![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Hamide Aslan,1,
Michelle Ainouze,1
Mathilde Guerbois,1
T. Fabian Wild,2
Robin Buckland,1* and
Johannes P. M. Langedijk3
Molecular Basis of Paramyxovirus Entry,1 Immunobiology of Viral Infections, INSERM U404, Immunité et Vaccination, Centre d’Etudes de Recherche en Virologie et Immunologie, IFR 128 Biosciences Lyon-Gerland, Université Claude Bernard, Lyon, France,2 Pepscan Systems Inc., Lelystad, The Netherlands3
Received 27 January 2006/ Accepted 15 May 2006
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Prophylaxis against these viruses is not available at present, but we have recently shown that vaccinia recombinants expressing NiV-G (the attachment protein) and NiV-F (the fusion protein) induce an immune response in hamsters which protects against a lethal NiV challenge (8). We have also shown that sera containing polyclonal antibody from hamsters vaccinated with vaccinia recombinants expressing the NiV glycoproteins can be used to passively protect naïve hamsters against a lethal NiV infection (8). We have recently extended these studies by producing banks of monoclonal antibodies (MAbs) specific for both NiV-G and NiV-F. We have shown that these MAbs can produce a sterilizing immunity in hamsters against a lethal NiV challenge (9), which suggests that there is a potential for using passive immunotherapy for populations that have been exposed to NiV infection.
With NiV outbreaks occurring only sporadically it will be hard to convince populations to be vaccinated, but if the local community could receive immunotherapy following an outbreak, this would potentially contain the virus by forming a barrier. The murine MAbs we have produced will have to be "humanized," without any loss of efficacy, in order to be used in this way. Novel antivirals that block entry of the virus could also fulfill such a barrier function. Their development will require a better understanding of the molecular basis of the entry of NiV into the host cell.
As a preliminary to the localization of receptor-binding sites on NiV-G, we have undertaken the identification of residues whose mutation has the effect of diminishing the protein's fusion promotion function. Current models of paramyxovirus entry (14) postulate that the attachment and fusion proteins are physically associated in the viral envelope (and the infected cell's plasma membrane) and that receptor binding triggers a series of conformational rearrangements, particularly in the fusion protein, that lead to viral fusion. As cell-cell fusion (syncytia formation) is dependent upon receptor binding, it can be used as an assay to identify residues in the attachment protein's globular head, which is potentially responsible for interaction with the cellular receptor.
We have used two strategies to identify such residues. First, NiV and HeV have a similar cellular tropism (3), and this suggested that they share a common receptor. We reasoned that in this case, residues functioning in receptor attachment could be expected to be conserved between the NiV-G and HeV-G proteins. Our initial strategy was to mutate charged residues (which can be expected to be at the surface of the protein) conserved between the NiV-G and HeV-G globular heads and to test the mutated NiV-G proteins in a fusion assay. This approach has been justified by the recent identification of the protein ephrinB2, which is expressed on endothelial cells and neurons, as a cellular receptor not only for NiV (21, 2) but also for HeV (2). Our second strategy was to use two anti-NiV-G MAbs, shown in protection studies to neutralize NiV both in vitro and in vivo (9), to generate NiV variants that escaped their neutralization. Neutralizing antibodies specific for the NiV attachment protein presumably have their effect by blocking the interaction with the receptor. Sequencing of viral variants that escape neutralization can identify amino acids present in an epitope specific for a particular MAb. Although such neutralizing antibodies can have an effect at a distance, their epitopes usually coincide, or at least overlap, with receptor-binding sites. Residue changes found in the escape mutants were reconstituted in vitro and tested in a fusion assay. Once we had identified residues by the two strategies, we needed to localize them on the NiV-G globular head. For this purpose, a new structural model for the NiV-G protein (J. P. M. Langedijk; www.pepscan.nl/downloads/nipahG.pdb) was devised.
A three-dimensional (3D) model of NiV-G was built based on sequence homology with several paramyxovirus HN proteins for which the crystal structure is known and which could be used as templates for model building. The G protein folds as a typical ß-propeller and has the same large loops at B3L23 and B5L01, typical of the attachment proteins of the paramyxovirineae. However, as for the morbillivirus H protein, the NiV-G protein has lost the typical active site residues in the center of the ß-propeller.
Our results suggest that NiV-G residues whose mutation results in a loss of the protein's fusion promotion function form a contiguous site that is implicated in attachment to ephrinB2. This site on the top surface of the barrel-like globular head is very reminiscent of the SLAM site on another paramyxovirus attachment protein, the measles virus hemagglutinin (MV-H).
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Antibodies. The production of anti-NiV-G monoclonal antibodies (MAbs) and the measurement of their neutralizing activities has been described previously (9). The secondary antibody used for flow cytometry was a fluorescein isothiocyanate-conjugated goat anti-mouse antibody (Dako).
Generation and analysis of anti-NiV-G MAb 3B10 escape mutants. Vero cells in 6-well plates were infected with NiV at a multiplicity of infection of 2 x 10–4 for 1 h. The virus inoculum was then removed, and the cells were washed twice with medium. Two milliliters of medium supplemented with anti-NiV.G MAb (3 µl of ascites fluid) was added to the wells. The medium was changed every 2 days, and then the virus was plaque purified on Vero cells in the presence of MAb 3B10. Total RNA was extracted from Vero cells using the RNeasy Mini kit (Qiagen Inc., Valencia, CA) following the manufacturer's instructions. The extracts were resuspended in 50 µl of RNase-free water, aliquoted, and stored at –80°C before use. The full-length NiV-G gene was amplified for DNA sequence analysis. The reverse transcriptase PCR was conducted in one step using High Fidelity PCR enzyme blend (Roche Applied Sciences, Germany) following the manufacturer's instructions, with the forward primer 5'-CGCGGATCCAGTCATAACAATTCAAG-3' and the reverse primer 5'-CGCGGATCCGAGGTTGATTTTTATG-3'. The reverse transcriptase PCR products were then sequenced using a cycle sequencing reaction using the ABI Prism Big Dye terminator cycle sequencing ready reaction kit (Applied Biosystems) on an ABI Prism 3000 automatic sequencer (Perkin-Elmer).
Production and expression of NiV-G mutants. Site-directed mutagenesis (QuickChange; Stratagene) was performed on the gene encoding NiV-G cloned into the phCMV plasmid (25). The mutated plasmids were amplified and purified as described previously (19). Typically, 5 x 105 Vero cells in OPTIMEM medium were transfected with 0.2 µg of mutated phCMV-NiV-G mixed with Lipofectamine (Invitrogen).
NiV-G expression analysis by flow cytometry. For the analysis of the NiV-G mutants, confluent Vero cells in 6-well plates were transfected with phCMV.NiV-G as described above. Twenty-four hours later, the cells were harvested and stained for 30 min with either an anti-NiV-G MAb or an anti-NiV-G polyclonal, washed, and then stained for 30 min with fluorescein isothiocyanate-conjugated goat anti-mouse antibody (Dako). The stained cells were analyzed by flow cytometry using a FACScan machine (Becton Dickinson). Cells positively stained with propidium iodide were excluded from the analysis.
Analysis of fusion promotion function of NiV-G mutants. On day 0 (d0), Vero cells (5 x 105 cells) in 6-well plates were transfected with 0.2 µg phCMV.NiV-G (nonmutated or point-mutated NiV-G) together with phCMV.NiV-F (2 µg). The cells were observed with a light microscope (Zeiss Axiovert 200M) for the presence of syncytia 20 h later (d1) and during the following three days (d2, d3, and d4). On d3, four fields from the well of each NiV-G point mutant were photographed (using a no. 10 objective), and the total number of nuclei present in syncytia were counted. Fusion promotion efficiency was expressed with the formula (number of cells in syncytia/total number of cells in the well) x 100, adjusted according to the value obtained for the positive control (unmutated NiV-G). Each mutant was analyzed at least three times. A similar procedure was followed for Chinese hamster ovary (CHO) cells transiently expressing human ephrinB2. A total of 5 x 105 CHO cells in 6-well plates were cotransfected with 1 µg human ephrinB2 plasmid (pCMV6; OriGene) and 0.2 µg phCMV.NiV-G (nonmutated or point mutated), together with 2 µg phCMV.NiV-F. The cells were observed for the presence of syncytia and photographed 48 (d2) later.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The fusion promotion capacity of NiV-G is largely abrogated by the mutation E533Q. We used site-directed mutagenesis to introduce the mutation E533Q into the gene encoding NiV-G expressed from the eukaryotic expression vector phCMV (25). As such mutations can have an effect on transport to the cell membrane, we used flow cytometry to monitor the expression of the E533Q mutant at the cell surface. We found that the E533Q mutant was expressed at levels similar to those of unmutated NiV-G (Fig. 1A). Coexpression of the E533Q mutant with NiV-F (also expressed from phCMV) in Vero cells allowed us to monitor the effect of this mutation on the fusion promotion capacity of NiV-G. We found that the in vitro reconstitution of the E533Q mutation present in the MAb 3B10 escape mutant resulted in a pronounced reduction in the level of viral fusion compared to that of unmutated NiV-G (Fig. 1B).
|
Mutations W504A, Q530A, T531A, A532K, and N557A also reduce the fusion promotion capacity of NiV-G. E533 was localized on the structural model for NiV-G, and residues predicted to be its near neighbors in 3D and conserved with HeV-G (W504, S528, N529, Q530, T531, A532, N534, Y547, N557, and Y581) were mutated and tested as described above. FACScan analysis revealed that the majority of the mutants had a cell surface expression comparable to that of the positive control (unmutated NiV-G), but three mutants, N534A, Y547A, and Y581A, exhibited a much reduced surface expression and were eliminated from the study. Coexpression of the other mutants with NiV-F in Vero cells allowed us to monitor the effect of their mutations on the fusion promotion capacity of NiV-G. The level of viral fusion for each point mutation was estimated as described in Materials and Methods. We found that mutations W504A, Q530A, T531A, A532K, and N557A also reduce the fusion promotion capacity of NiV-G by more than 50%. All had a cell surface expression comparable to that of the positive control (Fig. 1A), and a new ranking, in descending order in terms of the effect on fusion promotion, was established: E533 > W504 > A532 > N557 > E505 > T531 > Q530 (Fig. 3).
|
|
|
The structural model of NiV-G is based on sequence homology with Newcastle disease virus hemagglutinin-neuraminidase (NDV-HN), human parainfluenza type 3 HN, and parainfluenza virus type 5 HN, for which the crystal structures have been solved (6, 16, 26). Sequence homology and the number of templates used was higher than in the case of the homology modeling of MV-H (19), which resulted in a relatively accurate 3D model. Just as predicted for the morbillivirus H proteins, the NiV-G protein folds as a so-called ß-propeller similar to the neuraminidases but has "lost" all seven neuraminidase active site residues. Even the arginine residue on the B2L23 loop (R533 in MV-H), which is the only active site residue conserved in the morbillivirus hemagglutinin, is lost in NiV-G (and in both the Menangle and Mossman G proteins). Although the active site is lost, the top of the ß-propeller still has a shallow cavity, equivalent to the sialic acid-binding pocket of the paramyxovirus HN.
The residues that are completely conserved are exclusively important for folding. Five of the six disulfides shared with avula- and rubulaviruses are conserved. Two other disulfides are shared with rubula-, avula-, and morbilliviruses. The most conserved continuous stretch of amino acids is the sequence R487P488G489. Although this sequence is not part of the sheets of the ß-propeller, it has a structural role. In NDV, R449 (homologous to R487 in NiV-G) is a buried charged residue that makes hydrogen bonds with the carbonyl atoms of G417, S419, and C455. The hydrogen bond network around the conserved R449 (and R487 in NiV) is thus involved in structural stabilization of the very large ß5L01 loop, together with the conserved cystine bridge between C455 and C461 (C493 and C503 in NiV-G).
The potential glycosylation sites of NiV-G are all located on loops (N306 on ß2L23, N378 on ß3L23, N417 on ß3L34, and N529 on ß5L23) or on an outer strand on the side of the globular head (N481 on ß4S4) and are all located on the surface of the globular head, even when a possible tetramerization site is considered according to the homologous parainfluenza virus type 5 HN tetrameric structure.
The most variable regions in the globular head of NiV-G can best be analyzed when the sequence is compared with the other member of the Henipavirus genus, the closely related HeV-G sequence. Most variation within HeV-G is on the outside in loop regions, especially at the edges of sheets 2 and 3. The region at the top of the globular head, lining the shallow depression, is remarkably conserved, suggesting a functional role for this region. Also, the potential receptor-binding site described in this paper is completely conserved in hepinaviruses, which in itself suggests shared receptor usage.
We have interpreted the results generated by the mutagenesis expression studies of NiV-G in terms of the structural model (Fig. 4). Two hotspots are identified that are potentially involved in receptor binding: residues 504 and 505 on loop B5L01, the 530 to 533 group of residues on B5L23, and 557 on ß6S1. These residues are within a radius of 8 to 10 Å at the top and rim of the globular head, and part of the potential receptor binding site, implicated in ephrinB2 binding and formed by residues 530 to 533, is in the equivalent location of the SLAM-binding site in MV-H (19).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
By reference to the NiV-G structural model, our results suggest that the seven residues W504, E505, N557, Q530, T531, A532, and E533 form a contiguous site on the top surface between the central shallow depression and the rim of the globular head. Recently, it was reported that ephrinB2 acts as a receptor for both NiV and HeV (2, 21). In order to determine whether the site is implicated in ephrinB2 binding, we transfected CHO cells (which are nonpermissive for NiV) with a plasmid expressing human ephrinB2 and tested the seven NiV-G mutants for their fusion promotion capacity in these cells. As with the results obtained with Vero cells, all seven mutations reduced fusion by more than 50%, and once again the E533Q mutation was responsible for the highest reduction. These results strongly suggest that the site formed by the seven residues is implicated in binding ephrinB2.
The topology of this site, on the top surface of the globular head, is strikingly reminiscent of that for the binding site for SLAM (19) on another paramyxovirus attachment glycoprotein, the hemagglutinin of measles virus (MV-H). The structural models for MV-H and NiV-G are very similar. MV has two receptors, SLAM (CD150), whose expression is restricted to lymphoid cells, and CD46, whose expression is quasi-ubiquitous. Our previous results (19) suggest that the binding sites for SLAM and CD46 overlap in such a way that residues forming the binding site for CD46 (18) are a continuation of the SLAM site, but they localize on the side of the barrel-like globular head below the rim rather than on the top surface. In the present study, none of the residues whose mutations affect fusion promotion localize on the side of the globular head.
The contention that the contiguous site on MV-H does indeed coincide with the SLAM- and CD46-binding sites was recently corroborated by a study in which retargeted oncolytic MV was engineered to be blind to both receptors (20). Mutation of a single residue (R533) in the proposed SLAM-binding site was sufficient to render the MV recombinant SLAM blind. For CD46 ablation, the authors mutated three residues: Y481, S548, and F549. The latter two residues, previously identified as a second CD46-binding site on MV-H (18), form the major part of the CD46 site contiguous with that of SLAM (19). The mutation N481Y is required to change the specificity of a wild-type MV from SLAM to CD46 (1, 17), but there is no evidence that Y481 interacts directly with CD46. We have speculated that mutation of this residue induces a conformational change in the globular head that favors CD46 interaction via S548 and F549 (19). If NiV, like MV, uses two cellular receptors, we would speculate that the residues involved in the second binding site are likely to be localized as a continuation of the site described here, over the top rim onto the side of the globular head, with utilization of one site or the other being conformation dependent.
A potential glycosylation site (N529/Q530/T531) overlaps partly with the proposed binding site. Mutation of N529 or T531 could therefore be expected to abrogate glycosylation at this region of the globular head, with potential deleterious consequences regarding NiV-G surface expression and hence fusion promotion. Interestingly, although the fusion promotion function is adversely affected by the T531A mutation, there is no reduction in the surface expression of this mutant (Fig. 1A). The N529A mutant, which had no effect on fusion promotion, also expressed at levels equivalent to that of unmutated NiV-G (data not shown). These results suggest that this particular glycosylation site is not utilized, possibly because a sugar moiety at this location could hinder interaction with the cellular receptor.
A recent publication used escape mutants to identify the epitopes of anti-HeV-G neutralizing MAbs (24), but the authors did not report the effect of the mutations found in the variants on fusion promotion. None of the residues that we have identified as being important for NiV-G fusion promotion are present in these epitopes, but interestingly two of their escape mutants map to residues 385 and 386 of HeV-G. Although the escape mutant that we generated with anti-NiV-G MAb 3B10 mapped to residue 533, escape mutants from other neutralizing MAbs mapped to residues 391 to 393, which are close to 385 and 386. As previous researchers mentioned, this region, conserved between the morbilliviruses and henipaviruses, which in MV is called the "noose" epitope (28) by virtue of the cystine loop it contains, has been proposed as a potential host receptor-binding region (5, 15, 13, 23). However, our in vitro reconstitutions of the 391 to 393 mutants showed that changes in this region have no effect on fusion promotion. An examination of the NiV structural model provides an explanation for these results: antibodies binding the "noose" region would overlap with the contiguous site we have identified. Furthermore, an examination of the MV-H structural model (19) suggests that an antibody binding to the "noose" region of MV-H would obstruct access to both the CD46- and SLAM-binding sites. Following an examination of the literature, we have accumulated evidence that mutations in the noose region of morbillivirus and henipavirus attachment proteins could play a role in the persistence and virulence of these viruses (10).
In another recent study (27), 11 alanine-scanning mutants selected to represent different regions of the HeV-G protein were used in an attempt to localize the epitopes of human Fab specific for HeV-G. It was found that mutation of residues (P185, Q191, and K192) near the stalk/head junction of HeV-G decreased binding to two Fabs by more than 50% in Western blots. Considering their localization, it is perhaps unlikely that these residues form part of a receptor-binding site. Alternatively, they could participate in the interaction with HeV-F or induce conformational changes in the globular head upon mutation deleterious for receptor interaction.
In summary, it is remarkable that the NiV-G residue E533, which is located at the tip of the B5L23 loop, is topologically exactly analogous with R533 of MV-H, a residue that is central in the binding site for the SLAM receptor. It is interesting that two paramyxovirus attachment proteins that are distinct from an evolutionary point of view but with similar structures appear to use exactly the same coordinate on the surface for receptor binding.
|
ACKNOWLEDGMENTS |
|---|
V.G. was supported by a scholarship from the Direction Générale de l'Armée and the Fondation pour la Recherche Médicale. H.A. is the recipient of a scholarship from the French Government (BGF). R.B. is a CNRS scientist.
|
FOOTNOTES |
|---|
These authors contributed equally to this study.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| J. Bacteriol. | Mol. Cell. Biol. | Microbiol. Mol. Biol. Rev. |
|---|
| Clin. Vaccine Immunol. | ALL ASM JOURNALS |
|---|
) and strands (S) in each ß-sheet. Residues within 1 distance unit from the consensus are boxed. Gaps are indicated by dashes. Charged NiV-G residues conserved with HeV-G are in boldface type. MoV, Mossman virus, accession no. Q91MJ8; TPMV, Tupaia virus, accession no. Q9JFN4; NiV, Nipah virus, accession no. Q91H62; MeV, Menangle virus, accession no. Q6WGMO; NDV, Newcastle disease virus, accession no. Q71SA4.
