Structure of the Recombinant Alphavirus Western Equine Encephalitis Virus Revealed by Cryoelectron Microscopy

Abstract

Western equine encephalitis virus (WEEV; Togaviridae, Alphavirus) is an enveloped RNA virus that is typically transmitted to vertebrate hosts by infected mosquitoes. WEEV is an important cause of viral encephalitis in humans and horses in the Americas, and infection results in a range of disease, from mild flu-like illnesses to encephalitis, coma, and death. In addition to spreading via mosquito vectors, human WEEV infections can potentially occur directly via aerosol transmission. Due to its aerosol infectivity and virulence, WEEV is thus classified as a biological safety level 3 (BSL-3) agent. Because of its highly infectious nature and containment requirements, it has not been possible to investigate WEEV's structure or assembly mechanism using standard structural biology techniques. Thus, to image WEEV and other BSL-3 agents, we have constructed a first-of-its-kind BSL-3 cryoelectron microscopy (cryoEM) containment facility. cryoEM images of WEEV were used to determine the first three-dimensional structure of this important human pathogen. The overall organization of WEEV is similar to those of other alphaviruses, consistent with the high sequence similarity among alphavirus structural proteins. Surprisingly, the nucleocapsid of WEEV, a New World virus, is more similar to the Old World alphavirus Sindbis virus than to other New World alphaviruses.

The alphaviruses comprise a genus of single-stranded, plus-sense, enveloped RNA viruses that, together with rubella virus, comprise the family Togaviridae. The current classification of the genus Alphavirus includes 29 different species, with multiple subtypes and/or varieties represented within some species (30). These species can be grouped into 8 different complexes based on antigenic and/or genetic similarities (20). Most viruses from the New World are found in the Eastern, Venezuelan, and Western equine encephalitis (EEE, VEE, and WEE, respectively) complexes and cause encephalitis in humans and a variety of domesticated animals. Old World alphaviruses, on the other hand, typically cause only an arthralgia and rash syndrome that is rarely life threatening (5, 24). Among the New World alphaviruses, EEE, VEE, and WEE viruses (EEEV, VEEV, and WEEV, respectively) are potential biological weapons as well as naturally emerging pathogens and are therefore included on the category B Priority Pathogens list of the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (http://www.niaid.nih.gov/topics/biodefenserelated/biodefense/research/pages/cata.aspx).

Alphaviruses replicate in the cytoplasm of infected cells after entry via receptor-mediated endocytosis (8). Following internalization, fusion of the viral envelope with the endocytic membrane is mediated by a low-pH-induced conformational change that exposes a fusion peptide found in the E1 envelope glycoprotein. The nucleocapsid then disassembles upon interactions with ribosomes, and an open reading frame (ORF) found in the 5′ two-thirds of the genome is translated. The resultant polyprotein is cleaved into 4 nonstructural proteins (nsP1 to -4) that mediate viral RNA replication, RNA capping, and polyprotein processing (Fig. 1). The structural proteins, including the two envelope glycoproteins E2 and E1 as well as the capsid protein, are encoded in a second ORF that is translated from a subgenomic message often referred to as 26S RNA. Following auto-cleavage of the capsid protein in the cytoplasm, the remaining polyprotein is inserted into the endoplasmic reticulum, where it is cleaved by host cell proteases and then processed through the secretory pathway, where the glycosylation of E2 and E1 occurs. Virion maturation occurs after E2/E1 heterodimers are inserted into the plasma membrane and 240 copies of the capsid protein interact with one copy of the genomic RNA to form nucleocapsids. These nucleocapsids then interact with a cytoplasmic domain of the E2 protein to initiate budding. The mature virion thus includes 240 copies of the capsid protein and 240 E2/E1 heterodimers arranged as trimeric spikes on the surface of the virus (8).

FIG. 1.

Diagram of the alphavirus genome, showing the 5′ cap, 5′ untranslated region, nonstructural polyprotein open reading frame, and major functions of the individual proteins, subgenomic promoter, structural polyprotein open reading frame, 3′ untranslated region, and poly(A) tail.

The structures of several different alphaviruses, including Sindbis virus (SINV) (13), Ross River virus (RRV) (3, 35), Semliki Forest virus (SFV), (11), and VEEV (16), have been solved to subnanometer resolution using cryoelectron microscopy (cryoEM), and the X-ray crystallographic structure of the E1 protein from Semliki Forest virus has been determined to atomic resolution (9). The alphaviruses are ca. 700 Å in diameter, with 80 trimeric spikes on their surfaces. By fitting the E1 crystal structure into cryoEM reconstruction maps of whole viruses, the orientations of both envelope proteins within the spikes have been estimated (36). The E1 and E2 proteins are similar in shape, and the E2 proteins extend to the tips of the spikes, where most glycosylation and antibody-binding sites have been mapped (13). The underlying T=4 icosahedral capsid is constructed from regularly ordered capsomers arranged as hexons and pentons. These pentons and hexons consist of capsid protein monomers that apparently represent only the C-terminal half of the protein. Crystal structures of alphavirus capsid proteins also indicate that only the C terminus, including the protease domain, is ordered (25). cryoEM reconstructions of VEEV nucleocapsids isolated from virions have a less ordered structure, with density redistributed from the 3-fold to the 5-fold axis, suggesting that the envelope and/or the envelope glycoproteins constrain and stabilize the nucleocapsid in a compressed structure (15). Additionally, the VEEV nucleocapsids within viruses differ from those of Old World alphaviruses, with a counterclockwise rotation of the pentameric and hexameric capsomers in VEEV (16). Similar differences were observed in the capsid of Aura virus (AURAV), another New World alphavirus (34).

In addition to being an important human and equine pathogen, WEEV is one of three alphaviruses that descended from a recombinant ancestor (6, 31). This ancestor derived its nonstructural and capsid protein genes from an ancestral EEEV strain, whereas its envelope glycoprotein genes were provided from an ancestral SINV. The recombination event was apparently followed by compensatory mutations in the cytoplasmic domain of the E2 protein that restored efficient interactions with the EEEV-like capsid protein (6). If this interpretation of the WEEV ancestral recombination event is correct, its nucleocapsids, constructed from capsid proteins derived from the New World EEEV ancestor, would be expected be more similar to those of the New World VEEV than to those of the Old World SINV, RRV, and SFV. To test this hypothesis and to investigate other structural features of interest related to its recombinant history and pathogenicity, we determined the structure of WEEV to a 13-Å resolution using cryoEM image reconstruction.

Nucleotide sequence accession number.

The WEEV map has been deposited in the EmDep database under the accession number EMD-5210.

MATERIALS AND METHODS

Virus production and purification.

All virus manipulations were carried out in biosafety level 3 (BSL-3) laboratories using recommended procedures (26). WEEV strain CO92-1356, isolated from Culex tarsalis mosquitoes in Larimer County, CO, on 30 July 1992, was purified and examined microscopically. The virus was passaged once in Vero cells and then used to infect baby hamster kidney (BHK-21) cells at a multiplicity of ca. 1 PFU/cell. Following the appearance of cytopathic effects, the cell culture medium containing 2% fetal bovine serum was harvested and clarified by centrifugation at 2,000 × g for 10 min. Next, polyethylene glycol and NaCl were added to 7% and 2.3% (wt/vol) concentrations, respectively, and the virus was allowed to precipitate overnight at 4°C. Following centrifugation at 4,000 × g for 30 min at 4°C, the pelleted precipitate was resuspended in TEN buffer (0.05 M Tris-HCl [pH 7.4], 0.1 M NaCl, 0.001 M EDTA) and loaded onto a 20-to-70% continuous sucrose (wt/vol) gradient in TEN buffer. Following centrifugation at 270,000 × g for 1 h, the visible virus band was harvested using a Pasteur pipette and centrifuged 4 times through an Amicon Ultra-4 100-kDa-cutoff filter (Millipore, Billerica, MA), with resuspension in 4 ml of TEN buffer at each step to remove residual sucrose.

Grid preparation and freezing.

All manipulations with WEEV, including sample vitrification, microscopy, and grid transfers, were performed using personal protection equipment (PPE), including a powered air-purifying respirator (PAPR), surgical gloves, and a seamless gown. In addition, the preparation of frozen grids was done inside a class II biosafety cabinet.

Purified WEEV was transferred to a unique, BSL-3 cryoelectron microscopy facility described recently (22). The viruses were then vitrified as reported previously (4, 23) on holey carbon film grids (R2x2 Quantifoil; Micro Tools GmbH, Jena, Germany; or C-flat; Protochips, Raleigh, NC). Briefly, purified concentrated suspensions of viruses were applied to the holey films in a volume of ca. 3.5 μl, blotted with filter paper, and plunged into liquid ethane cooled in a liquid nitrogen bath. Frozen grids were stored under liquid nitrogen and transferred to a cryo-specimen 626 holder (Gatan, Inc., Pleasanton, CA) under liquid nitrogen before being loaded into a JEOL 2200FS electron microscope, equipped with an in-column energy filter (omega type) and a field emission gun (FEG) operating at 200 keV. Grids were maintained at near-liquid nitrogen temperature (−172 to −180°C) during imaging.

Particle imaging.

Imaging was carried out with the operator wearing all PPE either inside the containment laboratory or outside the containment lab using SIRIUS remote microscope control software (JEOL Inc.) through a computer network.

Viruses were imaged at a ×50,000 nominal magnification either with a 4,000- by 4,000-pixel slow-scan charge-coupled device (CCD) camera (UltraScan 895; Gatan, Inc., Pleasanton, CA) or with photographic film (SO-163; Kodak, Rochester, NY) and a low-dose imaging procedure. Images were acquired with an ∼20-electron/Ų dose; the CCD pixel size corresponded to 2.1 Å on the specimen scale. Images acquired on film were digitized using a Nikon Coolscan 9000 scanner with a step size of 1.3 Å/pixel. We used a 0.4- to 1.85-μm defocus range for imaging. An in-column omega electron energy filter was used during imaging, with a zero-loss electron energy peak selected with a 20-eV slit. The electron energy filter was essential for data acquisition, allowing high-contrast imaging at close to focus conditions. It substantially reduced the effect of chromatic aberrations of the electron optics as well as improved the quality of the images.

Overall, 7,452 individual virus images from 49 micrographs were processed, with 4,586 particles used in the final reconstruction.

Image processing.

Individual virus images were boxed from micrographs using the BOXER program from the EMAN suite (10). Icosahedral image processing software (1) was used for contrast transfer function (CTF) determination, particle alignment via projection matching, and three-dimensional (3D) reconstruction using Fourier-based methods. Selected images of individual WEEVs were normalized to the same mean density and standard deviation values and corrected for the contrast transfer function (CTF) of the microscope. The interactive graphics program RobEM (http://cryoem.ucsd.edu/programs.shtm) was used to estimate the defocus and other characteristics of the CTF for each CCD frame or film image. CTF correction of images with Wiener-type filtering (2, 32) was performed using a stand-alone program, CTFCOR (23), prior to performance of orientation/origin searches or refinement.

An initial model of the virus was obtained using the IMAGIC-5 package (28). A subset of WEEV particle images was centered, and the angular reconstitution technique (27) was then used to determine the relative orientations of the images; these orientations were used to calculate an initial 3D reconstruction. The resulting 3D map was used as an initial model for a projection-matching-based orientation search and first alignment of the images (1) (PFTSEARCH program; http://cryoem.ucsd.edu/programs.shtm). Subsequent refinement of initial orientation/origin parameters was done using a modified version of the PFT protocol (33). Iterative origin/orientation refinement cycles were performed until they converged. At each iteration step, all images were aligned relative to the projections of the 3D map obtained in previous search/refinement cycles. The best images (showing the highest correlations with reference map projections) were then selected to calculate a new 3D map, which was used as the reference for the next round of refinement. The final 3D map was reconstructed from the 4,586 best virus images. Particles with sizes different from that of the reference map were excluded from the reconstructions based on lower correlation with the reference. The effective resolution of the map was 13 Å according to a 0.5 Fourier shell correlation function (FSC) between maps calculated from independent half data sets (Fig. 2) (21). 3D density maps were calculated using P3DR (7) or I3DR (14), and PSF (7) was used to calculate the FSC. The 3D maps were surface rendered and displayed with a 1-standard-deviation (1σ) threshold in RobEM and CHIMERA (18), which accounted for ∼100% of the particle volume.

FIG. 2.

Resolution curve showing Fourier shell correlations calculated using two independent half data sets of the WEEV images. Resolution of the WEEV reconstruction was 13 Å, based on a threshold of 0.5.

To validate the rotation of the WEEV capsomers relative to those of other New World viruses (VEEV and AURAV), we used an AURAV 3D map (34) as the initial model and then PFTSEARCH and refinement of origin/orientations to obtain independent reconstructions.

RESULTS

Structure of WEEV.

Individual virus particles in the WEEV sample exhibited significant structural heterogeneity regardless of the purification methods used (density fractionation and gradient purification). The largest fraction of similarly sized virus images was thus selected for image analysis/reconstruction. A typical EM field showing vitrified WEEV is shown in Fig. 3. Virus sizes ranged from ∼660 to ∼760 Å.

FIG. 3.

WEE viruses embedded in vitreous ice. The particles were ∼700 Å in diameter, with surface protrusions clearly visible. There were aberrant WEEVs present in the sample (encircled); some particles were of drastically different sizes (black arrows point to very small viruses, while the white arrow shows a large broken WEEV). Insets show aberrant viruses with several nucleocapsids within common envelopes.

Figure 4 shows the 3D cryoEM reconstruction of WEEV. Particles were ∼700 Å in diameter and exhibited T=4 quasi-symmetry within the icosahedral lattice on the virus surface. As in other alphaviruses, the E1 and E2 proteins formed heterodimers that in turn oligomerized into 80 prominent trimeric spikes protruding perpendicularly to the virus surface. The spikes were ca. 88 Å high and 120 Å in diameter. In Fig. 4B, the front half of the virus is removed to show the density distribution within the virus. The outer glycoprotein shell (gold in color) of the virus was formed by the E1/E2 glycoproteins; the E1 proteins were oriented almost tangentially to the surface. Their positions and orientations in the map were derived using a homology model based on the E1 structure from Semliki Forest virus (9) and were similar to those of E1 in SINV (36). The rest of the spikes' density was attributed to E2, forming the “stem” of the spikes. The E2 protein apparently interacts with cell receptors and conceals the fusogenic peptide of E1 (8). The glycoprotein shell was separated from the nucleocapsid core by a lipid bilayer derived from the host cell (Fig. 4B, green). The 42-Å-thick membrane was ca. 230 Å from the particle center (Fig. 4B). The E2 cytoplasmic domain penetrated through the bilayer and appeared to bind the capsid protein, organizing the latter into well-ordered icosahedral particles with T=4 quasi-symmetry; in contrast, nascent alphavirus capsids have been shown to be plastic (12). The nucleocapsid core had pronounced capsomers protruding by as much as 40 Å from its surface. Hexons were skewed (distorted from local 6-fold symmetry), with a diameter of ∼116 Å; pentons were 104 Å across. The capsomers arose from a continuous layer of density positioned at a radial distance of ∼154 Å from the particle center. Inside the capsid shell, the organization of the viral RNA was not resolved in the map, presumably due to its lack of icosahedral symmetry.

FIG. 4.

(A) 3D map of WEEV, with the 2-fold axis oriented perpendicularly to the page. Six trimer spikes are clearly visible. Green layers visible through the holes in the outer glycoprotein layer represent the outer leaflet of the lipid membrane separating the glycoprotein shell from the inner nucleocapsid. (B) The front half of the map is removed to reveal a radial arrangement of the virus particle. Virus RNA (vRNA, red) occupies the center of the map and extends to ca. 150 Å radially. The capsid protein shell starts at ca. 154 Å, where the N terminus interacts with vRNA, and extends to ca. 230 Å, adjacent to the inner leaflet of the lipid membrane separating the capsid and the outer glycoprotein shell. The membrane is ca. 42 Å in width. The outer shell and spikes protruding outwards from it are formed by E1 and E2 glycoproteins. The overall size of the virus particle is ca. 700 Å. (C) WEEV nucleocapsid cut out from the map, demonstrating hexons and pentons protruding from its surface. Cytoplasmic tails of the E2 glycoproteins protrude through the lipid bilayer and interact with the capsid protein, organizing the latter into a well-ordered icosahedral shell with T=4 quasi-symmetry.

Comparison to other alphaviruses.

The overall structure of WEEV was similar to that of other alphaviruses that have been studied by cryoEM (SINV, VEEV strain TC-83, RRV, SFV, and AURAV), although details of the glycoprotein envelope and the inner nucleocapsid differed. For example, WEEV lacked the characteristic nodules at the tips of the trimeric spikes of E1/E2 heterodimers that are present in VEEV and SINV maps (16, 36).

To further compare WEEV with the VEEV and SINV 3D maps, we calculated differences among them using RobEM (http://cryoem.ucsd.edu/programs.shtm) or IMAGIC-5 (28). The maps were pair-wise scaled relative to each other both in size (to equalize pixel sizes) and in density. In Fig. 5 A, a WEEV spike is shown with the difference map (WEEV minus VEEV) superimposed. Figure 5B presents the same spike rotated by 90°, with the front part removed to demonstrate the radial positions of the differences. Positive difference density (features present in the WEEV structure but not in VEEV) is shown in red and negative difference density (features present in VEEV but absent in WEEV) in blue. The major differences in the glycoprotein shells were at the tips of the spikes (red and arrows in Fig. 5B), in the cavities below the E2 portion of the spikes (Fig. 5A and B, blue), and in the capsid region. The differences in the spikes could, in part, be attributed to differences in E2 glycosylation sites (VEEV has 3 N-linked sites, while WEEV has only 2) and to differences in E2 amino acid sequences between WEEV and VEEV, resulting in differences in their structures (Table 1). The capsomers in the nucleocapsids (WEEV versus VEEV) were slightly rotated relative to each other (∼7°) but otherwise were quite similar. In Fig. 5C, a WEEV hexon is shown with one vertex labeled with an asterisk. Figure 5D shows the corresponding VEEV hexon, and Fig. 5E demonstrates the superposition of both maps. Their relative rotations are indicated by blue (for WEEV) and cyan (for VEEV) lines running through centers of corresponding capsomers and their vertices. The superposition of pentons from WEEV (brown) and VEEV (magenta) is shown in Fig. 5F. Except for the rotation, the capsomers were quite similar in both maps. To avoid a model bias and to validate the rotation of the WEEV capsomers relative to those of VEEV, we aligned the images and determined their orientations using the AURAV map (34) as an initial model. Subsequent calculation of the WEEV 3D reconstruction from these images demonstrated a rotation of the capsomers that was the same as those of VEEV and AURAV (data not shown).

FIG. 5.

WEEV versus VEEV. (A) A WEEV trimeric spike cut from the WEEV map is superimposed with the WEEV-VEEV difference map. (B) Same as panel A but rotated by 90°, with the front part removed. Positive differences of >2σ are shown in red, and negative differences of the same amplitude are in blue. The major differences were at the tips of the spike (red blobs, arrows), in the cavities below the E2 part of the spikes (blue). (C) Hexon cut from the WEEV capsid, with one of the vertices indicated by an asterisk. (D) Hexon from the VEEV capsid, with a vertex indicated by an asterisk. (E) Superposition of two hexons, showing their slight rotation relative to each other. Relative orientations are indicated by blue (WEEV) and magenta (VEEV) lines running through capsomer centers and their vertices. (F) Similar to panel E but with pentons superimposed.

TABLE 1.

Percentages of amino acids among alphavirus structural proteins that are identical to those of WEEV

Structural protein	Virus	% of amino acids that are identical in:
		AURAV	VEEV	SINV	SFV
Capsid	WEEV	46	65	51	50
	AURAV		44	67	50
	VEEV			46	50
	SINV				53
E2 envelope	WEEV	46	40	68	36
glycoprotein	AURAV		39	56	36
	VEEV			40	36
	SINV				39
E1 envelope	WEEV	59	49	76	48
glycoprotein	AURAV		48	61	47
	VEEV			50	50
	SINV				48

The outer shells of WEEV and SINV were quite similar except for the tips of the spikes; SINV had very distinct extensions, whereas WEEV had flat tips. This difference cannot be attributed to different glycosylation sites (13, 19, 35, 36) because the corresponding E2 proteins of both viruses have potential N-linked sites at residues 196 and 318 and most likely originated from variations in the amino acid sequences of E2 glycoproteins (Table 1). The differences between WEEV and SINV occurred mainly within part of the spike corresponding to E2 glycoprotein (Fig. 6, with both positive peaks in red and negative peaks in blue). The E2 protein of WEEV may differ from that of SINV due to its C-terminal contacts with the capsid, which appeared different from those observed in SINV. Overall, the orientations of the E1 and E2 glycoproteins were similar. There were no strong densities observed in the region of the difference map corresponding to E1 glycoprotein, suggesting no major differences between the E1 conformations in WEEV and SINV.

FIG. 6.

WEEV versus SINV. (A and B) Similar to Fig. 4A and B but with an SINV difference map superimposed. Major differences are at the tips of the spikes (red and arrows in panel B), inside the spike (blue), and at the base of the spike (red) close to the holes at the 2-fold axes. (C and D) Capsids in both maps are similar; the difference peaks largest in size were at the 2-fold-symmetry axes, where in WEEV there was a density, while in the SINV map, the space was empty. (C) WEEV hexon with the WEEV-SINV difference map superimposed. The density in the center of the capsomer (red) is present in WEEV but absent in the SINV map. Smaller red and blue peaks are concentrated at the capsid protein subunit interfaces and indicate possible differences in capsid organization between WEEV and SINV. (D) The pentons are quite similar in both maps.

Surprisingly, there were no differences in the orientations of the capsomers despite WEEV deriving its capsid protein from the New World EEEV and SINV, which belongs to an Old World clade (16). There were, however, structural differences between WEEV and SINV nucleocapsids; there was no density at the 2-fold-symmetry axes in the SINV map, while WEEV exhibited substantial density at those locations. Smaller peaks in the difference map occurred at the subunit interface, suggesting differences in capsid protein folding between WEEV and SINV (Fig. 6C and D). The inner parts of the nucleocapsids were not resolved well enough to identify distinct differences between the maps.

DISCUSSION

We found the stoichiometric ratios and spatial organization of WEEV proteins to be identical to those of other alphaviruses studied by EM, further substantiating evolutionary links among members of the genus. Both the structural and nonstructural proteins of WEEV are similar in sequence to those of other members of the Alphavirus genus, and thus WEEV was not expected to differ significantly in structure from other alphaviruses studied earlier by cryoEM (SINV, VEEV, RRV, SFV, and AURAV). Because it originated as a hybrid or chimeric virus from two distinct alphavirus lineages, namely, ancestral EEEV and SINV, it was not surprising to find substantial similarities in the 3D reconstructions of WEEV, SINV, VEEV, and AURAV. All alphaviruses studied so far by electron microscopy exhibit similar features and differ primarily in fine details only. These details, however, may reflect variation in amino acid sequences in the proteins as well as differences in the glycosylation states of envelope glycoproteins. These differences may mediate host range and pathogenicity differences among the alphaviruses.

Earlier studies indicated that at least some New and Old World alphaviruses differ in the relative orientations of the capsomers within their nucleocapsids (17, 34) while maintaining similar appearances in their glycoprotein shells. Despite the overall high levels of amino acid sequence similarity among structural alphavirus proteins (>35% amino acid sequence identity among the mosquito-borne alphaviruses [Table 1]), the WEEV glycoproteins are more closely related to those of SINV than to those of VEEV. However, the WEEV nucleocapsid protein sequence is more similar to that of VEEV. Therefore, we expected to observe some rotation of the WEEV capsomers relative to those of SINV and a very similar organization of the WEEV capsid relative to that of VEEV. To our surprise, nucleocapsids of WEEV and SINV, as well as their outer envelopes, were quite similar, without any significant differences in orientation, while WEEV, VEEV, and AURAV capsids differed markedly in capsomer orientation. This finding suggests that the envelope glycoproteins, especially E2, impose an orientation on the capsomers rather than the capsid protein sequence/structure itself dictating this orientation. Additional 3D reconstructions of other alphaviruses, particularly EEEV, which is most similar to WEEV in its capsid sequence, should be determined to deduce common rules governing the variations and commonality in nucleocapsid as well as virus organization.

Because of the overall similarity of WEEV and SINV, and in particular the high degree of sequence identity between their E1 glycoproteins (76% of their amino acids are identical [Table 1]), we did not expect to see significant differences between their glycoprotein structures. Indeed, pseudo-atomic modeling of the WEEV E1 protein based on a homology model derived from the X-ray structure of SFV's E1 (9) did not reveal any differences at our current resolution level (13 Å) between SINV (36) and WEEV (data not shown). That finding corroborated our analysis of the difference maps (see above), where no significant differences were observed for the densities attributed to E1. On the other hand, the E2 glycoproteins are less conserved among alphaviruses, suggesting that some differences in their structures might be visualized by cryoEM. Unfortunately, there is no atomic resolution structure available for any alphavirus E2 protein. This precluded us from analyzing visible differences between the SINV and WEEV E2 proteins in detail using the corresponding portions of the density maps. Several artificial and natural amino acid substitutions in the E2 protein of several alphaviruses are known to mediate changes in virulence and host range (5). Therefore, greater resolution of the E2 protein will be required to fully understand the host range differences as well as tropisms and diseases that these viruses cause. One significant question that could be answered with higher-resolution E2 structures is why different strains of the same virus, for example VEEV (29), have different infectivities and pathogenicities both in vitro and in vivo. Since E1 is more highly conserved and because E2 is responsible for the recognition and interaction of host cell receptors, the E2 structure is probably more critical for viral tropism. High-resolution structures of alphaviruses would allow us to analyze why and how E2 performs these functions.

To our knowledge, WEEV is the first BSL-3 virus to be imaged using cryoEM within a full BSL-3 containment facility, using recommended safety procedures (22, 26). The development of this facility along with experimental methods for high-containment imaging of BSL-3 agents should have a major impact on the study of emerging infectious diseases as well as on developing measures to counter the use of highly virulent viruses as biological weapons.