Free access
Spotlight Selection
Research Article
15 February 2009

Rotavirus Architecture at Subnanometer Resolution

ABSTRACT

Rotavirus, a nonturreted member of the Reoviridae, is the causative agent of severe infantile diarrhea. The double-stranded RNA genome encodes six structural proteins that make up the triple-layer particle. X-ray crystallography has elucidated the structure of one of these capsid proteins, VP6, and two domains from VP4, the spike protein. Complementing this work, electron cryomicroscopy (cryoEM) has provided relatively low-resolution structures for the triple-layer capsid in several biochemical states. However, a complete, high-resolution structural model of rotavirus remains unresolved. Combining new structural analysis techniques with the subnanometer-resolution cryoEM structure of rotavirus, we now provide a more detailed structural model for the major capsid proteins and their interactions within the triple-layer particle. Through a series of intersubunit interactions, the spike protein (VP4) adopts a dimeric appearance above the capsid surface, while forming a trimeric base anchored inside one of the three types of aqueous channels between VP7 and VP6 capsid layers. While the trimeric base suggests the presence of three VP4 molecules in one spike, only hints of the third molecule are observed above the capsid surface. Beyond their interactions with VP4, the interactions between VP6 and VP7 subunits could also be readily identified. In the innermost T=1 layer composed of VP2, visualization of the secondary structure elements allowed us to identify the polypeptide fold for VP2 and examine the complex network of interactions between this layer and the T=13 VP6 layer. This integrated structural approach has resulted in a relatively high-resolution structural model for the complete, infectious structure of rotavirus, as well as revealing the subtle nuances required for maintaining interactions in such a large macromolecular assembly.
Rotaviruses are double-stranded RNA viruses belonging to the family Reoviridae. They are the major causative agents of severe gastroenteritis in young children and animals (24). These relatively large (∼1,000 Å) nonenveloped icosahedral viruses consist of 11 segments of double-stranded RNA. Each of the 11 segments codes for one protein, with the exception of segment 11, which codes for two proteins. Of these 12 proteins encoded by the viral genome, six are structural and six are nonstructural. The proteins encoded by the rotavirus genes are well established, and their properties have been reviewed (13).
Electron cryomicroscopy (cryoEM) structural analysis has shown that rotavirus has three concentric capsid layers that enclose the genomic RNA (35, 36, 42, 45). Two outer icosahedral layers surround an inner T=1 icosahedral (25, 45) layer composed of VP2, similar to the case for other members of the Reoviridae such as bluetongue virus (17), orthoreovirus (37), and rice dwarf virus (29, 46). The outermost capsid layer, composed of VP7, and the middle capsid layer, composed of VP6, both exhibit T=13 icosahedral organization. Rotavirus architecture also features 132 aqueous channels, ∼140 Å deep, spanning the outer two capsid layers and 60 spikes composed of VP4, each 120 Å long, emanating from the virion surface (36, 39, 41). The 132 channels are grouped into three types based on their locations in the T=13 icosahedral lattice. The type I channels, which have narrow openings, are located at the icosahedral fivefold axes, whereas the type II channel, located at the quasi-sixfold position close to the type I channel, and the type III channels, located at the quasi-sixfold position close to icosahedral threefold axis, have significantly larger openings. The spikes are located at the edge of the type II channels.
The outer capsid protein VP7 (∼37 kDa) and the spike protein VP4 (∼88 kDa) are both targets for neutralizing antibodies (30, 31). Although early studies implicated VP7 in the cell entry process (14, 38), subsequent studies increasingly have indicated that VP4 is the major player in this process (7). VP4 is susceptible to proteolysis. Proteolytic cleavage of VP4 enhances viral infectivity severalfold (1, 12) and facilitates virus entry into cells (23). During proteolysis, VP4 (88 kDa) is cleaved into VP8* (28 kDa, amino acids [aa] 1 to 247) and VP5* (60 kDa, aa 248 to 776), and the cleavage products remain associated with the virion (1, 15). Crystallographic analyses of VP8* and a major portion of VP5* (aa 246 to 477) have indicated that the former constitutes the distal globular density in the spike and the latter is localized to the central body of the spike (8, 9). In addition to the density that projects out from the capsid surface, the spike has a large internal globular density that is tucked inside the type II channel (39, 41).
Several recent studies have shown that rotavirus cell entry is a multistep process involving sialic acid-containing receptors in the initial cell attachment step and integrins such as αvβ3, α4β1, and α2β1 during the subsequent postattachment steps (4, 18, 19, 44). How VP4 facilitates such a multistep entry process involving multiple receptors remains unclear (7). In this process, the VP8* domain is involved in the interactions with sialic acid (27), whereas VP5* is suggested to be involved in the interactions with integrins. The latter, with a putative fusion peptide similar to that found in enveloped viruses such as Sindbis virus and Semiliki Forest virus, exhibits membrane-permeabilizing activity (6, 10). Based on the crystallographic studies of VP5*, in both the dimeric and trimeric forms, and cryoEM studies of nontrypsinized, trypisnized, and high-pH-treated rotavirus particles, it has been hypothesized that the VP4 spike undergoes a series of unique structural rearrangements from a disordered state prior to trypsinization to a dimeric state upon trypsinization and a trimeric state during the cell entry process (5, 8, 32, 43). The underlying assumption in this hypothesis is that spikes are trimers instead of dimers as indicated by previous cryoEM studies of native mature virions and virions labeled with VP4-specific antibodies (34, 40). Upon trypsinization, two subunits associate to form dimers as seen in the reconstructions of the trypsinized particles, and the other subunit remains disordered and associates to form a trimer only during the cell entry processes. How such a structural transformation takes place is not clear.
We report here a cryoEM structure of the mature triple-layer rotavirus particle at 9.5-Å resolution. Besides providing a more stringent framework for combining the available X-ray structures of VP8* (9), VP5* (8, 43), and VP6 (28) to delineate the intermolecular and interlayer interactions at the amino acid residue level, these studies have revealed several novel aspects of the VP4 spike structure in relation to its domain organization and oligomeric state and better structural definition of the other major components, VP7 and VP2, for which there are no X-ray structures available.

MATERIALS AND METHODS

CryoEM specimen preparation and data collection.

Rotavirus mature triple-layer particles (SA11-4F strain), grown in the presence of trypsin in MA-104 cells, were purified as described previously and suspended in TNC buffer (10 mM Tris, 150 mM NaCl, 10 mM CaCl2, pH 7.5) (33) for cryoEM specimen preparation. About 4 μl of the sample was applied to a R 2/2 Quantifoil copper grid (400 mesh), which had been glow discharged for 35 s just before use. The excess sample was blotted off by using a piece of filter paper, and the grid was then immediately plunged into liquid ethane that was cooled with surrounding liquid nitrogen, and a very thin layer of vitreous ice of virus suspension was formed over the grid holes. The frozen grid was imaged on a JEOL 3000SFF electron microscope equipped with a liquid helium stage at a magnification of ×60,000 with an accelerating voltage of 300 kV. Images were collected on Kodak SO-163 film and developed in Kodak full-strength D-19 developer at 20°C for 12 min, followed by fixation for 10 min at the same temperature. The data were recorded mainly as focal pair images, with close-to-focus images ranging from 0.5 μm to 2.0 μm defocus and far-from-focus images ranging from 1.5 μm to 3.5 μm defocus.

Image processing and three-dimensional reconstruction.

All the micrographs were first inspected visually for ice contamination, drift, or charging problems. Images with drift or charging problems were discarded. A total of 197 micrographs were scanned on a Zeiss SCAI microdensitometer (Carl Zeiss, Inc., Englewood, CO) with a step size of 7 μm and then averaged by 1.5, corresponding to a scan size of 1.75 Å/pixel. Preprocessing of the data was performed using two programs from EMAN (26): BOXER was used to box 8,582 particles from the scanned micrograph, and CTFIT was used to calculate the fast Fourier transform of boxed particles and to evaluate the contrast transfer function parameters. The initial orientation search, orientation refinement, and three-dimensional reconstruction were done using SAVR (Semi-Automated Virus Reconstruction package) (22). The final structure was calculated from 3,748 images to a resolution of 9.5 Å based on the 0.5 Fourier shell correlation criterion. The phase residual when refined to 9.13 Å was 39.7 to 64.0.

Structural analysis.

Segmentation of the individual subunits was done manually using UCSF's Chimera and Amira (16). When appropriate, known X-ray structures of rotavirus structural proteins were fitted into the cryoEM map first to identify subunit boundaries. Fitting of all atomic models to the density map were done using FOLDHUNTER (20), a rigid-body fitting routine from the AIRS toolkit of EMAN (26).
Identification of secondary structure elements in the component density maps was done using SSEHunter and SSEBuilder (2), feature recognition tools from the AIRS toolkit in EMAN. Only resolution, Å/pixel, and threshold were required inputs to SSEHunter (2). Appropriate isosurface thresholds were selected by determining the correct isosurface that corresponded to the mass of the segmented subunits using the program “volume” from EMAN.
Symmetry searches were carried out using the program SymAxisSearch from EMAN (26). Essentially, this program searches for a user-defined symmetry, aligns the map along this axis, and calculates an n-fold correlation score for each xy plane of the map. The result of this search, in which characteristic planes are shown, was illustrated using UCSF's Chimera (16).
A bilateral filter (one iteration with σ1 = 2, σ2 = 3, and width = 3) (21) was applied to a region of the cryoEM density map containing all the components surrounding a type II channel. This filtered map was used only in the localization of the flexible VP4 subunit.

Building the VP5*-t/VP8* connection.

To reveal the location of the missing connection between the VP5* and VP8* subunits, a difference map was first calculated. The fitted models were first blurred to density maps with the same resolution and sampling of the cryoEM density map. These maps were then turned into binary masks and subtracted from the segmented spike protein, revealing regions in the cryoEM density map that were unaccounted for by the fitted models. Measuring the distance between the C-terminal end of VP8* and the N-terminal end of VP5*-t (VP5* structure with a C-terminal deletion), combined with the difference map, revealed only one potential path to connect these two proteins. Using COOT (11), individual Cα atoms were placed at intervals of ∼3.8 Å along the path identified by the density map. Minor optimization of the Cα positions in this connecting loop and, to an even lesser extent, residues from VP5*-t and VP8* was preformed by hand to better fit the density map.

Sequence analysis.

Secondary structure prediction for the capsid proteins was based on a consensus prediction using PsiPred, JPRED, and SamT02. Correlation of sequence-predicted helices and those found with SSEhunter (2) was accomplished by comparing helix length. In the case of VP2A and VP2B, structural homologues from bluetongue virus and rice dwarf virus were fit to the segmented density maps of VP2. These templates, combined with the topological skeleton from SSEHunter (2), allowed for the assignment of all α-helices and a subsequent topological model.
All figures were made using UCSF's Chimera (16).

RESULTS AND DISCUSSION

The reconstruction of the mature rotavirus was computed from 3,748 virus particle images extracted from 197 focal pair images with defocus values ranging from 0.5 to 3.5 μm (Fig. 1A). The final resolution of the reconstruction was estimated to be 9.5 Å using the 0.5 Fourier shell correlation criterion. The overall architectural features of the virion, with a diameter of ∼1,000 Å, are consistent with previous lower-resolution structural analyses. The two outer capsid layers exhibit T=13 icosahedral symmetry with 132 aqueous channels, at the fivefold and quasi-sixfold locations, and 120-Å spikes projecting away from the capsid. The icosahedral lattice in the innermost capsid layer deviates significantly from the outer two layers, exhibiting T=1 icosahedral symmetry (Fig. 1B). Recently, a high-resolution (∼5.6-Å) cryoEM structure of the double-layer particle (i.e., without the outer VP7 layer and VP4 spikes) has been reported by Zhang et al. (45). Although no models of the protein subunits (VP6 and VP2) or the intersubunit interactions were reported in their paper, a qualitative comparison of the densities in the VP6 and VP2 layers between this higher-resolution cryoEM structure and our structure shows excellent agreement (data not shown).

The VP4 spike.

The VP4 spikes, located at the type II channels, visually appear to be dimers, each exhibiting four unique structural domains (Fig. 2A). The large globular heads (VP8*) and central bodies (VP5*) extend outwards from the capsid surface. In addition, each spike has a globular internal domain (base) that is tucked inside the type II channels. An elongated, bridging domain joins this internal domain to the two protruding domains.

The globular head domain, VP8*.

Previously, the two globular head domains were shown to accommodate the structure of the VP8* (1KRI and 1KQR; residues 65 to 224). The fitting of VP8* to each of the two globular domains in our 9.5-Å resolution reconstruction revealed that the C terminus of VP8* points toward the central body domain (Fig. 2B). In each of the two globular domains, a small depression corresponding to the sialic acid binding pocket in the VP8* (9) is also observed (Fig. 2B, left panel).

The central body domain, VP5*.

Each of the globular head domains, composed of VP8*, is attached to a central domain that extends outwards from the capsid surface. Previous structural analyses have localized a portion of the VP5* to this domain. X-ray structures of VP5* in both dimeric (2B4H, aa 250 to 476) and trimeric (1SLQ, aa 253 to 522) states have been reported (8, 43). For convenience, these VP5* structures with a C-terminal deletion (VP5*-t) will be referred to as VP5*-dt and VP5*-tt, respectively, to distinguish them from the full-length VP5*. Except for a long C-terminal α-helix in the trimeric structure, the overall VP5*-t structure is similar in both forms and was used in the fitting. Localization of the dimeric and trimeric structures to the density map failed to completely fit the central body domains. Based on the dimensions and the twofold nature of the central body, it was clear that the entire trimeric VP5*-tt structure would not fit; however, the density could accommodate two monomeric subunits taken either from the dimeric or the trimeric X-ray structure. Starting from these positions, monomers from the dimer and trimer structures were fit separately to each of the two central body domains (Fig. 2C). Although roughly similar to the crystallographic dimer (VP5*-dt), the new fit contains a slight rotation of the monomers with respect to each other as well as increased angular separation (Fig. 2D). This conformational change may be a result of interactions with VP8* and VP7, which were absent in the dimeric and trimeric VP5*-t structures.

Connecting VP5* and VP8*.

While the VP5*-t structure represents the major portion of the central body, ∼25 residues connecting the globular VP8* and VP5*-t domains are not structurally defined (Fig. 3A and B). As stated above, the C terminus of VP8* points toward the most distal portion of the central body domain. For VP5*-t, however, both the N and C termini are located in a portion of the central body domain that is proximal to the capsid surface, more than 50 Å away from the C terminus of VP8*. No immediate connection was initially visible linking the VP8* C terminus to the N terminus of VP5*-t.
Utilizing the fitted structures of VP5*-t, a mask was created for the central body domain, revealing regions of the density that were not occupied by the VP5*-t X-ray structure. As seen in Fig. 3A, a path extends from the base of the central domain outwards along the surface of the central body toward its distal end. Similarly, unoccupied density is also seen extending outwards from VP8* to the central body domain. While not completely unambiguous, a general path connecting the C-terminal end of VP8* and the N-terminal end of VP5*-t can be formalized.
As mentioned above, X-ray crystallographic structures of VP5*-t exist in both dimeric and trimeric conformations. In the dimeric conformation, the N terminus of one VP5*-t molecule extends across to the neighboring VP5*-t subunit, contributing one strand to a β-sheet in that subunit. In contrast, in the trimeric structure, the N terminus folds back on itself and forms an intrasubunit β-sheet. In our density map, neither conformation is distinguishable. However, given the predicted secondary structure for the sequence between VP5*-t and VP8* (Fig. 3B) and knowledge of average Cα-Cα distances, it is possible to formulate a plausible model for the approximate path connecting the two structures (Fig. 3A and C). A model based on the dimer structure is most likely, such that the N terminus of one VP5*-t subunit extends to the neighboring VP5*-t subunit, forming an intersubunit β-sheet. In fact, this β-sheet extends further than the aforementioned VP5*-t/VP8* connection, following a long β-sheet on the exterior surface of the central body domain. As this path approaches the apex of the central body domain, it crosses over, connecting with the VP8* C terminus via a very clear bridging density. Thus, residues 225 to 249 in VP4 connect the distal globular domains to the central body of the spike when trypsinized.

VP5*-t/VP8* interactions.

It would appear that the major force in stabilizing the VP4 dimer and keeping VP8* and VP5* together in the spike following trypsinization is the aforementioned cross-subunit connection (Fig. 3D). This connecting segment, with the trypsin-sensitive residue R247 located at the base of the central body domain, allows sufficient interaction to maintain the dimeric structure.
Residues 227 through 233 in the VP8* appear to run across the top of the VP5*-t domain in between four loops at the distal end of VP5*-t. However, the majority of the VP5*-t/VP8* interactions appear to be within a single subunit. Again, loops at the distal end of the VP5*-t structure interact with the VP8* domain. Specifically, two loops (residues 285 to 290 and 331 to 336) from the VP5*-t structure appear to interact with residues 65 to 80 and 204 to 208 of the VP8* domain primarily through hydrophobic interactions. It is possible that the N-terminal residues (1 to 64), not seen in the VP8* structure, contribute to the stabilization of the VP8* domain and spike dimer; however, localization of these residues in the density map was not possible at this resolution.

A trimeric base?

Extending inward through the type II channels, the VP4 spike forms a relatively large globular base situated between the outer two capsid layers (Fig. 4A). This internal globular density, visualized in the previous lower-resolution cryoEM reconstructions, has been indicated to be part of VP4 (39), but its exact composition is unknown. As the aforementioned structural analysis of the spike domain has already localized the first ∼500 amino acids of VP4 (VP8* and VP5*-t), the majority of the remaining amino acids likely make up this globular base domain and its connection to the central body of the spike.
Unlike the case for the rest of the spike structure, visual observation of the base domain suggests that it exhibits strong threefold symmetry. Computationally examining serial slices of the base domain bears out this observation. Individual slices have an average threefold cross-correlation score of 0.733; 30 of the 40 slices most proximal to the surface have an average threefold correlation of greater that 0.84 (Fig. 4B and C).
Secondary structure prediction of this domain using SSEHunter (2), clearly identifies the presence of α-helices and β-sheets. These secondary structure elements also exhibit the observed threefold symmetry (Fig. 5). While the unassigned sequence segments in VP5* are predicted to be primarily helical, no sequence-to-structure assignment was possible.

Connecting the base and central body domains.

Bridging the main body of the twofold symmetric central body (VP5*-t/VP8*) and the threefold symmetric base is an elongated density that runs nearly tangential to the capsid surface (Fig. 2A and 6). At one end, this density connects to the dimeric spike body in close proximity to the C termini of the VP5*-t structure (residue 476). At the other end, the density dips inward slightly, contacting the capsid surface at the trimeric VP7 layer. In the middle of this elongated density, a threefold protrusion extends to the trimeric base; only two of the three “staves” are directly contacting the elongated, bridge density (Fig. 2A and 6).
The elongated density itself does not appear to demonstrate either a twofold or a threefold symmetry. Rather, two long “rods” seem run on the outer surface, while a third possible “rod” runs underneath these two rods. Initial analysis with SSEHunter (2) indicated that these rods may be helical. Interestingly, in the X-ray structure of VP5*-tt, three long helices, one from each subunit, occur at the end of VP5* and run up the middle of the VP5* trimer (8). These helices are linked together by a leucine zipper. Excising this helix bundle and fitting it to the elongated bridge domain revealed that this domain is virtually the same size as the VP5* bridging domain (Fig. 6). In fact, the two rod densities observed on the outer surface of this domain correspond almost perfectly to two of the three helices in the trimeric helix bundle. The third helix in the bundle approximated the position of the third rod density in this domain, although the exact match to the density was considerably worse than for the other two helices.

VP4: a dimer or a trimer?

In our current reconstruction, the VP4 spike clearly adopts a dimeric appearance above the capsid surface but maintains a trimeric appearance below the capsid surface. Previous X-ray and cryoEM structures have indicated the possibility that the spike could indeed be composed of three VP4 molecules. Although not entirely unequivocal, observation of the strong threefold symmetry in the spike base appears to further substantiate this hypothesis. It is possible that the spike base is an anchor that holds portions of all three VP4 subunits together (Fig. 7A) and beyond which one of the VP4 molecules remains flexible or disordered. While a complete third molecule of VP4 could not be observed in our reconstruction, a large amount of “noise” was observed at a radius consistent with the VP4 spike. Applying a bilateral filter revealed the presence of several regions of “ordered noise” at the point where the bridging domain reaches the capsid surface and contacts the base domain (Fig. 7B). This density in fact could be the third subunit, extending from the trimeric base. Spike density is not visible in cryoEM reconstructions of untrypsinized rotavirus. It is therefore possible that all three VP4 subunits are flexible until trypsinization, at which time dimers are formed about the central body domain, leaving the third subunit flexible, and that due to icosahedral averaging during the reconstruction, only portions of this third molecule are visible in the reconstruction. It is possible that during the cell entry process this flexible subunit, prompted by an unknown triggering event, associates with the dimeric spike, following a large conformational reorientation, to form a trimeric spike with the membrane interaction region exposed as proposed by Yoder and Dormitzer (43).
The combined volume of the base and the bridging density is consistent with the possibility that the base includes a portion of the “third” VP4 molecule. Using an appropriate density threshold based on the available portions of VP5*-t and VP8* structures, the trimeric base accounts for ∼45 kDa, while the bridging density accounts for ∼26 kDa. A combination of a dimeric bridging domain, with 13 kDa per monomeric portion, and a trimeric base domain, with 15 kDa per monomeric portion, accounts for nearly all the ∼276 amino acids unassigned in the VP4 structure.

T=13 capsid layers.

As mentioned above, the VP4 spikes extend inward through the type II channels formed by the T=13 icosahedral layers, which consist of VP7 and VP6. Both proteins are trimers, and each VP7 trimer sits on top of a VP6 trimer (Fig. 8A). The structure for the VP6 trimer has previously been determined by X-ray crystallography (28) and therefore provides an accurate mask for the identification of protein boundaries between these capsid layers as well as an effective means for evaluating the secondary structure assignments by SSEHunter (2) in this region of the map (Fig. 8B).
Each subunit of the VP7 trimer interacts with a VP6 subunit at two locations (Fig. 8C). The primary point of interaction in the VP7 subunit appears to be with a rod-like density that projects inward toward the VP6 trimers (Fig. 8C). SSEHunter (2) identified this density and two other regions in the VP7 subunit as α-helices (Fig. 8D). Five helices (Y6 to F47, W98 to L107, M152 to L164, T214 to A119, and W289 to V309) based on a consensus secondary structure prediction were identified for VP7 (2) (see sequence at http://www.uniprot.org/uniprot/P03532 ). Helices H1 (W98 to L107), H2 (M152 to L164), and H4 (W289 to V309) were found to have a correspondence with the three helices identified using SSEHunter (2). Correlating the secondary structural prediction and SSEHunter feature detection, the rod-like density likely represents the C-terminal helix of VP7 (designated H4). Such an assignment is consistent with previous studies indicating that the extension of the C-terminal helix of VP7 into the cytoplasm from the endoplasmic reticulum-bound VP7 allows its interactions with nascent double-layer particles during virus assembly (3). The corresponding contacts on VP6, after fitting the X-ray structure of the VP6 trimer in the cryoEM density, are localized to two regions in VP6: S240 to T245 and L294 to M300. A second interaction site occurs between the distal ends of H1 and H2 in the VP7 subunit and a short helix (residues 302 to 308) in the VP6 subunit. Both of these locations appear to be primarily uncharged, suggesting that the VP6 and VP7 trimer association is stabilized by hydrophobic interactions.
In addition to interactions between the T=13 layers, both VP6 and VP7 interact with VP4 about the type II channels. Before the VP4 spike extends inward into the type II channel, the long bridging density of the spike crosses the channel almost horizontally, tangential to the capsid surface. In doing so, the bridging density in the spike makes contact with three of the six VP7 timers surrounding the type II channel (Fig. 9A and B). With one of the VP7 trimers, the contact is between its outer surface near H4 and the proximal end of the central domain corresponding to the C termini of VP5-t*. A second VP7 trimer makes extensive contacts through one of its corners (Fig. 9B) with the end of the bridging density that dips slightly into the type II channel. Sandwiched between these two VP7 trimers, a third VP7 trimer, again at one of its corners, interacts with the side of the bridging density (Fig. 9B).
While the interactions with VP7 essentially involve the bridging density of the VP4 spike, interactions with VP6 involve exclusively the base domain of the spike (Fig. 9C and D). Three of the VP6 trimers make symmetrical contacts with the spike base, each involving the same loop region between Y353 and P362 in one of their respective VP6 subunits (Fig. 9C and E). Due to their proximity to this loop, additional points of interaction could include another loop, I51 to P55, and a β-strand, Q268 to Q274 (Fig. 9E). Along with these three VP6 trimers, a fourth VP6 trimer interacts with the spike base. While it is not completely unambiguous, this interaction appears to be centered on the interaction interface of VP6 and VP7 (VP6, S240 to T245 and L294 to M300; VP7, H4).
A double-layer particle, composed of only the VP2 and VP6 layers, is known to exist during the viral life cycle. This raises several questions: at what point during capsid assembly is VP4 added, and how does it maintain its relative stability and attachment to the triple-layer particle? The base of the VP4 spike has a diameter of ∼70 Å (Fig. 7A). At the surface of the VP6 layer, the type II channel has a diameter of ∼78 Å which narrows down to 50 Å, whereas at the VP7 layer, it has a diameter of ∼54 Å to ∼58 Å. Taken together, these dimensions suggest that VP4 may be physically anchored to the capsid such that the VP7 layer locks the VP4 spike within the type II channel. This also implies that the VP4 spikes must first be attached to the double-layer particle about the type II channels. VP7 trimers would be subsequently added during virus assembly, locking in the spike and forming the complete, stable triple-layer particle.

T=1 inner capsid layer.

As alluded to previously, the two T=13 outer capsid layers sit on top of a T=1 capsid layer formed by 120 copies of VP2 (25), with two VP2 molecules (designated VP2A and VP2B) in the icosahedral asymmetric unit. Although this layer is relatively thin, the boundaries between the VP2 subunits are clearly resolved in the cryoEM map (Fig. 10A).
SSEHunter (2) revealed the presence of 28 distinct α-helices and four β-sheets (Fig. 10B). While the α-helices are fairly well distributed in the structure, three of the four β-sheets are located in the lower portion of VP2. Based on these secondary structure elements and the overall appearance, VP2 (VP2A and VP2B) likely shares a common fold with the VP3 of bluetongue virus (17) and other reovirus inner capsid proteins (29, 37) consisting of the apical, carapace, and dimerization domains (Fig. 10C). This rough structural similarity, combined with sequence-based secondary structure prediction, allowed us to assign an overall topological model to VP2A and VP2B, utilizing the α-helices as “anchor” points for sequence-to-structure assignment (Table 1). Of particular interest is helix 1, an N-terminal helix not observed in other reovirus inner capsid protein structures, which protrudes toward the fivefold vertex and crosses over to an adjacent VP2 (helix 1 in Fig. 10C). At the end of helix 1, another helix-like density (helix 0 in Fig. 10D) can be seen crossing underneath VP2 and extending toward the fivefold vertex, where it may interact with the transcription enzyme complex (VP1, the RNA-dependent RNA polymerase, and VP3 capping enzyme). However, at this resolution, this helix cannot be unambiguously assigned to either VP2 subunit. No structural model for VP2 was proposed in the higher-resolution studies by Zhang et al. (45), and hence no comparison could be made with our VP2 model.
Additionally, interactions between VP2 and T=13 VP6 layers can be clearly resolved. VP6 appears to contact VP2 primarily through a single loop, D62 to L73 (green loop in Fig. 8D and ball-and-stick representations in Fig. 10A and B), though most of the interface with VP2 seems to be located at residues 69 to 73 of VP6. The VP6 trimers surrounding the type I channels and nearest to the fivefold axis (P in Fig. 10A and E) contact the apical domains of the adjacent VP2A and VP2B molecules. A second VP6 trimer, which is located at the icosahedral threefold axis and surrounding the type III channels (T in Fig. 10A), contacts only the VP2B. The remaining three types of trimers (Q, R, and S in Fig. 10A and E) surrounding the type II channels contact VP2A and/or VP2B. This pattern of interactions at various sites throughout VP2A and VP2B is likely due to the symmetry mismatch between the inner two capsid shells and result in an intricate series of interactions that stabilize the double-layer capsid particle.

Conclusion.

While the 9.5-Å resolution cryoEM structure of rotavirus falls short of the near-atomic resolution required to build a complete model, the integration of existing high-resolution structures and structural analysis tools has allowed us to dissect and analyze the structure in sufficient detail to understand its structural organization and interactions. In fact, this analysis has allowed us to propose topological models for both VP7 and VP2A/B, for which no previous structures were known. Furthermore, the analysis of VP4 has led to new insight into how the spike structure is organized and potentially functions during receptor recognition and cell entry. Beyond just the structural description of these capsid proteins, a complex network of interactions was observed that link the various structural proteins and provide the rigidity and flexibility required during the rotavirus life cycle. As such, this work represents the first “high-resolution” look into the structure and function of infectious rotavirus.
FIG. 1.
FIG. 1. Structure of mature rotavirus particle. (A) A representative region from a micrograph is shown. Individual particles (indicated in red circles) were extracted and processed. (B) The resulting 9.5-Å resolution reconstruction is shown. The capsid layers are radially colored such that red represents VP4 spikes, yellow is the VP7 layer, blue is the VP6 layer, green is the VP2 layer, and orange is the internal density (RNA and polymerase complex).
FIG. 2.
FIG. 2. Structure of the VP4 spike. (A) A VP4 spike segmented from the cryoEM map is shown. The domains of VP4 (VP8*, VP5*-t, bridge, and base domains) are indicated on the left. The asterisk indicates a breakpoint between the base and the bridge domains. An arrow indicates the bridging domain shown in Fig. 6. (B) Fitting of VP8* crystal structure (PDB ID: 1KQR) into the cryoEM density map shown in two orientations. A clear depression corresponding to the sialic acid binding site is visualized in the VP8* domain (denoted by a circle and shown in the inset image). The sialic acid molecule can be seen extending out of the density. (C) Fitting of the individual monomers from the VP5*-td crystal structure into the spike density. (D) Comparison of the fitting of the VP5*-td dimer as a whole (in cyan), VP5*-tt (in yellow; two of the three subunits in the trimer are shown), and the monomers from VP5*-td (red).
FIG. 3.
FIG. 3. Connecting VP8* and VP5*-t. (A) Fitted structures of VP8* and VP5*-t are shown in one of the dimeric subunits of the VP4 spike. The polypeptide chain is colored from the N (blue) to C (red) terminus. A difference map corresponding to the density unaccounted for by the VP8* and VP5-t crystal structures is shown in yellow. (B) The amino acid sequence of the linker region from residue 225 to 249 with the predicted β-strand in blue is shown. (C) This sequence was putatively modeled in the context of the difference map, shown in green (arrow in panel A and as a stereo view in panel C. This linker interacts with the other VP5*-t subunit (in gray) in the VP4 spike. (D) The interface between the VP8* and VP5*-t domains is shown in two views.
FIG. 4.
FIG. 4. Base domain of VP4. (A) The density corresponding to the base domain, which sits inside the type II channel, is shown. The base domain becomes visible at a radius of 305 Å and terminates at a radial distance of ∼375 Å. (B) Serial sections of the base domain, along its length from bottom to top (vertical line in panel A). The slices are contoured from high density (blue) to low density (red). The threefold correlation coefficient is also indicated. (C) Plot of threefold correlation coefficient over the entire base domain.
FIG. 5.
FIG. 5. Secondary structure elements in the base domain. Secondary structure elements, as detected using SSEHunter (2), are shown. Helices are indicated as green cylinders, while β-sheets are shown as cyan polygons. A clear threefold pattern is again evident.
FIG. 6.
FIG. 6. Bridging domain of VP4. (A) Fitting of the helix bundle from the trimeric VP5*-tt crystal structure to the bridging domain between the base and VP5*-t is shown. The helix bundle was fit both as a dimer (left) and as a trimer (right). While not unambiguous, the dimer appears to fit the density better. An arrow and an asterisk indicate density features illustrated in Fig. 2.
FIG. 7.
FIG. 7. VP4 model. (A) A side view of the type II channel shows that the base of VP4 is clearly wider than the surface opening in the VP7 layer. However, the trimeric VP4 base appears to rest in the VP6 layer. Interactions of VP4 with VP6 and VP7 can be seen. (B) Application of the bilateral filter revealed the presence of additional density (denoted by a red circle) that is contiguous with a small protrusion emanating from the base domain (indicated by an asterisk in Fig. 2A and 7A). This density could represent a portion of the possible third flexible VP4 subunit. The third subunit protrudes outward from the top of the base domain.
FIG. 8.
FIG. 8. Interactions between VP7 and VP6 layers. (A) Interacting VP7 and VP6 trimers, colored in yellow and blue, respectively, are shown in top and side views. (B) The crystal structure of VP6, shown in the left panel, was fit to the segmented T=7 layer subunit, revealing the boundaries between VP6 and VP7. Comparison of the SSEHunter results in VP6 was used to assess the accuracy of secondary structure identification in this region. In the right panel, the secondary structure elements identified with SSEHunter compare favorably to those from the X-ray structure. (C) The contact regions between VP6 and VP7 are shown in two views. Two interaction points between VP6 and VP7 subunits are indicated by arrows. (D) Secondary structure elements in VP7 identified using SSEHunter are shown along with the VP6 crystal structure. SSEHunter identified three helices, which are labeled H1, H2, and H4; numbering is based on the sequence-based predicted secondary structure described in “T=13 capsid layers” in the text.
FIG. 9.
FIG. 9. VP4 interactions with the VP7 and VP6. VP5* makes contacts with three of the six VP7 trimers, at a point indicated by an asterisk in Fig. 8B, surrounding the type II channel. (A) One interaction is between the C-terminal residues of VP5*-t and a VP7 trimer (indicated by the asterisk). (B) The other two interactions, with two other VP7 trimers, occur at the end of the bridging domain (indicated by the number sign and asterisk). (C and D) VP4 also contacts the VP6 layer through its base domain. (C) Only three of the six VP6 trimers, indicated by asterisks, surrounding the type II channel, show interactions with the base domain. (D) A side view of the type II channel depicting the VP4 interactions with both VP7 and VP6. (E) A zoomed-in view showing the points of interaction between VP4 and the fitted VP6 crystal structure. The VP6 X-ray structure is colored in magenta, corresponding to the color scheme in panel A; regions of the VP6 X-ray structure that contact VP4 are colored in gray.
FIG. 10.
FIG. 10. T=1 VP2 layer. (A) VP2A (pink) and VP2B (light green) subunits in the icosahedral asymmetric unit are shown. The VP6 trimers that sit atop VP2 subunits are indicated by triangles. The VP6 residues, as deduced from fitting of the VP6 crystal structure, that interact with VP2 are shown. (B) The SSEHunter results are shown for both VP2A and VP2B subunits. (C) Comparison of the secondary structure elements to the structural homologue in bluetongue virus (VP3, right) allowed for the assignment of the VP2 sequence to the visualized helices (Table 1). (D) Helix 0, indicated with a red circle, protrudes toward the fivefold vertex. However, which of the VP2 subunits that helix 0 is associated with could not be determined. (E) Zoomed-in view of the interactions between VP6 residues and the VP2 layer.
TABLE 1.
TABLE 1. Assignment of helixes, as indicated in Fig. 10C, to VP2 sequencea
Helix Position
0 S71-K79
1 E82-Q94
2 K103-D112
3 D163-E178
4 E205-F214
5 K220-V233
6 I249-L252
7 V259-R270
8 Y275-M279
9 F315-W319
10 L336-D353
11 S373-L388
12 V398-V414
13 M419-N434
14 F467-Q483
15 V485-L494
16 G501-S515
17 Y526-S538
18 L544-V569
19 K574-I589
20 T600-N606
21 Y615-R649
22 D660-R669
23 E676-M688
24 M690-G703
25 L736-R741
26 Y745-L753
27 S770-K777
28 A781-L790

Acknowledgments

We acknowledge the use of cryoEM facilities at the National Center for Macromolecular Imaging at Baylor College of Medicine.
The National Center for Macromolecular Imaging at Baylor College of Medicine is supported by National Institutes of Health (NIH) grant RR002250 to W. Chiu. We also acknowledge support from NIH grants AI36040 (to B.V.V.P.) and DK30144 (to M.K.E.), National Science Foundation grant IIS-0705474 (to M.L.B.), and the Robert Welch Foundation (to B.V.V.P.)

REFERENCES

1.
Arias, C. F., P. Romero, V. Alvarez, and S. Lopez. 1996. Trypsin activation pathway of rotavirus infectivity. J. Virol.70:5832-5839.
2.
Baker, M. L., T. Ju, and W. Chiu. 2007. Identification of secondary structure elements in intermediate-resolution density maps. Structure15:7-19.
3.
Clarke, M. L., L. J. Lockett, and G. W. Both. 1995. Membrane binding and endoplasmic reticulum retention sequences of rotavirus VP7 are distinct: role of carboxy-terminal and other residues in membrane binding. J. Virol.69:6473-6478.
4.
Coulson, B. S., S. L. Londrigan, and D. J. Lee. 1997. Rotavirus contains integrin ligand sequences and a disintegrin-like domain that are implicated in virus entry into cells. Proc. Natl. Acad. Sci. USA94:5389-5394.
5.
Crawford, S. E., S. K. Mukherjee, M. K. Estes, J. A. Lawton, A. L. Shaw, R. F. Ramig, and B. V. Prasad. 2001. Trypsin cleavage stabilizes the rotavirus VP4 spike. J. Virol.75:6052-6061.
6.
Denisova, E., W. Dowling, R. LaMonica, R. Shaw, S. Scarlata, F. Ruggeri, and E. R. Mackow. 1999. Rotavirus capsid protein VP5* permeabilizes membranes. J. Virol.73:3147-3153.
7.
Dormitzer, P. R. 2008. Rotavirus cell entry, p. 189-214. In J. Patton (ed.), Segmented double-stranded RNA viruses—structure and molecular biology. Caister/Academic Press, New York, NY.
8.
Dormitzer, P. R., E. B. Nason, B. V. Prasad, and S. C. Harrison. 2004. Structural rearrangements in the membrane penetration protein of a non-enveloped virus. Nature430:1053-1058.
9.
Dormitzer, P. R., Z. Y. Sun, G. Wagner, and S. C. Harrison. 2002. The rhesus rotavirus VP4 sialic acid binding domain has a galectin fold with a novel carbohydrate binding site. EMBO J.21:885-897.
10.
Dowling, W., E. Denisova, R. LaMonica, and E. R. Mackow. 2000. Selective membrane permeabilization by the rotavirus VP5* protein is abrogated by mutations in an internal hydrophobic domain. J. Virol.74:6368-6376.
11.
Emsley, P., and K. Cowtan. 2004. Coot: model-building tools for molecular graphics. Acta Crystallogr. D60:2126-2132.
12.
Estes, M. K., D. Y. Graham, and B. B. Mason. 1981. Proteolytic enhancement of rotavirus infectivity: molecular mechanisms. J. Virol.39:879-888.
13.
Estes, M. K., G. Kang, C. Q. Zeng, S. E. Crawford, and M. Ciarlet. 2001. Pathogenesis of rotavirus gastroenteritis. Novartis Found. Symp.238:82-96.
14.
Fukuhara, N., O. Yoshie, S. Kitaoka, and T. Konno. 1988. Role of VP3 in human rotavirus internalization after target cell attachment via VP7. J. Virol.62:2209-2218.
15.
Gilbert, J. M., and H. B. Greenberg. 1998. Cleavage of rhesus rotavirus VP4 after arginine 247 is essential for rotavirus-like particle-induced fusion from without. J. Virol.72:5323-5327.
16.
Goddard, T. D., C. C. Huang, and T. E. Ferrin. 2007. Visualizing density maps with UCSF Chimera. J. Struct. Biol.157:281-287.
17.
Grimes, J. M., J. N. Burroughs, P. Gouet, J. M. Diprose, R. Malby, S. Zientara, P. C. P. Mertens, and D. I. Stuart. 1998. The atomic structure of the bluetongue virus core. Nature395:470-478.
18.
Guerrero, C. A., E. Mendez, S. Zarate, P. Isa, S. Lopez, and C. F. Arias. 2000. Integrin alpha (v) beta(3) mediates rotavirus cell entry. Proc. Natl. Acad. Sci. USA97:14644-14649.
19.
Hewish, M. J., Y. Takada, and B. S. Coulson. 2000. Integrins alpha2beta1 and alpha4beta1 can mediate SA11 rotavirus attachment and entry into cells. J. Virol.74:228-236.
20.
Jiang, W., M. L. Baker, S. J. Ludtke, and W. Chiu. 2001. Bridging the information gap: computational tools for intermediate resolution structure interpretation. J. Mol. Biol.308:1033-1044.
21.
Jiang, W., M. L. Baker, Q. Wu, C. Bajaj, and W. Chiu. 2003. Applications of a bilateral denoising filter in biological electron microscopy. J. Struct. Biol.144:114-122.
22.
Jiang, W., Z. Li, Z. Zhang, C. R. Booth, M. L. Baker, and W. Chiu. 2001. Semi-automated icosahedral particle reconstruction at sub-nanometer resolution. J. Struct. Biol.136:214-225.
23.
Kaljot, K. T., R. D. Shaw, D. H. Rubin, and H. B. Greenberg. 1988. Infectious rotavirus enters cells by direct cell membrane penetration, not by endocytosis. J. Virol.62:1136-1144.
24.
Kapikian, A. Z. 2001. A rotavirus vaccine for prevention of severe diarrhoea of infants and young children: development, utilization and withdrawal. Novartis Found. Symp.238:153-171. (Discussion, 238:171-179.)
25.
Lawton, J. A., C. Q. Zeng, S. K. Mukherjee, J. Cohen, M. K. Estes, and B. V. Prasad. 1997. Three-dimensional structural analysis of recombinant rotavirus-like particles with intact and amino-terminal-deleted VP2: implications for the architecture of the VP2 capsid layer. J. Virol.71:7353-7360.
26.
Ludtke, S. J., P. R. Baldwin, and W. Chiu. 1999. EMAN: semiautomated software for high-resolution single-particle reconstructions. J. Struct. Biol.128:82-97.
27.
Mackow, E. R., J. W. Barnett, H. Chan, and H. B. Greenberg. 1989. The rhesus rotavirus outer capsid protein VP4 functions as a hemagglutinin and is antigenically conserved when expressed by a baculovirus recombinant. J. Virol.63:1661-1668.
28.
Mathieu, M., I. Petitpas, J. Navaza, J. Lepault, E. Kohli, P. Pothier, B. V. Prasad, J. Cohen, and F. A. Rey. 2001. Atomic structure of the major capsid protein of rotavirus: implications for the architecture of the virion. EMBO J.20:1485-1497.
29.
Nakagawa, A., N. Miyazaki, J. Taka, H. Naitow, A. Ogawa, Z. Fujimoto, H. Mizuno, T. Higashi, Y. Watanabe, T. Omura, R. H. Cheng, and T. Tsukihara. 2003. The atomic structure of rice dwarf virus reveals the self-assembly mechanism of component proteins. Structure11:1227-1238.
30.
Offit, P. A., and G. Blavat. 1986. Identification of the two rotavirus genes determining neutralization specificities. J. Virol.57:376-378.
31.
Offit, P. A., G. Blavat, H. B. Greenberg, and H. F. Clark. 1986. Molecular basis of rotavirus virulence: role of gene segment 4. J. Virol.57:46-49.
32.
Pesavento, J. B., S. E. Crawford, E. Roberts, M. K. Estes, and B. V. Prasad. 2005. pH-induced conformational change of the rotavirus VP4 spike: implications for cell entry and antibody neutralization. J. Virol.79:8572-8580.
33.
Pesavento, J. B., J. A. Lawton, M. E. Estes, and B. V. Prasad. 2001. The reversible condensation and expansion of the rotavirus genome. Proc. Natl. Acad. Sci. USA98:1381-1386.
34.
Prasad, B. V., J. W. Burns, E. Marietta, M. K. Estes, and W. Chiu. 1990. Localization of VP4 neutralization sites in rotavirus by three-dimensional cryo-electron microscopy. Nature343:476-479.
35.
Prasad, B. V., R. Rothnagel, C. Q. Zeng, J. Jakana, J. A. Lawton, W. Chiu, and M. K. Estes. 1996. Visualization of ordered genomic RNA and localization of transcriptional complexes in rotavirus. Nature382:471-473.
36.
Prasad, B. V., G. J. Wang, J. P. Clerx, and W. Chiu. 1988. Three-dimensional structure of rotavirus. J. Mol. Biol.199:269-275.
37.
Reinisch, K. M., M. L. Nibert, and S. C. Harrison. 2000. Structure of the reovirus core at 3.6 A resolution. Nature404:960-967.
38.
Sabara, M., A. Barrington, and L. A. Babiuk. 1985. Immunogenicity of a bovine rotavirus glycoprotein fragment. J. Virol.56:1037-1040.
39.
Shaw, A. L., R. Rothnagel, D. Chen, R. F. Ramig, W. Chiu, and B. V. Prasad. 1993. Three-dimensional visualization of the rotavirus hemagglutinin structure. Cell74:693-701.
40.
Tihova, M., K. A. Dryden, A. R. Bellamy, H. B. Greenberg, and M. Yeager. 2001. Localization of membrane permeabilization and receptor binding sites on the VP4 hemagglutinin of rotavirus: implications for cell entry. J. Mol. Biol.314:985-992.
41.
Yeager, M., J. A. Berriman, T. S. Baker, and A. R. Bellamy. 1994. Three-dimensional structure of the rotavirus haemagglutinin VP4 by cryo-electron microscopy and difference map analysis. EMBO J.13:1011-1018.
42.
Yeager, M., K. A. Dryden, N. H. Olson, H. B. Greenberg, and T. S. Baker. 1990. Three-dimensional structure of rhesus rotavirus by cryoelectron microscopy and image reconstruction. J. Cell Biol.110:2133-2144.
43.
Yoder, J. D., and P. R. Dormitzer. 2006. Alternative intermolecular contacts underlie the rotavirus VP5* two- to three-fold rearrangement. EMBO J.25:1559-1568.
44.
Zarate, S., R. Espinosa, P. Romero, C. A. Guerrero, C. F. Arias, and S. Lopez. 2000. Integrin alpha2beta1 mediates the cell attachment of the rotavirus neuraminidase-resistant variant nar3. Virology278:50-54.
45.
Zhang, X., E. Settembre, C. Xu, P. R. Dormitzer, R. Bellamy, S. C. Harrison, and N. Grigorieff. 2008. Near-atomic resolution using electron cryomicroscopy and single-particle reconstruction. Proc. Natl. Acad. Sci. USA105:1867-1872.
46.
Zhou, Z. H., M. L. Baker, W. Jiang, M. Dougherty, J. Jakana, G. Dong, G. Lu, and W. Chiu. 2001. Electron cryomicroscopy and bioinformatics suggest protein fold models for rice dwarf virus. Nat. Struct. Biol.8:868-873.

Information & Contributors

Information

Published In

cover image Journal of Virology
Journal of Virology
Volume 83Number 415 February 2009
Pages: 1754 - 1766
PubMed: 19036817

History

Received: 3 September 2008
Accepted: 18 November 2008
Published online: 15 February 2009

Permissions

Request permissions for this article.

Contributors

Authors

Zongli Li
National Center for Macromolecular Imaging
Present address: Harvard Medical School, Department of Cell Biology, Boston, MA 02115.; § Present address: Markey Center for Structural Biology, Department of Biological Sciences, Purdue University, West Lafayette, IN 47907.
Matthew L. Baker
National Center for Macromolecular Imaging
Wen Jiang
National Center for Macromolecular Imaging
Present address: Harvard Medical School, Department of Cell Biology, Boston, MA 02115.; § Present address: Markey Center for Structural Biology, Department of Biological Sciences, Purdue University, West Lafayette, IN 47907.
Mary K. Estes
Department of Molecular Virology and Microbiology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030
B. V. Venkataram Prasad [email protected]
Verna and Marrs Mclean Department of Biochemistry and Molecular Biology

Metrics & Citations

Metrics

Note:

  • For recently published articles, the TOTAL download count will appear as zero until a new month starts.
  • There is a 3- to 4-day delay in article usage, so article usage will not appear immediately after publication.
  • Citation counts come from the Crossref Cited by service.

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. For an editable text file, please select Medlars format which will download as a .txt file. Simply select your manager software from the list below and click Download.

View Options

Figures and Media

Figures

Media

Tables

Share

Share

Share the article link

Share with email

Email a colleague

Share on social media

American Society for Microbiology ("ASM") is committed to maintaining your confidence and trust with respect to the information we collect from you on websites owned and operated by ASM ("ASM Web Sites") and other sources. This Privacy Policy sets forth the information we collect about you, how we use this information and the choices you have about how we use such information.
FIND OUT MORE about the privacy policy