An Alphavirus E2 Membrane-Proximal Domain Promotes Envelope Protein Lateral Interactions and Virus Budding

ABSTRACT Alphaviruses are members of a group of small enveloped RNA viruses that includes important human pathogens such as Chikungunya virus and the equine encephalitis viruses. The virus membrane is covered by a lattice composed of 80 spikes, each a trimer of heterodimers of the E2 and E1 transmembrane proteins. During virus endocytic entry, the E1 glycoprotein mediates the low-pH-dependent fusion of the virus membrane with the endosome membrane, thus initiating virus infection. While much is known about E1 structural rearrangements during membrane fusion, it is unclear how the E1/E2 dimer dissociates, a step required for the fusion reaction. A recent Alphavirus cryo-electron microscopy reconstruction revealed a previously unidentified D subdomain in the E2 ectodomain, close to the virus membrane. A loop within this region, here referred to as the D-loop, contains two highly conserved histidines, H348 and H352, which were hypothesized to play a role in dimer dissociation. We generated Semliki Forest virus mutants containing the single and double alanine substitutions H348A, H352A, and H348/352A. The three D-loop mutations caused a reduction in virus growth ranging from 1.6 to 2 log but did not significantly affect structural protein biosynthesis or transport, dimer stability, virus fusion, or specific infectivity. Instead, growth reduction was due to inhibition of a late stage of virus assembly at the plasma membrane. The virus particles that are produced show reduced thermostability compared to the wild type. We propose the E2 D-loop as a key region in establishing the E1-E2 contacts that drive glycoprotein lattice formation and promote Alphavirus budding from the plasma membrane.

contacts with the E1 glycoprotein that are necessary for particle budding. We propose that H348 and H352 are critical for the formation of the E1/E2 lattice at the plasma membrane, thus highlighting the importance of glycoprotein interactions in virus budding.

RESULTS
Initial characterization of E2 H348 and H352 mutants. In order to test the role of H348 and H352 in the E2 glycoprotein D-loop, we created single and double alanine substitutions in the pSP6-SFV4 infectious clone (24), subjected the viral RNAs to in vitro transcription, and electroporated them into BHK-21 cells to test the phenotypes. These mutants are referred to here as H348A, H352A, and H348/352A and the wild-type virus produced from the SFV infectious clone as WT. Immunofluorescence of cells fixed at 8 h postelectroporation (hpe) showed efficient cell surface expression of E1 and E2 for all mutants (see Fig. S1 in the supplemental material). Analysis of cocultures of electroporated cells and nonelectroporated cells showed that by 24 hpe all of the mutants were able to mediate secondary infection (data not shown).
We then performed growth assays by electroporation of WT and mutant RNAs into BHK cells. Growth of all three mutants at 37°C was significantly inhibited, with a 1.6-to-2-log reduction in growth compared to the WT at 24 h ( Fig. 2A). The decreases in growth of the three mutants were similar, with no significant differences among their titers at 24 h and thus no apparent additive effects of the mutations ( Fig. 2A). While some Alphavirus mutants with impaired virus production are rescued by growth at 28°C (25,26), the growth of the H348A, H352A, and H348/352A mutants remained inhibited when RNA-electroporated BHK cells were incubated at 28°C (Fig. 2B).
Role of H348 and H352 in virus entry. To test the hypothesis that H348 and H352 play an important role in E2 dissociation from E1 during virus entry, we first determined the pH threshold of fusion of the mutant viruses, since changes in dimer dissociation The images in panels A and B were prepared using PyMOL software (the PyMOL Molecular Graphics System, v1.2r2; Schrödinger, LLC). (C) Alignment of E2 proteins from members of the Alphavirus genus, demonstrating that H348 and H352 are highly conserved (numbering as described for SFV). Red highlighting indicates high sequence conservation, and yellow highlighting indicates some sequence conservation, with conserved residues shown in red text. Alignment was performed using NPS@ (58) with CLUSTAL W (59). Alignment formatting was carried out using ESPript (60 (18,21,27,28). Mutant and WT viruses were bound to BHK cells and pulsed at low pH to trigger virus-plasma membrane fusion, and the resultant infection was quantitated by immunofluorescence. Fusion of WT SFV had a pH threshold of~6.0 in this assay, while the mutant viruses all had a pH threshold of~6.2 (Fig. 3A). Although this difference was experimentally significant, such minor shifts in FIG 2 Growth properties of WT and mutant viruses. BHK cells were electroporated with WT or mutant viral RNA and incubated at 37°C (A) or 28°C (B). Cell media were collected at the indicated times, and virus production was quantitated by plaque assay. (A) Growth kinetics at 37°C. The graph shows averages and standard deviations of data from 3 to 4 independent experiments; two independent clones of each mutant were assayed in each experiment. At the time points of 8, 12, and 24 h, results from all 3 mutants were significantly different from WT results (P Ͻ 0.0001). The titers produced by the H348A, H352A, and H348/352A mutants were not significantly different at 24 h (P Ͼ 0.2). (B) Growth kinetics at 28°C. The graph shows averages of data from 2 independent experiments, each performed using two independent clones for each mutant. The bars indicate the ranges.

FIG 3
Effects of H348 and H352 mutations on virus fusion and infectivity. (A) WT and mutant viruses were prebound to BHK cells for 90 min on ice and then treated with media at the indicated pH for 3 min at 37°C to trigger virus fusion with the plasma membrane. Cells were cultured at 28°C for 16 h in the presence of 20 mM NH 4 Cl to prevent secondary infection. The percentage of infected cells was quantitated by immunofluorescence and normalized to the maximal fusion for each virus. The graph shows averages and standard deviations of data from 3 independent experiments. The difference between the results from the WT strain and all three mutants at pH 6.2 is statistically significant (P Ͻ 0.0001). (B) The specific infectivity of WT and mutant viruses was calculated by determining the ratio of infectious virus (as quantitated by plaque assay) to the number of virus particles (as quantitated by Western blotting of the E2 glycoprotein). The difference in specific infectivity between WT and H348/ 352A was not statistically significant (P Ͼ 0.1). Data shown represent averages and standard deviations of data from 3 independent experiments.

Byrd and Kielian
® fusion pH threshold have not been observed to affect virus growth or infectivity (see, e.g., references 29 and 30, and we note that the difference is based on a single pH point. A small increase in the mutant pH threshold could reflect somewhat reduced stability in the E1/E2 dimer of the mutant virus particles, contrary to what would be predicted by the dimer dissociation hypothesis. Given that the three mutants have similar properties, our further investigations focused on the H348/352A double mutant. We determined virus-specific infectivity, since it would be decreased if virus entry and fusion were less efficient in the mutant. WT and mutant SFV stocks were harvested from 8-h infections of BHK cells, and virus-specific infectivity was calculated as the ratio of the plaque-forming unit number to the virus particle E2 protein number. No significant difference was observed between the specific infectivity of WT SFV and that of the H348/352A mutant (Fig. 3B). Taken together, our results indicate that although mutation of the D-loop histidine residues strongly inhibited virus growth, it did not significantly affect dimer dissociation or virus infectivity.
Induction of intercellular extensions. Previous work from our laboratory and others showed that Alphavirus infection induces long (Ͼ10-m) actin-and tubulinpositive intercellular extensions that can transmit virus from infected cells to uninfected cells (31)(32)(33)(34). Induction of extensions is abrogated by mutation of a critical tyrosine residue in the E2 cytoplasmic tail, which blocks E2-capsid interaction and virus budding (31,32). We hypothesized that structural changes in the H348/352A mutant could perturb E2-capsid interactions, thus reducing formation of intercellular extensions and cell-to-cell virus transmission. We therefore evaluated the formation of extensions in Vero cells infected with WT or H348/352A virus. Vero cells were optimal for these experiments as their flat morphology permits efficient imaging of extensions (32). Cells were fixed at 8 h postinfection (hpi); stained with antibodies against tubulin, E1, and E2; and imaged by confocal microscopy (Fig. 4). Both WT and H348/352A virus induced long, tubulin-positive extensions, arguing that the route of cell-to-cell virus transmission per se would not be impaired by the D-loop mutations. Similar results were obtained using WT-and mutant-infected BHK cells (data not shown).
Assembly of H348/352A virus particles. The lack of effects of the D-loop mutations on particle infectivity and extension formation suggested that the mutant growth defects could be due to inhibition of particle assembly. To test this, BHK cells were infected, pulse-labeled with [ 35 S]methionine/cysteine at 5 hpi, and chased in the absence of label for 0 to 4 h. The cell media and lysates were collected, immunoprecipitated with a polyclonal Ab (pAb) against E1 and E2, and evaluated by SDS-PAGE. The cell lysate samples showed that WT-and mutant-infected cells produced the viral structural proteins and processed p62 to E2 with similar efficiencies and kinetics (Fig. 5A). The cell medium samples were immunoprecipitated in the absence of detergent to allow retrieval of intact virus particles. Increasing amounts of the viral structural proteins were recovered from the chase media of WT-infected cells with increasing chase times (Fig. 5A). In contrast, the mutant-infected cells produced negligible amounts of virus particles and primarily released E1s, a soluble fragment of E1 produced under budding-defective conditions (35,36). In agreement with the results of growth assays at 28°C (Fig. 2B), incubation at 28°C did not rescue mutant virus assembly (data not shown). Thus, the H348/352A mutant is strongly impaired for virus assembly and this is the primary cause of its significant reduction in growth.
The p62/E2 protein acts as a chaperone for E1 folding and protects it from premature fusion in the secretory pathway (19,20). Alphavirus assembly defects can occur due to failure of the E1 and E2 dimers to associate after synthesis (22) or to decreased stability of the dimers during virus budding at the cell surface (25). To test the stability of the H348/352A dimer during virus biogenesis, infected BHK cells were pulse-labeled and chased. Aliquots of the cell lysates were then immunoprecipitated with pAb to E1/E2 or with an E1 monoclonal antibody (MAb) to evaluate dimer stability by coimmunoprecipitation (Fig. 5B). The MAb to E1 retrieved comparable levels of associated p62/E2 from WT-infected and mutant-infected cells. These results suggest that the stability of the mutant E1/E2 dimer is not responsible for its assembly defect.
Mutant budding and nucleocapsid production. Transmission electron microscopy (TEM) analysis of WT-and H348/352A-infected cells was used to directly visualize virus particle assembly. The WT virus sample showed nucleocapsids at the plasma membrane in the process of budding and numerous released virus particles ( Fig. 6A and Fig. S2A and C). In contrast, H348/352A nucleocapsids associated with the plasma membrane, but little or no apparent evidence of budding or released virus particles was observed ( Fig. 6B and Fig. S2B and D). Immunofluorescence analysis showed similar diffuse localization of capsid protein in the cytoplasm and intercellular extensions of WT-and H348/352A-infected cells (Fig. S3). In addition, gradient sedimentation studies showed comparable levels of cytoplasmic nucleocapsid production by the WT and the H348/ 352A mutant (Fig. S4). These data suggest that the mutant budding defect does not involve E2-capsid interactions or indirect effects on nucleocapsid formation, since nucleocapsids were produced and localized to the plasma membrane.
Morphology and stability of H348/352A virus particles. We also used negativestain electron microscopy to compare WT and mutant virus particles pelleted from the culture medium. This analysis showed that although H348/352A-infected cells produced many fewer virus particles than WT-infected cells, the mutant virus particles had apparently normal morphology (Fig. 6C versus D), in keeping with their unaltered specific infectivity. However, conformational differences in virus glycoproteins can affect the stability of the virus particle, which can have significant implications in virus pathogenesis and virulence (37). To compare the levels of WT and mutant stability, we incubated WT and mutant virus stocks at 50°C for 0 to 30 min and quantitated their infectivity after temperature treatment. While both viruses were increasingly inactivated by elevated temperature over time, the H348/352A mutant was significantly less

Byrd and Kielian
® stable than the WT virus, with a 2-log decrease in infectivity after only 5 min (Fig. 7). Thus, in addition to the mutant budding defect, alanine substitutions of H348 and H352 decrease virus particle stability.

DISCUSSION
We report here on the role of the conserved histidine residues in the juxtamembrane D-loop of the Alphavirus E2 protein. Both single and double alanine substitutions of H348 and H352 caused a significant decrease in virus growth. Detailed analysis of the H348/352A mutant showed that E2/E1 synthesis, dimerization, processing, and transport, as well as virus fusion and specific infectivity were not significantly affected. Although the hypothesis suggested by the virus structure was that H348 and H352 are involved in the low-pH-dependent dissociation of E1 and E2 during virus fusion (23), our data indicated that H348 and H352 do not have an important role in virus entry and fusion. Electron microscopy studies of mutant-infected cells showed that viral nucleocapsids were associated with the plasma membrane, diagnostic of E2-capsid interaction. However, there was little or no evidence for budding or release of virus particles. Thus, the major effect of the mutations was to inhibit a late stage in the virus exit pathway.
Negative-stain electron microscopy studies of released virus particles showed that the WT and mutant viruses had comparable spherical morphologies, with spike complexes visible on the virus envelope. We did note that even after adjusting dilutions to compensate for the reduced production of mutant virus, it was relatively difficult to visualize intact H348/352A particles. Temperature inactivation studies at 50°C showed that the H348/352A mutant was significantly more thermolabile, suggesting that the mutations also affect the architecture and stability of the mature virus particle. Alphavirus E2 Lateral Interactions and Virus Budding ® Because the mutant is blocked at a stage of virus assembly after E2-nucleocapsid interaction at the plasma membrane, we propose that the E2 D-loop is important for correct formation of the E2/E1 lattice during budding. To analyze this possibility further, we considered the previously described atomic models of the envelope protein shell derived from cryo-EM reconstructions of Chikungunya virus-like particles (CHIKV VLPs, see reference 38) and VEEV (see reference 23). The CHIKV VLP model reveals contacts between the E2 D-loop and the E1 protein stem region, with E2-H348 interacting with E1-S403 and E2-H350 stacking against E1-W409 (see details in Fig. 8A and B). The VEEV model shows similar interactions, with E2-H348 interacting with E1-S403 and E2-H352 stacking against E1-W409 (Fig. 8C). While the sequence of the membrane-proximal E1 stem region is not generally conserved, the two interacting residues E1-S403 and E1-W409 are highly conserved (see, e.g., reference 9). In spite of the limits in the resolution of the CHIKV and VEEV maps, the predicted interactions of the E2 D-loop and the conserved E1 stem residues are very similar between the two viruses, supporting the idea of a functional role in establishing the glycoprotein lattice. These models may also explain the apparent lack of additive effects of the E2-H348A and E2-H352A mutations in SFV, as there may be some plasticity in the interactions of the D-loop, with the SFV E2-W350 being able to stack with E1-W409, similarly to the stacking of E2-H350 with E1-W409 in CHIKV.
Previous studies of a Sindbis virus mutant, ts103, showed that it is inhibited at a late stage of virus budding and produces aberrant multicored particles that are less thermostable (39). The ts103 phenotype is due to a mutation of E2-A344 to V (40), a position at the N-terminal boundary of E2 subdomain D (23). Unlike H348 and H352, A344 is not highly conserved (Fig. 1C) and does not appear to make specific contacts with E1. While the effects of the E2-A344V mutation may be more indirect than those  The positioning of E2-H350 and E2-P351 helps to orientate the juxtamembrane helices N-terminal to the E2 transmembrane domain. Note that SFV residue numbering is used here for CHIKV E2. (C) Expanded view of the D-loop in the VEEV structure. E1 is shown in gray and blue and E2 in yellow. E2-H348 is in position to make a polar interaction with E1-S403. E2-H352 stacks against E1-W409 and together with E2-P351 helps to orientate the juxtamembrane helices N-terminal to the E2 transmembrane domain. Note that SFV residue numbering is used here for VEEV E2. The E2 residue numbering for CHIKV and VEEV is H349, P352, and H353 (see Fig. 1C). The E1 residue numbering is the same for SFV, CHIKV, and VEEV, and E1 residues S403 and W409 are highly conserved. Although the coordinates derived from the 4.8-Å to 5-Å maps are not very accurate, the fact that the same residues were found in interactions in two independent maps, together with the functional data provided here, suggests that these interactions are real. All images were prepared using PyMOL software (the PyMOL Molecular Graphics System, v1.2r2; Schrödinger, LLC). of the D-loop mutations, its partial rescue by a second-site mutation in the membranedistal tip of E1 (40) is also consistent with effects on the E2/E1 lattice.
The envelope and capsid proteins both form organized lattices during Alphavirus assembly. These lattices are connected via the binding of a Tyr-X-Leu motif in the E2 cytoplasmic domain to the capsid hydrophobic pocket (14,15), an interaction that is required for virus budding (13,41). The functional importance of the capsid lattice versus the E2/E1 lattice during assembly is not clear. The capsid protein can assemble with RNA into nucleocapsid-like structures in vitro (42)(43)(44)(45). Nucleocapsids can also assemble in the cytoplasm in the absence of envelope protein expression or virus infection (46,47). However, while self-assembly information is clearly contained within the capsid protein, under infection conditions there is evidence that nucleocapsid formation is regulated by the expression of the envelope proteins and the E2 cytoplasmic domain (47). Moreover, capsid mutants that block capsid-capsid interactions and cytoplasmic nucleocapsid assembly are still able to bud at the plasma membrane, where the interaction with the E2/E1 lattice through the E2 cytoplasmic domain is sufficient to drive formation of the nucleocapsid (48,49). There is also evidence that cytoplasmic nucleocapsids undergo additional structural changes or maturation in response to glycoprotein binding during budding (26,50). To add to the complexity of Alphavirus assembly, expression of the small hydrophobic proteins 6K and transframe appears to promote budding at the plasma membrane (24,51), although their relative roles and mechanisms are unclear. While many viruses use host cell machinery such as the endosomal sorting complexes required for transport (ESCRT) pathway to promote budding (52), Alphavirus budding is ESCRT independent (53). The role of other host proteins, however, is largely unknown.
While it is clear that many unanswered questions on Alphavirus assembly and budding remain, our work identifies new interactions in E2 and E1 that promote virus budding. The E2 D-loop residues H348 and H352 have specific interactions with the E1 stem, particularly with respect to residues S403 and W409. While the D-loop did not show the hypothesized function in dimer dissociation during entry, its interaction with its E1 dimeric partner proved important in promoting virus exit. Our results suggest that the E2 D-loop has a significant role in formation of the glycoprotein lattice necessary for membrane curvature and Alphavirus budding, highlighting the importance of E1-E2 lateral protein interactions in Alphavirus exit.

Cells and viruses. BHK-21 cells were propagated in complete BHK medium (Dulbecco's modified
Eagle's medium [DMEM] plus 5% fetal bovine serum, 10% tryptose phosphate broth, and 100 U penicillin/ml and 100 g streptomycin/ml) at 37°C unless otherwise noted. All plaque assays were performed using BHK cells. Vero cells were cultured at 37°C in DMEM plus 10% fetal bovine serum and 100 U penicillin/ml and 100 g streptomycin/ml. Mutant viruses were created by site-directed mutagenesis of the DG-1 plasmid, a subclone of the pSP6-SFV4 infectious clone (24), as previously described (54). In summary, mutations were introduced into the DG-1 plasmid using PrimeSTAR HS DNA polymerase (TaKaRa Bio Company, Kusatsu, Japan) using the following primers: for H348A, 5= CCACTGAAGGGAAACCGGCCGGCTGGC 3= (forward) and 5= CTGAT GCGGCCAGCCGGCCGGTTTCCC 3= (reverse); for H352A, 5= CCGCACGGCTGGCCGGCTCAGATCGTAC 3= (forward) and 5= GTACTGTACGATCTGAGCCGGCCAGCCGTG 3= (reverse); and for H348/352A, 5= GGGAA ACCGGCCGGCTGGCCGGCTCAG 3= (forward) and 5= GTACGATCTGAGCCGGCCAGCCGGCCGG 3= (reverse). DG-1 clones containing the correct mutation as validated by sequencing were subcloned into the pSP6-SFV4 infectious clone using the NsiI and SpeI restriction endonuclease sites. Infectious clones were verified by sequencing of the entire NsiI/SpeI fragment (Genewiz, South Plainfield, NJ). Infectious clone RNA was generated by in vitro transcription (24). Virus stocks were prepared by electroporation of BHK cells with infectious clone RNA and collection of the cell media at 24 h postelectroporation. For each mutant, 2 infectious clones were independently generated and sequenced and used to verify results as indicated.
Immunofluorescence. BHK cells were electroporated with WT or mutant RNA, seeded on coverslips, and incubated at 37°C for 8 h. Cells were washed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (PFA) at room temperature for 20 min. Cells were permeabilized in 0.02% Triton X-100 when appropriate and stained with primary antibodies, including a pAb against E1 and E2, or with MAbs specific for E1 or E2 (55). Cells were then stained with appropriate secondary antibodies conjugated with Alexa Fluor 488 (Molecular Probes) and with Hoechst 33342 nuclear stain (Invitrogen). Images were acquired by immunofluorescence microscopy.
Virus growth assays. BHK cells were electroporated with WT or mutant RNA, mixed with an equal concentration of nonelectroporated cells, plated in 6-well culture dishes, and incubated at 37°C. After 2 h of incubation, cells were washed once with infection media (minimal essential medium [MEM] with 0.2% bovine serum albumin [BSA], 100 U penicillin/ml, 100 g streptomycin/ml, and 10 mM HEPES [pH 8.0]) and incubated at 37°C or 28°C in fresh media. At the indicated time points, the cell culture media were collected and pelleted at 20.8 K ϫ g for 10 min to remove cell debris, and the titer was determined by plaque assay. Where indicated, statistics were calculated by 2-way analysis of variance (ANOVA) with Tukey's multiple-comparison test using Prism (56).
pH dependence of fusion. BHK cells were prebound with a virus multiplicity of infection (MOI) of 0.2 PFU/cell for 90 min on ice to prevent endocytosis. The cells were then treated for 3 min at 37°C with media buffered at the indicated pH to trigger virus fusion with the plasma membrane. Cells were incubated at 28°C for 16 to 18 h in complete BHK media containing 20 mM NH 4 Cl to prevent secondary infection and were scored for infection by immunofluorescence to determine the percentages of infected cells, which were comparable between the WT and the mutants. Statistical significance was determined by 2-way ANOVA with Tukey's multiple-comparison test using Prism (56).
Specific infectivity. BHK cells were infected with WT or H348/352A virus at an MOI of 10, incubated at 37°C, and washed at 1 h postinfection with infection medium. At 8 hpi, supernatants were collected and centrifuged to remove cell debris, and an aliquot was saved for plaque assay. The remaining supernatant was layered over a 20% sucrose cushion and centrifuged at 35 K rpm for 3 h at 4°C using an SW 41 rotor. Pellets were resuspended identically in TN buffer (100 mM NaCl and 50 mM Tris, pH 7.4) containing 10 g/ml leupeptin, 100 g/ml pepstatin, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and 20 g/ml aprotinin and incubated on ice overnight. WT virus samples were diluted 1:100 to account for 2-log reduction in growth of H348/352A. Virus suspensions were serially diluted in TN buffer with protease inhibitors, heated to 95°C with SDS sample buffer, and analyzed by SDS-PAGE. Gels were transferred to nitrocellulose membranes and blotted with MAb to E2 and Alexa Fluor 680-conjugated secondary Ab. The amount of E2 in each dilution was then quantitated using a Li-COR Odyssey Classic fluorescence imaging scanner (Li-COR Biosciences). The specific infectivity was calculated as the ratio of the virus infectivity to the concentration of E2 (number of virus particles). Statistical significance was calculated by unpaired t test using Prism software (56).

Extension induction and capsid expression. Vero cells plated on
MatTek glass-bottom chambers were infected with 5 PFU/cell of WT or H348/352A virus. Cells were incubated at 37°C and fixed with 4% PFA at 8 hpi, permeabilized with 0.02% Triton X-100, and stained with the appropriate Abs (anti-tubulin, E1 and E2 or 568 Phalloidin, E1/E2 pAb, and capsid Ab). Cells were imaged by confocal microscopy using an LSM5 Live DuoScan confocal microscope (Carl Zeiss MicroImaging, Inc.).
Assembly and dimer stability assays. BHK cells were infected with 10 PFU/cell of WT or H348/352A virus and incubated at 37°C. At 4.5 hpi, cell media were changed to MEM without methionine and cysteine and cells were starved for 20 min. At 5 hpi, cells were pulse-labeled with 100 Ci/ml [ 35 S]methionine/cysteine for 30 min at 37°C. Following the pulse-labeling, cells were washed with media containing a 10ϫ excess of cysteine and methionine and were chased for the indicated times. For assembly assays, supernatants were collected and immunoprecipitated with an E1/E2 pAb in the absence of detergent to allow retrieval of intact virus particles. Cells were lysed in TN buffer containing 1% Triton X-100, 1 mM EDTA, 1 g/ml pepstatin, 50 g/ml leupeptin, 1 mg/ml BSA, 1% aprotinin, and 1 mM PMSF, and the lysates were immunoprecipitated with an E1/E2 pAb and washed with radioimmunoprecipitation assay (RIPA) buffer (57) or were coimmunoprecipitated with an E1 MAb (dimer dissociation assay) using a modified RIPA buffer without SDS and Na-deoxycholate. Samples were evaluated by SDS-PAGE and imaged using a Storm 860 Phosphorimager (Molecular Dynamics).
Electron microscopy. BHK cells were infected with virus at an MOI of 10 and incubated at 37°C. At 7 hpi, cells were fixed for 30 min at room temperature with 2.5% glutaraldehyde-2% paraformaldehyde-0.1 M cacodylate buffer and were embedded and processed by the Einstein Analytical Imaging Facility for transmission electron microscopy. Images were acquired on a JEOL 1200X electron microscope.
To image virus by negative staining, BHK cells were infected at an MOI of 10 and incubated at 37°C for 12 h. Cell media were harvested, centrifuged to remove cellular debris, and pelleted through a 20% sucrose cushion as described above. Virus pellets were resuspended in 20 mM Tris (pH 7.4), adsorbed to glow-discharged electron microscopy grids, and stained with uranyl acetate. Samples were imaged on a JEOL 1400Plus TEM (SIG no. 1S10OD016214).
Virus thermostability. Virus stocks were diluted in infection media to 1 ϫ 10 8 PFU/ml, incubated for the indicated times at 50°C, and immediately placed on ice. Infectious virus was quantitated by plaque assay. Statistical significance was determined by performing multiple t tests using Prism (56) and a Bonferroni-corrected alpha.
Nucleocapsid gradients. Cytoplasmic nucleocapsids were analyzed by sucrose gradient sedimentation as previously described (26,49) with minor adjustments. BHK cells were infected with 10 PFU/cell and incubated at 37°C for 5 h. Cells were labeled with 50 Ci/ml [ 35 S]methionine/cysteine for 14 h at 28°C, followed by a brief (30-min) chase. Cells were lysed (100 mM Tris [pH 7.4], 50 mM NaCl, 2 mM EDTA, 1% NP-40, 40 M N-ethylmaleimide, 250 U/ml RNase inhibitor, 1 g/ml pepstatin, 5 g/ml leupeptin, 1 mM PMSF, and 1% aprotinin) and pelleted at 20.8 K ϫ g for 10 min to remove cell nuclei. Lysates were then incubated on ice for 20 min with 25 mM EDTA to dissociate polysomes. Samples were loaded onto 7.5% to 20% (wt/wt) linear sucrose gradients in TN buffer with 2 mM EDTA and 0.1% NP-40 and centrifuged at 41 K rpm for 2 h at 4°C in an SW41 rotor. Fractions of approximately 500 l were collected and analyzed by SDS-PAGE and scintillation counting. Gels were imaged using a Storm 860 Phosphorimager (Molecular Dynamics).

ACKNOWLEDGMENTS
We thank Félix Rey for his insights into the interactions of E2 H348 and H352 with E1 and for Fig. 8, which he very kindly generated for us. We thank our laboratory members Maria Guadalupe Martinez for help with confocal microscopy and data acquisition and Rebecca Brown for acquiring the TEM images of infected cells and for her comments on the manuscript. We thank the Einstein Analytical Imaging Facility for EM sample preparation and their expertise with TEM and Hillel Cohen for advice on statistical analysis. This work was supported by a grant to M.K. from the National Institute of Allergy and Infectious Diseases (R01AI075647) and by Cancer Center Core Support Grant NIH/NCI P30-CA13330. E.A.B. was supported in part by MSTP training grant T32 GM007288.
The data in this paper are from a thesis submitted by E.A.B. in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Graduate Division of Medical Sciences of the Albert Einstein College of Medicine. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.