Structure and Organization of Paramyxovirus Particles

Abstract

The paramyxovirus family comprises major human and animal pathogens such as measles virus (MeV), mumps virus (MuV), the parainfluenzaviruses, Newcastle disease virus (NDV), and the highly pathogenic zoonotic hendra (HeV) and nipah (NiV) viruses. Paramyxovirus particles are pleomorphic, with a lipid envelope, nonsegmented RNA genomes of negative polarity, and densely packed glycoproteins on the virion surface. A number of crystal structures of different paramyxovirus proteins and protein fragments were solved, but the available information concerning overall virion organization remains limited. However, recent studies have reported cryo-electron tomography-based reconstructions of Sendai virus (SeV), MeV, NDV, and human parainfluenza virus type 3 (HPIV3) particles and a surface assessment of NiV-derived virus-like particles (VLPs), which have yielded innovative hypotheses concerning paramyxovirus particle assembly, budding, and organization. Following a summary of the current insight into paramyxovirus virion morphology, this review will focus on discussing the implications of these particle reconstructions on the present models of paramyxovirus assembly and infection.

Introduction

Together with the rhabdo-, filo-, borna- and pneumoviruses, the paramyxoviruses form the order mononegavirales that features enveloped virions with single-stranded, non-segmented RNA genomes of negative polarity. Common to all members of the paramyxovirus family are two membrane glycoprotein complexes, the attachment (H, HN, or G) and the fusion (F) proteins, that are responsible for receptor binding and cell entry through fusion of the viral envelope with target cell membranes, respectively [1] (Figure 1). The RNA genome is encapsidated by the viral nucleocapsid (N) protein, resulting in the formation of a helical ribonucleoprotein (RNP) complex that serves as the template for the viral RNA-dependent RNA-polymerase complex composed of the viral phospho- (P) and large (L) proteins. The matrix (M) protein organizes particle assembly through interaction with both N proteins in the RNP complex and the membrane-embedded glycoprotein complexes. Some members of the family such as pathogens of the rubulavirus genus contain a small hydrophobic (SH) transmembrane protein in addition to these six structural proteins. Only J paramyxovirus encode a fourth integral membrane protein, transmembrane (TM), that stimulates cell-to-cell fusion but not viral entry [2].

Figure 1.

A) Model of an MeV virion in which the matrix protein coats the nucleocapsid protein. The viral envelope is shown in orange. The nucleocapsid proteins are shown in cyan. The fusion protein trimers are shown in red. The attachment glycoprotein tetramers are shown as green. Matrix protein depicted is from Newcastle disease virus (PDB 4g1g). Viral glycoproteins are based on PIV5 (PDB: 4gip for the F protein; PDB ID: 4jf7 and 3tsi for the PIV5 HN ectodomain and stalk, respectively. The vesicular stomatitis virus L protein structure (yellow) was used to represent the unknown paramyxovirus L conformation (PDB ID: 5a22). The phosphoprotein (brown) was modeled using the oligomerization domain of measles P (PDB ID: 3zdo). B) A model of an MeV virion in which a matrix protein array is located at the inner leaflet of the viral membrane. The matrix protein is shown in blue. The matrix-coated nucleocapsids were created using Chimera [92] using electron density maps EMD-1973 and EMD-1974 [43]. PDB structures were created in PyMOL [93].

While electron microscopy has established a basic framework for the paramyxovirus virion organization [3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23] and a number of crystal structures of paramyxovirus proteins and protein fragments have been solved (for instance [6,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42], the reconstruction of the 3D ultrastructures of paramyxovirus particles is impaired by particle size and the pleomorphic nature of the virions, which prevents single particle reconstruction approaches. To overcome the problem, recent studies have applied cryo-electron tomography (cryo-ET) to the analysis of paramyxovirus particles. By providing insight into the 3D structures of SeV, MeV, and NDV particles, and the NiV glycoprotein organization displayed on VLPs [6,25,43,44], this work has returned two highly unexpected new proposals concerning paramyxovirus particle assembly and organization: i) coating of the MeV RNPs by M protein tubes, which may spotlight an alternative particle assembly mechanism; and ii) an organized assembly of NiV F protein trimers into a hexameric ring-like assembly that may contribute to F conformation stability and concerted triggering for efficient viral entry.

Virion morphology and substructures

Overall particle morphology appears to vary considerably depending on the paramyxovirus genus investigated. For instance, tomograms of SeV, a member of the respirovirus genus that also includes the human parainfluenzaviruses type 1 and 3, showed predominantly spherical particles [44], while reconstructions of NDV virions, a member of the avulavirus genus, revealed shapes ranging from spherical to elongated ellipsoidal [6], and particles of the morbillivirus type species MeV, schematically shown in figure 1A, showed a multitude of different configurations [43]. Independent of the predominant particle shape and consistent with previous negative-stain EM-based visualization of paramyxoviruses, each of these studies spotlighted large variations in particle size, ranging from approximately 110–540 nm in diameter for SeV, 100–250 nm diameter for largely spherical NDV particles, and 50–510 nm in the case of MeV. In particular elongated to filamentous particles frequently exceed 500 nm in length, resembling in appearance the largely filamentous organization of respiratory syncytial virus particles of the related pneumovirus family [45].

All paramyxovirus RNP genomes show a characteristic herringbone-like structure when examined by negative-stain EM [46]. Reconstructions of the nucleocapsids revealed a left-handed helical arrangement with a pitch varying from 4.6 to 7 nm, depending on the paramyxovirus family member analyzed [7,8,28,43,47,48]. Cross-sections through the nucleocapsids showed an inner diameter of 4–5 nm and an overall tube diameter of approximately 20 nm [6,43]. Virion reconstructions and functional studies have demonstrated that multiple genome copies can be packaged by a single particle [44,49], likely reflecting a poorly ordered assembly process. Like all other negative-polarity RNA viruses, isolated paramyxovirus genomes are not infectious and only the viral RNBA-dependent RNA-polymerase (RdRp) complex is capable of transcribing and replicating the RNP genomes [46]. In addition to the nucleocapsid, infectious particles must therefore package and deliver RdRp complexes to target cells to initiate a new infectious cycle.

Paramyxovirus entry is mediated under neutral pH conditions by a concerted action of the viral attachment and F glycoproteins. Cryo-electron tomograms of unstained paramyxovirus particles have revealed a dense array of glycoprotein spikes on the virion surface, but no higher order organization was apparent and the identification of individual glycoprotein oligomers was prevented by the tight packaging of the complexes [6,43,44,50].

Matrix protein assemblies

A crystal structure of the NDV M protein was recently solved and revealed a dimeric organization with 4-fold symmetry [6]. In addition to interacting with the N protein and the glycoprotein tails, the M protein contains positive charge patches on the surface that allow interaction with lipid membranes. Through multimerization of M dimers into a grid-like protein array with a 6° angle between the individual dimers, the M protein can introduce membrane curvature that is thought to promote virus budding [51,52]. Transient expression of M was shown to be sufficient in several cases to induce paramyxovirus VLPs formation [53,54,55,56,57,58,59,60,61,62]. However, this observation does not apply to members of the rubulavirus genus such as mumps virus, which require co-expression of M protein with the viral NP and glycoproteins to induce efficient particle production. In the case of the rubulaviruses, the M-N interaction not only recruits RNP to the sites of particle assembly, but is also thought to trigger particle release [21,63]. The ability to only release viral particles that contain RNP would be an advantage; allowing the virus to limit the release of noninfectious, empty virions. In addition to the M protein, also the glycoproteins, especially the F protein, have been implicated in modulating assembly and budding of at least some paramyxoviruses such as MuV [63], SeV [56], PIV5 [21] and NiV [59,64]. For NiV, autonomous formation of virus-like particles (VLP) by the F protein, and to a much lesser extent the G protein, was observed in addition to the more conventional M protein-mediated budding [59,64]. Although rare among the paramyxoviruses, autonomous F-induced VLP formation was also described for SiV and MeV [54,55,56]. However, efficient virion assembly requires in all of these cases the presence of the M protein and the biological function of F-mediated VLPs is still unclear.

Interaction of the M protein with the RNP complex

In NDV and SeV particle reconstructions, a distinct M layer of approximately 5 nm was observed below the envelope membrane, but only in a minority of virions examined and mostly covering only parts of the luminal membrane surface [6,44]. Interestingly, in membrane areas with detectable M protein layer, the NDV glycoproteins appeared to follow the pattern of the M array, positing the cytoplasmic tails of the HN and F protein complexes in the gaps between the M protein dimers. The low abundance of M protein arrays in the tomograms was hypothesized to represent disassembly of the arrays after budding is complete to facilitate subsequent membrane fusion and particle uncoating [6], but could alternatively also originate from specimen preparation, storage, and/or cryo-preservation. While the presence of functional M protein and intact glycoprotein cytoplasmic tails can down-modulate cell-to-cell fusion activity of some paramyxoviruses [65,66,67], it is unclear whether F refolding is indeed suppressed through F tail contact with intact M protein arrays as was speculated [6]. However, it is difficult to envision the subsequent introduction of extreme negative membrane curvature required for lipid merger and opening of a fusion pore [68] in the presence of an intact M protein lattice, necessitating the partial or complete breakdown of the arrays at some point prior to infection. The molecular driving force for disassembly of the ordered M layer is currently unknown. Receptor binding by the attachment protein was suggested as a possible impetus [6], but evidence is lacking that receptor binding translates to conformational rearrangements of the cytosolic tails and even if these occur it is not apparent how they could be sufficient to disturb highly ordered M protein arrays.

Unlike the matrix protein patches found below the envelope membrane in some of the NDV and SeV particle reconstructions, no significant protein density was detected in density profiles obtained from MeV tomograms. Rather, tubular structures with a diameter of approximately 30 nm were noted in some virions in addition to the 20 nm herringbone-like nucleocapsids (figure 1B) [43]. While the 20 nm tubes adhered to both anti-N and anti-M immunosorbent EM grids, the 30 nm tubes were precipitated only onto the latter grids, suggesting that the larger diameter structures consist of RNPs wrapped into M sheaths. The 30 nm M tubules formed a left-handed helix like the MeV RNP itself (figure 2), albeit with a pitch of 7.2 nm versus 6.4 nm calculated for the nucleocapsid. Within virions, the tubes packed into tight bundles that stood in lateral contact with the viral envelope. Interestingly, tubes of either diameter precipitated poorly onto anti-P grids, ruling out that assembled nucleocapsids are decorated by default with a high content of P molecules interacting with individual N protomers. By contrast, cylindrical M arrangements or structures equivalent to the 30 nm MeV tubes were not found in NDV particles [6]; unfortunately, SeV reconstructions lacked sufficient resolution of the RNPs to test for the presence of different types of tubular structures [44]. M tubes were likewise absent from HPIV3 particles [5]. However, the interpretation of the HPIV3 data is challenging, since none of the reconstructed HPIV3 particles that featured RNP and/or glycoprotein spikes contained any M density, while a particle proposed to contain M arrays lacked genome, glycoproteins, and was substantially smaller in size.

Figure 2.

Organization of the paramyxovirus nucleocapsid. A) Nucleocapsid proteins for all paramyxoviruses form helical assemblies of N proteins encapsidating the viral RNA (shown in this model are MeV RNPs, N proteins are depicted in cyan and forest green, the RNA is colored in red) (PDB ID: 4UFT). B–D) While RNPs for 20 nm diameter tubules, MeV RNPs were also found M protein wrapped in larger diameter 30 nm tubules [43]. Matrix proteins are shown in dark blue and nucleocapsids in cyan. Top view (B) and side view (C) of the 30 nm tubules depicting the distinct cylindrical M complex surrounding the MeV nucleocapsid. D) A clipped model of the 30 nm tubule structure. The matrix coated nucleocapsids were created using Chimera [92] based on electron density maps EMD-1973 and EMD-1974 [43].

The differences in results in particular between reconstructed NDV and MeV particles bring up the question of whether MeV nucleocapsids wrapped into M tubes represent a sample preparation artifact, dead-end complexes of a particle assembly process gone catastrophically wrong, or a physiologically relevant stage of the replication cycle of morbilliviruses, and perhaps of even a broader subset of paramyxovirus genera. Without further sightings of these M structures in additional MeV particle reconstructions that ideally follow particles through different stages of assembly and budding, this issue cannot be addressed definitively. However, we can examine the general plausibility of a physiological role of these M tubules by considering three basic questions: i) does a driving force exist for the formation of M tubules under native conditions; ii) is a physiological role of M tubules in the MeV life cycle conceivable; and iii) does the available phenotypic information support the existence of MeV M tubules?

1. The paramyxovirus N protein is composed of an N-terminal Ncore domain that is responsible for N homo-oligomerization and RNA encapsidation and a C-terminal Ntail that extends outwards from the RNP assembly. While Ntail is structurally intrinsically disordered, areas of high sequence conservation were identified that engage in multiple protein-protein interactions [69]. A molecular recognition element (MoRE) mediates transient binding of the P-L polymerase complex [70], and the terminal residues of the Ntail were demonstrated to mediate specific contacts with the MeV M protein [71], promoting genome packaging into nascent virions. Consequently, a high density of M binding sites is displayed on the surface of assembled MeV RNPs, which should be sufficient to trigger M multimerization around the nucleocapsid in infected cells. Disorder of the central Ntail section was furthermore proposed to provide the necessary flexibility to negotiate the different symmetries of the 20 nm inner RNP tubes and the 30 nm outer M cylinders [43]. However, a recent study has demonstrated that removal of the disordered tail region and relocation of the MoRE domain into Ncore does not abrogate replication of recombinant MeV, provided the M-binding terminal residues are added to the Ntail stump [72]. Although these MeV recombinants were not tested for the presence of the 30 nm tubes, they demonstrate that structural flexibility provided by the central Ntail section does not constitute a requirement for productive MeV particle assembly.

2. A number of candidate physiological effects of M tubules around the RNPs is conceivable. Certainly, RNP wrapping would prevent RdRp access and/or migration along the genome, blocking both genome transcription and replication. The tight arrangements of the 30 nm tubules in tomograms furthermore suggests that the wrapping may increase genome density, and it may ensure that genomes are packaged into nascent virions by recruiting the pre-wrapped nucleocapsids to budding sites. While all of these effects are poised to enhance the assembly of infectious particles, RNP wrapping by M tubules appears incompatible with a central step of paramyxovirus budding models, the introduction of membrane curvature through M protein arrays below the lipid membrane. Although tomograms of MeV particles showed association of the 30 nm tubules with the luminal surface of the viral envelope, large planar arrays rather than a cylindrical arrangement should be required to introduce sufficient membrane curvature for effective formation of progeny virions. On the other hand, MeV budding is inefficient [73], particle shape is more diverse than reported for members of some other paramyxovirus genera [6,43,44], and progeny virions remain largely cell-associated rather than being released into culture supernatants [74]. Also, it was suggested that MeV M could play different roles in RNP wrapping and the introduction of membrane curvature [43], although this would require additional interfaces between M proteins engaged in tubular and planar arrangements. In addition, tight spatial and temporal regulation of RNP wrapping would be essential, since the bulk of paramyxovirus proteins is generated through secondary transcription of progeny genomes [46] and premature shut-down of the transcriptase through genome wrapping would be catastrophic for the infection cycle. Equally important, the M tubules must disassemble to allow polymerase access to the genome after infection, and the molecular basis for both regulation of M tube formation and the subsequent induction of M depolymerization is unknown. A direct role of receptor binding, via conformational changes of the cytosolic glycoprotein tails, appears even less likely in the case of M tubules than for planar M arrays as was suggested for SeV uncoating.

3. Three lines of phenotypic evidence are consistent with a physiological role of M tubules in the MeV life cycle. Expression of the MeV M protein reduces genome transcription and replication [71,75] but no such effect was noted for SeV M [76]; intracellular transport of nucleocapsids of recombinant MeV harboring M proteins with reduced half life was inefficient in the absence of M accumulation at intracellular membranes [57]; and MeV N proteins lacking the disordered central region of the tail domain altered viral mRNA expression in the context of virus infection but not in polycistronic minigenome reporter assays [72]. Since M protein is only present in the infection but not the minigenome experiments, it is conceivable that removing the disordered tail section affects the frequency with which the tubular M structures form and thus the timing of shut-down of genome access by the RdRp.

Glycoprotein organization

The physiological oligomer of the paramyxovirus attachment protein is the tetramer, consisting of a dimer of homotypic dimers, whereas the F protein, a type I viral fusion protein, assembles into homotrimers [46]. The attachment proteins feature a globular head domain with the beta propeller fold characteristic for sialidases that attach to the transmembrane domain and cytosolic tails through a helical stalk domain [68]. A large body of evidence supports that the protein interface region responsible for specific hetero-oligomerization with homotypic F protein trimers resides in this stalk [50,77,78,79,80,81]. Upon receptor binding by the globular head domain, exposure of [79,82] and/or a conformational change in the membrane-proximal attachment protein stalk domain [77,83,84] is considered to trigger major conformational changes in the metastable prefusion F trimer, resulting in propelling of a membrane attack group or fusion peptide towards the opposing membrane, hairpin formation, and the assembly of a thermodynamically highly stable fusion core or six-helix bundle structure, which induces extreme local negative membrane curvature and brings the F trimer transmembrane domains and fusion peptides, and thus viral envelope and target membranes, into close proximity [68].

Although attachment and F glycoprotein oligomers could not be definitely separated in reconstructions of unstained paramyxovirus particles, tomograms of HPIV3 particles [5] and recombinant MeV particles displaying stalk-elongated attachment proteins allowed the identification of the attachment protein globular head domains in radial density distribution plots [50]. The MeV attachment proteins in particular were engineered to extend the stalk length by approximately 4 nm, which is equivalent to nearly 50% of the length of the unmodified MeV H stalk, while maintaining bioactivity. These reconstructions confirmed a spatial arrangement of functional fusion complexes in which the attachment protein head domain is positioned membrane distal and above the prefusion F trimers [78].

However, the tomograms failed to decipher the H to F oligomer stoichiometry in physiological fusion complexes. Multiple studies investigating different type I viral membrane fusion proteins have concluded that a concerted action of several fusion protein complexes is required to induce sufficient negative curvature in the viral envelope and cellular membrane to trigger local lipid disarray in the outer leaflets of the approaching bilayers, allowing merger of the disordered monolayers at the fusion tip and ultimately opening of a fusion pore [85,86,87,88,89,90,91]. If individual paramyxovirus glycoprotein homo-oligomers interact with each other for membrane fusion, each attachment protein tetramer would be sterically able to contact two F protein trimers in parallel. A recent study proposed an alternative F protein organization based on crystal structures of recombinant soluble NiV F protein trimers and tomograms of NiV glycoprotein-coated VLPs that both suggest a hexamer of F trimers ring-like assembly [25] (figure 3). In this arrangement, each F trimer contacts with its head domain two neighboring trimers, leaving only a single priming site available for interaction with the attachment protein. This F organization is provocative, since it may be able to reinforce the metastable prefusion conformation of the individual F trimers engaged in the ring structure prior to receptor binding, while activation of a single associated attachment protein tetramer may be sufficient to trigger the spatially and temporally highly coordinated refolding of all six F trimers locked into the ring structure. Since some F trimers appeared to be part of more than one hexameric structure, one could even envision a highly effective concerted wave-like refolding of numerous F complexes present on a viral particle after receptor binding by a small number of attachment protein tetramers. However, since hexameric F rings have so far been proposed only for NiV fusion proteins and NiV VLPs lacked the corresponding NiV G attachment proteins, it is unclear at present whether this F arrangement is germane only to the henipavirus genus and whether the presence of the attachment protein tetramers would alter the spatial organization of the F trimers.

Figure 3.

Models of alternative spatial organizations of the paramyxovirus glycoproteins on the virion surface. A) The overall organization of the paramyxovirus glycoproteins was thought to be random with an undetermined relative stoichiometry of individual fusion protein trimers (red) and attachment protein tetramers (green). B, C, D) In a recent study [25], an arrangement of NiV F protein trimers into hexamers of trimers and higher order complexes was proposed. Schematics of hexameric F trimer arrangements in contact with one (A) or multiple (B) attachment protein tetramers and higher order F assemblies consisting of interacting hexamers of trimers (C). Different hypothetical contacts of the F assemblies with attachment protein tetramers are shown, but the stoichiometry and positioning of the attachment protein oligomers relative to the F protein complexes has not yet been defined.

Conclusions

Although only a limited number of paramyxovirus virion reconstructions is available at present, the insight gained from these studies has substantially impacted the current view of particle organization, assembly, and budding. A number of unexpected and potentially paradigm-changing hypotheses concerning the possible roles of the matrix protein in particle formation, genome packaging, and transcription control and of higher order glycoprotein organization in mediating efficient viral entry has emerged from these reconstructions. In addition to spotlighting previously unappreciated features of the paramyxovirus life cycle, these studies have also reinforced that central parts of the paramyxovirus replication, assembly, and uncoating machinery are currently mechanistically not understood. Since a detailed structural and mechanistic understanding of these central steps of the viral life cycle will inform the targeted development of much needed improved prophylactic and therapeutic anti-paramyxoviruses strategies, obtaining additional high-resolution particle structures and substructures is a high priority to test the physiological relevance of individual observations and assess their applicability to clinically-relevant pathogens of different genera within the family.

Highlights.

• Recent cryo-electron tomography reconstructions of Sendai virus, Newcastle disease virus, measles virus, and Nipah virus-derived virus like particles have yielded novel insight into paramyxovirus particle assembly and organization

• The Newcastle disease virus matrix protein formed distinct layers below the viral envelope and the viral envelope glycoproteins followed the pattern of this matrix array, while tubular matrix protein assemblies were observed in measles virus particles that wrap the viral genome into matrix sheaths

• The physiological role of tubular matrix sheaths is unclear but could involve mediating transcription and replication shut-off at the time of particle assembly

• Nipah virus-based virus like particles have revealed a previously unappreciated hexameric ring-like organization of the viral fusion protein that if present on native virions may set the stage for efficient viral entry through concerted refolding of multiple F complexes