The Structural Biology of Eastern Equine Encephalitis Virus, an Emerging Viral Threat

Alphaviruses are arboviruses that cause arthritis and encephalitis in humans. Eastern Equine Encephalitis Virus (EEEV) is a mosquito-transmitted alphavirus that is implicated in severe encephalitis in humans with high mortality. However, limited insights are available into the fundamental biology of EEEV and residue-level details of its interactions with host proteins. In recent years, outbreaks of EEEV have been reported mainly in the United States, raising concerns about public safety. This review article summarizes recent advances in the structural biology of EEEV based mainly on single-particle cryogenic electron microscopy (cryoEM) structures. Together with functional analyses of EEEV and related alphaviruses, these structural investigations provide clues to how EEEV interacts with host proteins, which may open avenues for the development of therapeutics.


Introduction
Alphaviruses are enveloped icosahedral arboviruses that infect mammalian hosts, including humans [1][2][3][4][5][6][7]. Alphaviruses are broadly classified as arthritogenic and encephalitic based on the disease associated with their infection. For instance, Chikungunya virus (CHIKV), an arthritogenic alphavirus, was responsible for an epidemic that affected millions in the Indian subcontinent and islands in the Indian Ocean in 2006 [8]. In contrast to arthritogenic alphaviruses, infections of encephalitic alphaviruses are often associated with mortality, especially for EEEV, which is endemic mostly to the Americas [9][10][11][12][13][14][15][16]. EEEV is transmitted by the mosquito Culiseta melanura mostly to birds [17][18][19]. Although EEEV typically causes fewer than ten reported human cases per year, 34 cases of EEEV were reported in USA in 2019 and the virus has been detected in mosquitoes that feed on humans [16,18,20]. EEEV infections have a high mortality rate of up to 75%, and nearly 90% of survivors report permanent neurological sequelae [21][22][23]. It has been reported that EEEV can be spread as an aerosol, which has led to its classification as a select agent by United States Department of Agriculture (USDA) and Centers for Disease Control and Prevention (CDC) [24,25]. At present, there are no approved vaccines or therapeutics to prevent or contain EEEV infections in the general human population [17,26,27]. Hence, EEEV represents an emerging threat to human health. However, molecular insights into EEEV infection and interactions of its proteins with host factors are rather limited as compared to other alphaviruses such as CHIKV. In this review article, we provide a summary of recent publications on the structural biology of EEEV entry, disassembly, assembly, and interactions with host molecules. We highlight how EEEV escapes detection by receptors for viral glycans and provide a perspective on antibody interactions with viral envelope and interactions with host molecules. We highlight how EEEV escapes detection by receptors for viral glycans and provide a perspective on antibody interactions with viral envelope proteins. The structural questions on EEEV dynamics in host cells raised in this article have broad relevance to the understanding of alphavirus infection and interactions with host cell molecules.

Overview of Alphavirus Infection and Assembly Cycles
Alphaviruses such as EEEV have a 12kb positive-sense single-stranded RNA genome with five structural genes encoding capsid, E3, E2, 6K, Tf, and E1, and four non-structural genes encoding nsP1, nsP2, nsP3, and nsP4 [1,[28][29][30]. The present review article focuses on the structural proteins capsid, E2, and E1 that constitute viral particles [28,31,32]. Alphaviruses have a complex infection cycle that is summarized in Figure 1 [33]. This involves viral entry into host cells, low pH triggered conformational changes and subsequent fusion with endosomal membrane, release of the nucleocapsid in the cytosol, disruption of the capsid shell to release the viral RNA genome, biosynthesis of progeny RNA and proteins, assembly of progeny nucleocapsid, and assembly and release of progeny alphavirus particles [1,34,35]. Hence, multiple host organelles are involved in the alphavirus infection and assembly cycles. These distinct steps present potential targets for therapeutic intervention.  (4)), whereas progeny virus assembly and exit are from step (5) to (7). (Step 1) Receptor-mediated endocytosis allows alphavirus particles to enter cells (particles shown in blue, cyan). (Step 2) Viral and endosomal membranes undergo fusion upon acidification of the endosome lumen (pink). (Step 3) Cytosolic release of the nucleocapsid core (yellow). (Step 4) Disintegration of the core and release of the viral RNA genome (vRNA, red line). (Step 5) Cytosolic progeny RNA genome synthesis and nucleocapsid core assembly occur in the cytosol, whereas envelope proteins are synthesized in secretory ER

E1 and E2 Glycoproteins: Tools to Penetrate Host Membranes
The alphavirus envelope consists of three proteins, E1, E2, and E3 [28,30,50,51]. Substantial insights into the structure and function of these proteins have been obtained from prior investigations on other alphaviruses, using both purified proteins and intact virus particles [52][53][54][55][56]. The E2 protein is implicated in receptor binding, whereas E1 is involved in low pH fusion with the host membrane [44,[57][58][59][60][61][62]. This low pH-driven interaction of the alphavirus spike with host membranes raises an intriguing question, i.e., how is premature fusion inhibited during progeny spike biogenesis and assembly in the acidic Golgi lumen? The E3 protein has been shown to associate with the E1-E2 spike under acidic conditions to stabilize the hetero-dimer and to inhibit premature fusion [63]. Neutral pH in the extracellular medium drives dissociation of furin-cleaved E3 from the E1-E2 hetero-dimer [63][64][65][66][67]. The 4.4 Å resolution cryoEM structure of SINV-EEEV chimera identified the location of E1 and E2 proteins but not of E3 [33] (Figure 2a-e). Presumably, E3 was dissociated from the viral envelope during progeny assembly and purification. It is noted that E3 was reported in the cryoEM map of Venezuelan Equine Encephalitis Virus (VEEV) [48]. The SINV-EEEV cryoEM map corresponding to the E1-E2 ecto-domains has a resolution of approximately 3.5-6.0 Å (Figure 2a,b). The E1-E2 ecto-domains are organized in a hetero-dimer that closely resembles the hetero-dimer reported for SINV and CHIKV by X-ray crystallography [55,56]. The E1 and E2 ecto-domains are enriched in β-strands and demonstrate a three-domain organization, i.e., domains I, II, and III for E1 and A, B, and C for E2 (Figure 2c-e). Intra-chain disulfide bonds stabilize the respective ecto-domains of E1 and E2 proteins. Three E2 monomers in each trimeric spike demonstrate a radial arrangement, whereas three E1 monomers are organized tangentially, which is consistent with previous structural investigations (e.g., [41,48]).
The envelope layer in alphaviruses undergoes pH-driven conformational changes [68][69][70]. This includes a rearrangement of tangential E1 ecto-domain into a radial orientation for insertion into the endosomal membrane and subsequent release of the alphavirus genome into cytosol [54,58,71,72]. The residue-level details of this membrane interaction are currently not well understood. The cryoEM analysis of SINV-EEEV particles generated new hypotheses relevant to these spike conformational changes [33]. For instance, the cryoEM structure of SINV-EEEV showed an E1-E2 interface that is highly enriched in charged, protonatable residues (Figure 3a) [33]. This suggests a possible pH-sensing mechanism for E1-E2 dissociation and envelope layer disassembly as suggested for related alphaviruses [73][74][75][76]. Such a hetero-dimer dissociation event would release the E1 ecto-domain for interactions with the endosomal membrane. This investigation proposed conservation in the mechanism of hetero-dimer dissociation as the likely pH-responsive residues in the E1-E2 electrostatic interface are widely conserved in alphaviruses [33]. Low pH initiated changes also involve a radial to tangential re-orientation of the E2 ecto-domain, which requires disruption of E2-E2 contacts proximal to the spike 3-fold axis [53,69,72]. This E2-E2 interface in SINV-EEEV is enriched in basic residues that demonstrate limited sequence conservation  (Figure 3b-d). This suggests the possibility that the pH sensitivity of the E2-E2 interface is alphavirus-specific. However, experimental validation of the above-mentioned hypotheses and residue-specific testing are still lacking. Nevertheless, the involvement of distinct early to late endosomal compartments hints towards distinct pH requirements for entry of different alphaviruses [77][78][79][80]. The envelope layer in alphaviruses undergoes pH-driven conformational changes [68][69][70]. This includes a rearrangement of tangential E1 ecto-domain into a radial orientation for insertion into the endosomal membrane and subsequent release of the alphavirus genome into cytosol [54,58,71,72]. The residue-level details of this membrane interaction are currently not well understood. The cryoEM analysis of SINV-EEEV particles generated new hypotheses relevant to these spike conformational changes [33]. For instance, the cryoEM structure of SINV-EEEV showed an E1-E2 interface that is highly enriched in charged, protonatable residues ( Figure 3a) [33]. This suggests a possible pH-sensing mechanism for E1-E2 dissociation and envelope layer disassembly as suggested for related alphaviruses [73][74][75][76]. Such a hetero-dimer dissociation event would release the E1 ecto-domain for interactions with the endosomal membrane. This investigation proposed conservation in the mechanism of hetero-dimer dissociation as the likely pH-responsive Flaviviruses, such as Dengue and Zika viruses, are similar to alphaviruses in their icosahedral symmetry and the presence of an internal membrane, although there is no sequence relationship between these two distinct genera [50,81]. Structural analyses have demonstrated similarities in the fold and organization of E1 protein in alphaviruses and envelope (E) protein in flaviviruses [54,82,83]. It has been noted that the acidic character of E1 (theoretical isoelectric point of ecto-domain, 6.0-6.7) is shared by the flavivirus E protein (theoretical isoelectric point of ecto-domain, 5.8-6.8), suggesting likely conserved low pH neutralization of acidic residues in membrane insertion [33].

A Multifunctional Capsid Protein
The capsid protein of SINV-EEEV demonstrates a two-domain organization characteristic of alphaviruses and is divided into an N-terminal (NTD, residues 1-116) and a C-terminal domain (CTD, residues 117-261) [33,48,52,84]. The alphavirus capsid NTD is enriched in basic residues that have been suggested to interact with the viral RNA genome [85,86]. A recent investigation has identified multiple sites for capsid protein binding on the viral genomic RNA [87]. The capsid NTD in EEEV demonstrates an abundance of Gly and Pro residues, which are characteristic of intrinsically disordered proteins (reviewed in [88]). This is consistent with cryoEM reconstructions of intact alphavirus particles wherein a large segment of the capsid NTD is not visualized (e.g., [48]). It is inferred that this is due to disorder in the capsid NTD. However, cryoEM reconstructions of alphaviruses utilize icosahedral symmetry for averaging between 60 asymmetric units. This raises the question of whether there is symmetry mismatch between the outer layer of envelope proteins and inner capsid NTD, which would contribute to relatively poor reconstruction of the capsid NTD. This possibility has not been explored in detail. Nevertheless, it has been postulated that the intrinsic disorder of the capsid NTD may provide structural flexibility for interactions with dissimilar sequences in the viral RNA genome [33]. This is because the icosahedral protein shell of alphaviruses encloses a single copy of the viral RNA genome. The alphavirus genome sequence associated with each of the capsid proteins will be unique, leading to dissimilar genome-capsid interactions at these sites. Flexibility in the capsid NTD may facilitate these interactions while minimizing the genetic cost of encoding specialized sequences for each capsid position. Insights into the organization of the capsid NTD are derived from a comparison of a crystal structure of the SINV capsid protein and the cryoEM structures of SINV-EEEV and related VEEV [33,48,52,85]. Coordinates for short segments of the capsid NTD attached to the CTD have been refined in these maps. However, the coordinates of these short segments of capsid NTD show considerable differences (Figure 4a,b). For instance, in the SINV crystal structure, the NTD segment displays an elongated conformation that extends "radially" from the CTD [85]. An analysis of crystal packing shows that this short segment is involved in contacts with adjacent symmetry-related molecules [85]. In contrast, a comparable segment of the capsid NTD in the SINV-EEEV cryoEM structure has a more tangential organization, although the map is at a lower resolution [33]. However, in both structures the NTD short segment shows an extended conformation wherein the side-chains of the basic residues are available for interactions with the acidic viral RNA genome. The extended conformation of SINV-EEEV capsid NTD is consistent with a comparable, but shorter capsid NTD sequence mapped in the VEEV cryoEM structure [33,48]. Despite these advances in elucidating the structure of the capsid NTD, limited insights are available into the stoichiometry of genome-capsid interactions within an alphavirus particle, whether all 240 capsid protein copies bind the genome, and whether there is local "order" at the capsid-genome interaction site that is not visualized by current icosahedral averaging approaches. Addressing these questions is essential for a deeper understanding of genome packaging and alphavirus assembly.
The alphavirus capsid CTD has been characterized in substantially more detail than NTD. Functionally, the CTD encodes a protease activity that is essential for alphavirus polyprotein processing and is consistent with a chymotrypsin protease-like fold of CTD [52,85]. The CTD contains a mixed secondary structure content of short α-helices and β-strands. A groove in the capsid CTD provides the interaction surface for the C-terminal residues of the E2 glycoprotein [33,48,89,90]. In the cryoEM structure of SINV-EEEV, this CTD groove is enriched in polar interactions from side chains of Arg132, Tyr159, Tyr177, Trp244, and Thr250. Hydrophobic and Van der Waals interactions are provided by CTD Phe134 and Ile158. Mutagenesis of SINV capsid CTD residues in this groove has implicated Tyr162, which is equivalent to EEEV capsid Tyr159, and Lys252 in alphavirus assembly [89]. Structural superposition of SINV-EEEV and SINV capsid CTD (yellow surface) demonstrates difference in location of NTD residues between capsid of EEEV (cyan) and SINV (red). (c) In the internal capsid layer, pentamers and hexamers are arranged at the icosahedral 5-and 2-fold axes, respectively. These provide major stabilizing capsid-capsid contacts, whereas no interactions are observed near the 3-fold. In (a,b), coordinates for capsid protein (PDB ID 6MX7 for SINV-EEEV, 1SVP for SINV) were downloaded from PDB and figures were generated in PyMol (www.pymol.org, version 1.8.6.0). Panel (c) taken from [33].
The alphavirus capsid CTD has been characterized in substantially more detail than NTD. Functionally, the CTD encodes a protease activity that is essential for alphavirus polyprotein processing and is consistent with a chymotrypsin protease-like fold of CTD [52,85]. The CTD contains a mixed secondary structure content of short α-helices and βstrands. A groove in the capsid CTD provides the interaction surface for the C-terminal residues of the E2 glycoprotein [33,48,89,90]. In the cryoEM structure of SINV-EEEV, this CTD groove is enriched in polar interactions from side chains of Arg132, Tyr159, Tyr177, Trp244, and Thr250. Hydrophobic and Van der Waals interactions are provided by CTD The capsid protein in alphaviruses provides a protective shell around the viral genomic RNA [50]. An analysis of the SINV-EEEV and other alphavirus cryoEM structures shows differences in interactions between adjacent capsid CTD with respect to the icosahedral symmetry axes (Figure 4c) [33,48,50]. For instance, neighboring capsid CTDs proximal to the icosahedral 2-fold and 5-fold vertices are organized into hexagonal and pentagonal facets. Three capsid CTDs near an icosahedral 3-fold vertex belong to three adjacent capsid hexagons. These three capsid CTDs are more widely separated and do not demonstrate inter-capsid interactions, unlike the capsid pentagons and hexagons [33,50]. This suggests that stability in the nucleocapsid core is provided by CTD linkage with the E2 layer, which is closely packed with extensive inter-protein contacts [42,51,91]. In this model, disruption of envelope-capsid interactions upon alphavirus entry and low pH-triggered fusion releases the nucleocapsid core into the cytosol, devoid of stabilizing E2 contacts, followed by subsequent core disassembly. Although the experimental validity of this model of nucleocapsid disassembly remains to be tested, it is consistent with prior investigations that suggest nucleocapsid disassembly in the host cell cytosol follows alphavirus entry [92][93][94][95]. These structural data on EEEV reinforce several interesting questions that are yet to be fully addressed. For instance, if the nucleocapsid core is inherently unstable, then how is it assembled during progeny assembly in the cytosol prior to interactions with the E2 glycoprotein? Is the nucleocapsid core assembled initially in a metastable state(s)? Investigations of in vitro-assembled cores in alphaviruses such as RRV and WEEV have provided clues to this assembly process wherein the envelope-free cores maintain an overall arrangement similar to that inside viral particles [96,97]. In complementary structural analyses, icosahedral symmetry was demonstrated in newly assembled VEEV cores, although the diameter was substantially larger than in intact virus particles [98][99][100]. Furthermore, the isolation of envelope-free nucleocapsid cores by detergent treatment and gradient purification has been reported [101]. These data suggest that the cores are assembled in a somewhat stable, icosahedral state prior to complete alphavirus assembly. However, more recent data present a contrasting picture wherein a lack of nucleocapsid uniformity is noted even in fully assembled alphavirus particles [97]. These results suggest that while the envelope layer provides protection to the core, association with the envelope layer may be insufficient to generate an ideal capsid icosahedron. Presently, it is not known what causes this deviation in nucleocapsid core symmetry, the extent to which these differences in structural uniformity are important for assembly/disassembly, and if this non-uniformity is a widely conserved feature in alphaviruses. Recent data on genome-less capsid cores add another layer of complexity to the role of the viral genome in driving nucleocapsid assembly [102,103]. Overall, the understanding of uncoating [104], symmetry, organization, assembly of alphavirus nucleocapsid cores is limited despite recent advances. Progress in cryoEM imaging inside cells and improvement in resolution of such structures should help address several of the unanswered questions pertaining to alphavirus nucleocapsid cores.

Structural Basis of Heparan Sulfate (HS) Binding in EEEV
HS is a long anionic polysaccharide polymer linked to plasma membrane proteins and is involved in a variety of functions, such as cellular adhesion, signaling, and coagulation (reviewed in [105,106]). Cell culture adaptation of alphaviruses is associated with an HS binding phenotype and decreased pathogenesis [59,60,[107][108][109][110]. However, in EEEV, an HS binding phenotype is associated with host infection [111,112]. The sequencing of North American strains of EEEV host isolates identified key basic residues in the E2 ecto-domain (Lys71, Lys74, and Lys77) associated with HS binding. Of these three basic residues, Lys74 of E2 is widely conserved, Lys77 is least conserved, and residue Lys71 is semi-conserved and often replaced by His, which is also basic [33,111,112]. The 4.4 Å cryoEM structure of SINV-EEEV chimera placed these residues in a "linear triad" in E2 (Figure 5a,b) [33]. In this linear triad, access to the Lys71 side-chain is partly occluded by domain B, whereas Lys74 and Lys77 are well exposed. This linear triad is surrounded by a predominantly electrostatic protein environment. Hence, the putative binding site for HS is distributed over three adjacent symmetry-related E2 monomers. Structural insights into the interactions of the SINV-EEEV with HS were obtained from single-particle cryoEM analysis of SINV-EEEV particles complexed with heparin (Hp), a low-molecular-weight HS analog [113]. SINV-EEEV demonstrates specific interactions with Hp, which reduces infection of SINV-EEEV chimera in cell culture by almost 90% upon pre-incubation with Hp. In contrast, the sulfated polysaccharide polymer chondroitin sulfate (CS), which is similar to HS, only reduces infection by 30%, suggesting stereospecificity in SINV-EEEV for Hp and HS. The cryoEM structure of SINV-EEEV in the presence of Hp yielded surprising insights. No features corresponding to Hp were observed in the vicinity of the linear triad of Lys71-Lys74-Lys77. It was proposed that this could be due to a non-icosahedral arrangement of Hp, which likely results in a loss of features during reconstruction. Alternatively, this implies that the linear Lys triad is not directly involved in binding Hp, but is rather selective for endogenous host HS, which is more complex and larger [114]. In a trimeric spike, linear Lys triads from three monomers in a quaternary organization would provide multiple interactions with high avidity for HS, in contrast to a single linear Lys triad from one spike monomer. The substantially smaller chain length of Hp in the cryoEM analysis could have been insufficient to satisfy this complex interaction with multiple linear Lys triads. Hence, even though Hp molecules were not visualized by cryoEM analysis of SINV-EEEV, a role of the linear Lys triad in HS binding prior to viral entry cannot be excluded.
The cryoEM analysis of SINV-EEEV revealed four distinct sites for Hp binding that are unique with respect to icosahedral symmetry axes [113]. Two axial Hp sites overlap with the 3-fold rotational symmetry axis in q3 and i3 trimeric spikes, whereas two peripheral sites are in proximity to the E2 β-connector. The quaternary arrangement of residues in the axial sites provides multiple interactions for Hp binding. These axial sites are enriched in basic residues, whose side chains face the 3-fold rotational symmetry axes, and the bound Hp molecules. In the axial as well as peripheral sites, the bound Hp molecules appear to be spherical, which is likely a consequence of multiple binding poses and 3-fold averaging in the i3 spikes. Hence, this cryoEM analysis suggests that HS interactions in EEEV involve multiple sites in the viral E2 protein, which could provide multiple routes towards neurovirulence [111,112].

Glycosylation of E1 and E2 Proteins in EEEV
Post-translational modifications such as glycosylation play a key role in alphavirus infection and assembly cycles. Mutations in the alphavirus envelope protein glycosylation sites affect infectivity, membrane fusion, assembly, and yield of progeny virus [91,[115][116][117][118][119]. Elimination of E2 glycosylation sites enhances viral interaction with HS [118]. E2 glycosylation interacts with host cell DC-SIGN and L-SIGN molecules for viral entry [120]. DC-SIGN/L-SIGN are lectin molecules in macrophage and dendritic cell plasma membranes that demonstrate strong interactions with pathogen high-mannose glycosylation sites to activate downstream immune responses [121][122][123][124][125][126][127]. The structural analysis of SINV-EEEV chimera generated insights into the location and function of glycosylation sites ( Figure 6) [33]. A sequence analysis showed one N-linked glycosylation site each on E1 and E2 ecto-domains of EEEV, i.e., E1 Asn134 and E2 Asn315. These two sites conform to the well-established glycosylation motif, Asn-X-Thr (X = any residue except Pro), and have the sequence Asn134-Ile135-Thr136 and Asn315-Phe316-Thr317 for E1 and E2, respectively. The cryoEM analysis of SINV-EEEV particles was consistent with this sequence-based prediction. The cryoEM map demonstrated the presence of a feature connected to the Asn side-chain in both E1 and E2. One N-acetyl-glucosamine monosaccharide was fitted in the E2 site, whereas a disaccharide was accommodated in the E1 site. The E2 glycosylation site is not accessible on the viral surface, whereas the E1 site is well-exposed near the 2-fold and 5-fold symmetry axes, close to the base of the trimeric spike. In the context of E2, this presents a contrast as other alphaviruses demonstrate at least one exposed glycosylation site on the E2 ecto-domain [91,128,129]. In the case of EEEV, this lack of an accessible E2 glycosylation site may provide a biological route to escape immune detection via DC-SIGN/L-SIGN [33]. A comparative analysis of infectivity was performed between EEEV and SINV, which was previously shown to interact efficiently with DC-SIGN and L-SIGN [120]. It was observed that EEEV demonstrates more limited infectivity in this cellular system than SINV, thus implicating the poorly exposed E2 glycan in inefficient interactions with host lectins. This also suggests that the E1 glycan is not sufficient for viral entry even though it is surface exposed. This is consistent with the observation that glycan composition is a key determinant of viral tropism [130]. A chemical analysis of the E1 glycan in purified SINV-EEEV particles showed enrichment in pauci-mannose carbohydrates in the virus from mosquito C6/36 cell lines and in complex-type carbohydrates in the virus from mammalian BHK-15 cells [33]. Oligo-mannose glycans, which are essential for interactions with DC-SIGN and L-SIGN, constituted only one-tenth and one-sixth of the total glycans in E1 in the mosquito-and mammalian-derived SINV-EEEV particles, respectively. Hence, the E1 glycan is biochemically sub-optimal for interactions with DC-SIGN/L-SIGN even though it is exposed on the viral surface [131]. It would be informative to test if the EEEV envelope glycans are essential for efficient viral protein biogenesis. If this hypothesis is confirmed, it would suggest that EEEV has likely evolved a means for envelope protein biogenesis while avoiding a deleterious side-effect of glycosylation, i.e., immune detection. Figure 6. Glycosylation in EEEV E1-E2 proteins. The glycan groups are highlighted as sticks in yellow-red and are encircled for clarity. Color code, E1, cyan, E2, pink. Figure taken from [33].
In the first investigation [33,36], Fabs from five potent neutralizing mouse mAbs (EEEV-3, EEEV-5, EEEV-42, EEEV-58, and EEEV-69) were characterized in complex with SINV-EEEV chimera by single-particle cryoEM. These structures were determined to a resolution of 7.3-8.2 Å and all five Fabs were found to bind to the E2 ecto-domain ( Figure  7). These five Fabs were classified into two groups based on their footprints, i.e., domain A Fabs (EEEV-5, EEEV-42, and EEEV-58) and domain B Fabs (EEEV-3, and EEEV-69). Even though each respective group of Fabs demonstrated substantially overlapping footprints, large differences in average occupancies were reported. For instance, EEEV-5 had a relative occupancy of 45.4% with respect to the ecto-domain, whereas EEEV-58 had a relative occupancy of 97.2%. Both Fabs were incubated in excess with the SINV-EEEV particles prior to flash-freezing for cryoEM analysis. This raises an intriguing question about the factors that contribute to this difference in Fab occupancies. It was suggested that the angular orientation of the Fab may play a role in its occupancy (Figure 8). In a trimeric spike, three symmetry-related copies of E2 domain A are clustered close there near the 3-fold rotational axis. Hence, a Fab that is bound in a radial orientation on domain A, and hence parallel to the spike 3-fold axis, would face steric clash from its symmetryrelated partners. This would limit its average occupancy on the spike. However, domain A Fabs that deviate from this radial orientation and are more tangential would experience fewer steric restrictions. Hence, a tangential orientation of Fabs will be favored for high occupancy in domain A. Domain B Fabs present a contrasting scenario. Domain B of the E2 ecto-domain is located at the distal end of the trimeric spike, at a larger distance from the spike 3-fold axis than domain A. However, domain B from a trimeric spike is closer to
In the first investigation [33,36], Fabs from five potent neutralizing mouse mAbs (EEEV-3, EEEV-5, EEEV-42, EEEV-58, and EEEV-69) were characterized in complex with SINV-EEEV chimera by single-particle cryoEM. These structures were determined to a resolution of 7.3-8.2 Å and all five Fabs were found to bind to the E2 ecto-domain ( Figure 7). These five Fabs were classified into two groups based on their footprints, i.e., domain A Fabs (EEEV-5, EEEV-42, and EEEV-58) and domain B Fabs (EEEV-3, and EEEV-69). Even though each respective group of Fabs demonstrated substantially overlapping footprints, large differences in average occupancies were reported. For instance, EEEV-5 had a relative occupancy of 45.4% with respect to the ecto-domain, whereas EEEV-58 had a relative occupancy of 97.2%. Both Fabs were incubated in excess with the SINV-EEEV particles prior to flash-freezing for cryoEM analysis. This raises an intriguing question about the factors that contribute to this difference in Fab occupancies. It was suggested that the angular orientation of the Fab may play a role in its occupancy (Figure 8). In a trimeric spike, three symmetry-related copies of E2 domain A are clustered close there near the 3-fold rotational axis. Hence, a Fab that is bound in a radial orientation on domain A, and hence parallel to the spike 3-fold axis, would face steric clash from its symmetryrelated partners. This would limit its average occupancy on the spike. However, domain A Fabs that deviate from this radial orientation and are more tangential would experience fewer steric restrictions. Hence, a tangential orientation of Fabs will be favored for high occupancy in domain A. Domain B Fabs present a contrasting scenario. Domain B of the E2 ecto-domain is located at the distal end of the trimeric spike, at a larger distance from the spike 3-fold axis than domain A. However, domain B from a trimeric spike is closer to domain B from an adjacent spike. Hence, a Fab bound tangentially to domain B could potentially pose steric restrictions on the binding of a second Fab to domain B from an adjacent trimeric spike. However, Fabs that demonstrate a more radial orientation are likely to face fewer steric restrictions. Based on this analysis, it was suggested that the more radial orientation of domain A Fab, EEEV-5, i.e., 14.5 • with respect to the spike 3-fold axis, was partly responsible for this low 45.4% occupancy. Although this analysis was performed on SINV-EEEV, the conservation of E1-E2 organization and structure in alphaviruses suggests that this is widely applicable to investigations of steric limitations in Fab-alphavirus interactions.  A more recent structural analysis of SINV-EEEV with Fabs from two potent neutralizing human mAbs, EEEV-33 and EEEV-143, has provided insights into conformational epitopes [39]. These cryoEM structures showed that EEEV-33 is a domain A Fab, whereas EEEV-143 is a domain B Fab. This analysis demonstrated a radial orientation of EEEV-33 Fabs on the trimeric spike, which generates potential steric restrictions for binding of a divalent IgG. Here, it should be noted that Fabs from these two human mAbs possess strong neutralization activity, unlike monovalent Fabs from mouse mAbs with anti-EEEV neutralization activity. This suggests a likely interaction wherein the Fab from these human mAbs either sterically blocks access to the host membrane or cross-links the EEEV envelope. A binding analysis showed that EEEV-33 preferentially interacts with intact particles over purified E2 protein, indicating the involvement of a quaternary binding site. Hence, cross-linking of multiple adjacent subunits represents a potent mechanism for alphavirus neutralization.
It is likely that cross-linking of adjacent subunits by Fabs is a high probability event in alphaviruses from a structural perspective. In an alphavirus spike, the most highly exposed surfaces are of individual E1-E2 hetero-dimers. The likely sites on the alphavirus surface where adjacent subunits are in close contact are in proximity of the icosahedral 2, 3, and 5-fold axes. This is consistent with the cross-linking activity noted for EEEV-33, A more recent structural analysis of SINV-EEEV with Fabs from two potent neutralizing human mAbs, EEEV-33 and EEEV-143, has provided insights into conformational epitopes [39]. These cryoEM structures showed that EEEV-33 is a domain A Fab, whereas EEEV-143 is a domain B Fab. This analysis demonstrated a radial orientation of EEEV-33 Fabs on the trimeric spike, which generates potential steric restrictions for binding of a divalent IgG. Here, it should be noted that Fabs from these two human mAbs possess strong neutralization activity, unlike monovalent Fabs from mouse mAbs with anti-EEEV neutralization activity. This suggests a likely interaction wherein the Fab from these human mAbs either sterically blocks access to the host membrane or cross-links the EEEV envelope. A binding analysis showed that EEEV-33 preferentially interacts with intact particles over purified E2 protein, indicating the involvement of a quaternary binding site. Hence, cross-linking of multiple adjacent subunits represents a potent mechanism for alphavirus neutralization.
It is likely that cross-linking of adjacent subunits by Fabs is a high probability event in alphaviruses from a structural perspective. In an alphavirus spike, the most highly exposed surfaces are of individual E1-E2 hetero-dimers. The likely sites on the alphavirus surface where adjacent subunits are in close contact are in proximity of the icosahedral 2, 3, and 5-fold axes. This is consistent with the cross-linking activity noted for EEEV-33, which recognizes a quaternary epitope near the spike 3-fold axis and in a cryoEM struc-ture of CHIKV complexed with a bivalent IgG [155]. Overall, cross-linking of adjacent protein subunits is expected to be more efficient if the viral surface were "smoother", wherein neighboring subunits are available for simultaneous binding by Fabs or mAbs. An example is mature flaviviruses whose envelope protein is arranged in a smooth, herringbone pattern [81,[156][157][158][159][160][161]. In these enveloped icosahedral viruses, multiple sites for subunit cross-linking are accessible. This is noted by a relatively larger abundance of Fabs and mAbs whose epitopes are located close to the subunit-subunit interface (reviewed in [162,163]).

Conclusions
The investigations of SINV-EEEV chimeric particles and EEEV VLPs described above have greatly advanced the understanding of the structural basis of EEEV-host interactions, especially in the context of entry and antibody neutralization, while setting the stage for further investigations. For instance, it is not known which protein receptors in the host membrane are hijacked during EEEV entry. What are the intermediate states of E1-E2 ecto-domains during pH-triggered transitions in internalized EEEV, and more broadly in alphaviruses? Are these E1-E2 transition states conserved in alphaviruses and can they be exploited in the design of broad-spectrum inhibitors? Does association with host receptors affect E1-E2 conformational changes and disassembly? When Fabs and mAbs are provided in sub-stoichiometric quantities, what are the preferred epitope sites for Fab and mAbs binding on the crowded viral surface in EEEV and other alphaviruses? Is the binding of Fabs or mAbs at a few selected sites sufficient for neutralization? Addressing these questions will require a combination of cellular and functional approaches that build on high-resolution information derived from structural studies to gain further insights into alphavirus-host interactions.