The advent of reverse-genetics coupled with the in vitro
techniques of VLP production has made it possible to examine the cell biology of viruses in increasing detail. However, certain highly pathogenic viruses such as the henipaviruses are restricted to BSL-4 containment, making such studies impractical, and in the absence of a henipavirus reverse-genetics system there are considerable obstacles in place to dissecting out the details of an individual protein's role(s) in particle formation. To circumvent some of these obstacles, we have established a VLP system to allow us to investigate certain details of NiV assembly and release. Our first attempts made use of a recombinant poxvirus platform using MVA because of its high efficiency in gene delivery and expression, and for its reduced level of cytopathic effect as compared to vaccinia virus. Further, poxviruses have been successfully employed in reverse-genetics systems and virus budding assays with success [39
]. Using rMVAs to deliver NiV genes we observed membrane-associated release of M, F, and G. The ability of an M protein to be released from cells when expressed alone has also been observed for some other paramyxoviruses including SeV [35
], hPIV-1 [37
], and NDV [38
], as well as viruses in other related virus families including Ebola [42
] and VSV [39
]. In addition, Ciancanelli and Basler have recently independently reported that expression of NiV M leads to VLP formation [66
]. Release of F when expressed alone has been observed for SeV [35
], and release of HN has been noted for NDV [38
]. Ebola virus [44
], VSV [47
], and rabies virus [47
] each contain a single envelope glycoprotein that can also direct budding of vesicles, thus our observations here that NiV F and G can be independently released is not unprecedented. In contrast to these other viruses, individual viral proteins expressed alone from SV5 are not released from cells, and particle formation requires the expression of M, N, and at least one envelope glycoprotein in order for any significant VLP release to occur [50
Typically, viruses rescued using reverse-genetics techniques that employ vaccinia virus are subsequently purified from contaminating poxvirus and amplified by growth in cell culture. In such cases, the alteration of cellular metabolism by vaccinia virus is therefore of little concern because it is ultimately unimportant to the outcome. However, we were concerned that the effects of MVA infection might confound the interpretations of our VLP studies. Therefore, we employed a eukaryotic plasmid-based system as a means of gene expression in order to determine whether the results obtained using rMVAs faithfully reflected NiV biology.
When individual NiV proteins were produced in cells by transfection, M, F and G were each independently released into culture supernatants in a membrane-associated manner. In contrast, N was not detected in culture supernatants, as predicted, and these results were in agreement with those observations made using rMVAs. However, the percent of protein released was higher in the MVA system, which may reflect an effect of MVA on cellular metabolism or overall expression levels of the individual genes. When M was expressed in combination with various NiV proteins using rMVAs we observed little change in M release, irrespective of which other NiV proteins were present. Expression of M with N enabled greater release of membrane-associated N, which is consistent with the model that M and N interact, and that this interaction facilitates the incorporation of the genome into budding virions [31
]. The dynamics of protein release were somewhat different when a plasmid-based transfection and expression system was used. Under these conditions, we found that co-expression of M with N, or M with N, F and G lead to a reduction in M release. We obtained a similar result when F and M were co-expressed (data not shown). We also found that F and G co-expression lead to greater release of both proteins than when either was expressed independently. Notably, in either the MVA or plasmid-based expression systems we predominantly detected F0
in both cell lysates and supernatants (Fig. and Fig. ). In contrast, NiV virions appear to contain completely processed F [68
]. The reason for this difference is not known, but we and others have consistently observed greater levels of F0
when recombinant expression systems are employed which could bias its incorporation into particles [25
], or perhaps additional viral factors present during natural viral replication result in greater F processing or a biased incorporation of processed F.
The context-dependent variations in protein release and buoyant density suggest that the viral proteins interact either directly or indirectly to orchestrate the particle budding process. The differences observed in the protein release when a comparison is made between the transfected plasmid-based and rMVA systems are likely a result of the alterations of cellular metabolism brought about by MVA infection. Although VLPs can be produced using either system, our results suggest that the two methods have different applications. We believe the cell biology of NiV is more accurately reflected in the plasmid-based transfection system in the absence of poxvirus infection. Because of the efficiency of gene delivery and expression, and the greater percentage of viral protein released, the MVA system could be more useful when quantity of VLPs is of primary concern. VLPs have been reported to elicit a protective immune response against Ebola and Marburg viruses in animal models [70
], and might serve as a more efficient method of NiV VLP production for immunization studies or perhaps as a potential livestock vaccine, which we are exploring.
To ensure that NiV M was not released into culture supernatants by a non-specific means such as cell lysis, we assayed SV5 M in parallel as a control. As expected, we detected the release of M for NiV but not for SV5. However, SV5 M budding could readily be detected when it was co-expressed with its HN, F, and N genes (data not shown). Pulse-chase analysis of NiV M expression revealed detectable release from expressing cells by 2 to 4 h with a maximal release of at least 27% by 16 h. This percentage of release is, of course, greater than that observed during standard budding assays, which were not performed as a pulse-chase. The percentages of protein release reported in our budding assays here, as well as for SeV [35
] and SV5 [50
] have also been reported as non-pulse-chased systems. As a result, any nascent protein produced in the cell is included in the overall calculation. This method is useful as a relative measure between groups, but probably underestimates the true budding efficiency. Here, pulse-chase not only reveals the timing of release but also likely gives a more accurate picture of the kinetics of M production and its ultimate destination, that is its release from producing cells.
Ultrastructural studies of NiV have revealed pleiomorphic virions that range from 80–1900 nm [64
]. Here, when M was expressed alone or along with N, F and G, we observed VLPs containing the various proteins by immuno-gold labeling and electron microscopy. The VLPs ranged in size from approximately 100 to 700 nm, consistent with the range reported for actual virus. Using both the MVA and transfection systems we observed NiV proteins migrating as membrane vesicles at a density range of 1.11–1.19 g/ml. Authentic NiV migrated to a similar density range with peak detection at 1.15 g/ml. This range is consistent with the density of SeV and NDV VLPs as well as SeV virions [35
]. Using a plasmid transfection system we further evaluated the release of M alone, and with co-expression of N, F, and G. Although a greater percentage of M was detected in additional fractions when expressed alone, it was predominately concentrated in fractions 3 to 5 when co-expressed along with N, F, and G. Particles containing F and G migrated to denser fractions when additional NiV proteins were absent, but the position of both proteins in the gradient shifted to fractions 3 to 5 when M was present, suggesting that M interacts with F and G either directly or indirectly during assembly to facilitate the incorporation the envelope glycoproteins into the particles. The reason for the greater density of particles containing only F and G is unknown, but it is likely a reflection of altered lipid or protein incorporation relative to particle size. Taken together, the density and ultrastructural characteristics of the membrane-associated NiV proteins reported here are suggestive of authentic VLP formation.
Ciancanelli and Basler recently observed NiV VLPs when M was expressed alone or with envelope glycoproteins [66
], which is supportive of our data. Although they did not perform quantitative analysis in their study, M release was apparently unaffected by the presence of one or both of the envelope glycoproteins. As mentioned above, in our hands plasmid-based co-expression of M and F resulted in a distinct reduction in M release. This apparent discrepancy may be attributable to differences in the VLP purification techniques, especially in our inclusion of an additional particle flotation step. Another difference between the two studies is our inclusion of the N protein, which was found to also reduce M release and this was not examined in the Ciancanelli and Basler report. VLP formation has been evaluated for other paramyxoviruses including SeV, hPIV-1, SV5, and NDV [35
]. Of these, only SV5, SeV, and NDV have been addressed with any quantitative assessment. Two independent studies of SeV VLP formation reported that M and F can each be released independently and that when M and F are expressed together that the percentage of each protein released is greater [35
]. However the percent of protein released differed dramatically between the studies with Takimoto et al
], reporting approximately 50% of M released and Sugahara et al
], reporting 14.5% of M released. Using avian cells and a pulse-chase system, Pantua et al
. reported that NDV M is both necessary and sufficient for VLP release, with solo expression of NDV M resulting in 90% release efficiency [38
]. Whereas SV5 VLP release was most efficient (32% of M released) when N, M, F, and HN were expressed together [50
], there was a decrease in M release when the equivalent SeV or NDV proteins were co-expressed [36
], which is comparable to our results observed with NiV N, M, F and G.
Sugahara et al. [36
] have also shown that expression of the C protein with N, M, F, and HN leads to an increase in VLP release by 2- to 3-fold. This increase was subsequently shown to be due to the interaction of C with AIP1/Alix, a cellular protein involved in multivesicular body formation; however an interaction between AIP1/Alix and measles C was not detected, suggesting that this mechanism of budding is not applicable to all paramyxoviruses [71
]. Nevertheless, this raises the possibility that additional NiV proteins not evaluated here may increase VLP release, but further experiments will be required to determine whether this is the case. It is also important to recognize that because the various studies on several paramyxovirus VLPs each employed slightly differing methodologies in their analysis, the calculated percentage of released proteins among the various reports are not directly comparable. However, the qualitative differences seen between these viruses suggest that beneath the generalized model of paramyxovirus assembly and budding lie differences in the specific mechanisms employed, and these differences remain to be determined.