. The paramyxovirus F glycoproteins are class I fusion proteins with two α-helical domains, referred to as heptad repeats, that are involved in the formation of a trimer-of-hairpins structure or six-helix bundle during or immediately following fusion62,63,64. Peptide sequences that correspond to either heptad repeat of the F protein of several paramyxoviruses, including HeV and NiV, have been shown to be potent inhibitors of fusion65,66,67,68,69. Indeed, HeV and NiV peptide-fusion inhibitors have been shown effective in blocking live virus infection in vitro, and might offer an exploitable therapeutic avenue69.

F proteins are type I membrane proteins, and biologically active F consists of two disulphide-linked subunits, F1 and F2, that are generated by the proteolytic cleavage of a precursor known as F0 (Ref. 70). Most paramyxovirus F precursors are cleaved by furin, a ubiquitous, calcium (Ca2+)-dependent protease that is localized in the trans-Golgi network of many eukaryotic cells71. The minimum sequence requirement for efficient processing by furin in vitro is RXXR72, a sequence that is highly conserved in paramyxovirus F proteins. Viruses with this sequence, such as MeV, cause systemic infections after initial infection of the respiratory tract. By contrast, a small number of paramyxovirus F proteins are proteolytically cleaved by extracellular trypsin-like proteases that recognize a single basic residue at the cleavage site. Cleavage in vivo is achieved by trypsin-like proteases such as tryptase Clara and miniplasmin that have limited distributions and, as a result, viruses like Sendai virus (SeV) remain localized in the respiratory tract73,74. In view of the fact that henipaviruses generate systemic infections, it was surprising to find that the henipavirus F-protein cleavage site does not contain multiple basic residues. The cleavage site in the HeV F protein contains a single basic residue, lysine, in the sequence VGDVKLAG75. In NiV, the lysine is replaced by arginine11. A role for furin in cleavage was excluded when it was shown that LoVo cells, human colon-carcinoma cells that lack furin, support the replication of HeV and permit cleavage of the NiV F protein75,76. The involvement of an enzyme with specificity for single basic residues was also ruled out by the fact that activation of the HeV and NiV F proteins in cell culture did not require exogenous trypsin, a requirement for SeV replication in vitro11,38,75.

Studies using a range of protease inhibitors in conjunction with conditions that block the movement of glycoproteins through the secretory pathway indicated that the HeV F protein is cleaved in the secretory vesicles that bud from the trans-Golgi network76. However, more recent studies on the NiV F protein have shown that cleavage is not mediated in vesicles during transport along the secretory pathway, but only after endocytosis of the protein77,78. The 45-amino-acid cytoplasmic tail of henipavirus F proteins contains an endocytosis signal, which directs the protein expressed on the cell surface into clathrin-coated vesicles for ongoing transport to other cellular compartments79. Removal of the signal not only abrogates endocytosis of the F protein but, remarkably, also prevents its cleavage into F1 and F2 (Refs 77, 78). This observation provides an explanation for the finding that, although removal of the endocytosis signal causes an increase in the concentration of cell-surface-expressed NiV F, it also caused a decrease in the size of NiV glycoprotein-induced syncytia79 because the F protein remained in an uncleaved form.

The subcellular location, specificity, sensitivity to pH and decreased requirement for Ca2+ indicate that the protease responsible for processing henipavirus F proteins differs from proteases that have been previously implicated in the maturation of viral proteins76,77,78. Furthermore, the failure to abrogate cleavage by replacing the arginine of the NiV F-protein cleavage site with a non-polar residue was particularly surprising, and contrasts with the absolute need for a basic residue or residues in the F proteins of all other paramyxoviruses80. Recently, it was shown that the lysosomal cysteine protease, cathepsin L, is responsible for the proteolytic cleavage of the HeV F protein81.

The henipavirus P gene products. The IFN system is one of the first lines of innate immune defence against infection in mammals, and is designed to limit the spread of microorganisms from the source of infection82,83,84,85. There are two types of IFN. Type I IFNs are produced in response to virus and bacterial infection and comprise a family of related IFN-α proteins and IFN-β. The type II IFN, IFN-γ, is synthesized only by certain cells of the immune system. Here, we focus solely on the antiviral type I IFN response.

The transcriptional activation of type I IFN-α/β genes is a complex, bi-phasic process. The first phase, IFN induction, occurs in cells soon after infection (Fig. 3) and leads to the synthesis of IFN-β and a subset of IFN-α proteins84,86. The IFN induction pathway can be activated by double-stranded (ds)RNA87 or by virus infection, in which viral components other than dsRNA might be responsible88. For the sake of simplicity, we will refer here to the process by which IFN is induced as the dsRNA-signalling pathway. In the second phase, IFN signalling (Fig. 4), the IFNs that are induced as a result of virus infection bind to type-I-IFN receptors on the surface of both infected and uninfected cells, and activate hundreds of IFN-inducible genes, some of which have antiviral activity82,83,84.

Figure 3: Interferon (IFN) induction or double-stranded (ds)RNA signalling.
figure 3

The innate immune system depends on the ability of cells to detect the presence of unique, pathogen-specific molecules. The molecule considered most likely to be seen as foreign by virus-infected cells and activate the innate immune system is dsRNA, generated as a result of virus infection87. Several cellular sensors detect the dsRNA signal and respond by activating pre-existing transcription factors such as IFN-regulatory factor 3 (IRF-3) and the general transcription factor nuclear factor (NF)-κB83,97,144,145,146. Activated IRF-3 and NF-κB are redistributed to the nucleus, where they cooperate with other transcriptional activators to induce transcription of the interferon (IFN)-α/β genes. In one pathway, the sensor is an intracellular RNA helicase encoded by the retinoic inducible gene-1 (RIG1) or the melanoma differentiation-associated gene 5 (MDA5)102,147. RIG-1 and MDA5 proteins are DExD/H-box RNA helicases that unwind dsRNA by virtue of their intrinsic ATPase activity. They also contain caspase-recruitment domains (CARD). The binding of dsRNA to the helicase has been hypothesized to result in the activation of the ATPase leading to conformational changes in CARD102. An activated CARD acts as an interface between signalling molecules, and the CARD of RIG-1 and MDA5 has been shown to interact with the CARD-like domain of a protein called IFN-β promoter stimulator 1 (IPS-1)148 to transmit a signal downstream, resulting in the phosphorylation of IRF-3 by the kinases TANK-binding kinase 1 (TBK-1) and IκB kinase ε (IKK-ε)149,150. Activated IRF-3 dimerizes and is translocated to the nucleus. Intracellular dsRNA signalling through RNA helicases also activates NF-κB. The inhibitor of NF-κB, I-κB, is phosphorylated by an activated member of the IKK complex, and I-κB is destroyed in proteasomes. NF-κB is therefore released and translocated to the nucleus. A second IFN-induction pathway, likely to be activated after the helicase-dependant pathway, uses Toll-like receptor 3 (TLR3), which probably detects dsRNA released from virus-infected cells151. Signalling through TLR3 is mediated by an intracellular adaptor protein called TRIF, which signals two protein-kinase complexes, TBK-1–IKK-ε and IKK-α–IKK-β, and leads to the activation of both IRF-3 and NF-κB. The sites where henipavirus and other paramyxoviruses are known to interfere with dsRNA signalling and the viral proteins responsible are indicated (see text). HeV, Hendra virus; hPIV2, human parainfluenza virus 2; MuV, mumps virus; NiV, Nipah virus; SeV, Sendai virus; SV5, simian parainfluenza virus 5.

Figure 4: Interferon (IFN) signalling.
figure 4

Type I IFN (IFN-β and a subset of IFN-α) induced as a result of virus infection activates IFN-inducible genes using the Jak-STAT pathway, a signalling pathway shared by many cytokines and growth factors that use members of the Janus tyrosine kinase family (TYK2 and JAK1) and a family of proteins called signal transducers and activators of transcription (STAT)83,84,90,97. Type I IFN binds to two heterologous receptor subunits (IFNAR1 and IFNAR2) on the cell surface, and their dimerization leads to the activation of TYK2 and JAK1 tyrosine kinases bound to IFNAR1 and IFNAR2, respectively. TYK2 and JAK1 cross-activate each other and phosphorylate STATs. STAT1 and STAT2 form heterodimers and translocate to the nucleus where, in association with a DNA-binding protein p48 in a complex called IFN-stimulated gene factor 3 (ISGF3), they activate the transcription of genes containing IFN-stimulated response elements within their promoters. Antiviral activity and the additional, profound effects that IFNs have on cellular physiology are mediated by hundreds of IFN-induced proteins. The best characterized antiviral IFN-inducible gene products include dsRNA-dependent protein kinase (PKR), 2′,5′-oligoadenylate synthetases (2-5A) and RNase L and Mx proteins, which inhibit virus replication in various ways82. Paramyxoviruses block IFN signalling by targeting specific components of the Jak-STAT pathway (see text).

Over millions of years of co-evolution, almost all viruses have evolved ways to evade the IFN-induced antiviral responses of their hosts83,85,89,90. These mechanisms include the inhibition of host-cell transcription and translation and the consequent failure to synthesize IFN, inhibition of dsRNA-signalling and IFN-signalling pathways, and antagonizing the IFN-induced antiviral effector proteins. The anti-IFN activities of paramyxoviruses are encoded by the viral NS1 and NS2 genes in respiratory syncytial virus, a member of the Pneumovirinae subfamily91,92 and by the P gene in the Paramyxovirinae. Products of the P gene inhibit both dsRNA signalling93,94,95,96 and IFN signalling97,98,99. The anti-IFN strategies used by paramyxoviruses vary, both between genera and among viruses within a specific genus. This is primarily because the P gene and its encoded proteins are both organized and expressed in a genus-specific manner (Box 3). Recent analyses have revealed the unique way in which the P-gene products of HeV and NiV act as virulence determinants by antagonizing the IFN response of infected cells.

Inhibition of dsRNA signalling. In the henipaviruses, dsRNA signalling is inhibited not only by the accessory V protein, as observed for rubulaviruses and respiroviruses93,94,95,100,101 but surprisingly also by the accessory W protein (Fig. 3). The V protein of HeV targets the helicase encoded by the melanoma differentiation-associated gene 5 ( MDA5 ), like its rubulavirus and respirovirus counterparts102. Whether the binding activity of the HeV V protein resides in the C-terminal, cysteine-rich terminal domain, like the respirovirus simian parainfluenza virus 5 (SV5), or the N-terminal domain, the site of henipavirus anti-IFN-signalling activity (see below), is not yet known. The NiV V protein blocked activation of IFN-regulatory factor 3 (IRF-3)-responsive promoters in response to intracellular dsRNA signalling, but did not inhibit activation in response to dsRNA signalling through Toll-like receptor 3 (TLR3)103.

Although the V and W proteins share an extensive N-terminal component, the influence of their C-terminal domains can be seen in the fact that NiV V and W proteins exert antagonistic effects in different cellular compartments and target different steps in the dsRNA-signalling pathways. The W protein contains a nuclear-localization signal in the unique C-terminal domain — an unexpected observation for an RNA virus that replicates solely in the cytoplasm — and inhibits activation of IRF-3-responsive promoters in response to both intracellular dsRNA signalling and signalling through TLR3 (Refs 103105).

Inhibition of IFN signalling. In uninfected cells, signal transducers and activators of transcription 1 (STAT1) is found in both the cytoplasm and the nucleus, whereas STAT2 seems to have a solely cytoplasmic distribution. However, following IFN treatment, both STAT proteins are located in the nucleus as components of a complex called IFN-stimulated gene factor 3 (ISGF3), which in turn activates transcription of IFN-responsive genes (Fig. 4). Paramyxoviruses inhibit IFN signalling by targeting the STAT proteins and preventing their translocation to the nucleus83,85,89,90,97,99. Not surprisingly, the henipaviruses also target STAT proteins, but they use a novel strategy that extends the methods used by paramyxoviruses to abrogate IFN signalling.

In rubulavirus-infected cells, targeted STAT proteins are directed to proteasomes following their polyubiquitylation by a virus-induced ubiquitin ligase, a multi-subunit complex that contains the V protein, cellular cofactors and the target STAT protein. Whereas SV5 targets STAT1 for proteasomal destruction, the anti-STAT activity of human parainfluenza virus 2 (hPIV2) is directed at STAT2 (Refs 101, 106, 107). In mumps virus (MuV)-infected cells, both STAT1 and STAT3 are eliminated, and STAT2 is left intact108,109. Although STAT1 is destroyed in SV5-infected cells, there is an absolute requirement for STAT2 in the ubiquitin-ligase complex110,111. In the case of hPIV2, where STAT2 is removed, STAT1 is required for ubiquitin-ligase activity110,112. Although the cysteine-rich C-terminal domain of the MuV and hPIV2 V proteins can antagonize IFN signalling when expressed alone, and recombinant hPIV2 and SV5 V proteins lacking the C-terminal domain fail to inhibit IFN signalling, the cysteine-rich domain is not solely responsible for the inhibitory activity94,113; regions within the N-terminal region of the V protein are also required114,115.

By contrast, respiroviruses such as SeV and hPIV3 and morbilliviruses such as MeV use alternative strategies to block IFN signalling (Fig. 4). SeV and hPIV3 inhibit tyrosine phosphorylation of STAT1, STAT2 or TYK2, a process that requires the accessory C protein95,100,101,116. Several anti-IFN-signalling strategies have been proposed for MeV, all of which leave STAT proteins intact but prevent their translocation to the nucleus117. In MeV-infected cells, tyrosine phosphorylation of STAT1 and STAT2 is inhibited by the V protein and STAT1 is retained in a complex with the IFN receptor118,119. The MeV C protein has also been implicated as an inhibitor of IFN signalling, but its precise role has yet to be confirmed118,119.

The henipaviruses, on the other hand, broaden the paramyxovirus STAT-targeting strategies by sequestering them in high-molecular-weight complexes105,120,121 (Fig. 4). Remarkably, this activity does not reside in the cysteine-rich C-terminal domain of the V protein, but in an area upstream that is shared by the V, W and P proteins. This provides henipaviruses with a multi-pronged anti-IFN response in which STAT proteins are sequestered in complexes, consequently abrogating their biological activity104,122. The W protein is the most efficient, and the P protein the least efficient antagonist105. The V and W proteins of NiV showed IFN-antagonistic activity even when the C gene was mutated and when the domain downstream of the editing site was removed122. STAT tyrosine phosphorylation is also inhibited by all three NiV proteins105,120. The V and P proteins interact with STAT1 in the cytoplasm, whereas the W protein, armed with its nuclear-localization signal in the W-specific C-terminal domain, co-localizes with STAT1 in the nucleus103,105.

The henipavirus P gene is larger than any of its paramyxovirus counterparts, and the encoded P, V and W proteins have an N-terminal extension of approximately 100–200 amino acids compared with cognate proteins in the subfamily12,20 (Box 3). The minimum domain required for IFN-antagonist activity and STAT1 binding maps to this region, between amino acids 50 and 150 (Ref. 105). The NiV C protein also displays modest inhibition of IFN signalling, providing further depth to the multifaceted henipavirus strategy to abrogate IFN signalling, although the mechanism and target are unknown122.

Not surprisingly, given their role as IFN antagonists, P-gene-encoded proteins of paramyxoviruses have been shown to be virulence and host-range determinants. The pathogenicity of SeV, MeV, hPIV3 and Newcastle disease virus (NDV) depends on virus inhibition of the IFN response, and mutations in proteins expressing anti-signalling activities alter the virus–host relationship in favour of the host123,124,125,126. It is worth noting that information on the function of the henipavirus W and C proteins was obtained with cells transiently overexpressing the proteins, and it is not known if they are expressed in virus-infected cells. Immunoblot analysis using monospecific antibodies raised to the unique C-terminal region of the V protein confirms its presence in infected cells10. Any correlation between the virulence of henipaviruses and the wide range of anti-IFN strategies that they have acquired awaits the application of reverse genetics and the study of mutant-virus pathogenesis in vivo. Nevertheless, it can be speculated that the pathogenicity of henipaviruses in most terrestrial species, compared with their subclinical replication in Chiroptera, might be related to the ability of the virus to circumvent the host IFN response, and it will be of interest to determine the anti-IFN activities of the HeV and NiV P-gene products in chiropteran cells.

Conclusions

The routes by which henipaviruses emerged from obscurity in flying foxes to important pathogens in humans and livestock remain obscure. This is due primarily to our ignorance of virus ecology in bats. Although mechanisms of transmission from bat to livestock and human have been postulated, there are few data on which we can rely to develop models of virus transmission or risk-management strategies for control of diseases caused by henipaviruses in the future. The high virulence of henipaviruses, the absence of therapeutic intervention strategies and vaccines and their classification as BSL4 pathogens have undoubtedly impeded the rate at which information has been generated on the biology and pathogenesis of HeV and NiV. However, recent investigations into the structure and function of henipavirus proteins expressed by cloned P, F and G genes in cultured cells have provided valuable information on the nature of the relationship between henipaviruses and the cells they infect, and suggest explanations for the observed interaction between henipaviruses and their terrestrial hosts.

Two of the biological criteria that differentiate HeV and NiV from other paramyxoviruses are their wide host range and the virulence that they display in their hosts. The susceptibility to henipavirus infection of a range of mammalian species and the similarity in patterns of susceptibility to infection by HeV and NiV are now known to be due, at least in part, to the fact that both viruses use ephrin B2 as a cell receptor, a remarkably conserved surface glycoprotein of ancient lineage and widespread distribution among vertebrates. The widespread cellular distribution of ephrin B2, especially in vascular endothelial cells, also provides an explanation for one of the most frequently observed outcomes of henipavirus infection, namely systemic involvement of endothelial cells. However, it remains to be seen if ephrin B2 will be the universal henipavirus receptor used by all species and all naturally occurring HeV and NiV strains, or variants such as those implicated in the outbreaks of disease in Bangladesh where human-to-human transmission has been documented.

The recent molecular investigations have also revealed several other factors that probably contribute to virulence. The cleavage of the F protein by cathepsin L, a ubiquitous endosomal protease with a cleavage site that is unique among viral glycoproteins, facilitates virus dissemination in vivo. The widespread distribution of the protease among organs might also be crucial in the transmission of infectious virus within and between species. It is tempting to speculate that the virulence of henipaviruses is due at least in part to the multifaceted P-gene strategy that these viruses have developed to inhibit the IFN system, a strategy that is novel amongst the paramyxoviruses, having both cytoplasmic and nuclear components.

Several important questions have been raised by recent and varied in vitro studies on HeV and NiV. An appreciation of the factors that contribute to the virulence of henipaviruses in terrestrial hosts versus the outcome of virus infection of flying foxes might provide crucial clues. Is ephrin B2 the receptor in bats? In light of the highly conserved nature of murine and human ephrin B2 proteins, especially in the ectodomain, there will probably be significant homology in the ephrin B2 homologue from flying foxes. Does the ubiquity of cathepsin L, the F protease cleavage enzyme, extend to a range of flying-fox cells and tissues? Do henipaviruses use the same range of P-gene products to inhibit IFN in bats? If henipaviruses inhibit dsRNA signalling and IFN signalling in chiropteran cells, their limited replication observed in flying foxes could be due to other factors such as the nature, density and location of the bat cell receptors or the ability of the viral C protein to inhibit viral RNA synthesis in bat cells more effectively than has been observed in mammalian cells127. The C protein encoded by the respirovirus P gene has been shown to downregulate viral genome amplification128,129 and transcription127,130.

Alternatively, given the well known propensity of bats to tolerate infection with a wide range of viruses in the absence of clinical symptoms, more generic methods could be operative, such as the inhibition of virus replication by lectins such as mannose-binding protein and galectin-1 (Refs 131, 132). Galectin-1, an endogenous lectin secreted by various cell types, has been shown to inhibit henipavirus envelope-glycoprotein-mediated cell fusion, probably by aberrantly oligomerizing NiV F and G glycoproteins132. In addition to this direct effect on virus replication, galectin-1 might also act indirectly to limit NiV replication because it enhances dendritic-cell production of proinflammatory cytokines such as interleukin 6 (IL-6), which has an essential role in the final differentiation of B cells into antibody-secreting cells132. The capacity of henipavirus P-gene products to abrogate the STAT-dependent IL-6-signalling pathway in terrestrial or chiropteran cells remains to be determined. The development of a range of anti-IFN strategies by henipaviruses might have evolved to maximize virus replication under conditions of restricted growth in bats. Finally, the ability to conduct these studies using recombinant molecular biological techniques on otherwise highly pathogenic and dangerous viruses has provided important information on the biology of HeV and NiV, which should prove exploitable in the near future and offer new or novel approaches in treating or preventing henipavirus infection.

Timeline | Emergence of henipaviruses Download Timeline PDF