Baculovirus GP64 Disulfide Bonds: the Intermolecular Disulfide Bond of Autographa californica Multicapsid Nucleopolyhedrovirus GP64 Is Not Essential for Membrane Fusion and Virion Budding

Cells and transfections. Spodoptera frugiperda (Sf9) cells and the GP64-expressing cell line Sf9^Op1D (24) were cultured at 27°C in TNMFH medium (6) containing 10% fetal bovine serum (FBS). The cells were transfected using a standard CaPO₄ precipitation procedure (3).

Site-directed mutagenesis.The overlap PCR method was used to generate the cysteine-to-alanine substitution mutations with the plasmid pGEM3ZGP64 (14) as the template, as described previously (13). Primer sequences are available upon request. Briefly, the PCR products were digested with unique restriction enzymes, XbaI and NotI or NotI and EcoRI, and subcloned into the transient expression plasmid pBiepA (14). Recombinant baculoviruses expressing the modified GP64 proteins were generated by first subcloning overlap PCR products into plasmid pGEM3ZGP64 (KpnI and NotI or NotI and HindIII sites) and then excising the AcMNPV gp64 promoter and modified GP64 open reading frame (ORF) with KpnI and EcoRI and subcloning it into the donor plasmid pFastBac1 (Invitrogen). As a positive control, the wild-type (WT) gp64 gene was also cloned into the donor plasmid. The construct containing the WT or a modified gp64 gene was inserted into the polyhedrin locus of an AcMNPV GP64-null bacmid (vAc⁶⁴⁻) by Tn7-mediated transposition (17). The donor plasmid carries the β-glucuronidase (GUS) gene as a reporter gene under the transcriptional control of the AcMNPV p6.9 promoter. Constructs were confirmed by restriction enzyme analysis and DNA sequencing.

cELISA and fusion assays.The cell surface-localized GP64 protein was analyzed by a cell surface enzyme-linked immunosorbent assay (cELISA) as described previously (14). Briefly, at 36 h posttransfection (p.t.) or 40 h postinfection (p.i.), Sf9 cells were fixed in 0.5% glutaraldehyde. Relative levels of cell surface-localized GP64 were measured by cELISA using monoclonal antibody (MAb) AcV5 (at a dilution of 1:25) or B12D5 (at a dilution of 1:20) (8, 20). To determine whether GP64 constructs lacking an intermolecular disulfide bonds were capable of undergoing low-pH-induced refolding, cELISA was performed with MAb AcV1, which recognizes the native, but not the low-pH, conformation of GP64 (37). After 36 h p.t., transfected Sf9 cells were incubated in phosphate-buffered saline (PBS) adjusted to various pH values (pH 4.5 to 7) for 30 min and then fixed in 4% (wt/vol) paraformaldehyde. The cells were then processed for cELISA as described previously for MAb AcV5 (13).

To analyze membrane fusion by syncytium formation assays, Sf9 cells seeded into 12-well plates were transfected with plasmids encoding WT or modified forms of GP64, incubated for 36 h at 27°C, and then exposed to PBS at pH 5.0 for 3 min as described previously (13). To assess the possible effects of temperature on membrane fusion mediated by GP64 constructs with Cys-to-Ala substitutions, Sf9 cells transiently expressing GP64 constructs were shifted to 37°C at 36 h p.t. and then incubated for 30 min before fusion was induced by exposure to PBS at pH 5.0 for 3 min. The cells were then incubated at 37°C for 1 h and transferred to 27°C for another 3 h, and the numbers of syncytia were calculated as described previously (15). A syncytial mass was defined as a mass of fused cells containing ≥5 nuclei. Relative levels of fusion activity were determined by dividing the number of nuclei in syncytia by the total number of nuclei in a field and then normalizing to parallel data from cells expressing WT GP64 and localized to the cell surface at equivalent levels.

Transfection-infection assays.Bacmid DNAs encoding GP64 constructs with intermolecular disulfide bond substitutions (C24A, C372A, and C24A-C372A) were extracted using the Bac-to-Bac system protocol (Invitrogen) and then used to transfect Sf9 cells as described previously (13). Also, the gp64 knockout bacmid DNA was cotransfected with the transient-expression plasmids encoding wild-type GP64 or the intermolecular disulfide bond substitutions (C24A, C372A, and C24A-C372A). After 96 h p.t., the transfected or cotransfected cell supernatant was collected and cleared by centrifugation (10 min at 2,200 × g) and then used to infect Sf9 cells for 96 h as described previously (15). The transfected, cotransfected, and infected cells were stained with X-glucuronide (5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid) for detection of GUS activity. To confirm the viability of each bacmid genome and preparation, the same AcMNPV bacmids were used to transfect Sf9^Op1D cells (which express a wild-type GP64 protein). Supernatants from transfected Sf9^Op1D cells were then used to infect Sf9 cells to confirm production of infectious virus.

Metabolic labeling and virion budding.The effects of disrupting the GP64 intermolecular disulfide bond on virion budding efficiency were examined in the following experiments. First, viruses expressing the mutated GP64 constructs were amplified, and their titers were determined in Sf9^OP1D cells, which constitutively express Orgyia pseudotsugata MNPV (OpMNPV) GP64. The resulting virus stocks were used to infect Sf9 cells (5 × 10⁶ cells; multiplicity of infection [MOI], 5), and virion budding was examined as described previously (14). Briefly, at 15 h p.i., cells were starved for 1 h in 2 ml methionine-free Grace's medium (Grace^−met; Invitrogen), and then the medium was replaced with 2.2 ml of Grace^−met containing 200 μCi protein-labeling mix (³⁵S-EasyTag Express; 1,175.0 Ci/mmol; Perkin-Elmer Life Sciences). At 30 h p.i., the medium was supplemented with methionine by adding 0.8 ml of TNMFH. Supernatants were harvested at 40 h p.i. After cell debris was cleared by centrifugation for 10 min at 3,000 × g and 4°C, the virus-containing supernatant was loaded onto a 25% sucrose cushion and centrifuged at 80,000 × g for 90 min at 4°C (Beckman Instruments SW60 rotor). The virus pellets were resuspended in 200 μl Laemmli buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.04% bromophenol blue, 0.125 M Tris, pH 6.8) containing protease inhibitors (Complete; Roche Applied Science), and the proteins were separated on 10% SDS-PAGE gels. To identify and quantify labeled virion proteins, dried gels were exposed on PhosphorImager screens and scanned on a PhosphorImager (Molecular Dynamics), and individual protein bands were quantified using ImageQuant TL software (Amersham, GE Healthcare).

To compare virus budding mediated by various GP64 constructs under conditions that permit normalizing to cell surface levels of GP64, Sf9 cells were transfected with various quantities of plasmids expressing WT GP64 or GP64 constructs containing a disrupted intermolecular disulfide bond, and the cells were infected with a GP64-null virus. Progeny virion production was then analyzed. For these studies, Sf9 cells were transfected in 6-well plates (1 × 10⁶ cells/well), and after incubation for 12 h, the transfected cells were infected with a GP64-null virus (MOI, 5) that was previously prepared by amplification and titration in Sf9^OP1D cells. At 15 h p.i., the cells were starved for 1 h in 0.9 ml Grace^−met, and then the medium was replaced with 1.4 ml of Grace^−met containing 40 μCi protein-labeling mix (³⁵S-EasyTag Express; 1,175.0 Ci/mmol; Perkin-Elmer Life Sciences). At 30 h p.i., the medium was supplemented with methionine by adding 0.2 ml of TNMFH. Supernatants were harvested at 40 h p.i. Purification and analysis of progeny virions were performed as described above.

Western blot analysis.Analysis of GP64 proteins by reducing and nonreducing SDS-PAGE (6% or 10% polyacrylamide gels) was described previously (36). GP64 was detected on Western blots using MAb AcV5 (at a dilution of 1:1,000) or B12D5 (at a dilution of 1:200).

Determination of DSE.The dihedral strain energy (DSE) was determined as described previously (25, 26). Briefly, the DSE of each intramolecular disulfide bond was predicted from the magnitudes of the five dihedral torsion angles (χ) that define the disulfide using the following empirical formula (25, 26): DSE (kJ·mol⁻¹) = 8.37 (1 + cos3χ₁) + 8.37 (1 + cos3χ_1′) + 4.18 (1 + cos3χ₂) + 4.18 (1 + cos3χ_2′) + 14.64 (1 + cos2χ₃) + 2.51 (1 + cos3χ₃) (7). χ₁, χ₂, χ₃, χ_2′, and χ_1′ are the dihedral angles about the Cα—Cβ bond, the Cβ—Sγ bond, the Sγ—Sγ′ bond, the Sγ′—Cβ′ bond, and the Cβ′—Cα′ bond, respectively. The configuration of a disulfide bond is defined as previously described (25, 26). For example, a disulfide is a minus right-handed spiral (−RHSpiral) if the χ₁, χ₂, χ₃, χ_2′, and χ_1′ angles are −, +, +, +, and −, respectively. The disulfides are treated as symmetrical. For instance, a disulfide is a ±RHSpiral if the χ₁, χ₂, χ₃, χ_2′, and χ_1′ angles are +, +, +, +, − or −, +, +, +, +.

Disulfide bonds of the AcMNPV GP64 protein.The postfusion structure of the AcMNPV GP64 trimer is elongated and comprised largely of β-sheets, and it has a central core that contains three long coiled coils that are closely apposed (11). Each monomer contains five domains (DI to DV), and 6 intramolecular disulfide bonds are found among these domains. One additional disulfide, an intermolecular disulfide bond (Cys24—Cys372), connects one monomer to another in the trimer. Of the 6 intramolecular disulfide bonds, 5 are dispersed in the N-terminal region of the linear GP64 sequence, between amino acids 111 and 267 (structural domains I and II) (Fig. 1). Three disulfide bonds (Cys111—Cys164, Cys128—Cys158, and Cys177—Cys184) are located in domain I, while two disulfide bonds (Cys228—Cys246 and Cys262—Cys267) are found in domain II. Four of the five disulfide bonds at the N terminus of GP64 link cysteine residues in adjacent strands of antiparallel β-sheets. The sixth intramolecular disulfide bond (Cys382—Cys402) is located in the C-terminal region of GP64, presumably stabilizing domain IV.

• Open in new tab
• Download powerpoint

FIG. 1.

Disulfide bonds of AcMNPV GP64 in the postfusion structure (Protein Data Bank [PDB] ID, 3DUZ) (11). Cysteine residues are shown as spheres on a ribbon structure of the ectodomain of the GP64 monomer (A) or the GP64 trimer (B). The colors in panel A are used to designate GP64 structural domains (DI to DV). A black line and stem and TM (transmembrane) domains represent hypothetical positions of the C-terminal region, which is not visible in the crystal structure. The colors in panel B designate the three monomers of the GP64 trimer, and Cys residues are indicated as purple spheres. A top view of the intermolecular disulfide bond (C24—C372) is shown on the right (C). Panel D shows the relative positions of disulfide bonds on a linear map of the AcMNPV GP64 protein. (Note: position C24 is disulfide bonded to position C372 on an adjacent GP64 monomer.) The shaded box at the NH₂ terminus represents the signal peptide (SP), while the hatched box near the COOH terminus represents the transmembrane sequence (TM). The geometry of a disulfide bond is illustrated in panel E. The χ angles used to describe the configuration of a disulfide bond are labeled across the bond. The carbon atoms are shown in green, nitrogen in blue, oxygen in red, and sulfur in yellow. The table below lists the characteristics of each intramolecular disulfide bond. Locations are indicated as domains (DI to DV), loop sizes are in amino acids, and distances are in angstroms, based on the published GP64 postfusion structure (PDB 3DUZ). The structures represented in panels A, B, C, and E were generated using the Pymol program (http://pymol.org/ ). N-ter, N terminus; C-ter, C terminus; M, membrane.

The geometry of the intramolecular disulfides was characterized in terms of the five dihedral angles (χ) (Fig. 1E), and the potential energy of each bond was estimated using the equation described in Materials and Methods. The estimated DSEs of bonds Cys111—Cys164, Cys128—Cys158, and Cys382—Cys402 are about 34 to 40 kJ/mol; the distances of Cα—Cα′ for these bonds are about 6 to 6.4 Å; and the configurations for these bonds are RHSpiral (Fig. 1E). In comparison, the estimated DSEs of bonds Cys177—Cys184, Cys228—Cys246, and Cys262—Cys267 are calculated at 18 to 25 kJ/mol. The distances of Cα—Cα′ for these bonds are about 4.2 Å, and the configuration is +RHStaple or −RHStaple (25, 34). Thus, the lower strain energy of these bonds might suggest that they are more stable than the others. The intermolecular disulfide bond Cys24—Cys372 is located in the top region of the trimer and covalently links the N terminus of one GP64 monomer and the C terminus of another monomer (Fig. 1B and C).

Effects of cysteine-to-alanine substitutions on GP64 expression and oligomerization.We systematically disrupted each disulfide bond in the GP64 ectodomain and examined the effects on protein production or stability and trimerization. Each of the 14 cysteine residues in the ectodomain of GP64 was replaced with alanine, individually or as pairs for each disulfide bond (Fig. 1). GP64 proteins containing the cysteine-to-alanine substitutions were transiently expressed from plasmids in Sf9 cells. Cell lysates were examined at 36 h p.t. by Western blot analysis using either reducing or nonreducing conditions for SDS-PAGE to assess protein expression and/or stability, as well as trimerization. Because GP64 trimers contain the intermolecular disulfide bond (Cys24—Cys372), nonreducing conditions permit the assessment of GP64 trimerization. The GP64 trimer is typically observed as two electrophoretic forms (trimers I and II) (23).

Alanine substitutions that individually disrupted the 5 intramolecular disulfide bonds found near the N terminus of GP64 (Fig. 1, Cys111—Cys164, Cys128—Cys158, Cys177—Cys184, Cys228—Cys246, and Cys262—Cys267) resulted in defects in trimerization. GP64 trimers were not detected when the 5 N-terminal intramolecular disulfides were disrupted by either single or paired Cys substitutions (Fig. 2 A, top). None of these GP64 constructs formed higher-molecular-weight aggregates. For two of these intramolecular disulfide bonds (Cys128—Cys158 and Cys228—Cys246), the expression levels of the mutated GP64s were comparable to that of the wild type, while for the others, substitutions resulted in reduced levels of GP64 (Fig. 2A, bottom). The detection of lower levels of these GP64 constructs (compared with that of wild-type GP64) suggested the likelihood that the misfolded protein might be degraded (Fig. 2A, bottom). In the case of the intramolecular disulfide (Cys382—Cys402) located in the C-terminal region of GP64, single Cys-to-Ala substitutions in either position disrupted trimer formation or stability. In contrast, disruption of both residues (Cys382 and Cys402) resulted in trimerization similar to that of wild-type GP64 (Fig. 2A, gel 7, Cys382—Cys402), although present at lower levels.

• Open in new tab
• Download powerpoint

FIG. 2.

Transient expression and cell surface localization of GP64 constructs containing single and paired cysteine substitutions. (A) Analysis of transient expression and trimerization of GP64 constructs. At the bottom and top are shown Western blot analyses of GP64 proteins separated by nonreducing (NR; top) and reducing (R; bottom) SDS-PAGE. Each gel (1 to 7) represents analysis of cysteines associated with one disulfide bond. Sf9 cells were transfected with expression plasmids encoding either wild-type GP64 (WT) or a GP64 construct containing one or two Cys-to-Ala mutations (C##A, where ## represents the amino acid position of the Cys residue). Cells were lysed at 36 h posttransfection, and the proteins were analyzed. Oligomeric forms of GP64 (trimers I and II, dimer, and monomer) are indicated on the right. (B) cELISA analysis of cell surface levels of GP64 constructs. The cell surface-localized GP64s were detected in 12-well plates (4 × 10⁵ cells/well) by monoclonal antibody AcV5, as described previously (14). A standard curve generated from WT GP64 expression is shown on the left. The quantity (μg) of plasmid DNA used in each transfection is indicated below the graph. The relative cell surface level of each GP64 construct was normalized to that of cells transfected with 2 μg of the plasmid expressing WT GP64. The values represent the means from triplicate transfections, and the error bars represent the standard deviations from the mean.

When alanine substitution mutations were introduced into one or both Cys residues of the single intermolecular disulfide bond (Cys24—Cys372), no disulfide-linked trimer was observed on nonreducing SDS-PAGE, as expected. In addition, GP64 production and/or stability was largely unaffected (Fig. 2A, gel 1, lane 1 versus 2 to 4). Some higher-molecular-weight forms of GP64 were detected in one construct (C372A), but in the double mutation (C24A—C372A), the majority of the protein was observed as monomeric GP64 (Fig. 2A, gel 1, C24A—C372A). Thus, surprisingly, disruption of the GP64 intermolecular disulfide did not substantially affect GP64 production or stability. In contrast, disruption of the intramolecular disulfides resulted in (i) reduced protein production or stability in many cases and (ii) disruption of trimerization (as determined on reducing gels) in all except C382A—C402A.

Effects of cysteine-to-alanine substitutions on GP64 cell surface localization.Since cell surface levels of GP64 can affect the assessment of function, we examined the effects of cysteine substitutions on transport to the cell surface and stability there. Using a cELISA protocol, we compared the relative cell surface level of each GP64 construct to that detected from WT GP64. For comparisons of relative GP64 levels at the cell surface, we established a standard curve by transfecting Sf9 cells with decreasing quantities of a plasmid expressing WT GP64 (Fig. 2B). GP64 constructs containing substitutions that disrupted the intermolecular disulfide bond (Cys24—Cys372) were displayed at the cell surface at levels approximately 20 to 30% of that from WT GP64. For all of the GP64 constructs in which the intramolecular disulfide bonds were disrupted, cell surface levels of the GP64 constructs were less than 5% of that from WT GP64. Although the surface levels were severely reduced, the cell surface levels were sufficient for detection of membrane fusion activity from wild-type GP64 (Fig. 2B, 0.001; Fig. 3 A, WT, 0.001).

• Open in new tab
• Download powerpoint

FIG. 3.

Analysis of membrane fusion by GP64 constructs containing disrupted disulfide bonds. (A) Syncytium formation assays of transiently expressed GP64. Sf9 cells were transfected with plasmids expressing WT GP64 or the indicated GP64 construct containing one or two Cys substitutions. At 36 h posttransfection, the cells were incubated in PBS (pH 5.0) for 3 min to trigger fusion and then incubated in TNMFH medium for 4 h, stained with a HEMA3 stain kit (Fisher Scientific LLC), and photographed under phase-contrast microscopy (×200). By definition, each syncytial mass contains ≥5 nuclei, and examples are indicated by arrows. (B) Analysis of fusion activity. The relative fusion activity of each indicated construct was determined by analysis of syncytium formation. For each construct, five fields were randomly selected, and the syncytium formation efficiency was determined by dividing the number of nuclei in syncytia by the total number of nuclei in the field. The percentage was normalized to parallel syncytium formation data from wild-type GP64 that was localized to the cell surface at equivalent levels. The means and standard deviations of triplicate transfections are shown. (C) Analysis of temperature effects on syncytium formation efficiency. Sf9 cells were transfected, and GP64 constructs were displayed on the cell surface at 27°C as described for panel B and then shifted to the indicated temperature (either 27°C or 37°C) for 30 min prior to triggering fusion (by incubation with PBS, pH 5.0) and followed by 1 h of incubation in TNMFH, also at the indicated temperature. The relative fusion activity was determined as described above.

The intermolecular disulfide bond Cys24—Cys372 is not essential for membrane fusion.We evaluated the membrane fusion activities of GP64 constructs by measuring syncytium formation efficiency. Because surface levels of different GP64 constructs varied, the fusion activity of each modified GP64 construct was normalized by comparison with the activity from WT GP64 that was localized to the cell surface at a similar level. All constructs containing Cys substitutions (single or paired) that disrupted intramolecular disulfide bonds were defective for membrane fusion, as no syncytia were observed from any of those constructs (Fig. 3A). Surprisingly, and in contrast, syncytia were detected from GP64 containing Cys substitution mutations that disrupted the intermolecular disulfide bond (constructs C24A, C372A, and C24A-C372A) (Fig. 3A). The normalized fusion activities for these three constructs were 24 to 39% of that from WT GP64 (Fig. 3B). This result indicated that the intermolecular disulfide bond is not essential for GP64-mediated membrane fusion.

One possible role of the unusual intermolecular disulfide bond is to contribute to the thermal stability and function of the GP64 trimer. To examine thermal effects of disrupting the intermolecular disulfide bond on GP64-mediated membrane fusion, we performed syncytium formation assays at 37°C. The high-temperature treatment (37°C) had no substantial effect on detection of GP64 fusion activity (Fig. 3C). This result suggests that within this range of temperatures (27 to 37°C), the intermolecular disulfide bond is not necessary for stabilizing the GP64 trimer in its function as a fusion protein.

Low-pH-induced refolding of GP64 does not require the intermolecular disulfide bond.We also examined the potential effects of the intermolecular disulfide bond on the low-pH-triggered conformational change in GP64. To determine whether disruption of the intermolecular disulfide bond of the GP64 trimer had a measurable effect on the low-pH-induced conformational change, we used a conformation-specific monoclonal antibody (MAb AcV1). Monoclonal antibody AcV1 recognizes the native prefusion conformation of GP64, and the AcV1 epitope is lost upon exposure of GP64 to low pH (37) (Fig. 4, WT). In contrast to results from wild-type GP64 constructs, binding of AcV1 to constructs C24A and C24A-C372A increased at lower pH values. Lowering the pH value of the medium from 7.0 to 5.5 resulted in increased binding of MAb AcV1 to construct C24A-C372A, with binding peaking at pH 5.5 and then decreasing as the pH was decreased further (Fig. 4, C24A-C372A). A similar although less pronounced effect was observed for construct C24A. These AcV1 binding results suggest that in the absence of the intermolecular disulfide, AcV1 binding is enhanced at pH 5.5, but like WT GP64, binding is lost after the conformational change. Combined with the observation that fusion activity of GP64 is not inactivated by eliminating the intermolecular disulfide bond, these data suggest that the pH-triggered conformation change of AcMNPV GP64 does not require the intermolecular disulfide bond.

• Open in new tab
• Download powerpoint

FIG. 4.

Effects of the intermolecular disulfide bond on the GP64 conformational change. GP64 binding by MAb AcV1, which recognizes a conformational epitope in the prefusion structure of GP64, was measured at various pH values. AcV1 binding to three constructs that disrupt the intermolecular disulfide bond (C24A, C372A, and C24A-C372A) was compared with binding to WT GP64. Each GP64 construct was expressed by transfecting Sf9 cells with 2 μg of the appropriate plasmid DNA, and cELISA was performed using MAb AcV1 and methods described in the legend to Fig. 2. The values represent the means from triplicate transfections and cELISA analysis and are normalized to that of cells transfected with 2 μg of the plasmid DNA expressing WT GP64 at pH 7 (100%).

GP64 constructs containing a disrupted intermolecular disulfide bond failed to rescue the gp64 knockout bacmid.Because GP64 is important for viral entry and egress, we also examined the effects of disrupting the GP64 intermolecular disulfide bond on viral infection and propagation. Bacmid DNAs encoding GP64 constructs with single or double substitutions of the intermolecular disulfide bond (C24A, C372A, and C24A-C372A) were used to transfect Sf9 cells in a transfection-infection assay (see Materials and Methods). Transfected and infected Sf9 cells were stained for GUS activity as an indicator of viral replication. As shown in Fig. 5 A, transfection with bacmids expressing GP64 constructs C24A, C372A, and C24A-C372A did not result in rescue of viral infectivity and transmission by infection (Fig. 5A, bottom panels), although we previously found that these three GP64 constructs were capable of mediating membrane fusion (Fig. 3). The expression of these modified GP64 proteins was confirmed by Western blot analysis of bacmid-transfected Sf9 cells (Fig. 5B), although expression levels were very low. In addition, supernatants collected from bacmid-transfected Sf9^OP1D cells, which express GP64 from OpMNPV, were capable of transmitting infection to Sf9 cells (data not shown). These results indicate that the defect in infectivity was not caused by a second-site mutation in the bacmid DNA.

• Open in new tab
• Download powerpoint

FIG. 5.

Analysis of virus infectivity by transfection-infection assay. (A) Sf9 cells were transfected with bacmids in which the WT GP64 gene was deleted and its expression was replaced with a gene encoding GP64 substitution constructs C24A, C372A, and C24A-C372A. Controls included a GP64-null bacmid (KO), as well as a bacmid expressing wild-type GP64 (WT). (B) Western blot analysis of the expression of the wild-type and the mutated GP64s using MAb AcV5 from the transfected Sf9 cells in the top row of panel A. (C) Sf9 cells were cotransfected with GP64-null bacmid DNA in combination with plasmids transiently expressing WT GP64 or GP64s with a disrupted intermolecular disulfide bond. In panels A and C, the stained cells resulted from expression of a GUS reporter gene under the control of a p6.9 promoter in bacmid DNAs. The top row (Pt) represents cells at 96 h posttransfection. The bottom row (Pi) represents cells infected with supernatants from the above transfected cells and stained at 96 h p.i. (D) cELISA analysis of the cell surface expression of the indicated GP64 constructs. Cell surface-localized GP64s were detected in 6-well plates (1 × 10⁶ cells/well) as described above. The quantity (μg) of plasmid DNA used in each transfection is indicated below the graph. The relative cell surface level of each GP64 construct was normalized to that of cells transfected with 5 μg of the plasmid expressing WT GP64 (100%). The values represent the means from triplicate transfections, and the error bars represent the standard deviations from the mean.

Since the expression levels of the mutated GP64 constructs (in bacmid-transfected cells) were substantially lower than that of wild-type GP64 (Fig. 5B), failure to rescue viral infectivity could result from insufficient expression or insufficient GP64 at the cell surface. To address that possibility, we cotransfected gp64 knockout bacmid DNA with plasmids that expressed WT GP64 or GP64 constructs with a disrupted intermolecular disulfide bond. Transfection with 1 μg of the plasmid expressing WT GP64 resulted in sufficient levels of GP64 (Fig. 5D, WT) to rescue infectivity in the transfection-infection assay (Fig. 5C, WT, bottom). Transfection of Sf9 cells with 5 μg of each of the plasmids expressing GP64 with a disrupted intermolecular disulfide resulted in similar or higher levels of surface-localized GP64 (Fig. 5D, C24A, C372A, and C24A-C372A) yet failed to rescue viral infectivity in the transfection-infection assay (Fig. 5C, bottom, C24A, C372A, and C24A-C372A). Therefore, the failure to rescue infectivity by constructs C24A, C372A, and C24A-C372A does not appear to result from insufficient levels of cell surface-localized GP64.

The intermolecular disulfide bond is not necessary for virion budding and incorporation of GP64 into virions.GP64 is important for virion budding, and we previously found that virion budding efficiency was decreased by approximately 90% in the absence of GP64 (23). Therefore, we examined the effects of disrupting the intermolecular disulfide bond on virion budding. Two strategies were used to analyze virion budding efficiency. Initially, viruses encoding GP64 proteins that contained a disrupted intermolecular disulfide bond (constructs C24A, C372A, and C24A-C372A) were propagated in Sf9^OP1D cells and then used to infect Sf9 cells, where budding of progeny virions was analyzed. Analysis of GP64 expression and cell surface localization showed that GP64 constructs C24A, C372A, and C24A-C372A were found at very low levels compared to WT GP64 (data not shown). Relative quantities of progeny virions, as estimated by [³⁵S]methionine labeling (13), revealed that budding was reduced to approximately 2 to 6% relative to that mediated by WT GP64 (data not shown) and was indistinguishable from that of a GP64 knockout virus. Since cell surface levels of these GP64 constructs (C24A, C372A, and C24A-C372A) were extremely low, the results of those studies were inconclusive. Next, we used transient expression to increase the cell surface levels of GP64 constructs. Sf9 cells were transfected with the plasmids expressing GP64 constructs C24A, C372A, and C24A-C372A and WT GP64. After 12 h, the transfected cells were infected with a GP64-null virus. The cell surface levels of the transiently expressed GP64 constructs (containing intermolecular disulfide bond mutations) were approximately 40 to 46% of that from cells similarly transfected with WT GP64 (Fig. 6 A). Cell surface levels of WT GP64 were also titrated by transfection with various amounts of plasmid. By combining this procedure with measurements of budding (using [³⁵S]Met labeling as described previously [13]), we generated a standard curve of WT GP64 surface levels versus virion budding (Fig. 6A to C). By comparing data obtained from GP64 constructs C24A, C372A, and C24A-C372A (Fig. 6A and B) to the standard curve from WT GP64, we found that GP64 constructs C24A, C372A, and C24A-C372A mediated virion budding with normalized virion budding efficiencies of 39 to 88% (Fig. 6D) relative to WT GP64. In addition, by calculating the GP64/VP39 ratio in WT and mutated GP64 constructs, we found that GP64 constructs C24A and C24A-C372A were incorporated into virions at levels of approximately 72% and 95%, respectively (Fig. 6E), compared with WT GP64. Thus, we found that the intermolecular disulfide bond of GP64 had a significant effect on protein transport and localization at the cell surface in virus-infected cells. When we compensated for surface levels, we found that the intermolecular disulfide was not necessary for either virion budding or GP64 incorporation into the virion.

• Open in new tab
• Download powerpoint

FIG. 6.

Analysis of virion budding and GP64 incorporation into virions. (A) cELISA analysis of cell surface levels of GP64 constructs. Cell surface-localized GP64s were detected in 6-well plates (1 × 10⁶ cells/well) using monoclonal antibody AcV5 as described previously (14). A standard curve generated from transiently expressed WT GP64, generated by transfection of various quantities of plasmid DNA, is shown on the left. The quantity (μg) of plasmid DNA used in each transfection is indicated below the graph. The relative cell surface level of each GP64 construct was normalized to that of cells transfected with 5 μg of the plasmid expressing WT GP64. The values represent the means from triplicate transfections, and the error bars represent the standard deviations from the mean. (B) Analysis of virion budding efficiency by analysis of labeled progeny virions. Sf9 cells were transfected with plasmids expressing either WT GP64 or GP64 constructs C24A, C372A, and C24A-C372A and then infected at 12 h posttransfection with a GP64-null virus. Proteins were metabolically labeled with [³⁵S]methionine, labeled progeny baculoviruses were isolated, and proteins were analyzed by SDS-PAGE and PhosphorImager analysis. The positions of the GP64 and VP39 proteins are indicated on the right. The molecular mass markers are indicated on the left. NV, mock transfected and no-virus infection control; Mock, mock transfected and infected with GP64-null virus. Each lane represents virus purified from an equivalent volume of the cell culture supernatant. The quantities of plasmid DNAs used for transfections are indicated below each lane. (C) Standard curve representing the relationship between cell surface levels of WT GP64 and relative virion budding efficiency. The x axis represents the relative cell surface level of WT GP64 (panel A), and the y axis represents the relative budding efficiency (from panel B, lanes 1 to 8). The budding efficiencies were determined by measuring relative levels of [³⁵S]Met-labeled VP39 (panel B) generated from cells transfected with increasing quantities of plasmid DNA. Budding from cells transfected with 5 μg of plasmid DNA encoding WT GP64 was assigned a value of 100%. (D) Analysis of virion budding. The normalized budding efficiencies mediated by GP64 constructs C24A, C372A, and C24A-C372A were determined by comparison of the relative VP39 level for each construct (panel B) with the expected levels mediated by WT GP64 (panel C). (E) GP64 incorporation into virions. The relative efficiencies of GP64 incorporation into virions were determined by analysis of GP64/VP39 ratios in virion preparations containing WT or mutated (C24A, C372A, and C24A-C372A) GP64. Phosphorimager analysis was used to determine molar ratios of GP64 to VP39 (panel B), based on 16 methionine residues (GP64) or 9 methionine residues (VP39). Efficiencies for mutated GP64 constructs were determined by comparison to ratios derived from WT GP64 and generated from cells in which GP64 was localized to the cell surface at equivalent levels (panel A). The means and standard deviations determined from triplicate experiments are shown.

Intramolecular disulfide bonds.In the current study, we found that all of the intramolecular disulfide bonds of AcMNPV GP64 were critical for function. In all cases of single and double (pairwise) substitutions of cysteine residues involved in intramolecular disulfide bonds, protein transport and/or stability were severely and negatively affected. Cell surface levels of those GP64 constructs were ≤5% of that of WT GP64. Disruption of the intramolecular disulfides also resulted in almost complete disruption of detectable trimerization, except in one case (C382A—C402A). It was previously shown that in infected cells, monomers of wild-type GP64 that are not trimerized and transported to the cell surface are degraded within 30 to 45 min after synthesis (23). Thus, the reduced cell surface detection of the GP64 proteins containing mutations that disrupt the intramolecular disulfides may result from misfolding and/or failure to trimerize, followed by degradation. Most importantly, disruption of any of the intramolecular disulfides resulted in complete loss of membrane fusion activity, even though the low levels detected at the surface were sufficient for detection of fusion by WT GP64 (Fig. 2B and 3A). Thus, all of the intramolecular disulfides appear to be (i) important for folding, transport, and/or cell surface localization and (ii) required for GP64-mediated membrane fusion. It is noteworthy that among 5 intramolecular disulfide bonds at the N terminus of GP64, 3 (Cys177—Cys184, Cys228—Cys246, and Cys262—Cys267) are classified as cross-strand disulfides (CSDs). CSDs are an uncommon class of disulfide bonds that link cysteine residues in adjacent strands of parallel or antiparallel β-sheets. In a survey of protein structures, it was observed that CSDs are more frequently found in proteins involved in cell entry, such as bacterial toxins and viral fusion proteins, than in other proteins (35). CSDs store potential energy, are easily cleaved, and are more susceptible to reducing agents than other disulfide bonds. It is hypothesized that they may have a redox-related functional role rather than a strictly structural role (34). Cleavage of CSDs in viral entry proteins could release stored energy, which may be used for conformation changes that mediate the fusion activity of the protein (35). Previous studies showed that receptor binding caused the cleavage of one or more of the nine disulfide bonds in HIV-1 gp120 (2, 5). The reduction of the disulfide bonds is thought to facilitate the conformation change and the release of the fusion protein gp41. It was proposed that the reduction of disulfide bonds may include two of three CSDs in gp120 (35). In GP64, the three CSDs (C177—C184, C228—C246, and C262—C267) are located in a cluster in domains I and II, near the center and on the periphery of the elongated postfusion GP64 trimer. This observation, in combination with the similarity of the postfusion structures of GP64 and VSV G (1), suggests that domain II of GP64 could play the role of a hinge during the pre- to postfusion conformation change. Although the estimated DSE of these three bonds is lower than that of the other disulfides, the configurations and locations of the disulfide bonds Cys228—Cys246 and Cys262—Cys267 within domain II might suggest that these bonds play a critical functional role in the conformation change. Two intramolecular disulfide bonds, Cys111—Cys164 and Cys128—Cys158, are associated with the second fusion loop and may be involved in, or required for, stabilizing the position of the fusion loop in the postfusion (and possibly the prefusion) structure. Interestingly, these disulfides have relatively high estimated DSE values.

When the C-terminal intramolecular disulfide bond (Cys382—Cys402) was disrupted (by double Cys-to-Ala substitutions), both trimer I and trimer II forms were detected (Fig. 1A and 2A, gel 7), suggesting only a small or moderate effect on formation of the disulfide-bonded GP64 trimer. Because this disulfide may be the last intramolecular disulfide bond formed in the monomer structure, it may play a lesser role in stabilizing the structure prior to formation of the intermolecular disulfide bond in the trimer. Although the proteins containing a disruption in the Cys382—Cys402 disulfide form a trimer, protein levels at the cell surface were substantially reduced, and those mutated proteins had no detectable membrane fusion activity. Thus, while the Cys382—Cys402 disulfide may be dispensable for folding and trimerization, it appears to play an important role in transport and/or stability and a critical role in protein function during membrane fusion. It is possible that the functional defect may be in the stability of the prefusion structure or in the triggering or subsequent conformational change required for membrane fusion.

The intermolecular disulfide bond.In contrast to the results from disruptions of intramolecular disulfide bonds, disruption of the intermolecular disulfide bond Cys24—Cys372 (by either single or paired alanine substitutions) had only a moderate effect on expression of the mutated GP64 within cells (Fig. 2A, gel 1 versus 2 to 7). In our studies, trimers were detected on nonreducing gels by formation of the Cys24—Cys372 disulfide bond. Thus, it is not known whether the GP64 proteins containing a disrupted intermolecular disulfide are noncovalently associated in a trimer. However, some higher-molecular-weight forms were detected on nonreducing gels, suggesting the possibility of a low level of aberrant disulfide bond formation or aggregation. Although disruption of the intramolecular disulfides had substantial effects on GP64 stability and cell surface levels, disruption of the intermolecular disulfide bond did not have a dramatic effect on GP64 transport to the cell surface. Very notably, membrane fusion activity from transiently expressed GP64 was not eliminated by disruption of the intermolecular disulfide bond. In the postfusion structure (11), the large interface of the central helical bundle and the contacts of domains I and V likely contribute substantially or most significantly to the stability of the GP64 trimer. The observation that membrane fusion activity was not lost from GP64 constructs containing no intermolecular disulfide bond suggests that the GP64 trimer is stable in the absence of the intermolecular disulfide bond. Prior studies using dithiothreitol (DTT) reduction of surface-localized GP64 suggested that the intermolecular disulfides and a stable trimer were required for GP64-mediated fusion activity (19). However, structural data (11) and our current data from alanine substitutions that disrupt the Cys24—Cys372 intersubunit disulfide do not support that conclusion but rather indicate that the intersubunit disulfide is not critical for membrane fusion activity. However, trimers may be necessary, and stable trimers may form in the absence of the intermolecular disulfide. It is also possible that an intermolecular disulfide may form by disulfide isomerization in response to low-pH triggering, although we know of no experimental evidence that such isomers exist. In light of the current data, it is perhaps not surprising that DTT reduction of surface-expressed GP64 resulted in loss of membrane fusion activity (19), since reduction of several or many of the intramolecular disulfide bonds may have resulted in the fusion defect. In our analysis of intramolecular disulfide bonds by mutagenesis, we were unable to detect fusion activity when even one of the intramolecular disulfide bonds was disrupted.

Because disulfide bonds are frequently associated with protein stability or resistance to denaturing, we also examined the possibility that disrupting the intermolecular disulfide might destabilize the function of GP64 at higher temperatures. Using a conformation-specific MAb, our data suggest that while the intermolecular disulfide is not required for fusion activity, the structure may be somewhat moderately disrupted and that disruption causes an increase in the exposure of the conformational epitope at a pH around that required for triggering.

GP64 is necessary for both virus entry (binding and membrane fusion) and efficient virion budding. GP64 constructs containing single and paired substitutions that disrupt the intermolecular disulfide (C24A, C372A, and C24A-C372A) were capable of membrane fusion activity but were unable to rescue infectivity of a GP64-null virus (Fig. 5), even when GP64 cell surface levels were augmented by transient expression to levels comparable to those of WT GP64 (Fig. 5C and D). It is interesting that augmenting cell surface levels of GP64 constructs C24A, C372A, and C24A-C372 to mirror that of WT GP64 (Fig. 6A) resulted in virion budding that was approximately 39 to 88% of that measured in the presence of WT GP64 (Fig. 6D). In addition, the incorporation of GP64 into virions was not substantially negatively affected by the absence of the intermolecular disulfide (Fig. 6E). Thus, the defect in infectivity of viruses carrying a GP64 with a disrupted intermolecular disulfide bond cannot be easily explained by defects in budding or GP64 incorporation into virions.

Viral membrane fusion proteins, such as GP64, are folded into complex prefusion structures that change conformation in response to low-pH triggering. A number of critical questions are central to understanding the synthesis and functions of these proteins. How is the prefusion structure folded and stabilized, how is the conformational change triggered, and how is the postfusion structure formed and stabilized? Disulfide bond formation during protein synthesis represents a mechanism by which complex protein structures are generated and stabilized. Disulfide bonds may also be intimately involved in protein function (7, 18). Indeed, thiol/disulfide exchange has been shown to be necessary for membrane fusion in several viral fusion proteins (9, 27, 31). For example, in some retroviruses, the membrane fusion protein is a trimer of SU-TM (surface-transmembrane complex) monomers in which the SU and TM components are disulfide linked. In SU, the linking cysteine residue is located in an isomerization CXXC motif, and the other cysteine of this motif contains a free (non-disulfide-bonded) thiol. Receptor binding activates the free CXXC thiol to attack the intersubunit disulfide bond, resulting in isomerization and formation of a disulfide between the two cysteines of the CXXC motif. This facilitates SU dissociation from the TM subunit and results in activation of membrane fusion activity by TM (28, 31, 32). Thiol/disulfide exchange is also required for membrane fusion mediated by a paramyxovirus F protein. The reduction of one or more disulfide bonds in cell surface-localized F protein of Newcastle disease virus (NDV) is essential for membrane fusion (9). We cannot rule out mechanisms of disulfide exchange in the function of GP64, although data to suggest such a mechanism are currently lacking. In the current study, we examined the effects of disrupting each of the 7 disulfide bonds. Our data clearly demonstrate that the six intramolecular disulfide bonds are critical for GP64 expression, transport, and membrane fusion. The functional role(s) of these bonds may be related to structural stability or conformational change. The intermolecular disulfide bond of AcMNPV GP64 is an unusual feature not typically found in trimeric viral fusion proteins. We found that the intermolecular bond that bridges the subunits of the trimer is not essential for GP64-mediated membrane fusion, virion budding, or incorporation into the virion. However, GP64 constructs containing a disrupted intermolecular disulfide bond failed to rescue the GP64-null bacmid. Given the likely complex nature of structural changes during GP64-mediated membrane fusion, further experiments to examine the effects of disulfide bond disruptions on GP64 structure, stability, and conformation will be of great interest and should provide a better understanding of the roles of disulfide bonds in the structures and functions of viral membrane fusion proteins like GP64.