Results
- Quantitative analysis of Nipah virus proteins released as virus-like particles reveals central role for the matrix protein

MVA expression of NiV proteins

We first sought to develop a NiV VLP expression system using the MVA poxvirus as a means of gene delivery and expression. Here, the NiV N, M, F and G ORFs were sub-cloned into the pMC03-based vector [54] in which the vaccinia virus early-late promoter was replaced with the bacteriophage T7 promoter. These constructs were then used to create the various rMVAs containing the individual NiV genes under the control of the T7 promoter. To test for NiV protein expression, Vero cells were infected with individual rMVAs expressing N, M, F, or G, along with MVAGKT7 encoding the T7 RNA polymerase. Infected cells were metabolically labeled overnight. Cell lysates were prepared and the NiV proteins were immunoprecipitated with NiV-specific polyclonal rabbit serum or rabbit anti-F polyclonal serum (Fig. 1). Immunoprecipitated proteins revealed bands corresponding to NiV N and M (Fig. 1A, lanes 2 and 3) which migrated at the expected apparent molecular weights of ~58 kDa (N) and ~42 kDa (M) respectively [55,56]. The NiV F₀(~61 kDa), F₁(~49 kDa), and G (~74 kDa) shown in Fig 1A (lanes 5 and 7), were found to be consistent with patterns reported previously [25].

In subsequent experiments to evaluate whether expression of NiV proteins can lead to VLP formation, cells were infected with rMVAs and metabolically labeled for 44 h followed by collection of both the cells and culture supernatant. Vesicles in the culture supernatant were pelleted by centrifugation through a 10% sucrose cushion and then floated by centrifugation in a discontinuous sucrose gradient as described in the Methods. Membrane-associated proteins were collected from the top of the gradient and detected by immunoprecipitation and separation by SDS-PAGE followed by autoradiography. While little membrane-associated N was detected in culture supernatants, individual expression of M, F, and G resulted in detectable membrane-associated protein release (Fig. 1B). To characterize the protein release as genuine VLPs, the culture supernatant of metabolically labeled cells expressing N, M, F, and G together was layered onto a 5–45% sucrose gradient, which was then centrifuged to allow membrane-associated proteins to migrate to their buoyant density. Fractions of the gradient were then removed with a portion of each fraction set aside for sucrose density determination. The proteins in the remaining portion of the fractions were then detected by immunoprecipitation followed by SDS-PAGE and autoradiography analysis. NiV proteins were found predominantly in fractions 2 through 4, which corresponded to a density range of 1.12–1.19 g/ml (Fig. 1C). This density range was consistent with the density reported for SeV and NDV VLPs [35,36,38], as well as authentic NiV (see below). To examine the contribution of each viral protein to overall protein release, NiV proteins were expressed in different combinations and their release was quantified. A release of N was not detected when expressed alone, however co-expression of M with N resulted in release of both proteins into the supernatant (Fig. 2A). Quantified expression of M with other viral proteins resulted in a reduction in the mean M released (~12%) compared to its expression alone; however this reduction did not reach statistical significance among repeated experiments (Fig. 2B). We also confirmed that expression of multiple proteins simultaneously did not reduce the overall N, M, F, or G expression levels (data not shown). Quantitation of envelope glycoprotein release revealed that approximately 11% of F and 5% of G was released when expressed alone (Fig. 2C).

NiV protein release and VLP production is not dependent on MVA infection

Although less cytopathic than its wild-type parent, MVA retains the ability to block host protein synthesis and otherwise interfere with normal cellular metabolism [57]. In order to determine whether the protein release we observed was an accurate reflection of NiV biology rather than a potential artifact resulting from poxvirus infection, we employed a transfection plasmid-based expression system using the pCAGGS eukaryotic expression vector, which has also been used in other paramyxovirus VLP assays [35-38,50]. The NiV N, M, F, and G genes were sub-cloned into the pCAGGS vector and expression was verified in separate experiments by immunoprecipitation (data not shown).

To determine whether the various NiV proteins produced by this method were also released from expressing cells, cultures were transfected with individual plasmid constructs for 24 h, followed by metabolic labeling for another 20 h. Cells and culture supernatants were harvested and processed as described above and in the Methods. We again observed membrane-associated release of M, F and G, but no detectable release of N as predicted (Fig. 3A). Immunoprecipitation analysis of M expressed by plasmid-transfected cells consistently revealed a doublet of bands that were both released into the culture supernatant (Fig. 3A). When M was expressed by MVA, the doublet was usually only revealed by Western Blot (data not shown). We have been unable to determine the nature of the doublet, but it could reflect a post-translational modification of M. Although NiV M does contain a second potential AUG start codon 36 nt downstream of the first start codon [58], this does not account for the doublet appearance because a truncation mutant that begins at the second start codon, as well as HeV M, which lacks the second AUG codon, also appear as doublets (data not shown). Quantification of protein release revealed an overall reduction in protein release compared to that seen in the MVA system (Fig. 3B, compare with Fig. 2). Several studies have provided evidence for a physical interaction between paramyxovirus attachment and fusion proteins [59-61]. For NiV F and G this interaction is apparent from their ability to be co-precipitated without cross-linking (Bossart and Broder, unpublished data). We therefore sought to determine the effect of F and G co-expression on their release. Expression of F and G together resulted in greater release of both proteins together in comparison to expression of each protein individually (Fig. 3C). This observed increase of their release did not appear to be a result of increased protein expression or cell-surface expression (data not shown). We also noted that when N, M, F and G were all co-expressed using equivalent amounts of transfected plasmids, there was an overall reduction in protein release from cells (data not shown). Indeed, in certain other recombinant VLP expression systems, the amounts of transfected plasmids have been adjusted in order to more accurately reflect or achieve cellular expression levels of the individual viral proteins produced using infectious virus [36,38,50]. In an attempt to increase protein release in our system, we performed similar experimental variations in the amounts of tranfected plasmids and estimated envelope glycoprotein expression in authentic NiV infection as one third of M expression and adjusted plasmid ratios accordingly [55,62,63], and we further reduced N expression to 50 ng per well. However, even when the adjusted plasmid ratio was used, overall protein release remained low (Fig. 3D). We verified that this was not due to a reduction of protein expression due to multiple gene expression (data not shown), so we interpret this as indicative of a more organized assembly and budding process. In addition, the co-expression of N and M resulted in the release of N but with modest release of both proteins (Fig. 3E). This was in agreement with our previous results obtained using the MVA expression system, which demonstrated that M facilitated the release of N. However, in contrast to the MVA expression system, here we observed that N reduced overall M release. (Fig. 3E, compare with Fig. 2A and 2B).

Specific M release and kinetic analysis

In order to ensure that NiV M release was accomplished by a mechanism specific to itself, release of NiV M was again examined but in parallel with SV5 M, which requires co-expression of N and one of the envelope glycoproteins in order to form VLPs that are released into the culture supernatant [50]. Here the analysis of culture supernatants showed release of NiV M but no detectable release of SV5 M (Fig. 4A), suggesting that the observed release of NiV M was specific to its biological properties and not due to a non-specific mechanism such as cell lysis. In this respect, NiV M was similar to the M proteins of SeV, NDV, and hPIV1, all of which have been reported to form VLPs in the absence of other viral proteins [35-38].

We also wanted to explore the kinetics of M expression and its eventual release from cells. For this analysis, plasmid-transfected cells were starved for 45 min, ³⁵S-pulsed for 15 min, and then chased for varying lengths of time up to 32 h. At each time point cells and supernatants were harvested. Vesicles released into the supernatant were purified by centrifugation through a 10% sucrose cushion followed by sucrose gradient flotation. M derived from supernatants or cell lysates was immunoprecipitated with a monoclonal antibody and visualized by SDS-PAGE and autoradiography. Percent M release was quantitated by densitometry. The presence of membrane-associated protein in the supernatant was barely detectable at 2 h with 0.3% release, but then readily detected at 4 h with 1.7% release (Fig. 4B). Maximum M release of 27.4% was observed by 16 h.

Electron microscopy analysis of VLPs

The biochemical analysis of the NiV proteins released from expressing cells in a membrane-associated manner suggested that VLPs were being generated. To analyze the released material visually, VLPs were prepared and isolated by sucrose gradient floatation and the resulting fractions were immunolabeled using antibodies against NiV M or HeV G followed by a secondary antibody conjugated to gold beads. After negative staining, the samples were examined by immunoelectron microscopy. Expression of NiV M alone resulted in the release of VLPs that contained M and varied in size from approximately 100 to 700 nm in diameter (Fig. 5A–C). Expression of NiV N, M, F, and G resulted in the release of VLPs with detectable M (Fig. 5D) or G (Fig. 5E) with a diameter of approximately 100 to 300 nm. The size of the VLPs observed is consistent with observations made on authentic NiV virions which have been reported with sizes ranging from 40 to 1900 nm [64,65].

Interaction between proteins during VLP formation

In order to further investigate the interaction of F and G with M during VLP release we performed sucrose density gradient analyses to determine whether the buoyant density of released particles varied depending on the viral proteins present. M produced alone, F produced with G, or the combination of N, M, F and G were expressed in cells, metabolically labeled, and the culture supernatants were harvested and layered onto 5–45% continuous sucrose gradients. After centrifugation, the gradients were fractionated and analyzed by immunoprecipitation and SDS-PAGE. The M protein was again recovered predominantly in fractions 3 to 5, corresponding to a density range of 1.11–1.18 g/ml (Fig. 6A and 6B). We noted that the M protein could also be found in less dense fractions, especially when it was expressed in the absence of any other viral proteins. When F was co-expressed along with G, both proteins were recovered predominantly in fractions 1 to 3, which correspond to a density of 1.18–1.21 g/ml (Fig. 6A and 6B). However, the co-expression of N, M, F, and G led to a greater concentration of M in fraction 4 and, importantly, this also resulted in a shift of the fractionation profile of F and G distribution to fractions 3 to 5 (Fig. 6A and 6B). This density range was also consistent with the previous results obtained using rMVAs (Fig. 1C). Further, sucrose density gradient analysis of authentic infectious NiV revealed peak virus concentration at a density of 1.15 g/ml, which corresponded well with our VLP results (Fig. 6C). Taken together these findings suggest that although F and G can direct budding when produced alone, the co-expression of M facilitates F and G incorporation into VLPs with a density that is more consistent with that of an authentic virion.