INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Virus budding is a critical step in the life cycle of all enveloped viruses. Budding may be defined as the progressive envelopment of the virus core by a cellular membrane enriched with viral membrane proteins, culminating in a membrane fission reaction to release the completed virus particle. Different viruses use different host cell membranes as budding sites, including the plasma membrane and various membranes of the exocytic pathway. Viruses also differ in their requirements for virus proteins to drive the budding reaction (14). Viruses such as the alphaviruses and hepadnaviruses have a strict requirement for both nucleocapsids and spike proteins to permit virus budding (31, 57, 59). In contrast, the budding reactions of various other viruses can be driven by the capsid or core protein, by the matrix protein, or by the membrane proteins, without obligatory involvement of the other viral protein subunits (14). Virus budding reactions are an important area of research because of their key roles in virus replication, their potential as therapeutic targets, and their relevance to cellular membrane budding reactions. However, quantitative experimental methods to specifically assay budding have been limited. In the absence of such focused systems, broader studies of infectious particle production by necessity measure a wide range of reactions during the virus life cycle, including the biosynthesis of viral components and the traffic of viral membrane proteins through the exocytic pathway. The lack of more effective model systems has made it difficult to address fundamental questions about viral budding reactions, such as requirements for cellular components and energy sources.

Alphaviruses such as Semliki Forest virus (SFV) are simple, well-characterized enveloped animal viruses (see references 20 and 56 for a review). Each SFV particle contains 240 copies of four structural proteins: the capsid protein, which packages the single plus-stranded RNA genome into nucleocapsids, and three envelope proteins, the type I transmembrane polypeptides E1 and E2 (each about 50 kDa) and the peripheral E3 polypeptide (~10 kDa). The envelope proteins assemble into 80 spikes, each consisting of a trimer, (E1/E2/E3)₃. Both the spike protein layer and the viral nucleocapsid are arranged as T=4 icosohedral structures, which associate with each other via a one-to-one interaction of the E2 internal domain and the capsid protein (7, 13). The virus lipid bilayer is derived from the host cell plasma membrane during budding.

The life cycle of SFV and other alphaviruses has been studied in detail (20, 56). SFV enters host cells via receptor-mediated endocytosis, and virus membrane fusion is mediated by the spike protein and triggered by the low pH present in the endosome (15, 20). In addition to its low pH requirement, fusion of alphaviruses such as SFV is also strongly dependent on the presence of cholesterol and sphingolipid in the target membrane (21, 25, 37, 38, 65). Following fusion between the viral and endosome membranes, nucleocapsids are released into the cytoplasm and viral replication is initiated. Progeny RNA molecules associate with capsid protein in the cytoplasm to form new nucleocapsids. The spike protein E1 subunit and the E2 precursor, p62, are translated and translocated into the rough endoplasmic reticulum, where they are glycosylated and form a stable but noncovalently associated heterodimer. The dimer is transported through the secretory pathway and processing of p62 to mature E2 and E3 is carried out in the late secretory pathway by furin, a cellular protease. E1, E2, and E3 are then transported to the plasma membrane, where virus budding occurs.

Alphavirus budding is a clear example of a budding reaction that requires both nucleocapsids and spike proteins. This is due to a specific interaction between the capsid protein and a key tyrosine-containing motif in the cytoplasmic tail of E2 (56, 67). Structural studies indicate that this region of E2 binds to a hydrophobic pocket on the surface of the nucleocapsid (28, 52). Expression studies demonstrated that, although p62 could be transported to the cell surface in the absence of E1, stable association with nucleocapsids did not occur, suggesting that the dimeric interaction of E1 and E2 is important to maintain the correct conformation of the E2 tail for nucleocapsid binding (1). Further studies showed that lateral interactions between heterodimers are also required for efficient budding (12). In keeping with the known structure of the alphavirus particle (7, 13), it appears that the E1/E2 heterodimers assemble into trimers which associate via lateral interactions and that the resultant multivalency of E2-nucleocapsid binding promotes virus budding.

In addition to these requirements for viral proteins in budding, studies of alphavirus infection in cholesterol-depleted cells have revealed a role for cholesterol in efficient virus exit. Along with their well-characterized requirement for cholesterol in the membrane fusion reaction, both SFV and Sindbis virus (SIN) show strong requirements for cholesterol in exit (32, 34). In contrast, srf-3 (sterol requirement in function), an SFV mutant selected for more efficient growth in cholesterol-depleted cells, has markedly increased levels of both fusion and exit in the absence of cholesterol compared to those of wild-type (wt) virus (34, 62). The increased cholesterol independence of srf-3 fusion and exit were demonstrated to be caused by a single point mutation in the srf-3 E1 subunit, proline 226serine (P226S) (62). Mutagenesis studies of SIN demonstrated that the E1 226 region is also involved in the cholesterol dependence of its fusion and exit (32). However, although these studies reveal a role for cholesterol in the alphavirus exit pathway, the stage of the viral exit pathway affected and the mechanism of the cholesterol effect remain unclear.

We have here established a quantitative biochemical assay for SFV budding that is generally applicable to viruses that bud from the plasma membrane. The assay was based on biotin derivatization of radiolabeled virus spike proteins at the cell surface and subsequent retrieval of biotin-tagged virions using magnetic streptavidan particles. We have used the assay to define several basic properties of SFV budding, including kinetics, temperature dependence, and the role of cytoskeletal elements and exocytic transport. Using wt virus and the srf-3 mutant, we demonstrated that the previously observed cholesterol requirement for production of wt SFV was due to a block in spike protein budding from cholesterol-depleted cells. The block in wt virus budding correlated with the rapid degradation of the wt spike proteins throughout the infectious cycle in sterol-deficient cells.

(This research was conducted by Y.E.L. in partial fulfillment of the requirements for a Ph.D. from the Sue Golding Graduate Division of Medical Sciences, Albert Einstein College of Medicine, Yeshiva University.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. BHK-21 cells were cultured at 37°C in Dulbecco's modified Eagle's (DME) medium containing 100 U of penicillin/ml and 100 µg of streptomycin/ml (P/S), 5% fetal calf serum (FCS), and 10% tryptose phosphate broth (38). Control cholesterol-containing C6/36 mosquito cells were maintained at 28°C in DME medium with P/S and 10% heat-inactivated FCS (34). Cholesterol-depleted C6/36 cells were prepared by four passages at 28°C in DME medium containing P/S and 10% delipidated heat-inactivated FCS (32, 38) and used for experiments between passages 5 and 15 (32, 33). The levels of both free and esterified cholesterol in the cholesterol-depleted cells were less than 2% of those of control cells (38).

Virus infection, metabolic labeling, and cell surface budding assay. BHK cells, control C6/36 cells, or cholesterol-depleted C6/36 cells were cultured on 35-mm plates for 1 to 3 days until just confluent and then infected by the following protocols. BHK cells (ca. 1 × 10⁶ to 2 × 10⁶ cells/plate) were infected with wt SFV at 10 PFU/cell in 0.6 ml of minimal essential medium (MEM) containing P/S, 10 mM HEPES (pH 7.4), and 0.2% bovine serum albumin (BSA) for 1 h at 37°C. Cells were then washed to remove the input virus, and the incubation was continued for another 4 to 5 h (a total of 5 to 6 h) in 1 ml of the same medium. Control C6/36 cells (ca. 2 × 10⁶ to 6 × 10⁶ cells/plate) were infected with wt or srf-3 at the same multiplicities of infection (either 10 or 100 PFU/cell) in 0.7 ml of Opti-MEM containing 0.2% BSA (O/B) for 2 h at 28°C, washed with O/B, and incubated in 1.5 ml of O/B for the indicated times (1.5 to 5 h). To overcome the inhibition of wt virus fusion in cholesterol-depleted C6/36 cells and permit viral protein expression, depleted cells (ca. 3 × 10⁶ to 6 × 10⁶ cells/plate) were infected with wt SFV at 1,000 PFU/cell or srf-3 at 100 PFU/cell in 0.7 ml of O/B for 2 h at 28°C. The cells were then washed twice with O/B and incubated in 1.5 ml of O/B with or without 20 mM NH₄Cl for the indicated times.

Quantitative retrieval of biotinylated virus particles and spike proteins. One-quarter of the post-biotin incubation medium samples and one-quarter of the cell lysate samples were used for analysis. The retrieval method and quantitation of budding efficiency were essentially the same for BHK cells, control C6/36 cells, or cholesterol-depleted C6/36 cells. Medium samples were incubated with 25 µl of BioMag Streptavidin Ultra-Load Particles (mag-SA; PerSeptive Biosystems, Framingham, Mass.) for 30 min at 4°C on a nutator. The mag-SA particles were collected using a magnetic tube holder (Magnetic Particle Concentrator MPC-E; Dynal, Oslo, Norway), and washed three times with 0.5 ml of cold PBS containing 0.2% BSA and 1 mM PMSF and one time with cold PBS containing 1 mM PMSF. Retrieval of biotinylated spike proteins from the cell lysate samples was similar except that 30 µl of mag-SA was used and wash solutions contained 1% Triton X-100. Specific retrieval of biotinylated spike proteins from the medium was performed by adding 1% Triton X-100 to medium samples, followed by retrieval with 25 µl of mag-SA and washing as for cell lysate samples. mag-SA-bound material was then released by heating the samples to 95°C in 1× sodium dodecyl sulfate (SDS) sample buffer for 5 min. Samples were analyzed by electrophoresis on 10% acrylamide gels followed by fluorography (23). In order to accurately recover radiolabeled capsid protein from budded virus for SDS-polyacrylamide gel electrophoresis (PAGE) analysis, 3 µg of purified nonradiolabeled SFV per sample was included as carrier in the 1× SDS sample buffer. Gels were quantitated by phosphorimaging (ImageQuant, v. 1.2; Molecular Dynamics, Inc., Sunnyvale, Calif.).

Capsid protease protection assay. To assay virus particle and virus membrane integrity, biotinylated virus was retrieved from the post-biotin incubation medium using mag-SA as described above. After the final PBS wash, the mag-SA was resuspended in 25 µl of 10 mM Tris (pH 7.4)-50 mM NaCl. The samples were digested with 50 µg of trypsin XIII (Sigma) per ml for 10 min on ice in the presence or absence of 1% Triton X-100. Trypsin was then inhibited by the addition of soybean trypsin inhibitor and PMSF to final concentrations of 150 µg/ml. Then, 3 µg of cold purified SFV was added as carrier to the samples before boiling them in SDS gel buffer, followed by SDS-PAGE analysis. As a control, parallel trypsin digestion and SDS-PAGE analysis were performed using gradient-purified radiolabeled SFV.

Electron microscopy. Electron microscopic analysis was used to evaluate the morphology of biotinylated viruses retrieved by mag-SA. Two 35-mm plates of BHK cells were infected as described above, mock pulse-labeled, chased, and either derivatized with biotin or mock treated. Both samples were then postincubated for 1 h at 37°C, and one-half of the medium was retrieved by using 50 µl of mag-SA. After the final PBS wash, the mag-SA pellets were fixed in 1 ml of 2.5% glutaraldehyde in 0.1 M cacodylate for 30 min, dehydrated, embedded, processed, and evaluated by transmission electron microscopy, all as previously described for cell pellets (34). Parallel samples from radiolabeled cells were evaluated by retrieval and SDS-PAGE and showed efficient budding and specific mag-SA binding.

Assay of E1/E2 dimer association in C6/36 cells. Control or cholesterol-depleted C6/36 cells were infected with either wt SFV or the srf-3 mutant as described above. The cells were pulse-labeled, chased for 30 min, and lysed in 0.5 ml of lysis buffer containing 1% NP-40, 50 mM Tris-HCl (pH 7.4), 100 mM NaCl, 0.9 mM CaCl₂, 0.5 mM MgCl₂, 1 µg of leupeptin/ml, 1% aprotinin, 1 mM PMSF, and 1 mg of BSA/ml (19, 63). Monoclonal antibodies (MAb) that recognize either the E1 subunit (E1-1) or the E2 subunit (E2-3) (23) were used to coprecipitate the spike dimer, essentially as previously described (19, 63).

Analysis of spike protein transport to the cell surface of cholesterol-depleted cells. Due to the fusion block in cholesterol-depleted cells, only about 10 to 30% of the cells could be infected even at high multiplicities, and mosquito cells do not efficiently shut off host protein synthesis. Thus, to accurately measure the rate of spike protein transport to the cell surface, the spike proteins expressed in the cells were first quantitatively immunoprecipitated with an anti-spike protein antibody (23). Half of the immunoprecipitate was directly analyzed by SDS-PAGE to determine the amount of total radiolabeled spike protein in the cells. To quantitate the biotinylated spike proteins derivatized at the cell surface, the other half of the immunoprecipitate was boiled in SDS-containing buffer to release the spike proteins from the zysorbin (reference 10 and as described below). The released spike proteins were then retrieved by mag-SA binding and analyzed by SDS-PAGE.

Quantitation of virus budding and biotinylated spike protein degradation. Control or cholesterol-depleted C6/36 cells were infected with wt or srf-3 and tested at various times of infection using the budding assay. Due to the lower expression level of viral spike proteins early in the infection cycle and the lack of host protein shutoff, the analysis of both medium and cell lysate samples was modified by direct mag-SA retrieval of samples, followed by immunoprecipitation with an anti-spike protein antibody to remove non-spike protein background (10). The biotinylated proteins from one-quarter of the cell lysate or medium were first retrieved using 30 µl of mag-SA in the presence of detergent and then released from mag-SA by heating them twice at 95°C for 5 min in 50 µl of buffer containing 3% SDS, 50 mM NaCl, and 10 mM Tris (pH 7.4). The mag-SA was removed, the supernatant was diluted with 450 µl of buffer (50 mM NaCl; 10 mM Tris, pH 7.4; 1% Triton X-100) and immunoprecipitated with anti-spike protein antibody, and the samples were analyzed by SDS-PAGE. The biotinylated viral proteins released in the medium and the biotinylated spike proteins remaining in the cells were quantitated and used to determine the efficiency of budding and the spike protein degradation rate.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Development of an SFV budding assay. In order to measure SFV budding, a method to specifically monitor the incorporation of plasma membrane spike proteins into virus particles was required. During the initial phases of this work, we explored a variety of possible assays for SFV budding. We tested the use of temperature-sensitive SFV or SIN mutants (43, 53) or ionic conditions (64) to reversibly accumulate budding-competent virus spike proteins at the plasma membrane. None of these conditions gave a clearcut exit block that could be synchronously released to monitor budding (data not shown). We therefore developed a direct biochemical method to specifically tag SFV spike proteins at the plasma membrane and monitor their incorporation into virus particles.

View larger version (53K): [in this window] [in a new window]   FIG. 1.   Characterization of retrieval and integrity of biotinylated virus particles. BHK cells were infected with SFV at 10 PFU/cell for 5 h, labeled with [35S] methionine-cysteine at 100 µCi/ml for 15 min, and chased for 45 min. The cells were next either derivatized with biotin or mock derivatized and then incubated for 60 min at 37°C; the media were then collected for analysis. The total viral spike proteins in an aliquot of the media were determined by trichloroacetic acid precipitation (lanes 1 and 2). Equivalent aliquots of the media were retrieved with mag-SA in the absence of detergent and either directly analyzed by SDS-PAGE (lanes 3 and 4) or digested with exogenous trypsin before SDS-PAGE analysis (lanes 5 and 6) in the presence or absence of 1% Triton X-100 as indicated. Capsid protein is trypsin-sensitive and fully digested when the virus membrane is disrupted by detergent. [35S]methionine-[35S]cysteine-labeled gradient purified SFV was similarly processed in parallel as a control (lanes 7 to 9). A representative example of three experiments is shown, with all lanes exposed for the same length of time.

View larger version (112K): [in this window] [in a new window]   FIG. 2.   Electron microscopy of retrieved virus particles. Similar to Fig. 1, BHK cells were infected with SFV for 6 h, biotinylated (A) or mock treated (B), and further incubated at 37°C for 60 min. Virus particles in the medium were retrieved by mag-SA and processed for electron microscopy. Note that virus particles are only observed in the biotinylated sample. Bar, 0.1 µm.

Specificity of the assay for SFV budding from the plasma membrane. A key issue in the validation of this assay was its ability to specifically monitor the budding of newly formed progeny virus. Since SFV is known to bind to cell surface receptors (56), it was important to determine if the dissociation of newly budded virus particles from plasma membrane receptors played an important role in the kinetics of exit. In addition, it was also important to prove that the retrieved virus resulted from de novo budding of cell surface spike proteins during the post-biotin incubation rather than from the release of receptor-bound virus particles formed prior to biotinylation. We addressed these points both morphologically and biochemically. Electron microscopy was performed using streptavidin and immunogold to follow cell surface biotin (see Materials and Methods). Immediately after biotin derivatization, infected BHK cells showed abundant gold labeling at the plasma membrane (data not shown). Label was associated with numerous linear regions of the membrane and also with some regions overlying nucleocapsids or containing forming virus buds. Negligible gold labeling was observed in the absence of biotinylation. After post-biotin incubation of the biotinylated cells at 37°C for 45 min, a markedly increased number of gold-labeled virus particles and forming virus buds were observed. These morphological results suggested that the assay was detecting newly formed virus particles.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Comparison of the properties of release of prebound virus and nascent virus. (A) Release of prebound SFV. [35S]methionine-[35S]cysteine-labeled gradient purified SFV (~3 × 106 cpm/plate) was bound to unlabeled SFV-infected BHK cells on ice for 1 h. Using the standard protocol, unbound virus was removed by washing, and the cells with bound exogenous virus were biotin derivatized and quenched. (B) Release of nascent virus. BHK cells were infected with SFV for 5 h, radiolabeled, and biotin derivatized as in Fig. 1. Both the panel A and panel B samples were then incubated in post-biotin incubation medium (pH 8.0) at 4 or 37°C for 30 or 60 min to permit budding or release of radiolabeled virus. The media were then collected, and the biotinylated virus was retrieved using mag-SA. The samples were analyzed by SDS-PAGE and quantitation of the viral E2 protein. The amount of E2 release at 4°C was set to 1 for each panel. Note the different scales on the y axis of the A and B panels. Panel A is a representative example of two experiments; panel B shows averaged data and standard deviations from four separate experiments.

Kinetics and temperature dependence of SFV budding. We next used the assay to characterize basic features of virus budding in BHK cells. Retrieval was performed in the absence of detergent in order to recover virus particles, as described above. Such biotin-tagged virus samples would contain both biotinylated and nonbiotinylated radiolabeled spike proteins. Alternatively, retrieval could also be performed in the presence of detergent, and under these conditions the assay could specifically characterize the efficiency of cell surface spike protein budding, in the absence of retrieval of nonbiotinylated spike proteins.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Properties of SFV budding. (A) Kinetics. BHK cells were infected with SFV for 5 h, radiolabeled, and biotin derivatized as in Fig. 1. Post-biotin incubation was carried out at 37°C for the indicated times, and biotinylated virus particles in the medium were retrieved by mag-SA and analyzed by SDS-PAGE and phosphorimaging to quantitate the E2 subunit, graphed in arbitrary units. A representative example of five experiments is shown. (B) Temperature dependence. BHK cells were prepared as in panel A, the post-biotin incubation was carried out at the indicated temperature for 2 h, and budding was quantitated as in panel A. A representative example of four experiments is shown.

The role of membrane transport and cytoskeletal elements in SFV budding. The experiments described above indicated that a population of biotin-tagged spike proteins at the cell surface continued to be incorporated into virus particles for at least 2 h at 37°C. During this time, virus budding would cause the net loss of spike proteins, nucleocapsids, and membrane lipids from the cellular pool. We used the membrane transport inhibitors monensin and brefeldin A to test the importance of replenishment of plasma membrane components to continued budding (Fig. 5). Monensin inhibits exocytic transport from the medial Golgi to the plasma membrane (18), while brefeldin A inhibits transport from the endoplasmic reticulum to the Golgi and from the TGN to the plasma membrane (27, 44). The transport effects of both inhibitors are known to occur rapidly after their addition to cells, and control experiments demonstrated that the addition of monensin or brefeldin A to the chase medium blocked 90 to 95% of the transport of newly synthesized spike proteins to the cell surface during the 45-min chase (data not shown). The kinetics of inhibition thus made it possible to add the inhibitors at the start of the post-biotin incubation and observe their effects on subsequent virus budding. Budding of biotin-tagged spike proteins was specifically monitored in order to observe the effects of the transport inhibitors on virus budding, rather than on the delivery of additional radiolabeled spike proteins to the plasma membrane. SFV budding was unaffected by monensin or brefeldin A during the first 30 min of incubation, during which time about 3.5% of the total biotin-labeled E2 protein was released in virus particles (Fig. 5). A slight decrease was observed for both of the drugs after a 60-min incubation, from ~5.6% budding efficiency in untreated cells to ~4.5% in treated cells. Even after 2 h of incubation of cells in the presence of monensin or brefeldin A, SFV budding occurred at reasonable efficiency, although it was reduced (ca. 20 to 40%) compared to that in untreated cells. Thus, budding of spike proteins from the cell surface was relatively independent of continued exocytic transport to the plasma membrane. This result emphasizes that the assay focuses on the late steps of budding from the plasma membrane rather than on the general exocytic pathway involved in spike protein processing and transport.

View larger version (21K): [in this window] [in a new window]   FIG. 5.   Role of membrane transport in SFV budding. BHK cells were infected and biotinylated as in Fig. 4 and then incubated on ice for 10 min in post-biotin incubation medium containing 10 µM monensin or 5 µg of brefeldin A per ml. The cells were then shifted to 37°C in the same medium, and the incubation was continued until the indicated times. The medium was then treated with mag-SA in the presence of detergent to specifically retrieve biotinylated spike proteins. Samples were analyzed by SDS-PAGE and phosphorimaging to quantitate the E2 subunit. The biotinylated E2 released in the medium at each time point was expressed as the percentage of the total biotinylated E2 present on the cell surface at time zero. The points are the average of duplicate samples, and the bars show the range. A representative example of three experiments is shown.

Localization of the cholesterol requirement in SFV exit. Having established and characterized the SFV budding assay, we wished to use it to analyze the previously described block in the exit pathway of wt SFV from cholesterol-depleted cells (34). As a control, we used the SFV mutant srf-3 which, although not completely sterol independent, is more efficient at both fusion with and exit from cholesterol-depleted cells (6, 62). We optimized the infection and incubation conditions for the budding assay in control or sterol-depleted C6/36 mosquito cells and used these conditions to test wt and srf-3 budding. Spike proteins in both wt- and srf-3-infected control cells were efficiently biotin labeled at the cell surface, and biotin-tagged viruses budded from control cells infected with either virus (Fig. 6A). Quantitation showed comparable levels of budding for both viruses, with a budding index of about 0.6 after 2.5 h (graph in Fig. 6A). When the same assay was performed in cholesterol-depleted cells, efficient biotin labeling of the E1 and E2 spike protein subunits was observed for either wt or mutant-infected cells (Fig. 6B). However, the efficiency of the budding of wt virus was reduced to background levels, while abundant budding of srf-3 was observed, with a budding index of ~0.6 after 4 h of post-biotin incubation. These results thus suggested that the cholesterol requirement for the exit of wt SFV resided in a late stage of virus assembly, budding from the plasma membrane, and demonstrated that the budding of the srf-3 mutant was significantly less cholesterol dependent than that of wt virus.

View larger version (61K): [in this window] [in a new window]   FIG. 6.   Effect of cholesterol depletion on SFV budding. (A) Cholesterol-containing C6/36 mosquito cells were infected with wt SFV or srf-3 at 100 PFU/cell for 6 h, pulse-labeled for 15 min, chased for 30 min, and derivatized with biotin. The cells were then incubated at 28°C for the indicated times. The 0-h sample was incubated on ice for the duration of the experiment. Biotinylated virus particles in the medium and biotinylated spike proteins present in the cells were collected by mag-SA retrieval at each time point. Samples were analyzed by SDS-PAGE and phosphorimaging, and the data are expressed as a budding index, defined as the spike protein radioactivity retrieved from the medium divided by the spike protein radioactivity retrieved from both the medium and the cell lysate. (B) Cholesterol-depleted C6/36 cells were infected with wt SFV at 1,000 PFU/cell or with srf-3 at 100 PFU/cell for 24 h, pulse-labeled for 15 min, chased for 90 min, derivatized with biotin, and incubated at 28°C for the indicated times. Mag-SA retrieval and quantitation were performed as in panel A. Panels A and B show representative examples of three experiments in each cell type.

Characterization of spike protein dimerization and transport in sterol-depleted cells. Although the results presented above demonstrated that both the wt and the srf-3 spike proteins were accessible to biotin labeling at the plasma membrane of depleted cells, it was possible that in the absence of cholesterol the conformation or quantity of the wt spike protein was inadequate to support efficient budding. A decrease in wt spike protein transport might alter the plasma membrane spike protein pool and thus affect virus budding. Alternatively, the wt E1/E2 dimer interaction might be altered, similar to an SFV spike protein mutant that has a striking reduction in the stability of the E1/E2 dimer and greatly reduced budding but unimpaired transport (10). Thus, it was important to compare wt and srf-3 spike protein properties in cholesterol-depleted cells and to determine if the wt spike protein was altered in either dimer stability or delivery rate to the plasma membrane.

View larger version (21K): [in this window] [in a new window]   FIG. 7.   Transport of wt and mutant spike proteins to the plasma membrane in cholesterol-depleted cells. Cholesterol-depleted C6/36 cells were infected with wt SFV (1,000 PFU/cell) or srf-3 (100 PFU/cell) for 24 h. Transport of newly synthesized spike proteins to the plasma membrane was then measured by pulse-labeling for 5 min and chasing for the indicated times at 28°C. At each time point, the medium was collected, the cell surface spike proteins were derivatized with biotin, and the cells were harvested by detergent lysis. The radiolabeled spike proteins present on the cell surface at each time point were assessed by quantitative immunoprecipitation of an aliquot of the cell lysate using a polyclonal antibody to the SFV spike protein, followed by mag-SA retrieval. The amount of spike proteins derivatized by cell surface biotinylation, referred to as cell surface (CS) samples, was expressed as a percentage of the total radiolabeled spike proteins immunoprecipitated at the 5-min chase time. For comparison, at each time point the amount of radiolabeled spike proteins released into the medium due to virus budding was also determined by quantitative immunoprecipitation using the same antibody and is expressed as a percentage of the total spike proteins at the 5-min chase time (release samples). A representative example of two experiments is shown.

Comparison of spike protein budding and degradation. Little was known about the turnover of the cell surface spike protein in either control or sterol-depleted cells. Previous studies in BHK cells demonstrated that under conditions in which budding is blocked, such as the expression of spike proteins in the absence of capsid protein, spike proteins are rapidly degraded (10, 66). We examined the question of cell surface spike protein turnover first in control C6/36 cells, using biotin derivatization to mark spike proteins that had reached the plasma membrane and comparing cells at various times postinfection.

View larger version (27K): [in this window] [in a new window]   FIG. 8.   Degradation and budding of wt and srf-3 spike proteins in control cells. Cholesterol-containing C6/36 cells were infected with wt SFV (A) or srf-3 (B) at 10 PFU/cell for the indicated times. At each time of infection, the cells were pulse-labeled for 15 min with [35S]methionine-cysteine, using concentrations of 200 µCi/ml for the 3.5-h time point and 80 µCi/ml for the remaining time points. The cells were then chased for 30 min, derivatized with biotin (time zero), and incubated at 28°C for 2 h. The biotinylated spike proteins present in cell lysates and media were quantitated using mag-SA retrieval and immunoprecipitation as described in Materials and Methods. The amount of budding after 2 h (solid bars) was determined by comparing the biotinylated E2 subunit in the medium to the biotinylated E2 present in the cell lysate at time zero. The degradation of cell surface spike proteins after 2 h (hatched bars) was determined by subtracting the biotinylated E2 recovered in the cell lysate and medium from the biotinylated E2 in the cell lysate at time zero and is expressed as the percentage of the biotinylated E2 in the cell lysate at time zero. A representative example of two experiments is shown.

View larger version (29K): [in this window] [in a new window]   FIG. 9.   Degradation and budding of wt and srf-3 spike proteins in cholesterol-depleted cells. Cholesterol-depleted C6/36 cells were infected with wt SFV at 1,000 PFU/cell or with srf-3 at 100 PFU/cell for 2 h and then incubated in the presence of 20 mM NH4Cl to inhibit secondary infection. Thirty minutes before the completion of the indicated infection time, the cells were washed and the incubation was completed in the absence of NH4Cl. The cells were then pulse-labeled for 15 min with 35S-labeled methionine-cysteine using concentrations of 200 µCi/ml for wt infected cells and 40 µCi/ml for srf-3-infected cells. The cells were then chased for 90 min, derivatized with biotin, incubated at 28°C for 5 h, and analyzed for spike protein degradation (hatched bars) and budding (solid bars) as in Fig. 8. A representative example of two experiments is shown.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have developed a reliable biochemical method to measure the incorporation of cell surface viral spike proteins into budded virions. This assay quantitatively retrieved intact, morphologically normal virus particles. Our studies showed that the assay was specific for budding, defined as the final steps in the assembly and pinching off of a completed virus particle from the plasma membrane. In the case of SFV, such a budding reaction presumably would include the lateral association of the spike polypeptides into trimers and higher-order complexes, the formation of circumferential spike protein-nucleocapsid interactions, and the fission of the virus membrane to produce the enveloped virion. Biotin tagging of plasma membrane spike proteins permitted the collective evaluation of these steps in the budding reaction separate from the protein biosynthetic and secretory pathways.

Our studies characterized several aspects of virus budding, including kinetics, temperature dependence, and the role of membrane transport and cytoskeletal elements. The biotin-tagged spike protein pool was continuously incorporated into virus particles for at least 2 h at 37°C. Budding, although strongly temperature dependent, still occurred at temperatures such as 20°C which inhibit some membrane trafficking events. Our results with monensin and brefeldin A demonstrated that once the nascent spike protein was transported to the plasma membrane pool, budding during a 2-h incubation period at 37°C was largely independent of further exocytic delivery to the cell surface. Although we assume that at some point treatment with membrane transport inhibitors would limit budding, our data indicate that the amount of virus budding during a 2-h incubation time (estimated from growth curves as ~4 × 10³ PFU/cell) did not sufficiently deplete the cells of membrane components to produce a block in budding. This result also provides further evidence that the budding assay monitors incorporation of cell surface spike proteins into virus particles.

Actin microfilaments appear to be involved in the maturation and assembly of several enveloped viruses, such as measles virus (55, 61) and human immunodeficiency virus (30, 45), but not in the production of vesicular stomatitis virus and influenza virus (17). Our results from budding assays and growth curve analyses suggested that actin filaments and microtubules were not required for either SFV budding or for the overall production of infectious progeny virus, including the transport of spike proteins and nucleocapsid to the site of budding.

Our previous work has described a role for cellular cholesterol in the efficient production of progeny SFV and SIN, demonstrable either by infection at high multiplicity or by RNA transfection (32-34, 62). Cholesterol-depleted infected cells express the virus genome, translate the capsid and spike proteins, and process the p62 subunit but show inefficient virus exit. Virus mutants such as srf-3 that are less cholesterol dependent for fusion and entry are also less sterol dependent for virus exit. Using the budding assay developed in these studies, we were here able to demonstrate that the block in wt virus exit from cholesterol-depleted cells was due to an inhibition of virus budding from the plasma membrane. These studies thus localized the site of the cholesterol-dependent block to a late stage in virus exit at the cell surface and showed that budding from sterol-depleted cells was increased in the srf-3 mutant. Pulse-chase analysis of depleted infected cells demonstrated that the wt and mutant SFV spike proteins were equivalently dimerized and transported through the exocytic pathway to the plasma membrane. However, the wt spike protein was rapidly degraded (t_1/2 of <2 h) and showed a lower steady-state distribution on the plasma membrane compared to the more efficiently budding srf-3 mutant. Rapid turnover of cell surface spike proteins was not simply due to the absence of cholesterol, since at early times of infection both wt and srf-3 spike proteins were rapidly degraded in either control or depleted cells. As infection progressed, the spike protein degradation rate markedly decreased for wt virus in control cells and for srf-3 in either control or depleted cells, and abundant virus budding was observed. Thus, the behavior of the wt spike protein in depleted cells was unusual in that rapid spike protein degradation was maintained for the duration of infection. These results demonstrated a strong inverse correlation between the spike protein degradation rate and virus budding. Although the relationship between these two events was striking, it is not yet clear whether rapid degradation of cell surface spike proteins is a cause or an effect of decreased virus budding.

Rapid turnover of SFV spike proteins has been previously described under conditions in which budding is impaired, such as expression of the spike in the absence of capsid protein (10, 66). Although at least a proportion of the degradation occurs after the spike protein is delivered to the plasma membrane (10, 66), efficient endocytic uptake of the cell surface E2 subunit was not detected (66). Thus, the site of spike protein degradation in these experiments is not clear. In contrast, the gp160 envelope proteins of human and simian immunodeficiency viruses are efficiently endocytosed from the plasma membrane pool via interaction of a highly conserved tyrosine in the cytoplasmic tail with adapter complexes and clathrin-coated pits (2). The internalization of gp160 decreases the spike protein concentration at the plasma membrane and in the budded virus particle, reduces syncytia formation (46), and may be important in decreasing the susceptibility of virus-infected cells to the humoral immune response (46). Interestingly, at least one report suggests that gp160 internalization is blocked by association with the Gag precursor protein (11), raising the possibility that Gag interaction may suppress endocytosis and redirect gp160 into budding virions. In contrast, the wt form of the influenza virus hemagglutinin (HA) is normally endocytosed at a very low rate and has a half-life of ca. 8 to 10 h (40, 68), similar to that of typical plasma membrane proteins. Insertion of a tyrosine-based internalization motif into the HA cytoplasmic tail causes the protein to be very rapidly endocytosed, and further signals mediate the rapid intracellular degradation of HA in late endosomes and/or lysosomes (t_1/2 of ~2 to 3 h) (68). Thus, it is clear that viral spike proteins can go through several different pathways at the cell surface, including rapid degradation through endocytic clearance, and long-lived protein maintenance in the absence of endocytosis.

Importantly, our data demonstrated that even when the entire SFV wt virus genome was expressed, wt cell surface spike proteins were rapidly degraded at early times of infection or in sterol-depleted cells. This result was unexpected and differs from previous experiments in which SFV spikes were degraded when expressed in the absence of nucleocapsid (66) or in which mutant virus spike proteins were degraded (10). To address the relationship between the degradation of cell surface SFV spike proteins and virus budding, we attempted to block turnover of the wt spike protein in sterol-depleted cells. Cells were treated with the weak bases NH₄Cl or chloroquine or a protease inhibitor cocktail containing leupeptin, pepstatin, and aprotinin. No significant increase in the half-life of the spike protein or in virus budding was observed (Lu and Kielian, unpublished data). This experiment was complicated by the fact that, even if SFV spike proteins were being cleared from the cell surface by endocytosis, blocking degradation alone might not return the spike protein to the plasma membrane in quantities sufficient to promote budding. Thus, it is not yet clear what role endocytic uptake of the SFV spike protein may play in its degradation or whether degradation controls the budding-competent concentration of the spike protein at the plasma membrane.

How might cholesterol influence the budding of the SFV spike protein? One possibility is that the spike protein requires cholesterol-enriched regions of the membrane for budding and that the srf-3 mutation makes the spike less dependent on such regions. Within cellular membranes, cholesterol plays an important role in the formation of cholesterol and sphingolipid-enriched lipid domains termed "rafts," operationally assayed by their resistance to solubilization with the detergent Triton X-100 at 4°C (4, 5, 50). A variety of experiments have demonstrated that the influenza virus HA associates preferentially with rafts via the HA transmembrane domain (49) and that influenza virus buds from and is enriched in raft domains (48). In contrast, although SFV is dependent on cellular cholesterol for efficient budding, the SFV spike protein is not associated with detergent-resistant rafts and these domains are not enriched in the SFV particle (48; Lu and Kielian, unpublished data). Thus, the reason for the alphavirus cholesterol requirement in budding does not appear to be due to a requirement for raft domains, and a suggestion for the role of cholesterol may come from studies of the alphavirus srf mutants.

To date, all of the SFV and SIN mutants that display increased fusion and infection of cholesterol-depleted cells also show increased exit from depleted cells (32, 34, 62; P. K. Chatterjee and M. Kielian, unpublished data). These data suggest a connection between the sterol dependence of membrane fusion and that of virus exit. The SFV E1 protein undergoes several distinct conformational changes during the low pH-triggered fusion reaction, and the kinetics and efficiency of these changes are increased by the presence of cholesterol in the target membrane (6, 8). Studies of srf-3 have shown that the cholesterol independence of its membrane fusion reaction is controlled by the increased cholesterol independence of these E1 conformational changes (6). Thus, the presence of cholesterol in the target membrane during fusion functionally affects the conformation of the virus E1 subunit. It is possible that during exit cholesterol promotes the formation of a budding-competent conformation of the SFV spike protein and that the srf-3 mutant is less cholesterol dependent for production of such active spike proteins. This budding-competent, cholesterol-enhanced conformation of the spike protein could represent the spike protein trimer or higher-order, laterally interacting spike protein oligomers. The formation of such lateral interactions would be expected to be critical for the productive association of the spike with the nucleocapsid during budding. The rapid degradation of the wt and srf-3 spike proteins at early times of infection, or of wt spike protein in cholesterol-depleted cells, could be a secondary effect produced by the lack of such lateral interactions. In this scenario, decreased lateral interactions would be caused by either the low level of spike protein expression at the plasma membrane early in infection or by the absence of cholesterol in the case of the wt spike protein. Alternatively, it is also possible that the formation of lateral interactions is directly controlled by the degradation rate of the spike protein via its regulation of the amount of spikes at the plasma membrane.

Our studies on virus budding also offer insights into cellular membrane budding processes, including the role of lipids in the budding reaction. Vesicular trafficking mediates the transport and sorting of cargo and membrane proteins during endocytosis and exocytosis in eukaryotic cells. A key step in this traffic is the formation of vesicles through the process of vesicle budding (see reference 54 for review). Although our understanding of vesicle budding is incomplete, formation of exocytic transport vesicles has been shown to involve the formation of a complex containing coat components, priming membrane proteins, and small GTPases. Polymerization of the vesicle coat appears to lead to membrane deformation and release of a budded vesicle. Studies to date also suggest that specific lipids carry out important functions during the formation of membrane vesicles (9, 35). For example, the formation of clathrin-coated endocytic vesicles has been shown to be inhibited by depletion of cholesterol from the plasma membrane, the site of this budding reaction (39, 58). A recent study reported that the membrane protein synaptophysin specifically interacts with cholesterol and suggested that the generation of neurosecretory vesicles may be driven by lateral associations of cholesterol and synaptophysin in the forming membrane bud (60). Thus, a precedent for a role of cholesterol-protein interactions in the formation of highly curved membrane structures is found in the budding reactions of several types of cellular vesicles. As our mechanistic knowledge of these processes evolves, it will be interesting to follow the similarities and differences between the functions of cholesterol in alphavirus budding and in cellular membrane budding reactions.

Taken together, our studies have established an assay for the budding of SFV spike proteins from the cell surface into virus particles. We have also demonstrated that this assay can be adapted to other alphaviruses, such as SIN, as well as to other viruses, such as the rhabdovirus vesicular stomatitis virus (Lu and Kielian, unpublished data). Our results localized the inhibition of SFV exit in cholesterol-depleted cells to the virus budding step and demonstrated that a relatively cholesterol-independent SFV mutant is more permissive for budding from sterol-depleted cells. A striking correlation of efficient budding with decreased spike protein turnover was observed. Future experiments will address spike protein endocytic traffic and the mechanism of rapid spike protein turnover and their importance in the regulation of virus budding. In addition, the sustained budding of biotin-tagged spike proteins from SFV-infected cells and its relative insensitivity to continued cellular secretion suggest the possibility of adapting this assay for use in a semipermeabilized cell system. In vitro systems based on semipermeabilized cells have been used to reconstitute and characterize many cellular reactions involving the budding and fusion of vesicular carriers (reviewed in reference 41). Such an assay would be invaluable for the experimental manipulation and study of virus budding and has not yet been reported for any enveloped animal virus.