Single Amino Acid Insertions at the Junction of th
http://www.100md.com
病菌学杂志 2005年第12期
Department of Molecular and Structural Biochemistry, North Carolina State University, Raleigh, North Carolina 27695
ABSTRACT
The final steps in the envelopment of Sindbis virus involve specific interactions of the E2 endodomain with the virus nucleocapsid. Deleting E2 K at position 391 (E2 K391) resulted in the disruption of virus assembly in mammalian cells but not insect cells (host range mutant). This suggested unique interactions of the E2 K391 endodomain with the different biochemical environments of the mammalian and insect cell lipid bilayers. To further investigate the role of the amino acid residues located at or around position E2 391 and constraints on the length of the endodomain on virus assembly, amino acid insertions/substitutions at the transmembrane/endodomain junction were constructed. An additional K was inserted at amino acid position 392 (KK391/392), a KF substitution at position 391 was constructed (F391), and an additional F was inserted at 392 (FF391/392). These changes should lengthen the endodomain in the KK391/392 insertion mutant or shorten the endodomain in the FF391/392 mutant. The mutant FF391/392 grown in BHK cells formed virus particles containing extruded material not found on wild-type virus. This characteristic was not seen in FF391/392 virus grown in insect cells. The mutant KK391/392 grown in BHK cells was defective in the final membrane fission reaction, producing multicored or conjoined virus particles. The production of these aberrant particles was ameliorated when the KK391/392 mutant was grown in insect cells. These data indicate that there is a critical minimal spanning distance from the E2 membrane proximal amino acid at position 391 and the conserved E2 Y400 residue. The observed phenotypes of these mutants also invoke an important role of the specific host membrane lipid composition on virus architecture and infectivity.
INTRODUCTION
The alphavirus Sindbis is an excellent model for the study of the assembly of membrane-containing viruses. Unlike the majority of enveloped viruses, which are loosely structured protein modified membrane envelopes, the alphaviruses are highly organized icosahedral protein shells with an associated membrane (17, 36, 40). The three proteins of Sindbis virus are organized into two nested protein shells with T = 4 icosahedral symmetry (38). The outer protein shell contains 240 copies of glycoproteins E1 and E2. The inner protein shell contains 240 copies of the capsid protein C which are assembled around the plus-sense single-stranded RNA (48). The structure of the virus is stabilized through specific E1-E1 interactions, and anchoring of the outer protein shell to the inner protein shell occurs through specific associations of the E2 glycoproteins with the capsid proteins across the intervening host cell membrane (1, 22, 23). The E2-capsid interactions are acquired in the final stages of virus assembly as the virus capsid is enveloped in the E1-E2 modified plasma membrane (5, 27). These E2-capsid interactions occur for each of the 240 copies of E2 and capsid protein holding the mature virion in a high energy, metastable conformation (2, 8, 34). Energy stored in the E1 glycoprotein is likely used in the process of cell penetration (37).
Alphaviruses are hybrid structures containing protein and RNA encoded by the virus and lipid and carbohydrate derived from the host cell (21). The process of self-assembly of these relatively simple viruses has been the subject of intensive study. Although many of the biochemical details of virus protein processing, modification, and maturation have been described, protein-protein interactions at the molecular level are less well understood. The virus structural proteins are synthesized from a subgenomic polycistronic RNA to produce a 130-kDa polyprotein NH-C-PE2 (E3+E2)-6K-E1-COOH (45). While capsid proteins assemble with progeny 49S viral RNA into nucleocapsid structures within the cell cytoplasm, the glycoproteins mature via the secretory pathway (7, 8, 33, 35).
The interaction between the outer and inner protein shells occurs through the attachment of the 33-amino-acid endodomain of glycoprotein E2 (E2 tail) to a hydrophobic cleft in the capsid protein (22, 23) specifically interacting with E2 Y400. The process of E2-nucleocapsid association has been shown to involve a minimum of two steps. The first step has been postulated to involve sequence-specific recognition of the capsid protein by E2 amino acid sequences encoded in the carboxy-terminal endodomain (amino acids 402 to 420) (26, 28). The second interaction is proposed to position the E2 endodomain within the nucleocapsid structure in close apposition to the inner surface of the membrane (23). These interactions occur within a domain of E2 not precisely determined which has been proposed to involve the burial of the E2 hydrophobic T398/P399/Y400 domain within the capsid protein hydrophobic cleft (23). The process of repeatedly burying the amphipathic E2 endodomain within each of the 240 capsid proteins composing the nucleocapsid is predicted to be energetically favorable. These repeated interactions could provide sufficient energy to drive the budding process which incorporates the underlying icosahedral nucleocapsid structure (47).
The structure of wild-type capsid protein has been solved by X-ray crystallography (9). This structure has been fit into the electron cryomicroscopic reconstruction of Sindbis virus which positioned the hydrophobic pocket of the capsid protein against the membrane at a position consistent with the site where E2 emerges from the virus membrane (23, 53). The hydrophobic capsid protein pocket is comprised of two compartments: the first is between capsid Y180 and W247 and the second is between W247 and F166, with W247 central to the pocket. The distance from the inner surface of the virus membrane to the position between Y180 and W247 in the capsid hydrophobic pocket is 28 ?. This distance corresponds well to the length of 10 amino acids, which also corresponds to the distance between the membrane-proximal E2 K391 and the hydrophobic E2 Y400. These observations suggested that an interaction between Y400 of E2 and Y180-W247 of capsid may play a critical role in assembly.
Genetic and biochemical analysis of E2-nucleocapsid interactions occurring during envelopment provide a general picture of the maturational events which E2 and capsid undergo prior to the release of a virus particle. The 33-amino-acid E2 endodomain can be divided into several overlapping functional domains involved in virus assembly. Sequences responsible for the initial capsid recognition (E2 amino acids 402 to 420) comprise one domain (20). A second domain includes the membrane-proximal K391 through the conserved T398/P399/Y400 sequence proposed to be the region which mediates tight E2-capsid binding. A mutant deleting the E2 402 LAPNA 406 sequence had no effect on virus production (25), further defining the E2 tail first binding domain to amino acids 407 to 420. The wild-type E2 domain K391 to L402 is the portion of the E2 tail proposed to reach into the capsid binding pocket and interlock the two virus protein shells (23, 52). We have proposed that this domain contains a critical number of amino acids to place Y400 in register with the correct amino acids in the capsid protein structure (20, 23). We have tested this hypothesis by using a deletion mutant which truncated the E2 endodomain upstream of the TPY sequence. A nonconserved K at position 391 located at the membrane/cytoplasm junction of the E2 endodomain was deleted. This mutant, K391, produced very little virus from BHK cells, and nucleocapsids failed to associate with cell membranes (20). These results supported the hypothesis that a critical length is required between the membrane and Y400 for correct capsid binding. Remarkably, the K391 mutant produced virus at wild-type levels in insect cells, suggesting a host range effect of mutations at this position. This result revealed two important aspects of virus assembly involving the contiguous E2 transmembrane domain (TM) and endodomain. The failure of K391 to assemble virus only in mammalian cells suggested that manipulation of the amino acid sequence at the juncture between the TM and the endodomain could functionally, if not physically, extend residues into an adjacent domain. A second hypothesis resulting from these observations was that, because of biochemical differences of mammalian and insect host membranes, mutations made within the E2 TM would result in host range phenotypes. These mutations were predicted to identify virus domains necessary for assembly in mammalian or insect membranes.
To test the second hypothesis, nested deletion mutants in the E2 26-amino-acid TM were constructed and grown in mammalian or insect cells (21). Nine- and 10-amino-acid deletions in the E2 TM producing mutants TM 17 and 16, respectively, were significantly inhibited in infectious virus production when grown in mammalian BHK cells but not in insect cells. This defect in assembly was attributed to an inability of the shorter TM to efficiently interact with the relatively thicker mammalian membrane bilayers, suggesting a role of the membrane in the virus architecture (21). A single-deletion mutant TM25 (379 M) produced very low levels of virus from both BHK and U4.4 cells and a high particle-to-PFU ratio (105 particles/PFU). This phenotype was predicted to result from altered ectodomain conformations affecting infectivity. These data suggested subdomains in the E2 TM, as was previously demonstrated for the E2 endodomain.
In the present study, we address the possibility that amino acids adjacent to a membrane can become incorporated into the TM or, inversely, extend the length of the endodomain. To test this hypothesis, mutations at E2 positions 391 and 392 were constructed. These mutants were designed to probe the function of the membrane/cytoplasm-proximal amino acids of E2 in the assembly of Sindbis virus. Amino acid additions and substitutions were chosen to incorporate residues which are predicted to preferentially incorporate into the TM or the endodomain (3, 11, 12). An additional K residue at E2 392 (mutant KK391/392) was chosen as a residue which would remain in the cytoplasm, while F at positions 391/392 (mutants FF391/392 and F391) was chosen as a residue which would partition into the membrane lipid. The E2 KK391/392, FF391/392, and F391 endodomain mutants were designed not solely as substitution or insertion mutants but also to specifically exploit the hydrophobic nature of the Sindbis virus TM and the hydrophilic properties of the cytoplasmic portion of the E2 endodomain. The properties of these mutations were examined in both insect and mammalian cells to determine the effect of membrane structure and composition on the mutant phenotypes.
MATERIALS AND METHODS
Cell culture, plaque assay, and virus. The source and culture of baby hamster kidney cells (BHK-21) was as described previously (42). Briefly, these cells were grown and maintained in minimal essential medium containing Earl's salts (MEM-E) (Invitrogen, Carlsbad CA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen), 5% tryptose phosphate broth, and 2 mM glutamine as described previously (42). The Aedes albopictus clones were subclones derived from Singh's original larval isolates (31). The U4.4 cells were cloned from cells originally provided by Sonya Buckley (Yale Arbovirus Research Unit, New Haven, Conn.) and is the original line grown in Mitsuhashi and Maramorosch medium (M and M) supplemented with 20% FBS (32). The virus strain chosen for mutagenesis was produced from the cDNA Toto 1101 clone (43) containing a substitution in E2 at position 420, changing a serine to a tyrosine, Y420, and has been described previously (27). Y420 serves as the wild type, and all virus mutants were constructed using this cDNA. This construct was chosen over the original Toto1101 cDNA to facilitate identification of mutations which fail to retract the E2 tail from the endoplasmic reticulum membrane (20). We have shown previously that this mutation is silent (27). Titration of virus produced was done using BHK-21 cells as indicator cell monolayers. Titers using BHK cells were done using the standard plaque assay. The virus diluent was phosphate-buffered saline (PBS-D) supplemented with 3% FBS as described previously (21, 42). Mutants were made using the Quick Change site-directed mutagenesis protocol as detailed in reference 21. Mutants were made using a smaller subcloned template pGEM 3Z TE12 or using the full-length wild-type template. For all mutants, the most convenient unique restriction sites were used to subclone the mutagenized nucleotides into the wild-type vector to eliminate spurious changes acquired during the amplification process. Mutagenic oligonucleotides were designed as described previously (21). The mutagenic sequences for the coding strands of the mutants KK391/392, FF391/392, and F391, respectively, are as follows: 5'-GCAGTGTTATGTGCCTGTAAAAAAGCGCGCCGTGAG-3', 5'-GCAGTGTTATGTGCCTGTTTTTTTGCGCGCCGTGAG-3', 5'-GCAGTGTTATGTGCCTGTTTTGCGCGCCGTGAG-3'. The E2 mutant TM25 has been described previously (21). This mutant will be referred to as TM25 (379 M) to specify the amino acid deleted in this construct. TM25 (379 M) was used as the template to construct the mutant TM25 (379 M) FF391/392 essentially as described above. TM25 (379 M) FF391/392 combines the single-amino-acid deletion encoded in TM25 with the double insertion mutant FF391/392. The same TM25 (379 M) deletion mutant was also used to construct an analogous mutant by combining with the KK391/392 mutant to create TM25 (379 M) KK291/392. All mutant sequences were verified by sequencing both the DNA used to make transcripts and reverse transcription-PCR products from RNA extracted from transfected BHK cells or mutant virus.
In vitro transcription and RNA transfection. The mutant and wild-type cDNA constructs were prepared for transcription and transcribed in vitro as described previously (20). The phenotypes of the E2 endodomain mutants were determined for both insect U4.4 and BHK host cells. BHK and U4.4 cells were transfected with synthetic RNA transcripts produced in vitro. Production of the RNA transcripts was as described in reference 20, and transcripts were electroporated into both BHK-21 and U4.4 cells. For the BHK cells, the electroporation was performed essentially as described by Liljestrom and Garoff (24). The BHK-grown virus was harvested once cytopathic effect was evident or 24 h posttransfection, whichever came first. All mutant strains are supplemented with 10 mM HEPES-HCl, pH 7.4, when virus was grown in BHK cells. The U4.4 cell transfection was performed as described previously (20) with the addition of 10 mM morpholinepropanesulfonic acid (MOPS)-HCl, pH 7.2, to the culture medium to maintain an optimal pH until virus was harvested 48 h posttransfection. Virus harvested from transfected cells was resuspended in 10% glycerol, divided into 1-ml aliquots, and flash frozen in liquid N2.
Metabolic labeling of BHK and U4.4 cells. BHK cells were transfected by electroporation as described above, and at 6 h posttransfection, 5 ml of fresh MEM containing 4 μg/ml actinomycin D (ActD) (Calbiochem, San Diego, CA) was added to a 25-cm2 flask of cells (5 x 106 cells) and incubated at 37°C for 1 h. The flasks were then washed twice with 5 ml of room temperature PBS before 5 ml of starvation medium (MEM deficient in methionine and cysteine supplemented with 2 mM glutamine, 3% FBS) was added to the cells and returned to 37°C for 1 h. The transfected BHK cells were then labeled with 50 μCi/ml [35S]methionine-cysteine ([35S]Met/Cys) and incubated at 37°C for 2 h for cell labeling or 24 h for virus labeling. Virus grown in U4.4 cells was also metabolically labeled to facilitate concentration of low-titer mutants used in fusion from without (FFWO) assays (see Results). Cells were attached to flasks in M and M in the absence of FBS for 1 h at 28°C. Infection proceeded essentially as described for BHK cells above. Labeling of U4.4 cells was done in M and M in the absence of ActD treatment immediately after infection using 3 ml M and M containing 50 μCi/ml [35S]Met/Cys.
Isopycnic centrifugation analysis of mutant viruses. Subconfluent monolayers of BHK-21 cells in 75-cm2 flasks were treated with 5 ml MEM containing 4 μg/ml ActD for 1 h. The ActD was removed, and the cell monolayers were infected with each of the mutant viruses grown in BHK cells for 1 h at 25°C at a multiplicity of infection (MOI) of 1 PFU/cell, after which the inoculum was removed and 5 ml MEM-E was added. Infected monolayers were incubated at 37°C for an additional 6 h. The medium was removed, and the cell monolayers were washed once with PBS-D and starved for Met/Cys in starvation medium at 37°C for 1 h. The medium was then removed and replaced with starvation medium containing 50 μCi/ml [35S]Met/Cys protein labeling mix. The monolayers were incubated at 37°C for 18 to 24 h. The virus supernatant was harvested and spun to equilibrium on linear 15 to 35% potassium tartrate gradients in PBS-D buffer, pH 7.4, at 24,000 rpm in a Beckman SW-28 rotor overnight. The entire gradient was collected in 0.5-ml fractions, and 5 μl of each fraction was counted by scintillation spectrometry for detection of the labeled virus fractions. Linearity of the gradient was determined by a measurement of the refractive index of each fraction. Titration of the same virus fraction was performed on BHK-21 cells as described previously (42).
Transmission electron microscopy and negative staining. BHK and U4.4 cells were transfected with RNA transcribed from either wild-type Sindbis virus or each of the endodomain mutants. For BHK cells, incubation proceeded at 37°C for 16 to 18 h, and U4.4 cells were incubated at 28°C for 30 h. Cell monolayers were scraped from the flasks and pelleted by low-speed centrifugation. The specific protocols used in the fixation and embedding procedures are detailed in previous literature (17). The samples were examined at 80 kV in a JEOL JEM 100S transmission electron microscope. Virus particles to be viewed by negative staining were prepared by infection of four 75-cm2 flasks as described above (see "Metabolic labeling of BHK and U4.4 cells"). For these samples, the amount of [35S]Met/Cys was limited to 10 μCi/ml to reduce any possible damage from the isotope on the integrity of the virus sample. Virus preparations were then purified as described above on 35% to 15% potassium tartrate linear gradients. Virus collected from the gradients was directly attached to carbon-coated grids, washed three times with sterile H2O, and negatively stained with 1% uranyl acetate.
FFWI and FFWO assays. BHK cell monolayers for fusion from within (FFWI) and FFWO assays were prepared in 24-well plates. For FFWI assays, the cells were infected in PBS-D containing 3% FBS at 28°C with an MOI of 10 PFU/cell or mock infected for 1 h and returned to 37°C for an additional 7 h of incubation. The fusion process is initiated by the addition of fusion medium (1.8 mM CaCl2, 5.3 mM KCl, 0.1 μM MgSO4 · 6H2O, 116 mM NaCl, 2.9 mM sucrose, 1x MEM-E amino acids [Invitrogen, Carlsbad, CA], 1x MEM-E vitamins [Invitrogen, Carlsbad, CA], 2 mM L-glutamine, and 0.2% FBS). Fusion medium is adjusted to pH 5.2 with 10 mM morpholineethanesulfonic acid (MES) buffer, to neutral pH values by the addition of 10 mM HEPES, or to alkaline pH values by the addition of 75 mM NaHCO3 to the desired pH. The infected and mock-infected cells were then treated with pH 5.3 fusion medium for 5 min and retained at this pH or replaced with fusion medium of the pH series that follows: pH 6.4, 6.7, 7.0, 7.2, 7.4, 7.6, 7.8, 8.0, 8.3, 8.5. Control wells of mock-infected or infected cells treated only with fusion media corresponding to the pH series shown above were also included. The cells were incubated at 37°C for 1 h in a standard fusion protocol (30). FFWO assays were done using virus grown in U4.4 cells. Cell monolayers prepared for FFWO were prepared in 24-well plates as described above. Monolayers which were just subconfluent were placed on ice for 15 min. While still on ice, 100 PFU/cell of each of gradient-purified mutant was added to the appropriate wells. Control pH treatments and the pH series used were as for the FFWI assays except that the low-pH treatment was at pH 4.6. Previous work has shown that virus grown in U4.4 cells fused maximally when taken to a lower pH than that required for BHK-grown virus (14). The cell monolayers were then scored for the amount of fusion seen relative to the low-pH, high-pH, or mock-treated samples as a percentage of the wild-type control. Cell fusion into polykaryons was defined as those cells which contained three or more nuclei/cell and is shown as a percentage of fusion seen for the wild-type infection, pH 5.3 to pH 7.4 for FFWI and pH 4.6 to pH 7.4 for FFWO.
RESULTS
Our previously published results have indicated that amino acids adjacent to the membrane interface can play critical roles in the ability of the virus to assemble on vertebrate and invertebrate membranes (20). To further explore this hypothesis, we produced mutations at the membrane interface of the E2 glycoprotein endodomain. These mutants were designed to alter a critical length of the E2 endodomain from the membrane-proximal amino acid at position 391 to Y400, which is predicted to occupy the capsid protein hydrophobic pocket (23, 52). The length of the E2 endodomain was increased from 33 to 34 amino acids by incorporating an additional K at amino acid 392 (KK391/392). This insertion was predicted to lengthen the endodomain, since a second K should not be preferentially incorporated into a TM of sufficient length displaying no hydrophobic mismatch (12, 13). A second insertion, replacing K391 with F391 (K391F) and inserting a second F at amino acid position 392 produced the mutant FF391/392. This second type of insertion was designed to physically shorten the endodomain sequence by relocating these hydrophobic aromatic amino acids to the TM within the membrane (3). As the transmembrane domain is increased in length, the endodomain is reduced in length by two amino acids (FF) (11). This experiment was done, in part, in an effort to duplicate the host range phenotype of the previously described deletion mutant K391 which was proposed to retract the A at position 391 into the TM (20). A third mutant substituting F391 for K391 was constructed to test the effect of a single F at this position. As a genetic test for the partitioning of the KK or FF391/392 insertions into a cytoplasmic or lipid environment, we employed an E2 single-amino-acid deletion mutant at E2 379 (TM25 379 M) described previously (21). This mutant produces large numbers of noninfectious virus particles and low virus titers. Incorporation of the FF391/392 sequence into the TM25 virus was expected to restore the ability of the virus to assemble infectious virus by replacing the deleted amino acid from TM25 (379 M) into the carboxy terminus of the TM by inserting the hydrophobic FF residues. Conversely, incorporation of the TM25 (379 M) deletion into the KK391/392 mutant was not expected to restore amino acids to the TM if the insertion remained within the cytoplasmic tail. In addition to investigating the effects of these mutations on the interactions of the E2 endodomain with the nucleocapsid during the process of envelopment, these mutants may also prove of value in understanding the interaction of the virus membrane proteins with the different lipid bilayers it encounters in its broad host range.
Virus production from mammalian (BHK) and insect (U4.4) cells. Infectious virus production from transfected BHK and U4.4 cells was determined by assay on BHK cells as described in Materials and Methods. Titers of Sindbis virus heat-resistant strain (SVHR), parental wild-type virus, and mutant virus were measured from both cell lines and are shown in Table 1. Both insertion mutants, KK391/392 and FF391/392, produced significantly less infectious virus than wild-type virus in both hosts. The KK391/392 mutant grew approximately 10-fold better in BHK cells than the FF391/392 mutant. This result was the inverse compared to the amount of virus generated from U4.4 cells. The KK391/392 mutant was significantly growth attenuated in U4.4 cells, producing approximately 20-fold less virus than the FF391/392 mutant. By contrast, the KK391/392 mutant generated about 60-fold-higher virus titers in BHK cells than in U4.4 cells. The FF391/392 mutant produced similar amounts of virus in BHK and U4.4 cells. The F391 mutant grew to levels equivalent to that seen for wild-type virus in both cell lines. The parental Y420 virus yields equivalent virus titers from both cell lines. Virus production from the TM25 (379 M) FF391/392 mutant in BHK cells reproduced wild-type levels of virus, producing titers of 1.1 x 109 PFU/ml. This result provides indirect evidence that the FF residues can complement the TM25 (379 M) deletion by incorporating into the TM of E2. Interestingly, the TM25 (379 M) KK391/392 mutant also restored virus titers to 2 x 108 PFU/ml, a level 1,000-fold higher than the TM25 (379 M) mutant titer and 45-fold higher than the KK391/392 mutant titer.
Particle/PFU ratios of U4.4-grown endodomain mutant virus. Previous results obtained with E2 TM deletion mutants demonstrated that assembly mutants which produced low titers could assemble large quantities of noninfectious particles. Virus particles which are noninfectious were measured by determining a particle/PFU ratio. Large particle/PFU ratios in mutant virus populations are indicative of structural instability, resulting in a loss of infectivity. To further characterize the endodomain phenotypes, particle/PFU ratios were determined for each of the endodomain mutants grown in U4.4 cells and are shown in Table 1. Of particular interest are the number of noninfectious virus particles produced in insect U4.4 cells from TM25 (379 M), which encodes a shortened TM (7.8 x 104 particles/PFU), and the mutant KK391/392, which encodes a longer endodomain (4.3 x 103 particles/PFU). The mutant FF391/392, which is hypothesized to encode a longer TM, displayed a much lower particle/PFU ratio than either of these two mutants (116 particles/PFU), suggesting greater virus stability. Both mutants KK391/392 and FF391/392 produced fewer total particles (108 particles/ml) than the F391 mutant (109 particles/ml) and SVHR or Y420 (1010 particles/ml). These data suggest that the KK391/392 insertion results in the highest mutant virus instability of the endodomain mutations.
Electron microscopy of endodomain mutant-infected BHK and U4.4 cells and purified mutant virus. Electron micrographs of thin sections of transfected cells and negatively stained mutant virus of each of the mutants from BHK and U4.4 cells are presented in Fig. 1 to 7. In Fig. 1A and C and Fig. 2 to 4, images from BHK cells are presented, while Fig. 1B and D and Fig. 5 to 7 are images from U4.4 cells. All of the endodomain mutants were found to assemble nucleocapsids which are seen free in the cytoplasm and associated with cell membranes, as is typical of wild-type virus-infected BHK cells (Fig. 1A). There are, however, significant differences in the appearance of cells infected with the various mutants. Fewer cytoplasmic nucleocapsids are seen in BHK cells infected with FF391/392 than in wild-type infection. The reason for this is not known. Virus protein evaluated by immunoprecipitation of infected cells displayed normal processing and equivalent levels of protein compared to wild-type virus-infected cells (data not shown). While virus is seen budding from the plasma membrane (Fig. 2B), the outer virus membranes do not appear smooth and spherical but appear to be associated with nonvirus material (Fig. 2B). The virus cores, however (Fig. 2A and C), appear symmetrical. Deformations in the virus surface were seen more clearly in negatively stained preparations of purified FF391/392 virus (Fig. 2D and E). The FF391/392 virus displayed small extrusions of material, giving the particles a peculiar "lollypop" or "mouse ears" appearance. In thin sections (Fig. 2B and C), these particles did not appear uniform and symmetrical compared to virus produced from a wild-type infection (Fig. 1A). The possibility that this material was in part lipid is suggested by its appearance and the shift to a lower density in potassium tartrate gradients (Fig. 8A). Electron micrographs of cells infected with the endodomain mutant KK391/392 revealed a different phenotype (Fig. 3A to G). While this mutant appeared to produce normal numbers of nucleocapsids compared to the wild type and many of these capsids were found associated with membranes, this mutant is defective in the final stages of virus envelopment (Fig. 3A, B, and C). The KK391/392 mutant did not effectively recruit modified membrane to envelop single particles but rather incorporated multiple nucleocapsids into a bolus or tube of partially enveloped virus cores. In Fig. 3D, E, F, and G, negatively stained virus particles containing multiple cores in membranous vesicles of irregular size are shown (18). These larger multicored particles were found in the peak of virus infectivity shown in Fig. 8A and B. This fraction contained many double, triple, and quadruple conjoined particles. These multicored particles exhibit reduced density compared to wild-type virus (shown in Fig. 8A) and the FF391/392 mutant (compare the titers of fractions 4 and 5 in Fig. 8A and B). In accordance with the high virus titers measured for the mutant F391 in both BHK and U4.4 cells, electron micrographs of cells infected with this mutant (Fig. 4A and B) reveal large numbers of cytoplasmic capsids, membrane-associated capsids, and many particles budding from the plasma membrane. Likewise, the negative stains of virus produced by this mutant shown in Fig. 4C and D reveal many well-formed virus particles interspersed with few empty particles, which appear collapsed and electron dense. Some dimpled and collapsed particles are expected in negatively stained virus particles and are artifacts of the preparation.
A comparative analysis of the assembly and morphology of the endodomain mutants was also conducted in insect cells. A micrograph of wild-type virus-transfected U4.4 cells shown in Fig. 1B reveals large quantities of matured particles found within intracellular vesicles, a characteristic of Sindbis virus assembly in mosquito cells (18). Virus is released from the insect cells by a process similar to exocytosis, with few particles budding directly from the plasma membrane. Negative stains of wild-type virus produced from U4.4 cells (Fig. 1D) appear identical to those of wild-type virus produced from BHK cells (Fig. 1C). Any structural differences imparted by insect-specific lipids (29) and glycosylation patterns (6) are indiscernible at this resolution. Insect (U4.4) cells transfected with the FF391/392 mutant (Fig. 5A and B) produced many more virus particles than BHK cells (compare Fig. 2A, B, and C). This finding is in agreement with the titers of virus produced from each of the cell lines (Table 1). The mutant FF391/392 produced twofold-more infectious virus from U4.4 cells than BHK cells. A significant difference seen in thin sections and negative stains of the mutant FF391/392 from insect cells compared to those of BHK cells is the absence of extruded or external material associated with either cell-associated virus (Fig. 5A and B) or purified virus (Fig. 5C, D, E, F, and G). The negative stains of virus produced from insect cells show a small number of multicored and larger particles not seen in virus from BHK cells. These differences in FF391/392 virus structure were not the effect of revertants or adaptations in the virus sequence (data not shown) but are an insect host-specific effect on virus assembly. Of the two endodomain insertion mutants, the differences imparted by the U4.4 cells on the FF391/392 mutant produced virus particles with a particle-to-PFU ratio closest to that seen for the wild type (116 and 33 particles/PFU, respectively) (Table 1) compared to the KK391/392 mutant (4.3 x 103 particles/PFU) (discussed below).
Thin sections of KK391/392 in U4.4 cells (Fig. 6A, B, and C) revealed large quantities of virus particles. This observation is consistent with the large amount of noninfectious virus produced from U4.4 cells (Table 1). In thin sections, fewer virus particles and fewer elongated virus tubes (Fig. 6C) are found for this mutant in U4.4 cells than were seen in BHK cells (compare Fig. 3A, B, and C). The defect in the assembly of KK391/392 in U4.4 cells is illustrated in Fig. 6B (inset). Long stalks of membrane at the base of the budding viruses suggest a defect in the final stage of release of the virus from the cell membrane. The particle/PFU ratio was determined for KK391/392 mutant virus grown in insect U4.4 cells and is 4.3 x 103 particles/PFU compared to 33 particles/PFU found for wild-type virus (21). Negative stains of KK391/392 virus produced from insect cells (Fig. 6D, E, and F) reveal abnormalities on the virus surface. Compared to the wild type (Fig. 1C), three times more KK391/392 particles appear pocked (Fig. 6D and F) or dimpled (Fig. 6C), which may reflect instability in virus structure. Additionally, many more KK391/392 particles (37% of 315 particles examined) have lost the ability to exclude stain and appear empty (Fig. 6E) than have particles of the wild type (10% of 344 particles examined). These observations may reflect the loss of the nucleocapsids from less stable virus structures. The deformities seen in the mutant particles could also be the result of inefficient virus release, as was seen for this mutant produced from BHK cells (Fig. 3A, B, and C). Growth in insect cells, however, did not result in the production of the multicored particles assembled in BHK cells (seen in Fig. 3D, E, F, and G). This mutation may result in defective lateral protein associations required to seal the virus membrane during the final stages of budding from both host cells. In contrast to KK391/392 virus grown in BHK cells, defects in assembly of this mutant in U4.4 cells could be compounded by a thinner, less fluid insect membrane (4, 10) producing a more fragile, less infectious virus structure.
Thin sections of insect cells infected with the F391 mutant displayed typical virus particles in cytoplasmic vesicles (Fig. 7A and B) as well as a few virus particles budding from the cell surface (Fig. 7A), as is typically found in wild-type infections (Fig. 1B). The mutant F391 produces high virus titers in both cell lines, and this mutant's structural integrity is underscored by a particle/PFU ratio of 3 for virus produced in insect cells (Table 1).
Density gradient analysis of BHK cell-grown mutant viruses. Negative stains of mutant viruses grown in BHK cells revealed abnormal structures (Fig. 2 to 4). To evaluate what changes these defects in assembly may have on the buoyant densities of BHK-grown virus and to correlate any changes to altered infectivity, we determined the density of particles produced by the mutants in potassium tartrate density gradients. Virus from [35S]Met/Cys-labeled infected BHK cells were prepared as described in Materials and Methods, and their densities were determined. The results of these experiments are presented in Fig. 8A. Wild-type virus shows a large peak in fraction 3 corresponding to a density of 1.17 g/cm3. Figure 8B presents the virus titers of the peak fractions. Infectivity for wild-type virus is found predominantly in fraction 3 (2.3 x 1010 PFU/ml), with a small shoulder of infectivity in fraction 4 (1.0 x 107 PFU/ml). In Fig. 8A, the mutant F391, which grows to wild-type levels, has a density profile and distribution of infectivity similar to wild-type virus. The major FF391/392 virus peak in fraction 4 shows a slight shift in density corresponding to the density of the wild-type virus shoulder. This fraction also contained the peak virus titer of FF391/392 (7.0 x 107 PFU/ml) (Fig. 8B, fraction 4). The peak virus titer of FF391/392 (7.0 x 107 PFU/ml) is higher than that of the virus titer for the fraction of the same density of wild-type virus (1.0 x 107 PFU/ml), although the total counts per minute for this wild-type fraction was higher than that of the counts per minute/ml for FF391/392 fraction 4 (Fig. 8A). This result was due to a lower specific activity of the FF391/392 virus (data not shown). The mutant KK391/392 demonstrated the most pronounced shift in virus density (Fig. 8A, fraction 5). Most of the KK391/392 virus was found in fractions 4 and 5 (Fig. 8A), with the majority of infectious virus found in fraction 5 (Fig. 8B). This shift in virus buoyant density from the wild type (1.17 g/ml3) to a reduced density of 1.154 g/ml3 was found to be a consistent characteristic of the mutant KK391/392 and could be the result of a higher lipid content in the multicored particles.
FFWI and FFWO mediated by endodomain mutants. The ability of Sindbis virus glycoprotein to induce cell-cell fusion after brief exposure to acid pH has been well characterized (30, 49). In BHK cells, it has been shown that this process is a two-step event requiring exposure to pH 5.3 followed by a return to pH 7.2 for optimal fusion to be observed (15, 30, 37). These experiments can be conducted by the addition of virus particles to a cell monolayer (FFWO) or by using proteins exported to the cell surface after virus infection (FFWI). FFWI is a fusion assay which can be utilized for mutants which do not grow to the high titers required by FFWO (100 to 1,000 PFU/cell) (16). The pH requirements for the induction of FFWI and FFWO are the same for wild-type virus (30). We have previously used the property of Sindbis virus to fuse cell monolayers from within as an assay for export of functional E1-E2 heterodimers to the plasma membrane (30). Analysis of the presence of matured E1-E2 complexes at the cell surface by FFWI suggests that sufficient protein reaches the plasma membrane for 100% fusion at the midphase of the transfection (10 h posttransfection). We have employed FFWI as an assay for the functional conformation of the E1-E2 heterotrimers of the mutants described above. We reasoned that a significant change in the glycoprotein ectodomain conformations and/or lateral interactions at the cell surface would be reflected in changes in the pH requirements to yield optimum cell fusion. We hypothesize that conformational differences in the mutant virus particles are imparted by the host cell membrane component in addition to the heterodimer conformations. Mutant virus grown in BHK cells would present altered conformations compared to those produced in U4.4 cells. These changes would be reflected as a change in the pH required to achieve maximum fusion for BHK-grown virus or U4.4-grown virus. As discussed by Paredes et al. (37), exposure to low pH followed by return to neutral pH destabilizes virus structure. Conformational changes induced in the mutant virus particles may affect the pH at which the structural changes are seen. In separate experiments, BHK cells were transfected or infected, at an MOI of 10 PFU/cell, with wild-type virus or one of the three endodomain mutants. The infections were allowed to proceed as detailed in Materials and Methods, and fusion levels were scored as described in Materials and Methods and as described previously (14, 30). Fusion profiles (FFWI) of the endodomain mutants in BHK cells are shown in Fig. 9A. The values plotted represent the percentages of cell-cell fusion produced by each mutant and are compared to the percentages of fusion of wild-type virus. Mock infections were done for each of the pH treatments shown. In each experiment, the infected monolayer was exposed to pH 5.3 for 5 min to establish the conditions for fusion and then returned to the indicated pH. Both the control wild-type virus and the KK391/392 mutant displayed similar fusion profiles, with KK391/392 producing slightly less fusion at pH 6.4 (20% less fusion than wild type). Both wild-type virus and KK391/392 produce 100% fusion after return to a pH of 7.2. The mutant F391 displayed a slightly more convex curve than that of wild-type virus, showing 20% more fusion at pH 6.7 and 10% more fusion at pH 7.0. At a pH of 7.4, the F391 mutant reaches levels of 100% fusion, as seen with wild-type virus. These mutants displayed similar fusion profiles. The most interesting fusion profile of the endodomain mutants was the FF391/392 virus (Fig. 9A). The mutant FF391/392 displayed a complex curve, producing three separate plateaus of increasing cell fusion. The first curve is seen at pH values of 6.4 to 7.2, where fusion increased from 30% to 60%. At pH 7.2 to 7.4, the amount of fusion is maintained at 60%. This is followed by a second plateau from pH 7.6 to 7.8 which displays 80% fusion. The FF391/392 mutant ultimately achieves 100% fusion at pH 8.0 to 8.5. Compared to the wild type, FF391/392 requires a pH increase of 0.8 pH units to produce 100% BHK cell-cell fusion. Mock-infected monolayers displayed negligible fusion after the treatments described above compared to the wild-type controls. It is not possible to determine the exact amounts of E1 and E2 that are present on the cell surface in the fusion from within assay. It has been shown that the E1 and PE2 (precursor to E2) are paired as heterodimers in the endoplasmic reticulum. Formation of the heterodimer is required for export to the cell surface, suggesting that the proteins are present in equimolar quantities in the plasma membrane (35). We have shown that fusion of cells by mature virus requires as few as 1,000 viruses per cell (16). That the cells infected with the different mutants demonstrate 100% fusion from within indicates that they have delivered at least 1,000 virus equivalents of protein to the cell surface. The pH at which fusion is demonstrated is a property of the virus proteins (19) not the amount of protein on the cell surface (30).
The possibility that the endodomain mutant particles grown in insect U4.4 cells were structurally more similar to wild-type virus was suggested by the images produced in negatively stained preparations. This possibility was tested by comparing the ability of the mutants and wild-type virus to produce FFWO, reasoning that viruses with similar conformations would retain similar pH fusion profiles. FFWO, as apposed to FFWI, was necessary to evaluate U4.4-grown virus fusion with BHK cell membranes. Each of the endodomain mutants, in addition to wild-type virus, was gradient purified and concentrated to enable the addition of 1,000 PFU/cell. FFWO was performed as described in Materials and Methods. Shown in Fig. 9B are the FFWO profiles of U4.4-grown virus on BHK cells. Wild-type virus gave 100% fusion at all pH values tested, with the exception of pH 6.4, which gave 80% fusion. The fusion profile for virus produced from U4.4 cells differed from that seen in FFWI of BHK cells by a shift in the pH required for maximum fusion from pH 7.2 (BHK cells) to pH 6.7 (U4.4 cells). Differences in the pH requirements for fusion of BHK and U4.4 membranes have been previously described (14). F391 from U4.4 cells became 100% fusion competent at similar pHs in both host cells (pH 7.4 in BHK and pH 7.2 from U4.4). This mutant, however, differed in its response to the higher pH values (pH 8.0 to 8.5), showing 75% fusion compared to 100% fusion in BHK cells. The KK391/392 mutant, which most closely followed the wild-type virus curve in BHK cells, most closely resembled F391 when grown in U4.4 cells. This mutant showed 100% fusion at pH 7.2 to 7.6, began to decline to 75% fusion at pH 7.8 to 8.3 and, finally, only produced 50% fusion at pH 8.5. As with F391, KK391/392 displayed less fusion at the higher pH values (pH 7.8 to 8.5). Again, as seen with FFWI in the BHK cells, the most interesting result was seen from the FF391/392 mutant. In contrast to the results obtained from BHK cells, this mutant displayed 100% fusion at pH values from pH 6.4 to 7.8, while fusion efficiencies began to fall at pH 8.0 (75% fusion) and gave the least fusion (60%) at pH 8.3 to 8.5, pH values that gave 100% fusion in BHK cells (Fig. 9A). Cell monolayers not exposed to virus, as shown in Fig. 9A, displayed negligible fusion. This result is opposite that seen in BHK cells and suggests that the protein conformations and/or the protein lateral contacts may be different for FF391/392 when grown in BHK or U4.4 cells. All insect cell-grown viruses gave maximum fusion in the pH range between 7.2 and 7.6. These results demonstrate that this assay may be used as a relative probe of altered virus ectodomain conformations which may affect virus function.
DISCUSSION
The final event in the assembly of the alpha virion is the envelopment of the preassembled core structure within a virus protein-modified cell membrane. This interaction is mediated by a specific association between the endodomain of the E2 glycoprotein and a hydrophobic cleft in the capsid protein. The process is complicated by the broad host range exhibited by the alphaviruses. It is possible, therefore, to study virus assembly in two genetically and biochemically unrelated host cells, allowing the comparison of assembly events which are unique to each host and those structural interactions which are host independent (21). Sindbis virus has adapted to accommodate the biochemically distinct lipid environments of mammalian and insect cell membranes. Mammalian membranes contain cholesterol, are more viscous and ion impermeable, and are composed of longer-chain fatty acids than those found in insect cells (4, 29, 41). Insect cell membranes have very low levels of cholesterol (44) and display a lipid composition distinct from that found in mammals (10, 29). These important biochemical distinctions affect the thickness and fluidity of the lipid bilayers (51).
Installing aromatic amino acids at the TM lipid/cytoplasm junction was undertaken as a means of reorganizing the endodomain architecture by converting a hydrophilic portion of the E2 endodomain into a structure which would be more energetically favored as a part of the helical TM (3, 11, 12, 13). The phenotypes displayed by the endodomain mutants segregated the mutant viruses into two general categories. Mutants which displayed reproducible phenotypes from both mammalian and insect cells represent host-independent assembly mutants. These mutants would be of the type which are defective in some aspect of assembly because of a virus-encoded structural alteration and engage both host cells in an equivalent way. The second phenotype would be the result of mutant protein interactions with the lipid membrane acquired by the virus during the process of envelopment. The mutants KK391/392 and FF391/392 represent both categories.
The KK391/392 mutant extended the E2 endodomain by a second basic K residue and was expected to physically lengthen the endodomain (13). The length of the hydrophobic segment of an integral membrane protein should closely match the membrane which it spans, producing an energetically favorable structure with charged residues partitioned into more polar environments. A second K at E2 amino acid 392 would not be expected to be energetically stable within the lipid environment under conditions of no hydrophobic mismatch. The TM25 (379 M) KK391/392 mutant, however, restored virus production to levels above those of both individual mutants in both host cell lines. This result suggests that the membrane-proximal K at position 391 may be able to incorporate into the TM, although with 10-fold less efficiency than the F391 residue. Analysis of mutations in pseudorevertants of TM (379 M) show that compensatory insertions of hydrophobic amino acids restore the length of the TM to 26 amino acids (unpublished data). These results demonstrate a strong selective pressure to maintain 26 amino acids in the TM of Sindbis virus. The incorporation of the K at 391 in the mutant KK391/392 restored the TM length to 26 amino acids. Lysine residues are known to have the capacity to incorporate into the periphery of membrane bilayers by a phenomenon termed "snorkeling." This refers to the ability of the flexible side chain of lysine to stretch out of the membrane interior and place the charged amino group in the polar interface (46). These observations suggest that, for the KK391/392 mutant, both lysine residues are located within the cytoplasmic domain, producing the altered particle phenotypes and reduced titers.
The hypothesis that increasing/decreasing the distance between the membrane bilayer and E2 Y400 would define a critical distance between the membrane interface and the capsid binding pocket (20, 23, 52) was tested by inserting an additional K at the membrane/cytoplasm interface (KK391/392). The extended E2 endodomain resulted in significantly lower virus titers from both BHK and U4.4 cells. This effect was not the result of limited or altered protein expression in transfected cells, as determined by polyacrylamide gel electrophoresis (data not shown), or a failure to transport membrane glycoproteins to the plasma membrane (Fig. 9A). The KK391/392 virus density (1.154 g/cm2) compared to that of the wild type (1.17 g/cm2) provides evidence that the particles produced from this mutant are less dense and might contain more membrane than wild-type virus. Thin sections of infected BHK cells show tightly apposed binding of the nucleocapsids to the virus-modified membrane (Fig. 3). The initial association events, however, do not proceed to efficient envelopment; instead, many multicored virus tubes are formed. Negative stains of gradient-purified virus from this mutant reveal multilobed particles containing up to four virus cores. This observation suggests that the shift in virus density seen for the KK391/392 virus results from a higher membrane content in the conjoined particles.
In both the vertebrate and invertebrate cell systems, it would be expected that an additional lysine residue would extend the E2 endodomain without affecting the TM. If the defect produced by extending the E2 endodomain by an additional residue resulted in altering an assembly step in a manner which is host cell independent, the phenotypes in both mammalian and insect cells should be identical. That this is the case was demonstrated by the observation that virus assembly from U4.4 cells produced tubes of multicored viruses as well as single virus particles displaying discontinuity at the virus surface. Thin sections of KK391/392-infected U4.4 cells showed a much lower incidence of membrane tubes than seen for the same mutant grown in BHK cells. In U4.4 cells, this mutant produced completely enveloped, although distorted, virus particles compared to the conjoined particles produced in BHK cells. These particles, however, are noninfectious, as shown by the high particle/PFU ratio. The disposition of the E2 tail in this mutant, while affecting envelopment in both hosts, is able to complete the envelopment process and produce noninfectious virus when grown in U4.4 cells.
The FF391/392 mutant had a different phenotype in insect cells than that seen in BHK cells. FF391/392 grown in U4.4 cells appears to be wild type in morphology in both thin sections of transfected cells and in negative stains of gradient-purified virus. Mutant virus from insect cells assembles without the external aggregated material seen in virus from BHK cells and is found within vesicles typical of insect cell maturation (Fig. 6A). FF391/392 virus grown in U4.4 cells displayed a fusion phenotype most similar to that of wild-type virus (Fig. 9B). Membranes of the insect U4.4 cells might better accommodate the additional aromatic residues, perhaps through membrane lipid reorganization (11). Particle/PFU ratios of FF391/392 virus grown in U4.4 cells are 60% of the values seen for wild-type virus (116 to 33, respectively), evidence of a relatively stable structure, compared to the KK391/392 mutant (103 particles/PFU).
FF391/392 virus particles from BHK cells are deformed by the presence of extraneous material at the virus surface. Electron micrographs of thin sections of FF391/392 in BHK cells reveal a significant number of virus particles budding from the plasma membrane (Fig. 2B and C). These budding particles appear to be associated with extraneous or extruded material. Negative stains of gradient-purified FF391/392 virus particles from BHK cells contain significant numbers of broken particles and particles displaying distorted morphology (Fig. 2D and E). The composition of the virus appendages producing the lollypop appearance is not known, but particles from BHK cells are less dense than wild-type virus, suggesting an increase in the lipid-to-protein ratio. These "lobes" might result from additional material incorporated into virus associating with the outer shell through weaker protein interactions. This possibility is intriguing, since the negative stains of the same mutant grown in insect cells do not contain this extruded material (Fig. 5C to G). In U4.4 cells, lower membrane viscosity and fluidity of these membrane lipids may allow the bilayer to reorganize without distortion. One specific property demonstrated by this mutant is a change in the pH at which it demonstrates maximum membrane fusion in BHK cells. Maximum fusion by the FF391/392 mutant required exposure to pH 8.0 after initial exposure to pH 5.3 and provides biochemical evidence for a conformational difference of the E1 ectodomain residues in the E1-E2 heterotrimers. Virus grown in U4.4 cells demonstrated a larger pH range (7.2 to 7.8), resulting in 100% fusion, and produced 75% fusion at pH 8.0. The fusion pH profile data suggest conformational differences in the virus particles produced from BHK cells and those assembled in U4.4 cells.
We predicted that the FF residues would partition into the membrane-spanning domain of E2. This is supported by the observation that the mutant TM25 (379 M) FF391/392 restores normal virus production. This result suggests that at least one F residue is incorporated into the interface of the TM of TM25 (379). In a related study, Parks produced analogous insertions of L into the signal/anchor domain of paramyxovirus simian virus 5 hemagglutinin-neuraminidase. These mutants provided indirect evidence for incorporation of flanking hydrophobic residues into the TM of an integral membrane protein (39). Assuming that both FF391/392 are shifted in residence from the cytoplasm to the membrane, this mutant would reduce the distance between the membrane and E2 Y401 (FF391/392 numbering) to 9 amino acids (393ARRECLTPY401). This shift, however, did not appear to affect the attachment of nucleocapsids to protein-modified BHK membranes, as seen in Fig. 2A, B, and C. The present E2-C interaction model predicts that misalignment of the critical E2 Y400 residue with the capsid binding pocket may not affect primary binding of the nucleocapsid but might lead to weak capsid binding associations affecting virus assembly or infectivity (23). The phenotype of the FF391/392 mutant was consistent with defective assembly (Fig. 2). We were not able to determine a particle/PFU ratio for this mutant from BHK cells; however, the numbers of virus particles seen in the negative stains shown in Fig. 2D and E are sufficient to suggest that this number is much larger than the titer for this fraction would suggest (7 x 107 PFU/ml) (Fig. 8B), implying an effect on infectivity as well.
The substitution of F for the K at amino acid 391 in the mutant F391 resulted in virus with wild-type properties. This result suggests that E2 A392 may not be incorporated into the TM in the absence of the basic lysine residue. Such a shift would result in 31 amino acids in the endodomain, with 8 amino acids from the membrane to Y400. That assembly proceeds normally suggests that E2 A392 remains in the cytoplasmic domain, leaving R393 and R394 nonadjacent to the membrane. While we are unable to assign a specific domain for F391 in this construct, the possibility arises that, for Sindbis virus E2, a basic residue may not be a strict requirement at the membrane interface (50). If the single F391 substitution is incorporated into the membrane and becomes part of the TM, the effective spanning length of the E2 endodomain to the critical Y400 residue which produces efficient interactions with the capsid binding pocket is 9 amino acid residues.
Collectively, the data gathered from the endodomain mutants add to our knowledge of Sindbis virus assembly in several important ways. These observations suggest that interactions of the E2 endodomain with capsid protein involve several specific associations. (i) There is a critical spanning distance of 9 to 10 amino acids from the E2 amino acid 391 to the critical E2 Y400 for correct interaction with the capsid binding pocket. (ii) Close apposition of the nucleocapsid to the membrane-proximal amino acid 391 within the E2 tail must occur for efficient budding to proceed. (iii) Lateral protein interactions of E1-E2 ectodomains affect the efficient envelopment of virus particles after initial association with the nucleocapsid. (iv) An acidic residue may not be required at amino acid 391 of the E2 endodomain, although appropriate spacing from the membrane juncture to Y400 (9 amino acids) is required for assembly. (v) The composition of the viral membrane is important to virus architecture and assembly. Certain conformations of the virus glycoproteins were observed to be better accommodated in the thinner, less viscous insect membrane. The contiguous TM and endodomain display some flexibility for the amino acid residues found at the membrane-flanking positions, suggesting that certain structural constraints supersede the requirement for a specific amino acid residue. These constraints involve the specific host membrane and may represent the adaptations which these proteins have acquired to shuttle between such divergent host environments.
ACKNOWLEDGMENTS
This research was supported by grants from the National Institutes of Health (AI-42775), The Foundation for Research (Carson City, NV), and the North Carolina Agricultural Research Service (NCARS).
REFERENCES
Anthony, R. P., and D. T. Brown. 1991. Protein-protein interactions in an alphavirus membrane. J. Virol. 65:1187-1194.
Anthony, R. P., A. M. Paredes, and D. T. Brown. 1992. Disulfide bonds are essential for the stability of the Sindbis virus envelope. Virology 190:330-336.
Braun, P., and G. von Heijne. 1999. The aromatic residues Trp and Phe have different effects on the positioning of a transmembrane helix in the microsomal membrane. Biochemistry 38:9778-9782.
Bretscher, M. S. 1993. Cholesterol and the Golgi apparatus. Science 261:1280.
Brown, D. T., M. R. F. Waite, and E. R. Pfefferkorn. 1972. Morphology and morphogenesis of Sindbis virus as seen with freeze-etching techniques. J. Virol. 10:534-536.
Burke, D. J., and K. Keegstra. 1976. Purification and composition of the proteins from Sindbis virus grown in chick and BHK cells. J. Virol. 20:676-686.
Carleton, M., and D. T. Brown. 1996. Disulfide bridge-mediated folding of Sindbis virus glycoproteins. J. Virol. 70:5541-5547.
Carleton, M., H. Lee, M. Mulvey, and D. T. Brown. 1997. Role of glycoprotein PE2 in formation and maturation of the Sindbis virus spike. J. Virol. 71:1558-1566.
Choi, H. K., L. Tong, W. Minor, P. Dumas, U. Boege, M. G. Rossmann, and G. Wengler. 1991. Structure of Sindbis virus core protein reveals a chymotrypsin-like serine proteinase and the organization of the virion. Nature 354:37-43.
Clayton, R. B. 1964. The utilization of sterols by insects. J. Lipid Res. 5:3-19.
de Planque, M. R., J. W. Boots, D. T. Rijkers, R. M. Liskamp, D. V. Greathouse, and J. A. Killian. 2002. The effects of hydrophobic mismatch between phosphatidylcholine bilayers and transmembrane alpha-helical peptides depend on the nature of interfacially exposed aromatic and charged residues. Biochemistry 41:8396-8404.
de Planque, M. R., and J. A. Killian. 2003. Protein-lipid interactions studied with designed transmembrane peptides: role of hydrophobic matching and interfacial anchoring. Mol. Membr. Biol. 20:271-284.
de Planque, M. R., J. A. Kruijtzer, R. M. Liskamp, D. Marsh, D. V. Greathouse, R. E. Koeppe II, B. de Kruijff, and J. A. Killian. 1999. Different membrane anchoring positions of tryptophan and lysine in synthetic transmembrane alpha-helical peptides. J. Biol. Chem. 274:20839-20846.
Edwards, J., and D. Brown. 1984. Sindbis virus induced fusion of tissue cultured Aedes albopictus (mosquito) cells. Virus Res. 1:705-711.
Edwards, J., and D. T. Brown. 1986. Sindbis virus-mediated cell fusion from without is a two-step event. J. Gen. Virol. 67:377-380.
Edwards, J., E. Mann, and D. T. Brown. 1983. Conformational changes in Sindbis virus envelope proteins accompanying exposure to low pH. J. Virol. 45:1090-1097.
Ferreira, D., R. Hernandez, M. Horton, and D. T. Brown. 2003. Morphological variants of Sindbis virus produced by a mutation in the capsid protein. Virology 307:54-66.
Gliedman, J. B., J. F. Smith, and D. T. Brown. 1975. Morphogenesis of Sindbis virus in cultured Aedes albopictus cells. J. Virol. 16:913-926.
Glomb-Reinmund, S., and M. Kielian. 1998. The role of low pH and disulfide shuffling in the entry and fusion of Semliki Forest virus and Sindbis virus. Virology 248:372-381.
Hernandez, R., H. Lee, C. Nelson, and D. T. Brown. 2000. A single deletion in the membrane-proximal region of the Sindbis virus glycoprotein E2 endodomain blocks virus assembly. J. Virol. 74:4220-4228.
Hernandez, R., C. Sinodis, M. Horton, D. Ferreira, C. Yang, and D. T. Brown. 2003. Deletions in the transmembrane domain of a Sindbis virus glycoprotein alter virus infectivity, stability, and host range. J. Virol. 77:12710-12719.
Lee, H., and D. T. Brown. 1994. Mutations in an exposed domain of Sindbis virus capsid protein result in the production of noninfectious virions and morphological variants. Virology 202:390-400.
Lee, S., K. E. Owen, H. K. Choi, H. Lee, G. Lu, G. Wengler, D. T. Brown, M. G. Rossmann, and R. J. Kuhn. 1996. Identification of a protein binding site on the surface of the alphavirus nucleocapsid and its implication in virus assembly. Structure 4:531-541.
Liljestrom, P., and H. Garoff. 1991. Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65:147-154.
Liu, L. N., H. Lee, R. Hernandez, and D. T. Brown. 1996. Mutations in the endo domain of Sindbis virus glycoprotein E2 block phosphorylation, reorientation of the endo domain, and nucleocapsid binding. Virology 222:236-246.
Liu, N., and D. T. Brown. 1993. Phosphorylation dephosphorylation events play critical roles in Sindbis virus maturation. Virology 196:703-711.
Liu, N., and D. T. Brown. 1993. Transient translocation of the cytoplasmic (endo) domain of a type I membrane glycoprotein into cellular membranes. J. Cell Biol. 120:877-883.
Lopez, S., J. S. Yao, R. J. Kuhn, E. G. Strauss, and J. H. Strauss. 1994. Nucleocapsid-glycoprotein interactions required for assembly of alphaviruses. J. Virol. 68:1316-1323.
Luukkonen, A., M. Brummer-Korvenkontio, and O. Renkonen. 1973. Lipids of cultured mosquito cells (Aedes albopictus). Comparison with cultured mammalian fibroblasts (BHK 21 cells). Biochim. Biophys. Acta 326:256-261.
Mann, E., J. Edwards, and D. T. Brown. 1983. Polycaryocyte formation mediated by Sindbis virus glycoproteins. J. Virol. 45:1083-1089.
Miller, M. L., and D. T. Brown. 1992. Morphogenesis of Sindbis virus in three subclones of Aedes albopictus (mosquito) cells. J. Virol. 66:4180-4190.
Mitsuhashi, J., and K. Maramorosch. 1964. Leafhopper tissue culture: embryonic, nymphal and imaginal tissues from aseptic insects. Contrib. Boyce Thompson Inst. 22:435.
Mulvey, M., and D. T. Brown. 1996. Assembly of the Sindbis virus spike protein complex. Virology 219:125-132.
Mulvey, M., and D. T. Brown. 1994. Formation and rearrangement of disulfide bonds during maturation of the Sindbis virus E1 glycoprotein. J. Virol. 68:805-812.
Mulvey, M., and D. T. Brown. 1995. Involvement of the molecular chaperone BiP in maturation of Sindbis virus envelope glycoproteins. J. Virol. 69:1621-1627.
Paredes, A. M., D. T. Brown, R. Rothnagel, W. Chiu, R. J. Schoepp, R. E. Johnston, and B. V. Prasad. 1993. Three-dimensional structure of a membrane-containing virus. Proc. Natl. Acad. Sci. USA 90:9095-9099.
Paredes, A. M., D. Ferreira, M. Horton, A. Saad, H. Tsuruta, R. Johnston, W. Klimstra, K. Ryman, R. Hernandez, W. Chiu, and D. T. Brown. 2004. Conformational changes in Sindbis virions resulting from exposure to low pH and interactions with cells suggest that cell penetration may occur at the cell surface in the absence of membrane fusion. Virology 324:373-386.
Paredes, A. M., M. N. Simon, and D. T. Brown. 1992. The mass of the Sindbis virus nucleocapsid suggests it has T = 4 icosahedral symmetry. Virology 187:329-332.
Parks, G. D. 1996. Differential effects of changes in the length of a signal/anchor domain on membrane insertion, subunit assembly, and intracellular transport of a type II integral membrane protein. J. Biol. Chem. 271:7187-7195.
Pletnev, S. V., W. Zhang, S. Mukhopadhyay, B. R. Fisher, R. Hernandez, D. T. Brown, T. S. Baker, M. G. Rossmann, and R. J. Kuhn. 2001. Locations of carbohydrate sites on alphavirus glycoproteins show that E1 forms an icosahedral scaffold. Cell 105:127-136.
Renkonen, O., C. G. Gahmberg, K. Simons, and L. Kaariainen. 1972. The lipids of the plasma membranes and endoplasmic reticulum from cultured baby hamster kidney cells (BHK21). Biochim. Biophys. Acta 255:66-78.
Renz, D., and D. T. Brown. 1976. Characteristics of Sindbis virus temperature-sensitive mutants in cultured BHK-21 and Aedes albopictus (mosquito) cells. J. Virol. 19:775-781.
Rice, C. M., R. Levis, J. H. Strauss, and H. V. Huang. 1987. Production of infectious RNA transcripts from Sindbis virus cDNA clones: mapping of lethal mutations, rescue of a temperature-sensitive marker, and in vitro mutagenesis to generate defined mutants. J. Virol. 61:3809-3819.
Rietveld, A., S. Neutz, K. Simons, and S. Eaton. 1999. Association of sterol-and glycosylphosphatidylinositol-linked proteins with Drosophila raft lipid microdomains. J. Biol. Chem. 274:12049-12054.
Simmons, D. T., and J. H. Strauss. 1974. Translation of Sindbis virus 26 S RNA and 49 S RNA in lysates of rabbit reticulocytes. J. Mol. Biol. 86:397-409.
Strandberg, E., and J. A. Killian. 2003. Snorkeling of lysine side chains in transmembrane helices: how easy can it get? FEBS Lett. 544:69-73.
Strauss, J. H., E. G. Strauss, and R. J. Kuhn. 1995. Budding of alphaviruses. Trends Microbiol. 3:346-350.
Strauss, J. H., Jr., B. W. Burge, and J. E. Darnell. 1969. Sindbis virus infection of chick and hamster cells: synthesis of virus-specific proteins. Virology 37:367-376.
Vaananen, P., and L. Kaariainen. 1980. Fusion and haemolysis of erythrocytes caused by three togaviruses: Semliki Forest, Sindbis and rubella. J. Gen. Virol. 46:467-475.
Wahlberg, J. M., and M. Spiess. 1997. Multiple determinants direct the orientation of signal-anchor proteins: the topogenic role of the hydrophobic signal domain. J. Cell Biol. 137:555-562.
Yeagle, P. L. 1991. Modulation of membrane function by cholesterol. Biochimie 73:1303-1310.
Zhang, W., S. Mukhopadhyay, S. V. Pletnev, T. S. Baker, R. J. Kuhn, and M. G. Rossmann. 2002. Placement of the structural proteins in Sindbis virus. J. Virol. 76:11645-11658.
Zhao, H., B. Lindquist, H. Garoff, C.-H. von Bonsdorff, and P. Liljestrom. 1994. A tyrosine-based motif in the cytoplasmic domain of the alphavirus envelope protein is essential for budding. EMBO J. 13:4204-4211.(Raquel Hernandez, Davis F)
ABSTRACT
The final steps in the envelopment of Sindbis virus involve specific interactions of the E2 endodomain with the virus nucleocapsid. Deleting E2 K at position 391 (E2 K391) resulted in the disruption of virus assembly in mammalian cells but not insect cells (host range mutant). This suggested unique interactions of the E2 K391 endodomain with the different biochemical environments of the mammalian and insect cell lipid bilayers. To further investigate the role of the amino acid residues located at or around position E2 391 and constraints on the length of the endodomain on virus assembly, amino acid insertions/substitutions at the transmembrane/endodomain junction were constructed. An additional K was inserted at amino acid position 392 (KK391/392), a KF substitution at position 391 was constructed (F391), and an additional F was inserted at 392 (FF391/392). These changes should lengthen the endodomain in the KK391/392 insertion mutant or shorten the endodomain in the FF391/392 mutant. The mutant FF391/392 grown in BHK cells formed virus particles containing extruded material not found on wild-type virus. This characteristic was not seen in FF391/392 virus grown in insect cells. The mutant KK391/392 grown in BHK cells was defective in the final membrane fission reaction, producing multicored or conjoined virus particles. The production of these aberrant particles was ameliorated when the KK391/392 mutant was grown in insect cells. These data indicate that there is a critical minimal spanning distance from the E2 membrane proximal amino acid at position 391 and the conserved E2 Y400 residue. The observed phenotypes of these mutants also invoke an important role of the specific host membrane lipid composition on virus architecture and infectivity.
INTRODUCTION
The alphavirus Sindbis is an excellent model for the study of the assembly of membrane-containing viruses. Unlike the majority of enveloped viruses, which are loosely structured protein modified membrane envelopes, the alphaviruses are highly organized icosahedral protein shells with an associated membrane (17, 36, 40). The three proteins of Sindbis virus are organized into two nested protein shells with T = 4 icosahedral symmetry (38). The outer protein shell contains 240 copies of glycoproteins E1 and E2. The inner protein shell contains 240 copies of the capsid protein C which are assembled around the plus-sense single-stranded RNA (48). The structure of the virus is stabilized through specific E1-E1 interactions, and anchoring of the outer protein shell to the inner protein shell occurs through specific associations of the E2 glycoproteins with the capsid proteins across the intervening host cell membrane (1, 22, 23). The E2-capsid interactions are acquired in the final stages of virus assembly as the virus capsid is enveloped in the E1-E2 modified plasma membrane (5, 27). These E2-capsid interactions occur for each of the 240 copies of E2 and capsid protein holding the mature virion in a high energy, metastable conformation (2, 8, 34). Energy stored in the E1 glycoprotein is likely used in the process of cell penetration (37).
Alphaviruses are hybrid structures containing protein and RNA encoded by the virus and lipid and carbohydrate derived from the host cell (21). The process of self-assembly of these relatively simple viruses has been the subject of intensive study. Although many of the biochemical details of virus protein processing, modification, and maturation have been described, protein-protein interactions at the molecular level are less well understood. The virus structural proteins are synthesized from a subgenomic polycistronic RNA to produce a 130-kDa polyprotein NH-C-PE2 (E3+E2)-6K-E1-COOH (45). While capsid proteins assemble with progeny 49S viral RNA into nucleocapsid structures within the cell cytoplasm, the glycoproteins mature via the secretory pathway (7, 8, 33, 35).
The interaction between the outer and inner protein shells occurs through the attachment of the 33-amino-acid endodomain of glycoprotein E2 (E2 tail) to a hydrophobic cleft in the capsid protein (22, 23) specifically interacting with E2 Y400. The process of E2-nucleocapsid association has been shown to involve a minimum of two steps. The first step has been postulated to involve sequence-specific recognition of the capsid protein by E2 amino acid sequences encoded in the carboxy-terminal endodomain (amino acids 402 to 420) (26, 28). The second interaction is proposed to position the E2 endodomain within the nucleocapsid structure in close apposition to the inner surface of the membrane (23). These interactions occur within a domain of E2 not precisely determined which has been proposed to involve the burial of the E2 hydrophobic T398/P399/Y400 domain within the capsid protein hydrophobic cleft (23). The process of repeatedly burying the amphipathic E2 endodomain within each of the 240 capsid proteins composing the nucleocapsid is predicted to be energetically favorable. These repeated interactions could provide sufficient energy to drive the budding process which incorporates the underlying icosahedral nucleocapsid structure (47).
The structure of wild-type capsid protein has been solved by X-ray crystallography (9). This structure has been fit into the electron cryomicroscopic reconstruction of Sindbis virus which positioned the hydrophobic pocket of the capsid protein against the membrane at a position consistent with the site where E2 emerges from the virus membrane (23, 53). The hydrophobic capsid protein pocket is comprised of two compartments: the first is between capsid Y180 and W247 and the second is between W247 and F166, with W247 central to the pocket. The distance from the inner surface of the virus membrane to the position between Y180 and W247 in the capsid hydrophobic pocket is 28 ?. This distance corresponds well to the length of 10 amino acids, which also corresponds to the distance between the membrane-proximal E2 K391 and the hydrophobic E2 Y400. These observations suggested that an interaction between Y400 of E2 and Y180-W247 of capsid may play a critical role in assembly.
Genetic and biochemical analysis of E2-nucleocapsid interactions occurring during envelopment provide a general picture of the maturational events which E2 and capsid undergo prior to the release of a virus particle. The 33-amino-acid E2 endodomain can be divided into several overlapping functional domains involved in virus assembly. Sequences responsible for the initial capsid recognition (E2 amino acids 402 to 420) comprise one domain (20). A second domain includes the membrane-proximal K391 through the conserved T398/P399/Y400 sequence proposed to be the region which mediates tight E2-capsid binding. A mutant deleting the E2 402 LAPNA 406 sequence had no effect on virus production (25), further defining the E2 tail first binding domain to amino acids 407 to 420. The wild-type E2 domain K391 to L402 is the portion of the E2 tail proposed to reach into the capsid binding pocket and interlock the two virus protein shells (23, 52). We have proposed that this domain contains a critical number of amino acids to place Y400 in register with the correct amino acids in the capsid protein structure (20, 23). We have tested this hypothesis by using a deletion mutant which truncated the E2 endodomain upstream of the TPY sequence. A nonconserved K at position 391 located at the membrane/cytoplasm junction of the E2 endodomain was deleted. This mutant, K391, produced very little virus from BHK cells, and nucleocapsids failed to associate with cell membranes (20). These results supported the hypothesis that a critical length is required between the membrane and Y400 for correct capsid binding. Remarkably, the K391 mutant produced virus at wild-type levels in insect cells, suggesting a host range effect of mutations at this position. This result revealed two important aspects of virus assembly involving the contiguous E2 transmembrane domain (TM) and endodomain. The failure of K391 to assemble virus only in mammalian cells suggested that manipulation of the amino acid sequence at the juncture between the TM and the endodomain could functionally, if not physically, extend residues into an adjacent domain. A second hypothesis resulting from these observations was that, because of biochemical differences of mammalian and insect host membranes, mutations made within the E2 TM would result in host range phenotypes. These mutations were predicted to identify virus domains necessary for assembly in mammalian or insect membranes.
To test the second hypothesis, nested deletion mutants in the E2 26-amino-acid TM were constructed and grown in mammalian or insect cells (21). Nine- and 10-amino-acid deletions in the E2 TM producing mutants TM 17 and 16, respectively, were significantly inhibited in infectious virus production when grown in mammalian BHK cells but not in insect cells. This defect in assembly was attributed to an inability of the shorter TM to efficiently interact with the relatively thicker mammalian membrane bilayers, suggesting a role of the membrane in the virus architecture (21). A single-deletion mutant TM25 (379 M) produced very low levels of virus from both BHK and U4.4 cells and a high particle-to-PFU ratio (105 particles/PFU). This phenotype was predicted to result from altered ectodomain conformations affecting infectivity. These data suggested subdomains in the E2 TM, as was previously demonstrated for the E2 endodomain.
In the present study, we address the possibility that amino acids adjacent to a membrane can become incorporated into the TM or, inversely, extend the length of the endodomain. To test this hypothesis, mutations at E2 positions 391 and 392 were constructed. These mutants were designed to probe the function of the membrane/cytoplasm-proximal amino acids of E2 in the assembly of Sindbis virus. Amino acid additions and substitutions were chosen to incorporate residues which are predicted to preferentially incorporate into the TM or the endodomain (3, 11, 12). An additional K residue at E2 392 (mutant KK391/392) was chosen as a residue which would remain in the cytoplasm, while F at positions 391/392 (mutants FF391/392 and F391) was chosen as a residue which would partition into the membrane lipid. The E2 KK391/392, FF391/392, and F391 endodomain mutants were designed not solely as substitution or insertion mutants but also to specifically exploit the hydrophobic nature of the Sindbis virus TM and the hydrophilic properties of the cytoplasmic portion of the E2 endodomain. The properties of these mutations were examined in both insect and mammalian cells to determine the effect of membrane structure and composition on the mutant phenotypes.
MATERIALS AND METHODS
Cell culture, plaque assay, and virus. The source and culture of baby hamster kidney cells (BHK-21) was as described previously (42). Briefly, these cells were grown and maintained in minimal essential medium containing Earl's salts (MEM-E) (Invitrogen, Carlsbad CA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen), 5% tryptose phosphate broth, and 2 mM glutamine as described previously (42). The Aedes albopictus clones were subclones derived from Singh's original larval isolates (31). The U4.4 cells were cloned from cells originally provided by Sonya Buckley (Yale Arbovirus Research Unit, New Haven, Conn.) and is the original line grown in Mitsuhashi and Maramorosch medium (M and M) supplemented with 20% FBS (32). The virus strain chosen for mutagenesis was produced from the cDNA Toto 1101 clone (43) containing a substitution in E2 at position 420, changing a serine to a tyrosine, Y420, and has been described previously (27). Y420 serves as the wild type, and all virus mutants were constructed using this cDNA. This construct was chosen over the original Toto1101 cDNA to facilitate identification of mutations which fail to retract the E2 tail from the endoplasmic reticulum membrane (20). We have shown previously that this mutation is silent (27). Titration of virus produced was done using BHK-21 cells as indicator cell monolayers. Titers using BHK cells were done using the standard plaque assay. The virus diluent was phosphate-buffered saline (PBS-D) supplemented with 3% FBS as described previously (21, 42). Mutants were made using the Quick Change site-directed mutagenesis protocol as detailed in reference 21. Mutants were made using a smaller subcloned template pGEM 3Z TE12 or using the full-length wild-type template. For all mutants, the most convenient unique restriction sites were used to subclone the mutagenized nucleotides into the wild-type vector to eliminate spurious changes acquired during the amplification process. Mutagenic oligonucleotides were designed as described previously (21). The mutagenic sequences for the coding strands of the mutants KK391/392, FF391/392, and F391, respectively, are as follows: 5'-GCAGTGTTATGTGCCTGTAAAAAAGCGCGCCGTGAG-3', 5'-GCAGTGTTATGTGCCTGTTTTTTTGCGCGCCGTGAG-3', 5'-GCAGTGTTATGTGCCTGTTTTGCGCGCCGTGAG-3'. The E2 mutant TM25 has been described previously (21). This mutant will be referred to as TM25 (379 M) to specify the amino acid deleted in this construct. TM25 (379 M) was used as the template to construct the mutant TM25 (379 M) FF391/392 essentially as described above. TM25 (379 M) FF391/392 combines the single-amino-acid deletion encoded in TM25 with the double insertion mutant FF391/392. The same TM25 (379 M) deletion mutant was also used to construct an analogous mutant by combining with the KK391/392 mutant to create TM25 (379 M) KK291/392. All mutant sequences were verified by sequencing both the DNA used to make transcripts and reverse transcription-PCR products from RNA extracted from transfected BHK cells or mutant virus.
In vitro transcription and RNA transfection. The mutant and wild-type cDNA constructs were prepared for transcription and transcribed in vitro as described previously (20). The phenotypes of the E2 endodomain mutants were determined for both insect U4.4 and BHK host cells. BHK and U4.4 cells were transfected with synthetic RNA transcripts produced in vitro. Production of the RNA transcripts was as described in reference 20, and transcripts were electroporated into both BHK-21 and U4.4 cells. For the BHK cells, the electroporation was performed essentially as described by Liljestrom and Garoff (24). The BHK-grown virus was harvested once cytopathic effect was evident or 24 h posttransfection, whichever came first. All mutant strains are supplemented with 10 mM HEPES-HCl, pH 7.4, when virus was grown in BHK cells. The U4.4 cell transfection was performed as described previously (20) with the addition of 10 mM morpholinepropanesulfonic acid (MOPS)-HCl, pH 7.2, to the culture medium to maintain an optimal pH until virus was harvested 48 h posttransfection. Virus harvested from transfected cells was resuspended in 10% glycerol, divided into 1-ml aliquots, and flash frozen in liquid N2.
Metabolic labeling of BHK and U4.4 cells. BHK cells were transfected by electroporation as described above, and at 6 h posttransfection, 5 ml of fresh MEM containing 4 μg/ml actinomycin D (ActD) (Calbiochem, San Diego, CA) was added to a 25-cm2 flask of cells (5 x 106 cells) and incubated at 37°C for 1 h. The flasks were then washed twice with 5 ml of room temperature PBS before 5 ml of starvation medium (MEM deficient in methionine and cysteine supplemented with 2 mM glutamine, 3% FBS) was added to the cells and returned to 37°C for 1 h. The transfected BHK cells were then labeled with 50 μCi/ml [35S]methionine-cysteine ([35S]Met/Cys) and incubated at 37°C for 2 h for cell labeling or 24 h for virus labeling. Virus grown in U4.4 cells was also metabolically labeled to facilitate concentration of low-titer mutants used in fusion from without (FFWO) assays (see Results). Cells were attached to flasks in M and M in the absence of FBS for 1 h at 28°C. Infection proceeded essentially as described for BHK cells above. Labeling of U4.4 cells was done in M and M in the absence of ActD treatment immediately after infection using 3 ml M and M containing 50 μCi/ml [35S]Met/Cys.
Isopycnic centrifugation analysis of mutant viruses. Subconfluent monolayers of BHK-21 cells in 75-cm2 flasks were treated with 5 ml MEM containing 4 μg/ml ActD for 1 h. The ActD was removed, and the cell monolayers were infected with each of the mutant viruses grown in BHK cells for 1 h at 25°C at a multiplicity of infection (MOI) of 1 PFU/cell, after which the inoculum was removed and 5 ml MEM-E was added. Infected monolayers were incubated at 37°C for an additional 6 h. The medium was removed, and the cell monolayers were washed once with PBS-D and starved for Met/Cys in starvation medium at 37°C for 1 h. The medium was then removed and replaced with starvation medium containing 50 μCi/ml [35S]Met/Cys protein labeling mix. The monolayers were incubated at 37°C for 18 to 24 h. The virus supernatant was harvested and spun to equilibrium on linear 15 to 35% potassium tartrate gradients in PBS-D buffer, pH 7.4, at 24,000 rpm in a Beckman SW-28 rotor overnight. The entire gradient was collected in 0.5-ml fractions, and 5 μl of each fraction was counted by scintillation spectrometry for detection of the labeled virus fractions. Linearity of the gradient was determined by a measurement of the refractive index of each fraction. Titration of the same virus fraction was performed on BHK-21 cells as described previously (42).
Transmission electron microscopy and negative staining. BHK and U4.4 cells were transfected with RNA transcribed from either wild-type Sindbis virus or each of the endodomain mutants. For BHK cells, incubation proceeded at 37°C for 16 to 18 h, and U4.4 cells were incubated at 28°C for 30 h. Cell monolayers were scraped from the flasks and pelleted by low-speed centrifugation. The specific protocols used in the fixation and embedding procedures are detailed in previous literature (17). The samples were examined at 80 kV in a JEOL JEM 100S transmission electron microscope. Virus particles to be viewed by negative staining were prepared by infection of four 75-cm2 flasks as described above (see "Metabolic labeling of BHK and U4.4 cells"). For these samples, the amount of [35S]Met/Cys was limited to 10 μCi/ml to reduce any possible damage from the isotope on the integrity of the virus sample. Virus preparations were then purified as described above on 35% to 15% potassium tartrate linear gradients. Virus collected from the gradients was directly attached to carbon-coated grids, washed three times with sterile H2O, and negatively stained with 1% uranyl acetate.
FFWI and FFWO assays. BHK cell monolayers for fusion from within (FFWI) and FFWO assays were prepared in 24-well plates. For FFWI assays, the cells were infected in PBS-D containing 3% FBS at 28°C with an MOI of 10 PFU/cell or mock infected for 1 h and returned to 37°C for an additional 7 h of incubation. The fusion process is initiated by the addition of fusion medium (1.8 mM CaCl2, 5.3 mM KCl, 0.1 μM MgSO4 · 6H2O, 116 mM NaCl, 2.9 mM sucrose, 1x MEM-E amino acids [Invitrogen, Carlsbad, CA], 1x MEM-E vitamins [Invitrogen, Carlsbad, CA], 2 mM L-glutamine, and 0.2% FBS). Fusion medium is adjusted to pH 5.2 with 10 mM morpholineethanesulfonic acid (MES) buffer, to neutral pH values by the addition of 10 mM HEPES, or to alkaline pH values by the addition of 75 mM NaHCO3 to the desired pH. The infected and mock-infected cells were then treated with pH 5.3 fusion medium for 5 min and retained at this pH or replaced with fusion medium of the pH series that follows: pH 6.4, 6.7, 7.0, 7.2, 7.4, 7.6, 7.8, 8.0, 8.3, 8.5. Control wells of mock-infected or infected cells treated only with fusion media corresponding to the pH series shown above were also included. The cells were incubated at 37°C for 1 h in a standard fusion protocol (30). FFWO assays were done using virus grown in U4.4 cells. Cell monolayers prepared for FFWO were prepared in 24-well plates as described above. Monolayers which were just subconfluent were placed on ice for 15 min. While still on ice, 100 PFU/cell of each of gradient-purified mutant was added to the appropriate wells. Control pH treatments and the pH series used were as for the FFWI assays except that the low-pH treatment was at pH 4.6. Previous work has shown that virus grown in U4.4 cells fused maximally when taken to a lower pH than that required for BHK-grown virus (14). The cell monolayers were then scored for the amount of fusion seen relative to the low-pH, high-pH, or mock-treated samples as a percentage of the wild-type control. Cell fusion into polykaryons was defined as those cells which contained three or more nuclei/cell and is shown as a percentage of fusion seen for the wild-type infection, pH 5.3 to pH 7.4 for FFWI and pH 4.6 to pH 7.4 for FFWO.
RESULTS
Our previously published results have indicated that amino acids adjacent to the membrane interface can play critical roles in the ability of the virus to assemble on vertebrate and invertebrate membranes (20). To further explore this hypothesis, we produced mutations at the membrane interface of the E2 glycoprotein endodomain. These mutants were designed to alter a critical length of the E2 endodomain from the membrane-proximal amino acid at position 391 to Y400, which is predicted to occupy the capsid protein hydrophobic pocket (23, 52). The length of the E2 endodomain was increased from 33 to 34 amino acids by incorporating an additional K at amino acid 392 (KK391/392). This insertion was predicted to lengthen the endodomain, since a second K should not be preferentially incorporated into a TM of sufficient length displaying no hydrophobic mismatch (12, 13). A second insertion, replacing K391 with F391 (K391F) and inserting a second F at amino acid position 392 produced the mutant FF391/392. This second type of insertion was designed to physically shorten the endodomain sequence by relocating these hydrophobic aromatic amino acids to the TM within the membrane (3). As the transmembrane domain is increased in length, the endodomain is reduced in length by two amino acids (FF) (11). This experiment was done, in part, in an effort to duplicate the host range phenotype of the previously described deletion mutant K391 which was proposed to retract the A at position 391 into the TM (20). A third mutant substituting F391 for K391 was constructed to test the effect of a single F at this position. As a genetic test for the partitioning of the KK or FF391/392 insertions into a cytoplasmic or lipid environment, we employed an E2 single-amino-acid deletion mutant at E2 379 (TM25 379 M) described previously (21). This mutant produces large numbers of noninfectious virus particles and low virus titers. Incorporation of the FF391/392 sequence into the TM25 virus was expected to restore the ability of the virus to assemble infectious virus by replacing the deleted amino acid from TM25 (379 M) into the carboxy terminus of the TM by inserting the hydrophobic FF residues. Conversely, incorporation of the TM25 (379 M) deletion into the KK391/392 mutant was not expected to restore amino acids to the TM if the insertion remained within the cytoplasmic tail. In addition to investigating the effects of these mutations on the interactions of the E2 endodomain with the nucleocapsid during the process of envelopment, these mutants may also prove of value in understanding the interaction of the virus membrane proteins with the different lipid bilayers it encounters in its broad host range.
Virus production from mammalian (BHK) and insect (U4.4) cells. Infectious virus production from transfected BHK and U4.4 cells was determined by assay on BHK cells as described in Materials and Methods. Titers of Sindbis virus heat-resistant strain (SVHR), parental wild-type virus, and mutant virus were measured from both cell lines and are shown in Table 1. Both insertion mutants, KK391/392 and FF391/392, produced significantly less infectious virus than wild-type virus in both hosts. The KK391/392 mutant grew approximately 10-fold better in BHK cells than the FF391/392 mutant. This result was the inverse compared to the amount of virus generated from U4.4 cells. The KK391/392 mutant was significantly growth attenuated in U4.4 cells, producing approximately 20-fold less virus than the FF391/392 mutant. By contrast, the KK391/392 mutant generated about 60-fold-higher virus titers in BHK cells than in U4.4 cells. The FF391/392 mutant produced similar amounts of virus in BHK and U4.4 cells. The F391 mutant grew to levels equivalent to that seen for wild-type virus in both cell lines. The parental Y420 virus yields equivalent virus titers from both cell lines. Virus production from the TM25 (379 M) FF391/392 mutant in BHK cells reproduced wild-type levels of virus, producing titers of 1.1 x 109 PFU/ml. This result provides indirect evidence that the FF residues can complement the TM25 (379 M) deletion by incorporating into the TM of E2. Interestingly, the TM25 (379 M) KK391/392 mutant also restored virus titers to 2 x 108 PFU/ml, a level 1,000-fold higher than the TM25 (379 M) mutant titer and 45-fold higher than the KK391/392 mutant titer.
Particle/PFU ratios of U4.4-grown endodomain mutant virus. Previous results obtained with E2 TM deletion mutants demonstrated that assembly mutants which produced low titers could assemble large quantities of noninfectious particles. Virus particles which are noninfectious were measured by determining a particle/PFU ratio. Large particle/PFU ratios in mutant virus populations are indicative of structural instability, resulting in a loss of infectivity. To further characterize the endodomain phenotypes, particle/PFU ratios were determined for each of the endodomain mutants grown in U4.4 cells and are shown in Table 1. Of particular interest are the number of noninfectious virus particles produced in insect U4.4 cells from TM25 (379 M), which encodes a shortened TM (7.8 x 104 particles/PFU), and the mutant KK391/392, which encodes a longer endodomain (4.3 x 103 particles/PFU). The mutant FF391/392, which is hypothesized to encode a longer TM, displayed a much lower particle/PFU ratio than either of these two mutants (116 particles/PFU), suggesting greater virus stability. Both mutants KK391/392 and FF391/392 produced fewer total particles (108 particles/ml) than the F391 mutant (109 particles/ml) and SVHR or Y420 (1010 particles/ml). These data suggest that the KK391/392 insertion results in the highest mutant virus instability of the endodomain mutations.
Electron microscopy of endodomain mutant-infected BHK and U4.4 cells and purified mutant virus. Electron micrographs of thin sections of transfected cells and negatively stained mutant virus of each of the mutants from BHK and U4.4 cells are presented in Fig. 1 to 7. In Fig. 1A and C and Fig. 2 to 4, images from BHK cells are presented, while Fig. 1B and D and Fig. 5 to 7 are images from U4.4 cells. All of the endodomain mutants were found to assemble nucleocapsids which are seen free in the cytoplasm and associated with cell membranes, as is typical of wild-type virus-infected BHK cells (Fig. 1A). There are, however, significant differences in the appearance of cells infected with the various mutants. Fewer cytoplasmic nucleocapsids are seen in BHK cells infected with FF391/392 than in wild-type infection. The reason for this is not known. Virus protein evaluated by immunoprecipitation of infected cells displayed normal processing and equivalent levels of protein compared to wild-type virus-infected cells (data not shown). While virus is seen budding from the plasma membrane (Fig. 2B), the outer virus membranes do not appear smooth and spherical but appear to be associated with nonvirus material (Fig. 2B). The virus cores, however (Fig. 2A and C), appear symmetrical. Deformations in the virus surface were seen more clearly in negatively stained preparations of purified FF391/392 virus (Fig. 2D and E). The FF391/392 virus displayed small extrusions of material, giving the particles a peculiar "lollypop" or "mouse ears" appearance. In thin sections (Fig. 2B and C), these particles did not appear uniform and symmetrical compared to virus produced from a wild-type infection (Fig. 1A). The possibility that this material was in part lipid is suggested by its appearance and the shift to a lower density in potassium tartrate gradients (Fig. 8A). Electron micrographs of cells infected with the endodomain mutant KK391/392 revealed a different phenotype (Fig. 3A to G). While this mutant appeared to produce normal numbers of nucleocapsids compared to the wild type and many of these capsids were found associated with membranes, this mutant is defective in the final stages of virus envelopment (Fig. 3A, B, and C). The KK391/392 mutant did not effectively recruit modified membrane to envelop single particles but rather incorporated multiple nucleocapsids into a bolus or tube of partially enveloped virus cores. In Fig. 3D, E, F, and G, negatively stained virus particles containing multiple cores in membranous vesicles of irregular size are shown (18). These larger multicored particles were found in the peak of virus infectivity shown in Fig. 8A and B. This fraction contained many double, triple, and quadruple conjoined particles. These multicored particles exhibit reduced density compared to wild-type virus (shown in Fig. 8A) and the FF391/392 mutant (compare the titers of fractions 4 and 5 in Fig. 8A and B). In accordance with the high virus titers measured for the mutant F391 in both BHK and U4.4 cells, electron micrographs of cells infected with this mutant (Fig. 4A and B) reveal large numbers of cytoplasmic capsids, membrane-associated capsids, and many particles budding from the plasma membrane. Likewise, the negative stains of virus produced by this mutant shown in Fig. 4C and D reveal many well-formed virus particles interspersed with few empty particles, which appear collapsed and electron dense. Some dimpled and collapsed particles are expected in negatively stained virus particles and are artifacts of the preparation.
A comparative analysis of the assembly and morphology of the endodomain mutants was also conducted in insect cells. A micrograph of wild-type virus-transfected U4.4 cells shown in Fig. 1B reveals large quantities of matured particles found within intracellular vesicles, a characteristic of Sindbis virus assembly in mosquito cells (18). Virus is released from the insect cells by a process similar to exocytosis, with few particles budding directly from the plasma membrane. Negative stains of wild-type virus produced from U4.4 cells (Fig. 1D) appear identical to those of wild-type virus produced from BHK cells (Fig. 1C). Any structural differences imparted by insect-specific lipids (29) and glycosylation patterns (6) are indiscernible at this resolution. Insect (U4.4) cells transfected with the FF391/392 mutant (Fig. 5A and B) produced many more virus particles than BHK cells (compare Fig. 2A, B, and C). This finding is in agreement with the titers of virus produced from each of the cell lines (Table 1). The mutant FF391/392 produced twofold-more infectious virus from U4.4 cells than BHK cells. A significant difference seen in thin sections and negative stains of the mutant FF391/392 from insect cells compared to those of BHK cells is the absence of extruded or external material associated with either cell-associated virus (Fig. 5A and B) or purified virus (Fig. 5C, D, E, F, and G). The negative stains of virus produced from insect cells show a small number of multicored and larger particles not seen in virus from BHK cells. These differences in FF391/392 virus structure were not the effect of revertants or adaptations in the virus sequence (data not shown) but are an insect host-specific effect on virus assembly. Of the two endodomain insertion mutants, the differences imparted by the U4.4 cells on the FF391/392 mutant produced virus particles with a particle-to-PFU ratio closest to that seen for the wild type (116 and 33 particles/PFU, respectively) (Table 1) compared to the KK391/392 mutant (4.3 x 103 particles/PFU) (discussed below).
Thin sections of KK391/392 in U4.4 cells (Fig. 6A, B, and C) revealed large quantities of virus particles. This observation is consistent with the large amount of noninfectious virus produced from U4.4 cells (Table 1). In thin sections, fewer virus particles and fewer elongated virus tubes (Fig. 6C) are found for this mutant in U4.4 cells than were seen in BHK cells (compare Fig. 3A, B, and C). The defect in the assembly of KK391/392 in U4.4 cells is illustrated in Fig. 6B (inset). Long stalks of membrane at the base of the budding viruses suggest a defect in the final stage of release of the virus from the cell membrane. The particle/PFU ratio was determined for KK391/392 mutant virus grown in insect U4.4 cells and is 4.3 x 103 particles/PFU compared to 33 particles/PFU found for wild-type virus (21). Negative stains of KK391/392 virus produced from insect cells (Fig. 6D, E, and F) reveal abnormalities on the virus surface. Compared to the wild type (Fig. 1C), three times more KK391/392 particles appear pocked (Fig. 6D and F) or dimpled (Fig. 6C), which may reflect instability in virus structure. Additionally, many more KK391/392 particles (37% of 315 particles examined) have lost the ability to exclude stain and appear empty (Fig. 6E) than have particles of the wild type (10% of 344 particles examined). These observations may reflect the loss of the nucleocapsids from less stable virus structures. The deformities seen in the mutant particles could also be the result of inefficient virus release, as was seen for this mutant produced from BHK cells (Fig. 3A, B, and C). Growth in insect cells, however, did not result in the production of the multicored particles assembled in BHK cells (seen in Fig. 3D, E, F, and G). This mutation may result in defective lateral protein associations required to seal the virus membrane during the final stages of budding from both host cells. In contrast to KK391/392 virus grown in BHK cells, defects in assembly of this mutant in U4.4 cells could be compounded by a thinner, less fluid insect membrane (4, 10) producing a more fragile, less infectious virus structure.
Thin sections of insect cells infected with the F391 mutant displayed typical virus particles in cytoplasmic vesicles (Fig. 7A and B) as well as a few virus particles budding from the cell surface (Fig. 7A), as is typically found in wild-type infections (Fig. 1B). The mutant F391 produces high virus titers in both cell lines, and this mutant's structural integrity is underscored by a particle/PFU ratio of 3 for virus produced in insect cells (Table 1).
Density gradient analysis of BHK cell-grown mutant viruses. Negative stains of mutant viruses grown in BHK cells revealed abnormal structures (Fig. 2 to 4). To evaluate what changes these defects in assembly may have on the buoyant densities of BHK-grown virus and to correlate any changes to altered infectivity, we determined the density of particles produced by the mutants in potassium tartrate density gradients. Virus from [35S]Met/Cys-labeled infected BHK cells were prepared as described in Materials and Methods, and their densities were determined. The results of these experiments are presented in Fig. 8A. Wild-type virus shows a large peak in fraction 3 corresponding to a density of 1.17 g/cm3. Figure 8B presents the virus titers of the peak fractions. Infectivity for wild-type virus is found predominantly in fraction 3 (2.3 x 1010 PFU/ml), with a small shoulder of infectivity in fraction 4 (1.0 x 107 PFU/ml). In Fig. 8A, the mutant F391, which grows to wild-type levels, has a density profile and distribution of infectivity similar to wild-type virus. The major FF391/392 virus peak in fraction 4 shows a slight shift in density corresponding to the density of the wild-type virus shoulder. This fraction also contained the peak virus titer of FF391/392 (7.0 x 107 PFU/ml) (Fig. 8B, fraction 4). The peak virus titer of FF391/392 (7.0 x 107 PFU/ml) is higher than that of the virus titer for the fraction of the same density of wild-type virus (1.0 x 107 PFU/ml), although the total counts per minute for this wild-type fraction was higher than that of the counts per minute/ml for FF391/392 fraction 4 (Fig. 8A). This result was due to a lower specific activity of the FF391/392 virus (data not shown). The mutant KK391/392 demonstrated the most pronounced shift in virus density (Fig. 8A, fraction 5). Most of the KK391/392 virus was found in fractions 4 and 5 (Fig. 8A), with the majority of infectious virus found in fraction 5 (Fig. 8B). This shift in virus buoyant density from the wild type (1.17 g/ml3) to a reduced density of 1.154 g/ml3 was found to be a consistent characteristic of the mutant KK391/392 and could be the result of a higher lipid content in the multicored particles.
FFWI and FFWO mediated by endodomain mutants. The ability of Sindbis virus glycoprotein to induce cell-cell fusion after brief exposure to acid pH has been well characterized (30, 49). In BHK cells, it has been shown that this process is a two-step event requiring exposure to pH 5.3 followed by a return to pH 7.2 for optimal fusion to be observed (15, 30, 37). These experiments can be conducted by the addition of virus particles to a cell monolayer (FFWO) or by using proteins exported to the cell surface after virus infection (FFWI). FFWI is a fusion assay which can be utilized for mutants which do not grow to the high titers required by FFWO (100 to 1,000 PFU/cell) (16). The pH requirements for the induction of FFWI and FFWO are the same for wild-type virus (30). We have previously used the property of Sindbis virus to fuse cell monolayers from within as an assay for export of functional E1-E2 heterodimers to the plasma membrane (30). Analysis of the presence of matured E1-E2 complexes at the cell surface by FFWI suggests that sufficient protein reaches the plasma membrane for 100% fusion at the midphase of the transfection (10 h posttransfection). We have employed FFWI as an assay for the functional conformation of the E1-E2 heterotrimers of the mutants described above. We reasoned that a significant change in the glycoprotein ectodomain conformations and/or lateral interactions at the cell surface would be reflected in changes in the pH requirements to yield optimum cell fusion. We hypothesize that conformational differences in the mutant virus particles are imparted by the host cell membrane component in addition to the heterodimer conformations. Mutant virus grown in BHK cells would present altered conformations compared to those produced in U4.4 cells. These changes would be reflected as a change in the pH required to achieve maximum fusion for BHK-grown virus or U4.4-grown virus. As discussed by Paredes et al. (37), exposure to low pH followed by return to neutral pH destabilizes virus structure. Conformational changes induced in the mutant virus particles may affect the pH at which the structural changes are seen. In separate experiments, BHK cells were transfected or infected, at an MOI of 10 PFU/cell, with wild-type virus or one of the three endodomain mutants. The infections were allowed to proceed as detailed in Materials and Methods, and fusion levels were scored as described in Materials and Methods and as described previously (14, 30). Fusion profiles (FFWI) of the endodomain mutants in BHK cells are shown in Fig. 9A. The values plotted represent the percentages of cell-cell fusion produced by each mutant and are compared to the percentages of fusion of wild-type virus. Mock infections were done for each of the pH treatments shown. In each experiment, the infected monolayer was exposed to pH 5.3 for 5 min to establish the conditions for fusion and then returned to the indicated pH. Both the control wild-type virus and the KK391/392 mutant displayed similar fusion profiles, with KK391/392 producing slightly less fusion at pH 6.4 (20% less fusion than wild type). Both wild-type virus and KK391/392 produce 100% fusion after return to a pH of 7.2. The mutant F391 displayed a slightly more convex curve than that of wild-type virus, showing 20% more fusion at pH 6.7 and 10% more fusion at pH 7.0. At a pH of 7.4, the F391 mutant reaches levels of 100% fusion, as seen with wild-type virus. These mutants displayed similar fusion profiles. The most interesting fusion profile of the endodomain mutants was the FF391/392 virus (Fig. 9A). The mutant FF391/392 displayed a complex curve, producing three separate plateaus of increasing cell fusion. The first curve is seen at pH values of 6.4 to 7.2, where fusion increased from 30% to 60%. At pH 7.2 to 7.4, the amount of fusion is maintained at 60%. This is followed by a second plateau from pH 7.6 to 7.8 which displays 80% fusion. The FF391/392 mutant ultimately achieves 100% fusion at pH 8.0 to 8.5. Compared to the wild type, FF391/392 requires a pH increase of 0.8 pH units to produce 100% BHK cell-cell fusion. Mock-infected monolayers displayed negligible fusion after the treatments described above compared to the wild-type controls. It is not possible to determine the exact amounts of E1 and E2 that are present on the cell surface in the fusion from within assay. It has been shown that the E1 and PE2 (precursor to E2) are paired as heterodimers in the endoplasmic reticulum. Formation of the heterodimer is required for export to the cell surface, suggesting that the proteins are present in equimolar quantities in the plasma membrane (35). We have shown that fusion of cells by mature virus requires as few as 1,000 viruses per cell (16). That the cells infected with the different mutants demonstrate 100% fusion from within indicates that they have delivered at least 1,000 virus equivalents of protein to the cell surface. The pH at which fusion is demonstrated is a property of the virus proteins (19) not the amount of protein on the cell surface (30).
The possibility that the endodomain mutant particles grown in insect U4.4 cells were structurally more similar to wild-type virus was suggested by the images produced in negatively stained preparations. This possibility was tested by comparing the ability of the mutants and wild-type virus to produce FFWO, reasoning that viruses with similar conformations would retain similar pH fusion profiles. FFWO, as apposed to FFWI, was necessary to evaluate U4.4-grown virus fusion with BHK cell membranes. Each of the endodomain mutants, in addition to wild-type virus, was gradient purified and concentrated to enable the addition of 1,000 PFU/cell. FFWO was performed as described in Materials and Methods. Shown in Fig. 9B are the FFWO profiles of U4.4-grown virus on BHK cells. Wild-type virus gave 100% fusion at all pH values tested, with the exception of pH 6.4, which gave 80% fusion. The fusion profile for virus produced from U4.4 cells differed from that seen in FFWI of BHK cells by a shift in the pH required for maximum fusion from pH 7.2 (BHK cells) to pH 6.7 (U4.4 cells). Differences in the pH requirements for fusion of BHK and U4.4 membranes have been previously described (14). F391 from U4.4 cells became 100% fusion competent at similar pHs in both host cells (pH 7.4 in BHK and pH 7.2 from U4.4). This mutant, however, differed in its response to the higher pH values (pH 8.0 to 8.5), showing 75% fusion compared to 100% fusion in BHK cells. The KK391/392 mutant, which most closely followed the wild-type virus curve in BHK cells, most closely resembled F391 when grown in U4.4 cells. This mutant showed 100% fusion at pH 7.2 to 7.6, began to decline to 75% fusion at pH 7.8 to 8.3 and, finally, only produced 50% fusion at pH 8.5. As with F391, KK391/392 displayed less fusion at the higher pH values (pH 7.8 to 8.5). Again, as seen with FFWI in the BHK cells, the most interesting result was seen from the FF391/392 mutant. In contrast to the results obtained from BHK cells, this mutant displayed 100% fusion at pH values from pH 6.4 to 7.8, while fusion efficiencies began to fall at pH 8.0 (75% fusion) and gave the least fusion (60%) at pH 8.3 to 8.5, pH values that gave 100% fusion in BHK cells (Fig. 9A). Cell monolayers not exposed to virus, as shown in Fig. 9A, displayed negligible fusion. This result is opposite that seen in BHK cells and suggests that the protein conformations and/or the protein lateral contacts may be different for FF391/392 when grown in BHK or U4.4 cells. All insect cell-grown viruses gave maximum fusion in the pH range between 7.2 and 7.6. These results demonstrate that this assay may be used as a relative probe of altered virus ectodomain conformations which may affect virus function.
DISCUSSION
The final event in the assembly of the alpha virion is the envelopment of the preassembled core structure within a virus protein-modified cell membrane. This interaction is mediated by a specific association between the endodomain of the E2 glycoprotein and a hydrophobic cleft in the capsid protein. The process is complicated by the broad host range exhibited by the alphaviruses. It is possible, therefore, to study virus assembly in two genetically and biochemically unrelated host cells, allowing the comparison of assembly events which are unique to each host and those structural interactions which are host independent (21). Sindbis virus has adapted to accommodate the biochemically distinct lipid environments of mammalian and insect cell membranes. Mammalian membranes contain cholesterol, are more viscous and ion impermeable, and are composed of longer-chain fatty acids than those found in insect cells (4, 29, 41). Insect cell membranes have very low levels of cholesterol (44) and display a lipid composition distinct from that found in mammals (10, 29). These important biochemical distinctions affect the thickness and fluidity of the lipid bilayers (51).
Installing aromatic amino acids at the TM lipid/cytoplasm junction was undertaken as a means of reorganizing the endodomain architecture by converting a hydrophilic portion of the E2 endodomain into a structure which would be more energetically favored as a part of the helical TM (3, 11, 12, 13). The phenotypes displayed by the endodomain mutants segregated the mutant viruses into two general categories. Mutants which displayed reproducible phenotypes from both mammalian and insect cells represent host-independent assembly mutants. These mutants would be of the type which are defective in some aspect of assembly because of a virus-encoded structural alteration and engage both host cells in an equivalent way. The second phenotype would be the result of mutant protein interactions with the lipid membrane acquired by the virus during the process of envelopment. The mutants KK391/392 and FF391/392 represent both categories.
The KK391/392 mutant extended the E2 endodomain by a second basic K residue and was expected to physically lengthen the endodomain (13). The length of the hydrophobic segment of an integral membrane protein should closely match the membrane which it spans, producing an energetically favorable structure with charged residues partitioned into more polar environments. A second K at E2 amino acid 392 would not be expected to be energetically stable within the lipid environment under conditions of no hydrophobic mismatch. The TM25 (379 M) KK391/392 mutant, however, restored virus production to levels above those of both individual mutants in both host cell lines. This result suggests that the membrane-proximal K at position 391 may be able to incorporate into the TM, although with 10-fold less efficiency than the F391 residue. Analysis of mutations in pseudorevertants of TM (379 M) show that compensatory insertions of hydrophobic amino acids restore the length of the TM to 26 amino acids (unpublished data). These results demonstrate a strong selective pressure to maintain 26 amino acids in the TM of Sindbis virus. The incorporation of the K at 391 in the mutant KK391/392 restored the TM length to 26 amino acids. Lysine residues are known to have the capacity to incorporate into the periphery of membrane bilayers by a phenomenon termed "snorkeling." This refers to the ability of the flexible side chain of lysine to stretch out of the membrane interior and place the charged amino group in the polar interface (46). These observations suggest that, for the KK391/392 mutant, both lysine residues are located within the cytoplasmic domain, producing the altered particle phenotypes and reduced titers.
The hypothesis that increasing/decreasing the distance between the membrane bilayer and E2 Y400 would define a critical distance between the membrane interface and the capsid binding pocket (20, 23, 52) was tested by inserting an additional K at the membrane/cytoplasm interface (KK391/392). The extended E2 endodomain resulted in significantly lower virus titers from both BHK and U4.4 cells. This effect was not the result of limited or altered protein expression in transfected cells, as determined by polyacrylamide gel electrophoresis (data not shown), or a failure to transport membrane glycoproteins to the plasma membrane (Fig. 9A). The KK391/392 virus density (1.154 g/cm2) compared to that of the wild type (1.17 g/cm2) provides evidence that the particles produced from this mutant are less dense and might contain more membrane than wild-type virus. Thin sections of infected BHK cells show tightly apposed binding of the nucleocapsids to the virus-modified membrane (Fig. 3). The initial association events, however, do not proceed to efficient envelopment; instead, many multicored virus tubes are formed. Negative stains of gradient-purified virus from this mutant reveal multilobed particles containing up to four virus cores. This observation suggests that the shift in virus density seen for the KK391/392 virus results from a higher membrane content in the conjoined particles.
In both the vertebrate and invertebrate cell systems, it would be expected that an additional lysine residue would extend the E2 endodomain without affecting the TM. If the defect produced by extending the E2 endodomain by an additional residue resulted in altering an assembly step in a manner which is host cell independent, the phenotypes in both mammalian and insect cells should be identical. That this is the case was demonstrated by the observation that virus assembly from U4.4 cells produced tubes of multicored viruses as well as single virus particles displaying discontinuity at the virus surface. Thin sections of KK391/392-infected U4.4 cells showed a much lower incidence of membrane tubes than seen for the same mutant grown in BHK cells. In U4.4 cells, this mutant produced completely enveloped, although distorted, virus particles compared to the conjoined particles produced in BHK cells. These particles, however, are noninfectious, as shown by the high particle/PFU ratio. The disposition of the E2 tail in this mutant, while affecting envelopment in both hosts, is able to complete the envelopment process and produce noninfectious virus when grown in U4.4 cells.
The FF391/392 mutant had a different phenotype in insect cells than that seen in BHK cells. FF391/392 grown in U4.4 cells appears to be wild type in morphology in both thin sections of transfected cells and in negative stains of gradient-purified virus. Mutant virus from insect cells assembles without the external aggregated material seen in virus from BHK cells and is found within vesicles typical of insect cell maturation (Fig. 6A). FF391/392 virus grown in U4.4 cells displayed a fusion phenotype most similar to that of wild-type virus (Fig. 9B). Membranes of the insect U4.4 cells might better accommodate the additional aromatic residues, perhaps through membrane lipid reorganization (11). Particle/PFU ratios of FF391/392 virus grown in U4.4 cells are 60% of the values seen for wild-type virus (116 to 33, respectively), evidence of a relatively stable structure, compared to the KK391/392 mutant (103 particles/PFU).
FF391/392 virus particles from BHK cells are deformed by the presence of extraneous material at the virus surface. Electron micrographs of thin sections of FF391/392 in BHK cells reveal a significant number of virus particles budding from the plasma membrane (Fig. 2B and C). These budding particles appear to be associated with extraneous or extruded material. Negative stains of gradient-purified FF391/392 virus particles from BHK cells contain significant numbers of broken particles and particles displaying distorted morphology (Fig. 2D and E). The composition of the virus appendages producing the lollypop appearance is not known, but particles from BHK cells are less dense than wild-type virus, suggesting an increase in the lipid-to-protein ratio. These "lobes" might result from additional material incorporated into virus associating with the outer shell through weaker protein interactions. This possibility is intriguing, since the negative stains of the same mutant grown in insect cells do not contain this extruded material (Fig. 5C to G). In U4.4 cells, lower membrane viscosity and fluidity of these membrane lipids may allow the bilayer to reorganize without distortion. One specific property demonstrated by this mutant is a change in the pH at which it demonstrates maximum membrane fusion in BHK cells. Maximum fusion by the FF391/392 mutant required exposure to pH 8.0 after initial exposure to pH 5.3 and provides biochemical evidence for a conformational difference of the E1 ectodomain residues in the E1-E2 heterotrimers. Virus grown in U4.4 cells demonstrated a larger pH range (7.2 to 7.8), resulting in 100% fusion, and produced 75% fusion at pH 8.0. The fusion pH profile data suggest conformational differences in the virus particles produced from BHK cells and those assembled in U4.4 cells.
We predicted that the FF residues would partition into the membrane-spanning domain of E2. This is supported by the observation that the mutant TM25 (379 M) FF391/392 restores normal virus production. This result suggests that at least one F residue is incorporated into the interface of the TM of TM25 (379). In a related study, Parks produced analogous insertions of L into the signal/anchor domain of paramyxovirus simian virus 5 hemagglutinin-neuraminidase. These mutants provided indirect evidence for incorporation of flanking hydrophobic residues into the TM of an integral membrane protein (39). Assuming that both FF391/392 are shifted in residence from the cytoplasm to the membrane, this mutant would reduce the distance between the membrane and E2 Y401 (FF391/392 numbering) to 9 amino acids (393ARRECLTPY401). This shift, however, did not appear to affect the attachment of nucleocapsids to protein-modified BHK membranes, as seen in Fig. 2A, B, and C. The present E2-C interaction model predicts that misalignment of the critical E2 Y400 residue with the capsid binding pocket may not affect primary binding of the nucleocapsid but might lead to weak capsid binding associations affecting virus assembly or infectivity (23). The phenotype of the FF391/392 mutant was consistent with defective assembly (Fig. 2). We were not able to determine a particle/PFU ratio for this mutant from BHK cells; however, the numbers of virus particles seen in the negative stains shown in Fig. 2D and E are sufficient to suggest that this number is much larger than the titer for this fraction would suggest (7 x 107 PFU/ml) (Fig. 8B), implying an effect on infectivity as well.
The substitution of F for the K at amino acid 391 in the mutant F391 resulted in virus with wild-type properties. This result suggests that E2 A392 may not be incorporated into the TM in the absence of the basic lysine residue. Such a shift would result in 31 amino acids in the endodomain, with 8 amino acids from the membrane to Y400. That assembly proceeds normally suggests that E2 A392 remains in the cytoplasmic domain, leaving R393 and R394 nonadjacent to the membrane. While we are unable to assign a specific domain for F391 in this construct, the possibility arises that, for Sindbis virus E2, a basic residue may not be a strict requirement at the membrane interface (50). If the single F391 substitution is incorporated into the membrane and becomes part of the TM, the effective spanning length of the E2 endodomain to the critical Y400 residue which produces efficient interactions with the capsid binding pocket is 9 amino acid residues.
Collectively, the data gathered from the endodomain mutants add to our knowledge of Sindbis virus assembly in several important ways. These observations suggest that interactions of the E2 endodomain with capsid protein involve several specific associations. (i) There is a critical spanning distance of 9 to 10 amino acids from the E2 amino acid 391 to the critical E2 Y400 for correct interaction with the capsid binding pocket. (ii) Close apposition of the nucleocapsid to the membrane-proximal amino acid 391 within the E2 tail must occur for efficient budding to proceed. (iii) Lateral protein interactions of E1-E2 ectodomains affect the efficient envelopment of virus particles after initial association with the nucleocapsid. (iv) An acidic residue may not be required at amino acid 391 of the E2 endodomain, although appropriate spacing from the membrane juncture to Y400 (9 amino acids) is required for assembly. (v) The composition of the viral membrane is important to virus architecture and assembly. Certain conformations of the virus glycoproteins were observed to be better accommodated in the thinner, less viscous insect membrane. The contiguous TM and endodomain display some flexibility for the amino acid residues found at the membrane-flanking positions, suggesting that certain structural constraints supersede the requirement for a specific amino acid residue. These constraints involve the specific host membrane and may represent the adaptations which these proteins have acquired to shuttle between such divergent host environments.
ACKNOWLEDGMENTS
This research was supported by grants from the National Institutes of Health (AI-42775), The Foundation for Research (Carson City, NV), and the North Carolina Agricultural Research Service (NCARS).
REFERENCES
Anthony, R. P., and D. T. Brown. 1991. Protein-protein interactions in an alphavirus membrane. J. Virol. 65:1187-1194.
Anthony, R. P., A. M. Paredes, and D. T. Brown. 1992. Disulfide bonds are essential for the stability of the Sindbis virus envelope. Virology 190:330-336.
Braun, P., and G. von Heijne. 1999. The aromatic residues Trp and Phe have different effects on the positioning of a transmembrane helix in the microsomal membrane. Biochemistry 38:9778-9782.
Bretscher, M. S. 1993. Cholesterol and the Golgi apparatus. Science 261:1280.
Brown, D. T., M. R. F. Waite, and E. R. Pfefferkorn. 1972. Morphology and morphogenesis of Sindbis virus as seen with freeze-etching techniques. J. Virol. 10:534-536.
Burke, D. J., and K. Keegstra. 1976. Purification and composition of the proteins from Sindbis virus grown in chick and BHK cells. J. Virol. 20:676-686.
Carleton, M., and D. T. Brown. 1996. Disulfide bridge-mediated folding of Sindbis virus glycoproteins. J. Virol. 70:5541-5547.
Carleton, M., H. Lee, M. Mulvey, and D. T. Brown. 1997. Role of glycoprotein PE2 in formation and maturation of the Sindbis virus spike. J. Virol. 71:1558-1566.
Choi, H. K., L. Tong, W. Minor, P. Dumas, U. Boege, M. G. Rossmann, and G. Wengler. 1991. Structure of Sindbis virus core protein reveals a chymotrypsin-like serine proteinase and the organization of the virion. Nature 354:37-43.
Clayton, R. B. 1964. The utilization of sterols by insects. J. Lipid Res. 5:3-19.
de Planque, M. R., J. W. Boots, D. T. Rijkers, R. M. Liskamp, D. V. Greathouse, and J. A. Killian. 2002. The effects of hydrophobic mismatch between phosphatidylcholine bilayers and transmembrane alpha-helical peptides depend on the nature of interfacially exposed aromatic and charged residues. Biochemistry 41:8396-8404.
de Planque, M. R., and J. A. Killian. 2003. Protein-lipid interactions studied with designed transmembrane peptides: role of hydrophobic matching and interfacial anchoring. Mol. Membr. Biol. 20:271-284.
de Planque, M. R., J. A. Kruijtzer, R. M. Liskamp, D. Marsh, D. V. Greathouse, R. E. Koeppe II, B. de Kruijff, and J. A. Killian. 1999. Different membrane anchoring positions of tryptophan and lysine in synthetic transmembrane alpha-helical peptides. J. Biol. Chem. 274:20839-20846.
Edwards, J., and D. Brown. 1984. Sindbis virus induced fusion of tissue cultured Aedes albopictus (mosquito) cells. Virus Res. 1:705-711.
Edwards, J., and D. T. Brown. 1986. Sindbis virus-mediated cell fusion from without is a two-step event. J. Gen. Virol. 67:377-380.
Edwards, J., E. Mann, and D. T. Brown. 1983. Conformational changes in Sindbis virus envelope proteins accompanying exposure to low pH. J. Virol. 45:1090-1097.
Ferreira, D., R. Hernandez, M. Horton, and D. T. Brown. 2003. Morphological variants of Sindbis virus produced by a mutation in the capsid protein. Virology 307:54-66.
Gliedman, J. B., J. F. Smith, and D. T. Brown. 1975. Morphogenesis of Sindbis virus in cultured Aedes albopictus cells. J. Virol. 16:913-926.
Glomb-Reinmund, S., and M. Kielian. 1998. The role of low pH and disulfide shuffling in the entry and fusion of Semliki Forest virus and Sindbis virus. Virology 248:372-381.
Hernandez, R., H. Lee, C. Nelson, and D. T. Brown. 2000. A single deletion in the membrane-proximal region of the Sindbis virus glycoprotein E2 endodomain blocks virus assembly. J. Virol. 74:4220-4228.
Hernandez, R., C. Sinodis, M. Horton, D. Ferreira, C. Yang, and D. T. Brown. 2003. Deletions in the transmembrane domain of a Sindbis virus glycoprotein alter virus infectivity, stability, and host range. J. Virol. 77:12710-12719.
Lee, H., and D. T. Brown. 1994. Mutations in an exposed domain of Sindbis virus capsid protein result in the production of noninfectious virions and morphological variants. Virology 202:390-400.
Lee, S., K. E. Owen, H. K. Choi, H. Lee, G. Lu, G. Wengler, D. T. Brown, M. G. Rossmann, and R. J. Kuhn. 1996. Identification of a protein binding site on the surface of the alphavirus nucleocapsid and its implication in virus assembly. Structure 4:531-541.
Liljestrom, P., and H. Garoff. 1991. Internally located cleavable signal sequences direct the formation of Semliki Forest virus membrane proteins from a polyprotein precursor. J. Virol. 65:147-154.
Liu, L. N., H. Lee, R. Hernandez, and D. T. Brown. 1996. Mutations in the endo domain of Sindbis virus glycoprotein E2 block phosphorylation, reorientation of the endo domain, and nucleocapsid binding. Virology 222:236-246.
Liu, N., and D. T. Brown. 1993. Phosphorylation dephosphorylation events play critical roles in Sindbis virus maturation. Virology 196:703-711.
Liu, N., and D. T. Brown. 1993. Transient translocation of the cytoplasmic (endo) domain of a type I membrane glycoprotein into cellular membranes. J. Cell Biol. 120:877-883.
Lopez, S., J. S. Yao, R. J. Kuhn, E. G. Strauss, and J. H. Strauss. 1994. Nucleocapsid-glycoprotein interactions required for assembly of alphaviruses. J. Virol. 68:1316-1323.
Luukkonen, A., M. Brummer-Korvenkontio, and O. Renkonen. 1973. Lipids of cultured mosquito cells (Aedes albopictus). Comparison with cultured mammalian fibroblasts (BHK 21 cells). Biochim. Biophys. Acta 326:256-261.
Mann, E., J. Edwards, and D. T. Brown. 1983. Polycaryocyte formation mediated by Sindbis virus glycoproteins. J. Virol. 45:1083-1089.
Miller, M. L., and D. T. Brown. 1992. Morphogenesis of Sindbis virus in three subclones of Aedes albopictus (mosquito) cells. J. Virol. 66:4180-4190.
Mitsuhashi, J., and K. Maramorosch. 1964. Leafhopper tissue culture: embryonic, nymphal and imaginal tissues from aseptic insects. Contrib. Boyce Thompson Inst. 22:435.
Mulvey, M., and D. T. Brown. 1996. Assembly of the Sindbis virus spike protein complex. Virology 219:125-132.
Mulvey, M., and D. T. Brown. 1994. Formation and rearrangement of disulfide bonds during maturation of the Sindbis virus E1 glycoprotein. J. Virol. 68:805-812.
Mulvey, M., and D. T. Brown. 1995. Involvement of the molecular chaperone BiP in maturation of Sindbis virus envelope glycoproteins. J. Virol. 69:1621-1627.
Paredes, A. M., D. T. Brown, R. Rothnagel, W. Chiu, R. J. Schoepp, R. E. Johnston, and B. V. Prasad. 1993. Three-dimensional structure of a membrane-containing virus. Proc. Natl. Acad. Sci. USA 90:9095-9099.
Paredes, A. M., D. Ferreira, M. Horton, A. Saad, H. Tsuruta, R. Johnston, W. Klimstra, K. Ryman, R. Hernandez, W. Chiu, and D. T. Brown. 2004. Conformational changes in Sindbis virions resulting from exposure to low pH and interactions with cells suggest that cell penetration may occur at the cell surface in the absence of membrane fusion. Virology 324:373-386.
Paredes, A. M., M. N. Simon, and D. T. Brown. 1992. The mass of the Sindbis virus nucleocapsid suggests it has T = 4 icosahedral symmetry. Virology 187:329-332.
Parks, G. D. 1996. Differential effects of changes in the length of a signal/anchor domain on membrane insertion, subunit assembly, and intracellular transport of a type II integral membrane protein. J. Biol. Chem. 271:7187-7195.
Pletnev, S. V., W. Zhang, S. Mukhopadhyay, B. R. Fisher, R. Hernandez, D. T. Brown, T. S. Baker, M. G. Rossmann, and R. J. Kuhn. 2001. Locations of carbohydrate sites on alphavirus glycoproteins show that E1 forms an icosahedral scaffold. Cell 105:127-136.
Renkonen, O., C. G. Gahmberg, K. Simons, and L. Kaariainen. 1972. The lipids of the plasma membranes and endoplasmic reticulum from cultured baby hamster kidney cells (BHK21). Biochim. Biophys. Acta 255:66-78.
Renz, D., and D. T. Brown. 1976. Characteristics of Sindbis virus temperature-sensitive mutants in cultured BHK-21 and Aedes albopictus (mosquito) cells. J. Virol. 19:775-781.
Rice, C. M., R. Levis, J. H. Strauss, and H. V. Huang. 1987. Production of infectious RNA transcripts from Sindbis virus cDNA clones: mapping of lethal mutations, rescue of a temperature-sensitive marker, and in vitro mutagenesis to generate defined mutants. J. Virol. 61:3809-3819.
Rietveld, A., S. Neutz, K. Simons, and S. Eaton. 1999. Association of sterol-and glycosylphosphatidylinositol-linked proteins with Drosophila raft lipid microdomains. J. Biol. Chem. 274:12049-12054.
Simmons, D. T., and J. H. Strauss. 1974. Translation of Sindbis virus 26 S RNA and 49 S RNA in lysates of rabbit reticulocytes. J. Mol. Biol. 86:397-409.
Strandberg, E., and J. A. Killian. 2003. Snorkeling of lysine side chains in transmembrane helices: how easy can it get? FEBS Lett. 544:69-73.
Strauss, J. H., E. G. Strauss, and R. J. Kuhn. 1995. Budding of alphaviruses. Trends Microbiol. 3:346-350.
Strauss, J. H., Jr., B. W. Burge, and J. E. Darnell. 1969. Sindbis virus infection of chick and hamster cells: synthesis of virus-specific proteins. Virology 37:367-376.
Vaananen, P., and L. Kaariainen. 1980. Fusion and haemolysis of erythrocytes caused by three togaviruses: Semliki Forest, Sindbis and rubella. J. Gen. Virol. 46:467-475.
Wahlberg, J. M., and M. Spiess. 1997. Multiple determinants direct the orientation of signal-anchor proteins: the topogenic role of the hydrophobic signal domain. J. Cell Biol. 137:555-562.
Yeagle, P. L. 1991. Modulation of membrane function by cholesterol. Biochimie 73:1303-1310.
Zhang, W., S. Mukhopadhyay, S. V. Pletnev, T. S. Baker, R. J. Kuhn, and M. G. Rossmann. 2002. Placement of the structural proteins in Sindbis virus. J. Virol. 76:11645-11658.
Zhao, H., B. Lindquist, H. Garoff, C.-H. von Bonsdorff, and P. Liljestrom. 1994. A tyrosine-based motif in the cytoplasmic domain of the alphavirus envelope protein is essential for budding. EMBO J. 13:4204-4211.(Raquel Hernandez, Davis F)