JOURNAL OF VIROLOGY, June 2011, p. 5644–5650 0022-538X/11/$12.00 doi:10.1128/JVI.00130-11 Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Vol. 85, No. 11
Activation of the Alphavirus Spike Protein Is Suppressed by Bound E3䌤 Mathilda Sjo ¨berg, Birgitta Lindqvist, and Henrik Garoff* Department of Biosciences and Nutrition, Karolinska Institute, S-141 57 Huddinge, Sweden Received 19 January 2011/Accepted 11 March 2011
Alphaviruses are taken up into the endosome of the cell, where acidic conditions activate the spikes for membrane fusion. This involves dissociation of the three E2-E1 heterodimers of the spike and E1 interaction with the target membrane as a homotrimer. The biosynthesis of the heterodimer as a pH-resistant p62-E1 precursor appeared to solve the problem of premature activation in the late and acidic parts of the biosynthetic transport pathway in the cell. However, p62 cleavage into E2 and E3 by furin occurs before the spike has left the acidic compartments, accentuating the problem. In this work, we used a furin-resistant Semliki Forest virus (SFV) mutant, SFVSQL, to study the role of E3 in spike activation. The cleavage was reconstituted with proteinase K in vitro using free virus or spikes on SFVSQL-infected cells. We found that E3 association with the spikes was pH dependent, requiring acidic conditions, and that the bound E3 suppressed spike activation. This was shown in an in vitro spike activation assay monitoring E1 trimer formation with liposomes and a fusion-from-within assay with infected cells. Furthermore, the wild type, SFVwt, was found to bind significant amounts of E3, especially if produced in dense cultures, which lowered the pH of the culture medium. This E3 also suppressed spike activation. The results suggest that furin-cleaved E3 continues to protect the spike from premature activation in acidic compartments of the cell and that its release in the neutral extracellular space primes the spike for low-pH activation. A similar problem for the maturation of the flavivirus spike has recently found an interesting solution. As in alphavirus, the flavivirus fusion subunit (E) is chaperoned from premature acid activation by a second transmembrane subunit (prM) (2, 26, 37, 43). Although prM cleavage by furin occurs in an acid compartment, E activation is prevented by the retention of the cleaved pr piece on E. Not until the virus is released into the neutral extracellular space is the pr removed and the virus primed for low-pH-triggered fusion (40, 41). Recently, these studies, which were done using virus particles, were supported by analyses with isolated pr and E fragments (44). In the present study, we analyzed whether the E3 protein of alphavirus plays a similar regulatory role. We used a Semliki Forest virus (SFV) mutant, SQL, which was defective in furin cleavage of p62 but could be cleaved by chymotrypsin in vitro, rendering SFVSQL infectious (4).
The alphavirus spike is a trimer of the E1-E2 heterodimer (6, 19, 42, 46). Both subunits are transmembrane glycoproteins (12). E1 carries the membrane fusion function of the virus, while E2 binds the virus to a still ill-defined receptor structure(s) on the cell surface (16, 17, 32). E2 also controls the fusion function of E1 so that it does not occur before the virus has entered the endosome (23, 38). There, the acidic pH dissociates the heterodimer and allows the E1 to interact with the endosomal membrane via its fusion loop (16, 33). This results in E1 homotrimerization and subsequent jack knife like back folding, which brings the viral and the endosomal membranes together for fusion (16, 34). The oligomerization of the spike subunits into the heterodimer and the subsequent trimerization of the heterodimers into spikes take place in the rough endoplasmic reticulum of the infected cell (20, 25, 45). The spikes are then transported to the cell surface via the Golgi complex and apparently also the early endosome (3, 28). At the cell surface, the spikes interact with the viral nucleocapsid and with each other, driving budding of virus particles with T ⫽ 4 icosahedral symmetry (6, 10, 31). The fact that the heterodimer is made as an acid-resistant E1-p62 precursor appears to be an elegant solution to avoid premature activation when passing the acidic conditions of the Golgi complex and the early endosome (8, 9, 11, 33). However, the cleavage of the p62 subunit into E2 and the small peripheral E3 protein by cellular furin occurs before the precursor form of the heterodimer has left these compartments, thereby exposing the mature heterodimer to activating conditions (7, 28). Thus, the question of how the heterodimer avoids premature activation in the producer cell remains.
MATERIALS AND METHODS Cell culture. BHK-21 cells (CCL-10; American Type Culture Collection, Manassas, VA) were maintained in BHK-21 medium, i.e., Glasgow minimal essential medium (GMEM) supplemented with 5% fetal calf serum (FCS), 20 mM HEPES, pH 7.3, 10% tryptose phosphate broth, and 2 mM Glutamax (all cell culture media were from Gibco BRL, Gaithersburg, MD). Sparse, almost confluent, and dense cell cultures were prepared by plating a series of increasing cell numbers, ranging from ⬃2 ⫻ 105 to 1.5 ⫻ 106 cells, in 25-cm2 bottles. Cultures of the desired density were used the day after they were plated. Virus infection. BHK-21 cells were infected with at least 10 PFU per cell of wild-type SFV (SFVwt) or activation-primed SFVSQL, as described previously (30). To prime SFVSQL for low-pH activation, an aliquot was incubated with 0.1 mg/ml chymotrypsin for 2 h on ice, and then 0.8 mg/ml aprotinin was added. Virus was bound to the cells for 1 h at 37°C. Then, the cultures were washed with phosphate-buffered saline (PBS) (containing Ca2⫹ and Mg2⫹) and incubated in BHK-21 medium for 3 or 3.5 h at 37°C. Metabolic labeling and isolation of radioactive virus. At 4.5 h postinfection, the cells were starved for Cys by incubation in MEM with low Cys content (25 M) for 30 min. Then, [35S]Cys (0.1 mCi/ml) in the same medium was added, and the cells were incubated for 16 h. The medium was collected and clarified by low-speed centrifugation, and the virus was isolated by centrifugation in a step
* Corresponding author. Mailing address: Department of Biosciences and Nutrition, Karolinska Institute, S-141 57 Huddinge, Sweden. Phone: 46-8-52481032. Fax: 46-8-7745538. E-mail:
[email protected]. 䌤 Published ahead of print on 23 March 2011. 5644
VOL. 85, 2011
FUSION ACTIVATION IN ALPHAVIRUS
gradient consisting of 8 ml 20% (wt/wt) sucrose and 1 ml 60% (wt/wt) sucrose in a buffer containing 50 mM HEPES, 100 mM NaCl, 1.8 mM CaCl2, pH 7.4 (HNC or pH 7.4 buffer), in a Beckman SW28.1 rotor (1.75 h at 9.3 ⫻ 104 ⫻ g, i.e., 22 krpm, and 4°C). The virus was collected from the 20%-60% sucrose interphase. In some cases, labeling was done for 90 min and the cells were chased for various times in MEM containing 2 mM Cys and 0.2% bovine serum albumin (BSA). After the pH of the medium was changed to 6.1 with a pretitrated volume of 50 mM MES (morpholineethanesulfonic acid), pH 3.7, and phenylmethylsulfonyl fluoride (PMSF) was added to 0.2 mM, the virus was isolated by centrifugation through 20% (wt/wt) sucrose in a buffer containing 25 mM MES, 25 mM HEPES, 100 mM NaCl, and 1.8 mM CaCl2, pH 6.1 (HNC-MES), in 5- by 41-mm tubes in a Beckman SW55 rotor (45 min; 1.1 ⫻ 105 ⫻ g, i.e., 36 krpm, and 4°C). For surface expression of labeled viral spike proteins, the infected cells were starved for Cys for 30 min at 4 h postinfection, pulse-labeled with [35S]Cys for 30 min, washed once with PBS, and then chased for 20 min in MEM containing an excess of Cys (2 mM). P62 cleavage in vitro and analysis of E3 binding to the virus. P62 of [35S]Cyslabeled SFVSQL was cleaved into E2 and E3 using 60 or 90 g/ml proteinase K (PK) for 20 min at 20°C in HNC, pH 7.4, or in HNC-MES, pH 6.1. Then, 10 mM PMSF was added, and the virus was reisolated in a 20% to 60% sucrose step gradient as described above, but using either HNC (pH 7.4) or HNC-MES (pH 6.1) as the gradient buffer and 5- by 41-mm tubes in a Beckman SW55 rotor (45 min; 1.1 ⫻ 105 ⫻ g, i.e., 36 krpm, and 4°C). The degree of p62 cleavage and the amount of virus-bound E3 was followed by nonreducing SDS-PAGE (13% acrylamide) and phosphorimaging. For cleavage of p62 on the cell surface, SFVSQL-infected cells were pulselabeled and chased (see above), washed with HNC-MES (pH 6.1 buffer), and then treated with 5 to 50 g/ml PK in pH 6.1 buffer for 10 min at 20°C. PMSF (10 mM) was added, and after 2 min, the cells were washed once with pH 6.1 buffer to remove the PK. Finally, the cells were washed twice with 500 l of either pH 6.1 or 7.4 buffer containing 10 M PMSF. The respective washes were combined and analyzed for E3 content by immunoprecipitation and subsequent nonreducing SDS-PAGE. Spike activation. Samples (5 l) of [35S]Cys-labeled SFVSQL or SFVwt were mixed with 30 l large unilamellar vesicles, prepared as described previously (34) but using a miniextruder (Avanti Polar Lipids, Alabaster, AL) and pH 6.1 buffer, and incubated for 5 min at 20°C. Activation was triggered by adjusting the pH to 5.7 or 4.5 using pretitrated aliquots of 0.1 M or 0.3 M sodium acetate, pH 3.8. After 10 min at 20°C, the process was terminated by neutralizing the samples with NaOH. The degree of spike activation was monitored by following E1 trimerization in nonreducing SDS-PAGE (34). Fusion from within. Spikes on the surfaces of SFVwt-infected cells or on SFVSQL-infected cells, which had or had not been primed for low-pH activation by PK cleavage of p62 (see above), were activated for membrane fusion by washing them at 20°C with MEM without carbonate containing 20 mM sodium succinate and adjusted to either pH 5.5 or pH 4.5. The cells were returned to BHK-21 medium and incubated for 2 h at 37°C to allow the fused cells to develop into polykaryons. The cells were stained with Giemsa stain. Immunoprecipitation and SDS-PAGE. Immunoprecipitation of E3 from pH 7.4 or 6.1 buffer washes of PK-treated cells infected with SFVSQL and labeled with [35S]Cys was done essentially as described previously (29) using a rabbit anti-E3 antiserum (␣E3; kindly provided by Leevi Ka¨¨aria¨inen, Helsinki, Finland). The precipitates were dissolved in SDS-PAGE sample buffer as described previously (18), except that the buffer was titrated to pH 8.0 and 0.8 mM methionine was included. Then, 1/10 volume of BHK cell lysate (10) was added and the sample was heated to 70°C for 3 min and subjected to nonreducing 13% SDS-PAGE. Virus and cell samples were analyzed by SDS-PAGE as the immunoprecipitates, but using a two-times-concentrated sample buffer. For E1 trimer analysis on gels, the samples were warmed either to 37°C for 10 min to preserve the trimers or for 3 min at 95°C to disrupt them. All gels were dried and exposed to phosphorimager screens (BAS-MS2025; Fujifilm; Science Imaging Scandinavia, Nacka, Sweden), and the labeled proteins were visualized and quantitated using a Molecular Imager FX and the QuantityOne program from Bio-Rad Laboratories (Hercules, CA).
RESULTS We first analyzed the pH dependence of the binding of cleaved E3 to SFVSQL. To this end, we produced [35S]Cyslabeled SFVSQL in BHK-21 cells and purified it by centrifugation in a sucrose step gradient, pH 7.4. The p62 was cleaved
5645
FIG. 1. pH-dependent binding of E3 to SFVSQL. 35S-labeled SFVSQL was cleaved with 60 or 90 g/ml PK in pH 7.4 or 6.1 buffer and then reisolated by centrifugation in a sucrose step gradient in either pH 7.4 or 6.1 buffer. The viral proteins were analyzed by nonreducing 13% SDS-PAGE. The conditions for PK treatment and virus isolation are indicated at the top, the viral proteins and MW standards on the sides, and the quantification of retained E3 as a percentage of E2 with SD (n ⫽ 3) at bottom. Shown is a phosphorimage of a dried gel. Note the retention of E3 when both the PK treatment and virus isolation were done in pH 6.1 buffer. NA, not applicable.
into E2 and E3 using 60 or 90 g/ml PK for 20 min at 20°C in HNC buffer, pH 7.4, or HNC-MES buffer, pH 6.1. PK was used instead of chymotrypsin because the latter became less specific at pH 6.1. After cleavage, the protease inhibitor PMSF was added, and the virus was reisolated by centrifugation in a 20% to 60% sucrose step gradient in HNC, pH 7.4, or HNC-MES, pH 6.1. The virus at the sucrose interphase was collected and analyzed by nonreducing 13% SDS-PAGE. Analysis of untreated SFVSQL showed the immature spike subunits p62 and E1 and the nucleocapsid protein C, but no mature E2 and E3 (Fig. 1, lane 7). Protease cleavage generated the mature subunits (Fig. 1, lanes 2 to 6), but only when both cleavage and virus isolation were done in acidic buffer did E3 remain bound to the virus (about 50% relative to E2) (Fig. 1, lanes 3 and 4). The E3, which has one heterogeneous complex sugar unit, was seen as a ladder of bands with masses around 17 kDa, as shown previously (15, 36). The cleavage was complete at pH 7.4 (Fig. 1, lane 2), but not at 6.1 (Fig. 1, lanes 3 to 6). We also analyzed SFVwt, which had been reisolated in the pH 7.4 sucrose step gradient by SDS-PAGE. This twice-isolated virus contained little detectable E3 (Fig. 1, lane 1), but we will return to the question of E3 in newly released virus below. We concluded that E3 remains bound to the virus spike under acidic, but not under neutral, conditions. We then studied whether the spike-bound E3 inhibited spike activation in vitro. To follow activation, we monitored E1 trimerization upon fusion with liposomes at low pH (5). [35S]Cyslabeled SFVSQL was treated with 90 g/ml PK to cleave p62 into E3 and E2 at pH 6.1, and the virus was reisolated in the
5646
¨ BERG ET AL. SJO
FIG. 2. E3 suppresses spike activation. 35S-labeled SFVSQL was PK treated at pH 6.1 and reisolated in the E3-releasing pH 7.4 or E3retaining pH 6.1 gradient. Uncleaved SFVSQL and SFVwt were reisolated in the pH 7.4 gradient and used as controls. All preparations were tested for activation by incubation with liposomes at pH 5.7 or 4.5. Spike activation was monitored by E1 trimer formation. This was revealed by nonreducing 13% SDS-PAGE of samples warmed for 10 min at 37°C and compared to samples heated for 3 min at 95°C, which dissociated the trimers into E1 monomers. (A) Phosphorimage of a representative gel analysis with sample treatment indicated at the top. Grad., gradient. (B) Quantification of E1 in trimeric form as a percentage of total E1 after activation with liposomes at pH 5.7 or 4.5 (n ⫽ 3). Note that PK-cleaved SFVSQL that had been reisolated under E3-releasing conditions did support E1 trimer formation with liposomes at pH 5.7 (lane 7), but not when reisolated under E3-retaining conditions (lane 5). The error bars indicate SD.
E3-retaining pH 6.1 gradient or the E3-releasing pH 7.4 gradient (see above). As controls, we used [35S]Cys-labeled SFVwt and uncleaved [35S]Cys-labeled SFVSQL, both reisolated in the pH 7.4 gradient. The virus preparations were mixed with liposomes in pH 6.1 buffer, and the mixtures were incubated for 5 min at 20°C. Activation was triggered by adjusting the pH to 5.7 or 4.5 by addition of pretitrated aliquots of 0.1 and 0.3 M sodium acetate, respectively. After 10 min at 20°C, the samples were neutralized with NaOH. They were then mixed with an SDS gel sample buffer containing N-ethylmaleimide (NEM) and BHK-21 cell lysate and incubated either at 37°C for 10 min to maintain the E1 trimers or at 95°C for 3 min to dissociate the trimers (5, 35). Nonreducing 13% SDS-PAGE analyses showed that the SFVwt formed E1 trimers as expected, at both pH 5.7 and 4.5 (Fig. 2A, lanes 1, 2, 9, and 10), whereas SFVSQL required the lower pH (Fig. 2A, lanes 3, 4, 11, and 12) (27). Quantification of E1 trimers (expressed as a percentage of total E1) is shown in Fig. 2B. Importantly, the cleaved SFVSQL with retained E3 formed significant amounts of E1 trimers only at the lower pH, like the uncleaved SFVSQL (Fig. 2A, lanes 5, 6, 13, and 14, and B), whereas the cleaved SFVSQL with re-
J. VIROL.
leased E3 was significantly activated at both pH values, like SFVwt (Fig. 2A, lanes 7, 8, 15, and 16, and B). Thus, we conclude that the bound E3 suppresses acid activation of the spike. We intended to study SQL virus-induced fusion from without with BHK-21 cells. Here, the spikes of cell-bound virus are used for fusion of uninfected cells. Unfortunately, the virus binding was insufficient for the assay. Therefore, we turned to the fusion-from-within assay, in which spikes on the surfaces of infected cells are used for fusion. First, we tested PK cleavage of SQL spikes on the cell surface and the possible pH dependence of E3 release. BHK-21 cells in six-well plates were infected with SFVwt or chymotrypsin-activated SFVSQL for 1 h and then incubated for 3.5 h before being depleted with Cys for 30 min, pulse-labeled with [35S]Cys for 30 min, and finally chased for 20 min. The cells were rinsed once with pH 6.1 buffer and then treated with 5 to 50 g/ml PK in pH 6.1 buffer for 10 min at 20°C. PMSF was added for 2 min, and the cells were washed once with pH 6.1 buffer to remove the PK. After this, the cells were washed twice with 500 l pH 6.1 or 7.4 buffer containing PMSF. Washes at the same pH were pooled and analyzed for E3 content using a rabbit anti-E3 antiserum, ␣E3, and subsequent nonreducing SDS-PAGE. The cells were lysed and analyzed directly. Analyses of the lysates showed that PK treatment generated some E2 that comigrated with E2 of SFVwt-infected cells, although most of the p62 was apparently degraded (Fig. 3A). Significantly, analyses of the immunoprecipitates showed that E3 was released from the PK-treated cells washed with the pH 7.4 buffer, but not those washed with the pH 6.1 buffer (Fig. 3B). This suggested that the PK treatment did generate some mature spikes on the surfaces of the SFVSQL-infected cells and that the cleaved E3 was released in a pH-dependent way. For the fusion-from-within assay, several cultures of BHK-21 cells were infected with SFVwt or SFVSQL. One set of SFVwt and SFVSQL cultures was washed twice with pH 7.4, 5.5, or 4.5 buffer at 20°C and then incubated with BHK-21 medium for 2 h at 37°C. As expected, there was no fusion in the cultures washed with pH 7.4 buffer (Fig. 4A and D). The pH 5.5 buffer wash induced fusion in the SFVwt-infected cells, but not in the SFVSQL-infected cells (Fig. 4B and E). The fusion was seen as extensive polykaryon formation in the Giemsa-stained cultures. The SFVSQL-infected cells required the pH 4.5 buffer wash for fusion (Fig. 4F). This agrees with earlier findings (27). A second set of SFVSQL infected cells was treated with 5 to 50 g/ml of PK in pH 6 buffer for 10 min at 20°C, which should cleave the p62 of spikes at the cell surface into E2 and E3. PMSF was added, and the cultures were washed once with pH 5.5 buffer. This should not dissociate E3, and therefore, triggering of the spike should be inhibited at this pH. Analysis for polykaryons after further incubation with BHK-21 medium for 2 h at 37°C was completely negative, thus confirming the expectations (Fig. 4G to I). A third set of SFVSQL-infected cultures was treated like the second one, but a pH 7.4 wash was included between the PK treatment and the pH 5.5 wash to release E3 and thus allow triggering of the spikes. In this case, extensive fusion was observed, suggesting that the pH 7.4 buffer wash did release the PK-cleaved E3 from the spikes and that this rendered them sensitive to pH 5.5 triggering, like SFVwt (Fig. 4J to L). Thus, these results support a model
VOL. 85, 2011
FIG. 3. PK-cleaved E3 is released from SFVSQL-infected cells in a pH-dependent way. BHK-21 cell cultures were infected with SFVSQL and SFVwt. The SFVSQL-infected cells were treated with 5 to 50 g/ml PK in pH 6.1 buffer for 10 min at 20°C. After a pH 6.1 buffer wash to remove the PK, the cultures were washed twice with either pH 7.4 or 6.1 buffer. (B) The SFVwt-infected cells were directly washed twice with the pH 7.4 or 6.1 buffer. Washes at the same pH were pooled and analyzed for E3 by ␣E3 immunoprecipitation and subsequent nonreducing SDS-PAGE. (A) The cells were lysed with Triton X-100 and analyzed directly. Sample treatments are indicated at the top and the molecular weight markers and the viral proteins on the sides of each panel, which show phosphorimages of the gels. Note the formation of some E2 by PK cleavage of p62 and the release of E3 from cells washed with nearly neutral, but not acidic, buffer.
FUSION ACTIVATION IN ALPHAVIRUS
5647
fluent, almost confluent, and dense BHK-21 cell cultures were infected, and after 4 h, they were labeled with [35S]Cys for 20 h before the media were collected and the pH was measured. The pH values had decreased from about 7.2 to 6.5, 6.0, and 5.7, respectively, in the three cultures. Each medium was divided in two, and the virus was purified in the E3-retaining pH 6.1 or the E3-releasing pH 7.4 step gradient. SDS-PAGE of virus purified under E3-retaining conditions showed that the virus produced in the denser cultures, with more acidic medium, contained much more E3 than that produced in the less dense culture and hence less acidic medium (Fig. 5A, lanes 4 to 6). The bound E3 was partly removed if the virus was purified in the pH 7.4 step gradient (Fig. 5A, lanes 1 to 3). Quantification of bound E3 (average ⫾ standard deviation [SD]), expressed as a percentage of E2, is given for three experiments in Fig. 5B. The virus preparations were then analyzed for their potential to become activated and form E1 trimers with liposomes at pH 5.7. We found that significant activation occurred with the virus produced in the sparse culture and purified in the pH 6.1 gradient, whereas much less activation was possible with the virus from the denser cultures, with more bound E3 (Fig. 5C, lanes 7 to 12). All preparations were possible to activate if a fraction of the bound E3 was first removed by virus isolation in the pH 7.4 step gradient (Fig. 5C, lanes 1 to 6). Quantification of E1 trimers as percentages of total E1 are shown in Fig. 5D (n ⫽ 3). We concluded that SFVwt produced in a dense culture retains more E3 than virus made in a sparse culture because of a more acidic pH and that the bound E3 suppresses spike activation. We were still interested in knowing how much E3 remained bound to newly formed SFVwt under neutral conditions and how fast it was released. To measure E3 in newly made virus, we pulse-labeled infected cells with [35S]Cys for 90 min and chased it for 4 min. The medium was collected, the pH was changed to 6.1, and the virus was isolated by sedimentation through 20% (wt/wt) sucrose, pH 6.1. SDS-PAGE showed that the amount of E3 was about 30% of the E2 in that virus preparation. The E3 release was measured using a [35S]Cyslabeled virus preparation produced by a 16-h labeling of infected cells followed by isolation in the pH 6.1 step gradient. The E3 content was 32% of the E2 in that virus preparation. Aliquots of the virus (in triplicate) were incubated at 37°C under neutral conditions for 0, 0.5, 1.5, and 3 h. Then, the virus was reisolated by sedimentation through 20% (wt/wt) sucrose, pH 6.1, and analyzed by SDS-PAGE (Fig. 6). This demonstrated that the E3 was gradually released by incubation at 37°C under neutral conditions. Quantification showed that the original E3 content decreased with a half-life (t1/2) of 0.5 h during the incubation. DISCUSSION
according to which a pH-dependent E3 association with the spike controls its potentiation for low-pH triggering in the infected cell. SFVwt is known to contain bound E3, though the exact amounts may vary. This should suppress spike activation. We reasoned that E3 binding might be correlated with the acidity of the culture medium during production. Therefore, we investigated SFVwt produced in cells of different densities. This should influence the pH of the culture medium. Thus, subcon-
The E3 protein has multiple functions during SFV replication. After capsid cleavage by autoproteolysis, the E3 part of the p62-E1 membrane polyprotein redirects the ribosomal translation of the viral structural 26S mRNA from the cytoplasm to the endoplasmic reticulum (ER) for synthesis of the spike (1, 14, 24). For this purpose, E3 carries a signal sequence in its N terminus (13). This has several unusual properties. First, it is not cleaved by signal peptidase. Second, it becomes glycosylated by the addition of an Asn-linked sugar unit. This
5648
¨ BERG ET AL. SJO
J. VIROL.
FIG. 4. E3 association with the spike suppresses spike-mediated cell-cell fusion. (A to F) One set of SFVwt- and SFVSQL-infected BHK-21 cells was washed with pH 7.4, 5.5, or 4.5 buffer and then further incubated in pH 7.4 medium for 2 h. After this, the cultures were stained with Giemsa stain. Note the fusion of pH 5.5 buffer-washed SFVwt-infected cells, but not SFVSQL-infected cells, which required a pH 4.5 buffer wash. In unfused cultures, the cells are seen as separate entities, whereas the fused cells show nuclei in a continuous cytoplasm. This difference is highlighted in the insets at higher magnification. (G to I) A second set of SFVSQL-infected cells was treated with 5 to 50 g/ml PK at pH 6.1 before being washed with pH 5.5 buffer. This treatment was expected to generate E3 by p62 cleavage but not to release E3 from the spike. Note the absence of fusion. (J to L) A third set of SFVSQL-infected cells was treated with 5 to 50 g/ml PK at pH 6.1, but before the pH 5.5 buffer wash, a pH 7.4 buffer wash was included. This was expected to release E3. Note the extensive fusion in all samples. The sample treatments are indicated at the top of each panel.
facilitates its cotranslational removal from the ER membrane and subsequent participation of E3 in the formation of the spike ectodomain in the ER lumen. Spike formation involves heterodimerization of the transmembrane spike subunits p62 and E1 and further trimerization of the heterodimers (25, 46). E3 appears to initiate the p62-E1 heterodimerization reaction (22). If E3 is swapped with a heterologous cleavable signal sequence, then heterodimerization between E2 and E1 is prevented. In the heterodimer, E3 increases the strength of the subunit association so that it resists the low pH of the late parts of the secretory pathway (21, 27, 33). This prevents premature activation of the spike. However, furin-mediated cleavage of
p62 into the mature E2 and E3 occurs before the spike has left the low-pH compartments, leaving it seemingly unprotected (7, 28). In the present study, we showed that E3, which has been cleaved in vitro with PK in the cleavage-deficient mutant SFVSQL, remains associated with the virus under acidic, but not under neutral, conditions and that the bound E3 suppressed low-pH-triggered spike activation. These results suggest that furin-cleaved E3 continues to protect the spike from premature activation in the late and acidic parts of the secretory pathway of the infected cell. Dissociation of E3 under neutral conditions outside the cell would thus prime the spikes of the virus for activation by low pH.
VOL. 85, 2011
FUSION ACTIVATION IN ALPHAVIRUS
5649
FIG. 6. Release of E3 from SFVwt. [35S]Cys-labeled SFVwt was isolated in the pH 6.1 gradient and then incubated under neutral conditions at 37°C for 0, 0.5, 1.5, or 3 h. The pH of each sample was readjusted to 6.1, and the virus was reisolated by centrifugation through 20% (wt/wt) sucrose for analysis by SDS-PAGE. Shown is a phosphorimage of the gel. The incubation times of the virus samples are indicated at the top, and the remaining E3 (the mean of three quantifications) is given at the bottom as a percentage of the E2 in the virus.
FIG. 5. E3 binding to SFVwt. A sparse, an almost confluent, and a dense BHK-21 cell culture were infected with SFVwt and incubated for 4 h before being labeled with [35S]Cys for 20 h. After this, the media with the released virus were collected. The pH values of the media at the time of harvest were 6.4, 6, and 5.7, respectively, in the originally sparse, almost confluent, and dense cultures. (A) Each medium was divided in two, and the virus was isolated in a sucrose step gradient with a pH of either 7.4 or 6.1 and analyzed by nonreducing SDSPAGE. (B) Quantification of E3, expressed as a percentage of E2 ⫾ SD, in virus from different preparations purified in either pH 7.4 or 6.1 gradients. The quantifications represent the means of three experiments. The ranges of pH values of the media at the time of virus harvest for the different preparations are indicated at the bottom of each panel. Note the increased binding of E3 to virus produced in the denser cultures with more acidic medium. (C) The virus preparations were mixed with liposomes and activated in pH 5.7 buffer for 10 min at 20°C and then, after neutralization, analyzed by nonreducing SDSPAGE for E1 trimerization. For the PAGE assay, samples were heated to 37°C for 10 min to maintain E1 trimers or to 95°C for 3 min to dissociate the trimers. The activation of spikes in virus purified in the pH 6.1 gradients is shown in lanes 7 to 12. Note the presence of much more E1 trimer in activated virus from the sparse culture than in virus from the denser cultures. The activation of the virus preparations purified in pH 7.4 step gradients, with less bound E3, is shown in lanes 1 to 6. Note the significant formation of E1 trimers in all these preparations after activation. (D) Quantification of E1 trimers (⫾SD) in the different preparations after low-pH activation in the presence of liposomes (n ⫽ 3).
Recently, the atomic structure of the p62-E1 ectodomain of chikungunya virus (CHIKV), an alphavirus of the same clade as SFV, was determined (39). This revealed the structure of the p62 subunit, which was previously unknown, and showed how it interacted with E1. Interestingly, the E3 part of p62 did not interact with E1 at all, but only with the E2 part of p62. The binding stabilized a region of p62 called the “acid-sensitive region,” which with its opposite face interacted with the fusion loop carrying domain II of E1. The latter interaction, which included a critical His 170, was shown to be dissociated under acidic conditions to facilitate E1 activation for membrane fusion, using Sindbis virus (SINV), a member of another clade of alphaviruses (20). In the crystal structure, the intact furinsensitive loop of the CHIKV p62 appeared to be important for this stabilizing effect (39). It was proposed that furin cleavage switched the heterodimer from an acid-resistant oligomer to a sensitive one. However, our present results with SFV suggest that the switch in pH sensitivity of the spike is not directly controlled by the cleavage but by the pH-sensitive removal of E3, as in the case of the pr protein control of flavivirus spike activation (40, 41). Analyses of SFVwt preparations in general show that they retain significant amounts of E3 (15, 36). Indeed, screening various preparations made in our laboratory using centrifugation in a single sucrose step gradient at pH 7.4, showed that the E3 content was 36% ⫾ 15% (n ⫽ 17) of that of E2. One reason for the large variation in E3 content that we found is the culture conditions of the SFVwt-infected cells. When sparse cell cultures were infected and virus was produced for 20 h, the pH of the medium remained closer to neutral than when dense cultures were used, and this correlated with virus-bound E3, i.e., virus produced at lower pH contained more E3. The amounts of E3 corresponding to 70% of the E2 was found when virus was harvested in media with pH values of 6 to 6.2. The E3 content of newly released SFVwt produced under neutral conditions was about 30% of that of the viral E2. The
5650
¨ BERG ET AL. SJO
bound E3 suppressed spike activation and was released at 37°C and under neutral conditions with a t1/2 of 0.5 h. These results suggested that the E3 produced by p62 cleavage in the SFVwtinfected cells also remains spike associated and that it protects E1 until it is gradually released in the neutral milieu of the extracellular space. SFV has apparently evolved to retain a significant fraction of spike-bound E3 in released virus for a limited time. This should protect the virus from premature activation but at the same time compromises infection of new cells. However, these effects of the bound E3 are probably diminished by the fact that only a subset of the viral spikes has to be activated to reach full infectivity. Indeed, only about half of the spikes of the furin cleavage-defective Semliki Forest Virus (SFV) mutants SFVSQL and SFVL had to be processed in vitro for maximal infectivity (4, 27). In the case of SINV, the E3 has been reported to be efficiently released (36). A model for the SINV E3 structure has been made, and it suggests a weaker interaction with E2 than SFV E3 (39). It is likely that the E3-E2 interaction represents an important regulator of alphavirus infectivity. Its strength has probably evolved to fit the conditions of natural infection in the host of each alphavirus. Thus, the interaction of E3 with the spikes in the different alphaviruses appears to be an important field of research, which also should have much potential for antiviral drug development. ACKNOWLEDGMENTS Swedish Science Foundation grant 2778 and Swedish Cancer Foundation grant 0525 to H.G. and the EU FP7-People-ITN-2008 Marie Curie actions Project Virus Entry 235649 supported this study. REFERENCES 1. Aliperti, G., and M. J. Schlesinger. 1978. Evidence for an autoprotease activity of Sindbis virus capsid protein. Virology 90:366–369. 2. Allison, S. L., et al. 1995. Oligomeric rearrangement of tick-borne encephalitis virus envelope proteins induced by an acidic pH. J. Virol. 69:695–700. 3. Band, A. M., J. Maatta, L. Kaariainen, and E. Kuismanen. 2001. Inhibition of the membrane fusion machinery prevents exit from the TGN and proteolytic processing by furin. FEBS Lett. 505:118–124. 4. Berglund, P., et al. 1993. Semliki Forest virus expression system: production of conditionally infectious recombinant particles. Biotechnology 11:916–920. 5. Bron, R., J. M. Wahlberg, H. Garoff, and J. Wilschut. 1993. Membrane fusion of Semliki Forest virus in a model system: correlation between fusion kinetics and structural changes in the envelope glycoprotein. EMBO J. 12:693–701. 6. Cheng, R. H., et al. 1995. Nucleocapsid and glycoprotein organization in an enveloped virus. Cell 80:621–630. 7. de Curtis, I., and K. Simons. 1988. Dissection of Semliki Forest virus glycoprotein delivery from the trans-Golgi network to the cell surface in permeabilized BHK cells. Proc. Natl. Acad. Sci. U. S. A. 85:8052–8056. 8. Demaurex, N. 2002. pH homeostasis of cellular organelles. News Physiol. Sci. 17:1–5. 9. Demaurex, N., W. Furuya, S. D’Souza, J. S. Bonifacino, and S. Grinstein. 1998. Mechanism of acidification of the trans-Golgi network (TGN). In situ measurements of pH using retrieval of TGN38 and furin from the cell surface. J. Biol. Chem. 273:2044–2051. 10. Forsell, K., L. Xing, T. Kozlovska, R. H. Cheng, and H. Garoff. 2000. Membrane proteins organize a symmetrical virus. EMBO J. 19:5081–5091. 11. Gagescu, R., et al. 2000. The recycling endosome of Madin-Darby canine kidney cells is a mildly acidic compartment rich in raft components. Mol. Biol. Cell 11:2775–2791. 12. Garoff, H., A. M. Frischauf, K. Simons, H. Lehrach, and H. Delius. 1980. Nucleotide sequence of cDNA coding for Semliki Forest virus membrane glycoproteins. Nature 288:236–241. 13. Garoff, H., D. Huylebroeck, A. Robinson, U. Tillman, and P. Liljestrom. 1990. The signal sequence of the p62 protein of Semliki Forest virus is involved in initiation but not in completing chain translocation. J. Cell Biol. 111:867–876. 14. Garoff, H., K. Simons, and B. Dobberstein. 1978. Assembly of the Semliki Forest virus membrane glycoproteins in the membrane of the endoplasmic reticulum in vitro. J. Mol. Biol. 124:587–600. 15. Garoff, H., K. Simons, and O. Renkonen. 1974. Isolation and characterization of the membrane proteins of Semliki Forest virus. Virology 61:493–504.
J. VIROL. 16. Gibbons, D. L., et al. 2004. Conformational change and protein-protein interactions of the fusion protein of Semliki Forest virus. Nature 427:320– 325. 17. Klimjack, M. R., S. Jeffrey, and M. Kielian. 1994. Membrane and protein interactions of a soluble form of the Semliki Forest virus fusion protein. J. Virol. 68:6940–6946. 18. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680–685. 19. Lescar, J., et al. 2001. The fusion glycoprotein shell of Semliki Forest virus: an icosahedral assembly primed for fusogenic activation at endosomal pH. Cell 105:137–148. 20. Li, L., J. Jose, Y. Xiang, R. J. Kuhn, and M. G. Rossmann. 2010. Structural changes of envelope proteins during alphavirus fusion. Nature 468:705–708. 21. Lobigs, M., J. M. Wahlberg, and H. Garoff. 1990. Spike protein oligomerization control of Semliki Forest virus fusion. J. Virol. 64:5214–5218. 22. Lobigs, M., H. X. Zhao, and H. Garoff. 1990. Function of Semliki Forest virus E3 peptide in virus assembly: replacement of E3 with an artificial signal peptide abolishes spike heterodimerization and surface expression of E1. J. Virol. 64:4346–4355. 23. Marsh, M., E. Bolzau, and A. Helenius. 1983. Penetration of Semliki Forest virus from acidic prelysosomal vacuoles. Cell 32:931–940. 24. Melancon, P., and H. Garoff. 1987. Processing of the Semliki Forest virus structural polyprotein: role of the capsid protease. J. Virol. 61:1301–1309. 25. Mulvey, M., and D. T. Brown. 1996. Assembly of the Sindbis virus spike protein complex. Virology 219:125–132. 26. Rey, F. A., F. X. Heinz, C. Mandl, C. Kunz, and S. C. Harrison. 1995. The envelope glycoprotein from tick-borne encephalitis virus at 2 A resolution. Nature 375:291–298. 27. Salminen, A., J. M. Wahlberg, M. Lobigs, P. Liljestrom, and H. Garoff. 1992. Membrane fusion process of Semliki Forest virus. II. Cleavage-dependent reorganization of the spike protein complex controls virus entry. J. Cell Biol. 116:349–357. 28. Sariola, M., J. Saraste, and E. Kuismanen. 1995. Communication of postGolgi elements with early endocytic pathway: regulation of endoproteolytic cleavage of Semliki Forest virus p62 precursor. J. Cell Sci. 108:2465–2475. 29. Sjo ¨berg, M., and H. Garoff. 2003. Interactions between the transmembrane segments of the alphavirus E1 and E2 proteins play a role in virus budding and fusion. J. Virol. 77:3441–3450. 30. Sjo ¨berg, E. M., and H. Garoff. 2006. Growth of Semliki Forest virus, p. 419–423. In J. E. Celis (ed.), Cell biology: a laboratory handbook, 3rd ed., vol. 1. Academic Press, New York, NY. 31. Smith, J. F., and D. T. Brown. 1977. Envelopments of Sindbis virus: synthesis and organization of proteins in cells infected with wild type and maturationdefective mutants. J. Virol. 22:662–678. 32. Strauss, J. H., and E. G. Strauss. 1994. The alphaviruses: gene expression, replication, and evolution. Microbiol. Rev. 58:491–562. 33. Wahlberg, J. M., W. A. Boere, and H. Garoff. 1989. The heterodimeric association between the membrane proteins of Semliki Forest virus changes its sensitivity to low pH during virus maturation. J. Virol. 63:4991–4997. 34. Wahlberg, J. M., R. Bron, J. Wilschut, and H. Garoff. 1992. Membrane fusion of Semliki Forest virus involves homotrimers of the fusion protein. J. Virol. 66:7309–7318. 35. Wahlberg, J. M., and H. Garoff. 1992. Membrane fusion process of Semliki Forest virus. I. Low pH-induced rearrangement in spike protein quaternary structure precedes virus penetration into cells. J. Cell Biol. 116:339–348. 36. Welch, W. J., and B. M. Sefton. 1979. Two small virus-specific polypeptides are produced during infection with Sindbis virus. J. Virol. 29:1186–1195. 37. Wengler, G., and G. Wengler. 1989. Cell-associated West Nile flavivirus is covered with E⫹ pre-M protein heterodimers which are destroyed and reorganized by proteolytic cleavage during virus release. J. Virol. 63:2521– 2526. 38. White, J., and A. Helenius. 1980. pH-dependent fusion between the Semliki Forest virus membrane and liposomes. Proc. Natl. Acad. Sci. U. S. A. 77:3273–3277. 39. Voss, J. E., et al. 2010. Glycoprotein organization of Chikungunya virus particles revealed by X-ray crystallography. Nature 468:709–712. 40. Yu, I. M., et al. 2009. Association of the pr peptides with dengue virus at acidic pH blocks membrane fusion. J. Virol. 83:12101–12107. 41. Yu, I. M., et al. 2008. Structure of the immature dengue virus at low pH primes proteolytic maturation. Science 319:1834–1837. 42. Zhang, W., et al. 2002. Placement of the structural proteins in Sindbis virus. J. Virol. 76:11645–11658. 43. Zhang, Y., et al. 2003. Structures of immature flavivirus particles. EMBO J. 22:2604–2613. 44. Zheng, A., M. Umashankar, and M. Kielian. 2010. In vitro and in vivo studies identify important features of dengue virus pr-E protein interactions. PLoS Pathog. 6:e1001157. 45. Ziemiecki, A., H. Garoff, and K. Simons. 1980. Formation of the Semliki Forest virus membrane glycoprotein complexes in the infected cell. J. Gen. Virol. 50:111–123. 46. Ziemiecki, A., and H. Garofff. 1978. Subunit composition of the membrane glycoprotein complex of Semliki Forest virus. J. Mol. Biol. 122:259–269.
