Background
Hendra virus (HeV), along with the closely-related Nipah virus (NiV) and Cedar virus (CedPV), form the
Henipavirus genus in the family
Paramyxoviridae. Bats are the reservoir host for the henipaviruses and have been the source of a number of spill-over events. HeV outbreaks have so far been restricted to northern Australia [
1], but NiV outbreaks have occurred in Bangladesh, Malaysia/Singapore and India [
2]. In these spill-over events domestic animals and humans are infected with significant mortality rates which, for NiV in particular, range from 40-100% [
3].
Paramyxoviruses replicate within the host cell cytoplasm and virus particles bud from the cell surface, incorporating a portion of the host cell membrane as the viral envelope. However, the precise mechanisms involved in viral protein intra-cellular trafficking and infectious particle assembly are not clear for many viruses [
4], including paramyxoviruses [
5].
The HeV genome codes for 6 major proteins. The nucleoprotein (N), phosphoprotein (P) and polymerase (L) proteins interact with the newly formed RNA genome to form a ribonucleoprotein complex (RNP). In addition to the P protein, the P gene also encodes several smaller proteins [
6]. Sub-cellular localisation of these V, W and C proteins has been demonstrated in infected cells with C and V proteins present throughout the cytoplasm and W protein in the nucleus (but not the nucleolus). All three proteins were also detected in purified virions [
7]. HeV V, W and C proteins are present in relatively low abundance and their functions remain unclear, although they have been shown to inhibit transcription and replication [
8].
Most work on henipavirus proteins in infected cells has focussed on the F and G glycoproteins found on the outside of the virions as they are key to the attachment and internalisation processes of the virus. The HeV G glycoprotein binds to its cell surface receptors ephrin B2 and ephrin B3 [
9]-[
11] which are most highly expressed on neurons, arterial endothelial and smooth muscle cells [
12]-[
14]. The F (fusion) glycoprotein undergoes a conformational change when G binds to a host cell and drives the fusion of the virion with the host cell membrane [
15] to initiate the process of virus replication. The F proteins of both NiV and HeV have been shown to be synthesised in an inactive form and need activation by cathepsins which may take place within the endosomal compartment [
16],[
17].
The HeV matrix protein (M), by analogy with other paramyxoviruses, is crucial for virion morphogenesis and along with the RNP constitutes the virion contents. The precise role that M protein plays in viral morphogenesis is unclear, although expression of NiV M protein in tissue culture cells leads to the formation of virus-like particles [
18] and in
E. coli the formation of round particles sized between 20 and 50 nm [
19]. Patch et al. [
20] identified a short sequence of NiV M protein that was critical for budding of viral–like particles. NiV M protein, along with the M protein of a small number of other paramyxoviruses [
21]-[
24] is found within the nucleus of infected cells, but the precise reason(s) for this are not clear. In their studies, [
25] Wang et al. observed NiV M protein first in the nucleus and then later in infection, within the cytoplasm and at the plasma membrane. Furthermore, this transit through the nucleus appeared to be essential for correct viral budding. These authors also demonstrated that ubiquitination of NiV M protein takes place within the nucleus, and that this appears to be important for virus budding. In cells infected with respiratory syncytial virus (RSV), there was a reduction in host cell transcription raising the possibility that this may be a function of nuclear localised M protein [
21].
An understanding of virion structure is a key stage in the process of unravelling henipavirus assembly. We used confocal and transmission electron microscopy (TEM) to compare HeV protein and virion production in different cell lines. In addition, two systems of super-resolution (SR) imaging were used to determine if sub-virion resolution of paramyxovirus proteins was feasible. These observations led to important conclusions regarding the morphology of HeV virions and the suitability of various cell lines as in vitro models of HeV replication.
Discussion
The process of paramyxovirus replication and virus assembly has remained ill-defined despite the importance of paramyxoviruses such as mumps and measles in human health. This has been due in part to inherent difficulties in carrying out structural analysis of pleomorphic paramyxovirus particles with proteins containing disordered regions that make them difficult to analyse. Thus, the first three dimensional structure of a paramyxovirus (Sendai virus) was not obtained until 2009 when the structure was determined using cryo-electron tomography [
29]. In this structure, the nucleocapsid was free inside the virion and the matrix protein was bound to the inside of the particle. Two years later the 3-D structure of measles was obtained and showed that the matrix protein formed helices coating the ribonucleocapsid rather than the inner membrane of the virus [
30]. How widespread either of these configurations is among the paramyxoviruses is not known.
Here we used a number of different techniques to help elucidate some aspects of the replication of a relatively new group of paramyxoviruses, the henipaviruses. Infection of a range of cell lines as well as primary endothelial cells gave surprisingly different results. At early stages (8 hpi) of HeV infection of Vero cells there was no sign of virus formation at the cell membrane but at 18 and 24 hpi the pattern of virus M and G protein labelling was difficult to interpret. In contrast, in A549, MDBK and HeLa cells infected with HeV and fixed at 18 hpi the majority of HeV M protein was in the nucleus with some labelling in the cytoplasm. Hendra virus G protein was present in the ER and at high levels at the cell membrane but with little co-localisation with the M protein. In these latter three cell lines, there appeared to be reduced trafficking of HeV M protein from the nucleus to the cell membrane. Whilst the defective interferon response in Vero cells [
31] may complicate the interpretation of these experiments, the reasons for the apparent accumulation of the HeV M protein in the nucleus of A549, MDBK and HeLa cells remain to be resolved. Clearly, immortalised cell lines are able to produce infectious virus particles, but the patterns of HeV M and G proteins seen in these cell lines were not observed in HeV-infected endothelial cell cultures at any of the time points from 8 – 48 hrs (data not shown). What role this may play in the titres of infectious virus produced in the different cell lines remains unclear.
As endothelial cells are a major target for HeV and NiV
in vivo[
26] the results obtained with infection of these cells were seen as of particular interest in the context of the
in vivo virus assembly process. Previous reports involving NiV have shown some differences in the infectivity of endothelial cells from different sites [
32] with aortic endothelial cells refractory to infection unless modified to express the ephrin B2 receptor. This is in contrast to the results presented here where two species of primary aortic endothelial cells were readily infected with HeV. Similar results have been obtained with NiV (data not shown).
In a recent study of henipavirus replication in a number of cell lines, Aljofan et al. [
27] reported that virus protein levels and viral titres varied considerably between cell types. The highest levels of virus production were obtained with Vero cells, HeLa cells being intermediate and A549 producing a lower viral titre. The authors did not address the question of the subcellular distribution of viral proteins. Our data broadly support their results and also indicate possible reasons for the differences in virus titre production they observed from different cell lines.
NiV M protein has been reported to transit the nucleus [
25], but this is the first report of this occurring in HeV infected cells. The function of nuclear trafficking of henipavirus M protein is unclear but a recent report by Wang et al. [
25] showed that for NiV, the M protein had both nuclear localisation and export motifs and was ubiquitinated in the nucleus. Preventing this process had dramatic negative effects on M protein trafficking within the cytoplasm and viral budding. Nuclear localisation of M protein has been reported for other paramyxoviruses including Sendai virus [
23], Newcastle disease virus [
24] and RSV [
22]. RSV M protein, whilst present in the nucleus, is excluded from the nucleolus and its export from the nucleus relies on the crm-1 exporter receptor [
21]. However the authors reported that the M protein was then found in the cytoplasm in contrast to HeV infected cells, where almost all detectable M protein appears in the form of virions at the cell membrane. The function of the process remains obscure but a role in inhibiting host cell transcription has been proposed for the M protein of RSV [
21]. An unexpected observation in this study was a tendency for the cells with the highest number of virus particles to have the lowest level of HeV M protein in the nucleus. One possible explanation is that rather than a continuous synthesis and trafficking of HeV M through the nucleus during the infection process, there is a wave of M synthesis; the protein moves through the nucleus but is not continuously replaced. The mechanism and function of such a process remains highly speculative.
Many questions remain concerning the arrangement of the virus proteins within the viral particle and the structure of the virus assembly site. The arrangement of the M protein has recently been studied for measles virus and it has been suggested that it is closely associated with the RNP rather than the inner face of the viral envelope [
30]. There are very few reports of the use of SR microscopy to image virus structure, although the value of SR imaging has recently been demonstrated for vaccinia virus [
33]. In this study, we investigated the value of two very different SR approaches. Localization microscopy using GSDIM/dSTORM relies on the detection of single fluorophores, whose position can be determined with nanometre precision. GSDIM/dSTORM renders the majority of fluorophores transiently non-fluorescent by performing ground state depletion and reversibly transferring the molecules to an “off-state” (Triplet- and Dark-states). Due to the reversible nature of this process, nearly all fluorophores can be imaged while being transiently in their fluorescent ‘on-state’ over the acquisition time. Based on the position information of the detected fluorophores, a single super-resolution image is calculated [
34],[
35]. In contrast, 3D-Structured Illumination Microscopy (3D-SIM) works by illuminating the sample with a known pattern (in this case a grid pattern). The grid pattern produces high frequency information in the form of moiré fringes that can be mathematically extrapolated to result in a two fold increase in resolution in x, y and z directions [
36]. Despite significant differences between the two SR systems, both gave very similar images of HeV virions and demonstrated for the first time sub-particle resolution of paramyxovirus proteins and the round morphology of HeV virions. Imaging of virions co-labelled for the presence of HeV M, N and G proteins provided a first indication that the M protein appeared be associated with the RNP rather than the virion envelope.
Overall these results indicated that HeV should be described as a predominantly round virus with a diameter of 280 – 310 nm depending on imaging method. As the samples remained hydrated throughout the labelling and imaging process of SR microscopy, a virus diameter of approximately 300 nm should offer a more accurate estimate of the virus in vivo dimension.
Despite clear serological evidence of HeV infection, actual isolation of the virus from bats has not proved simple [
37]. An additional objective of this work was to identify any differences in HeV M and G protein localisation between bat and non-bat cells. Comparison of the expression pattern of M and G proteins in HeV-infected endothelial cells with that seen in two bat-derived cell lines indicated that there was no observable difference between the two cell types in response to henipavirus infection, but a pronounced difference between these two cell types and other continuous cell lines.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
PM, DG, JP, LW designed the study and with PH provided the imaging. RK, CW undertook the experiments and the cell culture, ML, PMcM and LT undertook SR imaging and image analysis, JW provided key reagents and helped draft the manuscript. All authors read and approved the final manuscript.