Critical Residues and Contacts within Domain IV of Autographa californica Multiple Nucleopolyhedrovirus GP64 Contribute to Its Refolding during Membrane Fusion

Baculovirus GP64 is grouped with rhabdovirus G, herpesvirus gB, and thogotovirus glycoproteins as a class III viral fusion protein. In their postfusion structures, these proteins contain five domains (domains I to V). Distinct from domain IV of rhabdovirus G and herpesvirus gB proteins, which is composed of β-sheets, domain IV of GP64 is a loop region; the same domain in thogotovirus glycoproteins has not been solved. In addition, domain IV is proximal to domain I (fusion domain) in prefusion structures of vesicular stomatitis virus (VSV) G and human cytomegalovirus (HCMV) gB but resides at the domain I-distal end of the molecule in a postfusion conformation. In this study, we identified that highly conserved residues and contacts within domain IV of AcMNPV GP64 are necessary for low-pH-triggered conformational change and fusion pore expansion. Our results highlight the roles of domain IV of class III viral fusion proteins in refolding during membrane fusion.

ABSTRACT

Autographa californica multiple nucleopolyhedrovirus (AcMNPV) GP64 is a class III viral fusion protein that mediates low-pH-triggered membrane fusion during virus entry. Although the structure of GP64 in a postfusion conformation has been solved, its prefusion structure and the mechanism of how the protein refolds to execute fusion are unknown. In its postfusion structure, GP64 is composed of five domains (domains I to V). Domain IV (amino acids [aa] 374 to 407) contains two loops (loop 1 and loop 2) that form a hydrophobic pocket at the membrane-distal end of the molecule. To determine the roles of domain IV, we used alanine-scanning mutagenesis to replace each of the individual residues and the contact-forming residues within domain IV and evaluate their contributions to GP64-mediated membrane fusion and virus infection. In many cases, replacement of a single amino acid had no significant impact on GP64. However, replacement of R392 or disruption of the N381-N385, N384-Y388, N385-W393, or K389-W393 contact resulted in poor cell surface expression and fusion loss of the modified GP64, whereas replacement of E390 or G391 or disruption of the N381-K389, N381-Q401, or N381-I403 contact reduced the cell surface expression level of the constructs and the ability of GP64 to mediate fusion pore expansion. In contrast, replacement of N407 or disruption of contact D404-S406 appeared to restrict fusion pore expansion without affecting expression. Combined with the finding that these constructs remain in the prefusion conformation or have a dramatically less efficient transition from the prefusion to the postfusion state under acidic conditions, we proposed that domain IV is necessary for refolding of GP64 during membrane fusion.

IMPORTANCE Baculovirus GP64 is grouped with rhabdovirus G, herpesvirus gB, and thogotovirus glycoproteins as a class III viral fusion protein. In their postfusion structures, these proteins contain five domains (domains I to V). Distinct from domain IV of rhabdovirus G and herpesvirus gB proteins, which is composed of β-sheets, domain IV of GP64 is a loop region; the same domain in thogotovirus glycoproteins has not been solved. In addition, domain IV is proximal to domain I (fusion domain) in prefusion structures of vesicular stomatitis virus (VSV) G and human cytomegalovirus (HCMV) gB but resides at the domain I-distal end of the molecule in a postfusion conformation. In this study, we identified that highly conserved residues and contacts within domain IV of AcMNPV GP64 are necessary for low-pH-triggered conformational change and fusion pore expansion. Our results highlight the roles of domain IV of class III viral fusion proteins in refolding during membrane fusion.

INTRODUCTION

During infection, enveloped viruses enter into host cells via fusion of the virus envelope and cellular membranes (1). The virus-cell fusion reaction is catalyzed by one or more viral fusion proteins that are associated with the envelope of virus particles. Based on the characteristics of their three-dimensional structures, viral fusion proteins have been classified into three distinct categories, classes I to III (2–4). Class I fusion proteins, such as hemagglutinin (HA) of influenza virus, the coronavirus spike (S) glycoprotein, and human immunodeficiency virus type 1 (HIV-1) GP41, are homotrimers, and they mainly consist of α-helixes with a central coiled-coil. Class II fusion proteins, which include the E glycoprotein of flaviviruses and the E1 protein of alphaviruses, form an elongated homo- or heterotrimer mainly comprised of β-sheets with internal fusion loops. Class III fusion proteins, which include baculovirus GP64, herpesvirus glycoprotein B (gB), the thogotovirus glycoprotein (Gp), and the rhabdovirus glycoprotein (G), share the structural features of both class I and class II proteins, such as the α-helical coiled-coil in the heart of homotrimers, like class I proteins, and β-sheets holding the internal fusion loops, like class II proteins. Although viral fusion proteins from different classes show great diversity in structural features, they appear to mediate fusion via a common mechanism (3). In a well-established stalk-hemifusion-pore model for biological membrane fusion, the fusion process mediated by viral fusion proteins undergoes several distinct stages. Initially, upon being triggered by specific mechanisms (binding to cellular receptors and/or low pH, etc.), the viral fusion proteins transit from their prefusion to postfusion structures and expose the fusion peptide or fusion loops, which insert into the target cellular membrane. Then, refolding of the trimeric viral fusion proteins onto themselves brings the host membrane into close proximity to the viral envelope. Following contact between the two separate membranes, merger of the outer membrane leaflets of the membranes forms a hemifusion stage. Further refolding of the fusion proteins into a stable postfusion structure drives the merger of inner leaflets of both membranes to form a fusion pore and subsequent fusion pore expansion to release viral capsids into cells (3, 4).

Baculoviruses are enveloped, arthropod-specific viruses that are isolated from infected insects of the orders of Lepidoptera, Diptera, and Hymenoptera. These viruses are composed of one or more rod-shaped nucleocapsids that contain the circular double-stranded DNA genome (approximately 80 to 180 kb) (5, 6). Baculoviruses are widely used as biological pesticides in insect pest control and as vectors in protein expression and mammalian cell transduction (5, 7, 8). Autographa californica multiple nucleopolyhedrovirus (AcMNPV) is the archetype species of the Alphabaculovirus genus of the Baculoviridae family. In an infection cycle, AcMNPV replicates in the nucleus and produces two kinds of virions with distinct phenotypes: occlusion-derived virions (ODVs) and budded virions (BVs). ODVs infect midgut epithelial cells and may fuse with the plasma membrane of midgut cells with the aid of the viral PIF (per os infectivity factor) complex. BVs infect cells of other tissues and cells cultured in vitro (5, 9, 10). During entry, the major envelope glycoprotein GP64 of AcMNPV BVs has receptor binding activity and facilitates the attachment of BVs to the cell surface (11–13). Then, BVs penetrate into cells via clathrin-mediated endocytosis (9, 10, 14, 15). In the cellular endosomal system, low pH triggers the conformational change of GP64-mediated fusion of the envelope of BVs with endosomal membranes, which leads to the release of nucleocapsids into the cytosol (16–18). In addition to its necessary roles in virus entry (receptor binding and membrane fusion), GP64 is also required for nucleocapsid budding and release at the plasma membrane (19).

GP64 is a type I membrane protein that forms a disulfide-linked homotrimer and anchors at the apical and basal regions of AcMNPV BVs (18, 20, 21). Disruption of the intermolecular disulfide bond (C24-C372) of AcMNPV GP64 has a modest effect on the fusion activity (21). In the determined structure of a low-pH (postfusion) form of AcMNPV GP64, the protein exhibits an extended conformation that is similar to those of other class III viral fusion proteins from herpesviruses, rhabdoviruses, and thogotoviruses (18, 22, 23). Currently, vesicular stomatitis virus (VSV) G is the only class III fusion protein whose prefusion and postfusion structures have been solved in high resolution (24, 25). In low-pH-triggered conformational change, the transition of VSV G from the prefusion to the postfusion state may adopt monomeric stages in which the individual domains may retain their structures (25–27). However, the detailed molecular mechanisms of the conformation change of class III viral fusion proteins are not clear.

In its postfusion structure, GP64 is composed of five domains (domains I to V) (18). Domain I resides at the base of the molecule close to the membrane or viral envelope (18). This domain contains two fusion loops (loop 1 and 2) whose hydrophobic residues are essential for membrane interaction and the progression of distinct stages of membrane fusion (17, 28). At the opposite end of the molecule (top end), domain IV (residues 374 to 407) connects the central α-helical domain III and the C-terminal domain V (Fig. 1A) (18). Different from domain IV of herpesviruses gB and VSV G that are made entirely of β-sheets, domain IV of GP64 is a loop region in which few residues (amino acids [aa] 394 to 398) have not been solved in the structure (18, 24, 29, 30). Similarly, in thogotovirus (Thogoto virus, Dhori virus, and Bourbon virus) glycoproteins (Gps), domain IV in the postfusion structures has not been solved (23, 31). In the prefusion structure of VSV G and a recently determined low-resolution prefusion structure of human cytomegalovirus (HCMV) gB, domain IV is close to domain I (the fusion domain that contains fusion loops) (25, 32). The conformational changes of VSV G and HCMV gB from prefusion to postfusion structures may rearrange domain IV to the top end of the trimer, distant from domain I and the membrane (26, 32). The biological functions of domain IV of class III viral fusion proteins are not clear. In this study, to investigate the roles of domain IV in membrane fusion, we substituted alanine for individual residues and residues that form contacts within domain IV of AcMNPV GP64 and analyzed the effects of these mutations on the ability of GP64 to mediate membrane fusion. The modified gp64 genes were also introduced into a gp64 knockout AcMNPV bacmid, and their effects on virus infection were assessed in transfected and infected cells.

FIG 1.

The conformation and amino acid contacts of domain IV of AcMNPV GP64. (A, B) Postfusion structure of GP64 monomer and trimer (PDB ID 3DUZ) (18). Domains I to V (DI to DV) in the monomer are colored green, red, blue, cyan, and magenta, respectively. The conformation of residues V394 to D398 in domain IV that has not been solved in the structure is predicted and shown in gray. FL1, fusion loop 1; FL2, fusion loop 2. (C) Domain IV and a portion of domain III are shown both in ribbon and surface representation. The conserved amino acids in domain IV among baculovirus GP64s and thogotovirus glycoproteins are displayed in sticks. The E390-G391-R392 motif and the C382-C402 disulfide bond are shown as spheres. (D) The contacts in domain IV of GP64. (E) Amino acid sequence alignments of domain IV of baculovirus GP64s and thogotovirus glycoproteins. A schematic diagram of AcMNPV GP64 is shown at the top of the alignment. SP, signal peptide; PTM, pretransmembrane domain; TM, transmembrane domain. The color code for each domain (DI to IV) is the same as in panel A. Two epitopes in domain III and domain V that are recognized separately by MAbs AcV1 and AcV5 are shown. Virus name abbreviations are those used on the ICTV website (https://talk.ictvonline.org).

RESULTS

Domain IV of GP64 shields the top end of domain III to form a hydrophobic pocket.

To determine the roles of domain IV of GP64, we first performed an analysis of the postfusion structure of GP64 (PDB identification number [ID] 3DUZ) (18) by using the WHAT IF molecular modeling package (http://swift.cmbi.ru.nl/whatif/) and I-TASSER (iterative threading assembly refinement) approach (33) to identify the potential amino acid side chain contacts among residues within domain IV or the contacts between residues within domain IV and residues from other domains. In the postfusion structure of GP64, domain IV is located at the membrane-distal end of the molecule and close to the top end of domain III (Fig. 1A and B). This domain is composed of two adjacent loops that are connected by a conserved disulfide bond (C382-C402) (Fig. 1C and D) (21). At the top of the structure, two loops of domain IV fold back as a lid structure toward the center of the molecule and shield the top end of the α-helix of domain III to form a hydrophobic pocket (Fig. 1C). Within the pocket, the conserved YXEGRW motif (Y388-X-E390-G391-R392-W393) is located at the tip of the lid structure, whereas the conserved hydrophilic, hydrophobic, and charged residues T379, N381, N384, N385, Y388, K389, W393, I401, Q403, and F405 within domain IV form complex network contacts within loop 1 or between two loops, including T379-F405, N381-N385, N381-K389, N381-I401, N381-Q403, N384-Y388, N385-K389, N385-W393, and K389-W393. In addition, at the lateral side of loop 2, D398 forms contacts with S400 and Q401, and D404 forms a contact with S406 (Fig. 1C and D). All these contacts occur in the same monomer of GP64. Sequence alignment indicated that the residues involved in the interactions are highly conserved in GP64 proteins from different baculoviruses and in the GP64 family homologs from thogotoviruses (Fig. 1E). Based on structural analysis of AcMNPV GP64 and sequence conservation among the GP64 family proteins, we used site-directed mutagenesis to individually mutate the conserved residues to alanine or mutate the interacting residues in pairs to alanines and assessed the contribution of each residue or interaction to GP64 function (Table 1). Note that, in predicted structures of domain IV, individual substitutions of alanine for T379, N381, N384, N385, K389, Q401, or I403 did not disrupt the original paired interactions mediated by these residues (Table 1).

TABLE 1.

Phenotypes of virus constructs expressing wild-type protein and alanine substitution mutations in domain IV of AcMNPV GP64

Construct	Contactsa	Trimer	Surface level (%)	Fusion activity (%)	Dye transfer (%)b		Virus infectivityc
					R18	Calcein-AM	Rescued	Virus titer (TCID50/ml)
WT	✓	+	100.0 ± 2.9	97.9 ± 1.3	99.5 ± 0.3	99.2 ± 0.3	+	(2.4 ± 0.3) × 108
N374A	—	+	97.3 ± 2.7	81.0 ± 2.0	ND	ND	+	(1.8 ± 0.0) × 108
P375A	—	+	127.4 ± 1.3	81.8 ± 1.6	ND	ND	+	(1.5 ± 0.0) × 108
P376A	—	+	118.4 ± 1.1	87.9 ± 1.5	ND	ND	+	(2.2 ± 0.6) × 108
T379A	✓	+	96.8 ± 3.0	81.1 ± 3.5	ND	ND	+	(1.1 ± 0.1) × 108
S380A	—	+	135.4 ± 1.8	95.0 ± 0.7	ND	ND	+	(1.9 ± 0.5) × 108
N381A	✓	+	62.0 ± 2.9	97.7 ± 1.3	ND	ND	+	(2.2 ± 0.1) × 108
Y383A	—	+	53.3 ± 1.6	147.2 ± 2.3	ND	ND	+	(1.0 ± 0.2) × 108
N384A	✓	+	60.8 ± 5.7	141.6 ± 1.6	ND	ND	+	(2.0 ± 0.3) × 108
N385A	✓	+	76.3 ± 2.2	117.1 ± 0.2	ND	ND	+	(2.4 ± 0.3) × 108
S386A	—	+	85.1 ± 1.3	108.8 ± 0.8	ND	ND	+	(1.7 ± 0.2) × 108
I387A	—	+	116.8 ± 2.0	81.7 ± 1.6	ND	ND	+	(1.7 ± 0.2) × 108
Y388A	×	+	6.5 ± 0.2	0	0	0	−	NA
K389A	✓	+	166.6 ± 6.8	97.1 ± 0.4	ND	ND	+	(1.5 ± 0.4) × 108
E390A	—	+	6.1 ± 0.5	0	0	0	+	(1.4 ± 0.2) × 106
G391A	—	+	6.7 ± 0.3	0	0	0	+	(2.2 ± 0.0) × 106
R392A	—	+	6.0 ± 0.3	0	0	0	−	NA
W393A	×	+	7.3 ± 0.2	0	0	0	−	NA
V394A	—	+	107.5 ± 0.8	96.9 ± 1.4	ND	ND	+	(1.1 ± 0.1) × 108
N396A	—	+	94.0 ± 3.1	93.4 ± 2.5	ND	ND	+	(1.4 ± 0.2) × 108
T397A	—	+	128.9 ± 2.4	96.2 ± 1.3	ND	ND	+	(5.1 ± 0.7) × 108
D398A	×	+	106.4 ± 2.8	96.8 ± 1.1	ND	ND	+	(4.2 ± 0.6) × 108
S399A	—	+	101.4 ± 5.0	101.6 ± 1.6	ND	ND	+	(5.1 ± 0.7) × 108
S400A	✓	+	93.1 ± 2.3	98.3 ± 1.8	ND	ND	+	(2.2 ± 0.0) × 108
Q401A	✓	+	99.6 ± 7.0	104.5 ± 1.0	ND	ND	+	(6.2 ± 0.8) × 108
I403A	✓	+	100.1 ± 3.7	92.4 ± 1.1	ND	ND	+	(2.4 ± 0.3) × 108
D404A	×	+	88.3 ± 0.9	13.1 ± 0.5	81.2 ± 2.2	80.3 ± 2.2	+	(1.7 ± 0.2) × 108
F405A	×	+	96.7 ± 5.0	91.4 ± 8.8	ND	ND	+	(1.7 ± 0.2) × 108
S406A	✓	+	96.8 ± 0.3	96.6 ± 2.6	ND	ND	+	(2.7 ± 0.7) × 108
N407A	—	+	103.1 ± 4.9	13.7 ± 0.2	82.8 ± 0.9	81.95 ± 1.1	+	(1.1 ± 0.1) × 108
T379/F405A	×	+	99.4 ± 3.3	96.8 ± 1.0	ND	ND	+	(2.4 ± 0.3) × 108
N381/N385A	×	+	16.4 ± 1.3	0	0	0	−	NA
N381/K389A	×	+	8.9 ± 0.3	0	0	0	+	(8.4 ± 0.2) × 106
N381/Q401A	×	+	53.0 ± 2.3	7.6 ± 3.5	70.9 ± 2.4	69.7 ± 2.3	+	(1.3 ± 0.4) × 108
N381/I403A	×	+	51.8 ± 0.4	11.0 ± 0.3	65.7 ± 1.3	64.2 ± 1.1	+	(1.4 ± 0.2) × 108
N384/Y388A	×	+	7.0 ± 1.3	0	0	0	−	NA
N385/K389A	×	+	152.8 ± 0.7	94.1 ± 0.6	ND	ND	+	(3.9 ± 1.0) × 108
N385/W393A	×	+	6.1 ± 0.4	0	0	0	−	NA
K389/W393A	×	+	6.3 ± 0.4	0	0	0	−	NA
D398/S400A	×	+	106.7 ± 1.5	99.7 ± 2.2	ND	ND	+	(2.0 ± 0.3) × 108
D398/Q401A	×	+	76.6 ± 2.5	105.1 ± 4.0	ND	ND	+	(2.7 ± 0.7) × 108
D404/S406A	×	+	82.5 ± 1.9	11.2 ± 2.4	83.8 ± 1.3	82.6 ± 1.1	+	(3.5 ± 0.4) × 108

^aResidue contacts were analyzed in predicted postfusion structures of WT and modified GP64s. ✓, contacts remain; —, no contacts formed; ×, disruption of contacts.

^bND, not done.

^cVirus infectivity was determined by a transfection-infection assay. +, virus infection rescued; −, virus infection aborted; NA, not available.

Substitution in the YXEGRW motif or disruption of the contacts in domain IV affects cell surface expression and localization of GP64.

To examine the effects of alanine substitutions on GP64 expression and cell surface localization, Sf9 cells were transfected with plasmid transiently expressing wild-type (WT) or modified GP64 proteins under the control of the promoter of the AcMNPV ie1 gene. At 36 h posttransfection (p.t.), GP64 proteins in cell lysates were separated by reducing or nonreducing SDS-PAGE and then detected by Western blotting. Under nonreducing conditions, GP64 monomers are covalently linked by intermolecular disulfide bonds to form trimers. As shown by the results in Fig. 2A, the similar trimer forms (trimer I and trimer II) that are typically observed for WT GP64 were detected from all GP64 proteins containing substitutions. For most of the modified GP64 constructs, the sizes and intensities of bands corresponding to the WT and modified GP64 proteins were similar, suggesting that the alanine substitution did not substantially alter the expression or oligomerization of those GP64s. Similar to previous observations (34), a single substitution of alanine at the N-glycosylation site N385 (N385A) or double-alanine mutation of the paired residues containing N385 (N381/N385A, N385/K389A, and N385/W393A) resulted in a slight reduction of the molecular weight (MW) of the modified GP64s. In addition, disrupting the interaction of T379-F405 (F405A and T379/F405A) resulted in a slight increase of the MW of the modified proteins on either reducing or nonreducing gels (Fig. 2A). A possible interpretation is that replacement of F405 or T379 and F405 with alanine may alter the local hydrophobicity of the modified proteins and reduce the amount of SDS binding to the proteins on SDS-polyacrylamide gels.

FIG 2.

Expression and cell surface localization of modified GP64 proteins. (A) Western blot analysis of expression and trimerization of WT and modified GP64 proteins on nonreducing (NR) and reducing (R) gels. Sf9 cells were transfected with plasmid expressing each of the WT or modified GP64s. At 36 h p.t., the expression of GP64 was detected. (B) Relative cell surface levels of GP64 were determined by cELISA. A standard curve generated by transfecting Sf9 cells with decreasing quantities of plasmid expressing WT GP64 is shown on the left end. Cells were transfected with 2 μg of plasmid expressing each modified GP64. Error bars represent standard deviations (SD) of the mean values from three replicates. (C) A diagram of domain IV shows the contacts and residues identified as critical for cell surface localization of GP64. L1, loop1; L2, loop2.

Cell surface expression of the GP64 constructs was assessed by using a cell surface enzyme-linked immunosorbent assay (cELISA) and compared to that detected from WT GP64 (Fig. 2B). Of the 29 single-alanine substitution mutations examined, the cell surface levels for 20 of them were very similar to that of WT GP64, and the cell surface levels for another 4 constructs (N381A, Y383A, N384A, and N385A) were reduced 23.7% to 46.7%. In contrast, severe reduction of the cell surface expression level was detected for constructs with substitutions in 5 highly conserved amino acid positions (Y388A, E390A, G391A, R392A, and W393A). Similarly, poor cell surface expression was also observed for 5 of the 12 double alanine substitution constructs (N381/N385A, N381/K389A, N384/Y388A, N385/W393A, and K389/W393A) (Fig. 2B and C; Table 1).

To confirm the cell surface localization of the GP64 constructs with poor cell surface expression (below 20% of the cell surface expression of WT GP64), transfected Sf9 cells expressing WT or modified GP64 were examined using indirect immunofluorescence with monoclonal antibody (MAb) AcV1, which recognizes the native neutral-pH conformation (prefusion form) of GP64 (35). The results indicated that the GP64 proteins were present at the surface, and AcV1 binding indicated that they were in the native prefusion conformation (data not shown). Together, these results suggest that the YXEGRW motif and the intramolecular contact network mediated by N381, N384, N385, Y388, K389, and W393 are critical for cell surface localization of GP64 (Fig. 2C).

Substitution in the YXEGRW motif or disruption of the contacts in domain IV restricts membrane fusion.

To determine whether mutations in domain IV affect membrane fusion, we first evaluated the fusion activities of the WT and modified GP64s in a cell-cell fusion assay. Because the cell surface level of GP64 is correlated with its fusion activity and cell surface levels varied among different modified GP64 constructs, we initially established a standard curve for cell surface level and fusion activity of WT GP64 by transfecting Sf9 cells with decreasing quantities (0.005 to 2 μg) of plasmid expressing WT GP64 (Fig. 2B, left; data not shown). The fusion (syncytium formation) activity of each modified GP64 construct was determined and then normalized to that of WT GP64 that was localized to the cell surface at an equivalent level. Most of the substitution constructs, including four with significant reductions of cell surface expression (N381A, Y383A, N384A, and N385A constructs), mediated WT expression levels or minor reductions in the levels of fusion (Fig. 2B and Fig. 3A and B). In contrast, single substitutions in the YXEGRW motif, D404, and N407 or paired substitutions (N381/N385A, N381/K389A, N381/Q401A, N381/I403A, N384/Y388A, N385/K389A, N385A/W393A, K389/W393A, and D404/S406A) that disrupted the contacts formed by N381, N384, N385, Y388, K389, W393, Q401, I403, D404, and S406 resulted in loss or dramatic reductions of the fusion activities of the modified GP64 proteins (Fig. 3). Those constructs that impaired the fusion activity of GP64 can be further classified into three groups. (i) The D404A, N407A, and D404/S406A constructs were expressed at the cell surface at nearly WT levels but only induced few and small syncytia and maintained substantially low fusion activity (Fig. 2B and Fig. 3). (ii) Two constructs, the N381/Q401A and N381/I403A constructs, were localized at the cell surface at significantly low levels, and their normalized fusion efficiencies were reduced to 7.6% to 11.0% (Fig. 3; Table 1). (iii) The Y388A, E390A, G391A, R392A, W393A, N381/N385A, N381/K389A, N384/Y388A, N385/W393A, and K389/W393A constructs had poor cell surface localization and no detectable fusion activity (Fig. 2B and Fig. 3; Table 1).

FIG 3.

Cell-cell fusion mediated by WT and modified GP64 proteins. (A) Cell-cell syncytium formation assay. Sf9 cells were transfected with plasmid expressing WT or modified GP64 proteins. At 36 h p.t., syncytium formation was induced by low-pH treatment and photographed using phase-contrast microscopy. Arrows indicate syncytial masses. (B) Analysis of relative fusion activity for each construct. Error bars represent SD of the mean values from three replicates. (C) A diagram of domain IV shows the contacts and residues identified as critical for membrane fusion. L1, loop1; L2, loop2.

To further determine in which step the membrane fusion was restricted for the GP64 constructs with impaired fusion activity, we used a dye transfer assay to detect hemifusion (outer membrane leaflet merger) and pore formation (inner leaflet merger) by observing the transfer of a membrane-specific dye (lipophilic octadecyl rhodamine B chloride [R18]) and a cytosolic dye (calcein-AM) between red blood cells (RBCs) and Sf9 cells as described in prior studies (17, 36). In this assay, RBCs were dually labeled with R18 and calcein-AM and then attached to Sf9 cells that expressed the WT or modified GP64. After triggering under low-pH conditions, the transfer of each dye between RBCs and Sf9 cells was observed and dye transfer efficiencies were calculated. Based on the dye transfer efficiencies, the membrane fusion defect for constructs of GP64 can be classified as either (i) inhibiting the outer membrane leaflet merger (hemifusion) or (ii) reducing the efficiency of pore formation or expansion (Fig. 4A). As shown by the results in Fig. 4B, no dye transfer (neither R18 nor calcein-AM) was observed for the Y388A, E390A, G391A, R392A, W393A, N381/N385A, N381/K389A, N384/Y388A, N385/W393A, and K389/W393A constructs. This suggests that either replacement of the residues in the YXEGRW motif or substitutions that disrupt the contacts formed by N381, N384, N385, Y388, K389, and W393 impaired the ability of GP64 to catalyze the initial merger of the outer membrane leaflets. D404A, N407A, N381/Q401A, N381/I403A, and D404/S406A constructs, which showed substantial reductions of fusion activity (Fig. 3), appeared to impair membrane fusion at a different step. For these constructs, we observed the transfer of both R18 and calcein-AM dyes between RBCs and Sf9 cells (Fig. 4B), and dye transfer efficiencies were relatively high, ranging from 64.2% to 83.8% (Fig. 4C). Thus, D404A, N407A, N381/Q401A, N381/I403A, and D404/S406A constructs induced complete fusion pore formation, but the fusion pore may not be efficiently expanded. Combined, these results suggest that, while most of the residues in domain IV were not necessary for membrane fusion activity of GP64, the YXEGRW motif (Y388-X-E390-G391-R392-W393), N407, and the contacts formed by N381, N384, N385, Y388, K389, W393, Q401, I403, D404, and S406 were critical for the initial membrane merger or fusion pore expansion.

FIG 4.

Hemifusion and pore formation mediated by fusion-deficient GP64 proteins. (A) Schematic of the hemifusion and pore formation assay. Sf9 cells were transfected with plasmid expressing WT or modified GP64. At 36 h p.t., the cells were incubated with dual-dye (R18 and calcein-AM)-labeled RBCs. After removing the unattached RBCs, the cells were treated with PBS (pH 5.0) for 3 min to induce fusion. Transfer of R18 between RBCs and Sf9 cells indicates the merger of outer membrane leaflets of two kinds of cells (hemifusion), whereas transfer of calcein-AM between RBCs and Sf9 suggests the inner membrane leaflet merger of the cells (pore formation). (B) Hemifusion and pore formation assay. Sf9 cells were transfected with 2 μg or 0.005 μg plasmid expressing WT GP64 (WT2 or WT0.005) or with 2 μg of plasmid expressing each mutated GP64. At 36 h p.t., Sf9 cells were incubated with dual-dye-labeled RBCs, and the dye transfers between RBCs and Sf9 cells were observed and photographed. (C) The dye transfer efficiency was estimated by the ratio between the number of Sf9 cells to which R18 or calcein-AM was transferred and the number of Sf9 cells with bound RBCs. Error bars represent SD of the mean values from three replicates.

Substitution in the YXEGRW motif or disruption of the contacts in domain IV alters the conformational change of GP64.

To determine whether the membrane fusion defect for those constructs that could not induce hemifusion and fusion pore expansion resulted from a detectable effect on the low-pH-triggered conformational change in GP64, the binding of the conformation-specific MAb AcV1 on the cell surface was examined by cELISA under various pH conditions (Fig. 5). As is known, the AcV1 antibody recognizes only the neutral-pH (or prefusion) conformation of GP64, and the AcV1 epitope is lost upon the low-pH-triggered conformational change (35). To monitor the potential change in the cell surface level for each GP64 construct corresponding to low-pH treatment, the transfected cells expressing WT or modified GP64 proteins were also analyzed by cELISA with MAb AcV5, which binds to the denatured GP64 (37). As expected, treatment of the transfected cells by successively decreasing the pH of the phosphate-buffered saline (PBS) from 7.0 to 4.5 did not significantly change the cell surface level of the WT or each modified GP64 (data not shown). Low-pH treatments resulted in a dramatic loss of AcV1 binding to WT GP64 (Fig. 5, WT). For the modified GP64, the binding activity of AcV1 under various pH conditions could be subdivided into three categories. (i) The constructs (Y388A, E390A, G391A, R392A, W393A, N381/N385A, N381/K389A, N384/Y388A, N385/W393A, and K389/W393A constructs) that were detected at significantly low levels at the cell surface using MAb AcV5 (Fig. 2B; data not shown) showed significantly low binding to AcV1 under neutral-pH or moderately low-pH conditions. However, the binding activities of AcV1 for these constructs appeared to increase substantially after treatment with PBS at pH 4.5 (Fig. 5A and B). (ii) Two constructs, the N381/Q401A and N381/I403A constructs, that were present at the cell surface at dramatically low levels (Fig. 2B; data not shown) showed relatively high AcV1-binding activities under neutral-pH conditions and loss of AcV1 binding at various low-pH values (Fig. 5C). (iii) Three modified GP64 constructs, the D404A, D404/S406A, and N407A constructs, that were expressed at the cell surface at levels similar to that of WT GP64 (Fig. 2B; data not shown) also showed WT or slightly lower levels of AcV1-binding activities under neutral-pH conditions (Fig. 5D). Upon the low-pH trigger, these constructs showed a very modest pattern of loss of AcV1 binding as pH values were lowered from 7.0 to 5.0. Even after treatment with PBS at pH 4.5, the AcV1-binding activities of these constructs were significantly higher than that of WT GP64 (Fig. 5D). These results suggest that even though membrane fusion is affected at two distinct stages for different sets of constructs (Fig. 4), the low-pH-triggered conformational change appears to be dramatically affected in both cases. Thus, these mutations (except for N381/Q401A and N381/I403A) may alter the neutral-pH conformation of GP64 and generate the structural or energy obstacle (note that the refolding of the modified proteins from a prefusion to a postfusion conformation may require higher energy) to prevent the low-pH-triggered prefusion-to-postfusion conformational change of the modified GP64.

FIG 5.

Analysis of low-pH-triggered conformational change of the fusion-deficient GP64 constructs. The conformational change of cell surface-localized GP64 proteins upon low-pH trigger was measured by cELISA using MAb AcV1. Sf9 cells were transfected with 2 μg of plasmid for WT or each modified GP64. Each value represents the mean result from triplicate transfections and is normalized to that of Sf9 cells transfected with plasmid expressing WT GP64 at pH 7.0. Error bars represent SD.

Substitution in the YXEGRW motif or disruption of the contacts in domain IV affects infectious virus production.

To determine whether the introduction of alanine in domain IV to replace specific amino acids or paired residues affects infectious AcMNPV production, we constructed recombinant AcMNPV bacmids expressing WT GP64 or each modified GP64 under the control of the native promoter of the AcMNPV gp64 gene. At 96 h p.t., the cell culture supernatants were used to infect Sf9 cells. At 96 h postinfection (p.i.), the infectious virus titers were determined, and the levels of infectious BV production by the recombinant viruses that expressed modified GP64 proteins with no significant reduction or a minor reduction of fusion activities were similar to that of the virus expressing WT GP64 (Table 1). For constructs (D404A, N407A, N381/Q401A, N381/I403A, and D404/S406A constructs) with dramatic reductions of fusion activities (Fig. 3 and 4), we also observed a WT level of infectious virus production (Table 1). In contrast, no measurable infectious BVs were detected for each of the constructs expressing Y388A, R392A, W393A, N381/385A, N384/Y388A, N385/W393A, or K389/W393A (Table 1). It is expected that those modified GP64 proteins could not induce membrane fusion (Fig. 3 and 4). However, infectious BVs with about 30- to 180-fold reductions in their levels of production were detected for constructs expressing the E390A, G391A, and N381/K389A membrane fusion-inhibiting mutations in a cell-cell fusion assay (Fig. 3 and 4; Table 1).

To further assess the potential effects of the modified GP64 proteins with impaired fusion activities on infectious BV generation, we performed one-step and multistep virus growth curve analyses to evaluate virus replication kinetics. As shown by the results in Fig. 6A, no significant difference in infectious BV production was observed for viruses expressing modified GP64 protein with D404A, N407A, N381/Q401A, N381/I403A, and D404/S406A mutations or WT GP64 (V^D404A, V^N407A, V^N381/Q401A, V^N381/I403A, V^D404/S406A, and V^WT) at any time point p.i. In contrast, the infectious virus production for V^N381/K389A at each time point p.i. in both growth curves was significantly lower than that for the virus expressing WT GP64 (Fig. 6B). Similar but more dramatic reductions of virus production were observed for V^E390A and V^G391A (Fig. 6B). Thus, replacement of D404 or N407 or mutations disrupting the N381-Q401, N381-I403, and D404-S406 contacts had no apparent effect on infectious BV production. Rescue of the gp64 knockout virus infectivity by E390A, G391A, or N381/K389A, even though it occurred at a substantially low rate, suggests that these constructs may fold properly to perform the conformational change and restore the fusion activity in virus-infected cells.

FIG 6.

Growth kinetics of viruses expressing fusion-deficient GP64 constructs. Sf9 cells were infected with virus expressing the WT or each modified GP64 at an MOI of 5 (A, B, left) or at an MOI of 0.1 (A, B, right). At the indicated time points p.i., virus titers were determined by TCID₅₀ assays. Error bars represent SD of the mean values from three replicates.

Constructs E390A, G391A, and N381/K389A restore membrane fusion in virus-infected cells.

Because the E390A, G391A, and N381/K389A constructs that inhibited hemifusion in transiently expressing cells rescue the infectivity of gp64 knockout AcMNPV, we asked whether these proteins could induce membrane fusion in virus-infected cells. To examine this question, Sf9 cells were infected with virus expressing E390A, G391A, N381/K389A, or WT GP64, and then the expression and fusion activities of these proteins were examined. As shown by the results in Fig. 7, these constructs were expressed and oligomerized into trimers like WT GP64 (Fig. 7A). The cell surface levels of these modified GP64s ranged from 25.2% to 34.4% of the level for WT GP64, dramatically lower (Fig. 7B). Membrane fusion analysis showed that E390A, G391A, and N381/K389A constructs could only induce a few syncytia in virus-infected cells, and the percentages of cells with syncytia induced by these constructs were about only 5.5% to 12.8% (Fig. 7C and D).

FIG 7.

Effects of E390A, G391A, and N381/K389A mutations on membrane fusion and conformational change of the modified GP64 in virus-infected cells. Sf9 cells were infected with virus expressing either WT GP64 or E390A, G391A, or N381/K389A GP64 (MOI = 5). At 48 h p.i., the cells were subjected to analysis. (A) Western blot analysis of the expression of GP64 proteins under nonreducing (NR) and reducing (R) conditions. (B) Relative cell surface levels of GP64 proteins were determined by cELISA. (C, D) Cell-cell fusion assay. The infected cells were incubated with PBS (pH 5.0) for 3 min and then cultured at 27°C for 4 h. Then, the syncytium formation was observed and photographed (C) and the percentages of cells in syncytial masses (≥5 nuclei) were calculated (D). (E, F) Hemifusion and pore formation assay. The infected cells were incubated with dual-dye-labeled RBCs. After attachment, the cells were incubated with PBS (pH 5.0), and then the dye transfers between RBCs and Sf9 cells were observed and photographed (E) and the dye transfer efficiency was estimated as described above (F). (G) Low-pH-triggered conformational change of WT and modified GP64. The infected cells were incubated in PBS at various pH values (pH 4.5 to 7.0) and then fixed and analyzed by cELISA using MAb AcV1. Error bars represent SD of the mean values from three replicates.

To dissect in which step the membrane fusion activity for the modified GP64s was impaired, we used the dye transfer assay, as mentioned above, to detect hemifusion and pore formation induced by these constructs in virus-infected cells. As shown by the results in Fig. 7E, we observed transfer of both of the membrane and cytosolic dyes (R18 and calcein-AM) between RBCs and infected Sf9 cells induced by E390A, G391A, and N381/K389A constructs under low-pH conditions (pH 5.0), even though the dye transfer efficiencies for these proteins were significantly lower than that of WT GP64 (Fig. 7F). Using MAb AcV1, we also analyzed the antibody-binding activity for each of the constructs under various pH conditions (pH 7.0 to 4.5). The results showed that, in comparison with the loss of AcV1-binding activity for WT GP64 in response to low-pH treatments, the AcV1-binding activities of E390A, G391A, and N381/K389A constructs were substantially lower under neutral-pH conditions, and the binding activities for these constructs were not affected upon low-pH treatment at pH 5.7 but decreased slightly corresponding to more acidic treatments (pH 5.5 to 4.5). It is worth noting that the AcV1-binding activities for E390A, G391A, and N381/K389A constructs at pH 4.5 were significantly higher than that for WT GP64 (Fig. 7G). cELISA analysis with MAb AcV5 indicated that the cell surface levels of these proteins were not significantly changed upon various pH treatments (pH 7.0 to 4.5) (data not shown). Together, these results indicate that, upon the low-pH trigger, E390A, G391A, and N381/K389A constructs may partially refold from the prefusion to postfusion states and induce membrane fusion with low efficiency in virus-infected cells.

E390A, G391A, and N381/K389A constructs affect virus entry but not egress.

Since the E390A, G391A, and N381/K389A constructs could induce membrane fusion in virus-infected cells (Fig. 7), we next asked whether the impaired fusion activity for these proteins has a negative effect on virus binding to or penetration into cells or progeny virus egress and whether that in turn may result in slow virus growth kinetics (Fig. 6). First, Sf9 cells were incubated with the purified virus expressing WT or modified GP64 proteins at low temperature, and then the cell-binding efficiency of each virus was determined by measuring the virus genomic DNA using quantitative real-time PCR (qPCR) (Fig. 8A, panel a). As shown by the results in Fig. 8A, the binding efficiency of each of the viruses V^E390A-GFP, V^G391A-GFP, and V^N381/K389A-GFP was similar to that of V^WT-GFP, suggesting that mutation of E390, G391, or N381-K389 has no apparent effect on virus binding (Fig. 8A, panel b). To further determine virus entry, Sf9 cells were incubated with the purified virus at 4°C for binding. Then, the culture temperature was raised to 27°C to allow the virus to internalize into cells. At 1 h postinternalization, any virions remaining at the cell surface were inactivated and removed by treatment with citrate buffer (pH 3.1) as described in prior studies (15). Total DNA was extracted from the cells, and viral genomic DNA was quantified by qPCR. As shown by the results in Fig. 8A, the amounts of internalized V^E390A-GFP, V^G391A-GFP, or V^N381/K389A-GFP viruses were significantly lower than that of V^WT-GFP (Fig. 8A, panel c). Together, these results suggest that the single substitution of E390 or G391 or disruption of the N381-K389 contact affects the efficient entry of BVs.

FIG 8.

Effects of E390A, G391A, and N381/K389A mutations on BV binding and internalization (A) and infectious BV egress (B). (A) (a) Schematic representation of BV binding and internalization assays. Sf9 cells were incubated with virus expressing WT GP64 or E390A, G391A, or N381/K389A GP64 (MOI = 5) at 4°C. (b) After removing the virus inoculum and washing the cells, one set of the cells was lysed, and virus binding efficiency was determined by measuring viral genomic DNA using qPCR. (c) The other set of cells was cultured at 27°C for 1 h, and then the virus internalization was examined by qPCR of viral genomic DNA. (B) (a) Schematic representation of BV egress assay. Sf9 cells were transfected with 0.5 to 4 μg bacmid DNA expressing WT GP64 or with 4 μg bacmid DNA expressing either E390A, G391A, or N381/K389A GP64. (b, c) At 24 h p.t., the relative cell surface levels of GP64 proteins were determined by cELISA (b) and infectious virus production was measured by TCID₅₀ assays (c). Dashed lines indicate similar cell surface levels or infectious virus production of different constructs. Error bars represent SD of the mean values from three replicates. ***, P < 0.0005 (by paired two-tailed t test).

Since GP64 is necessary for efficient budding and release of progeny nucleocapsids (19), we next examined whether the E390A, G391A, and N381/K389A mutations affect virus egress. To circumvent the negative effect of these constructs on virion entry, we transfected Sf9 cells with recombinant AcMNPV bacmids encoding the WT or modified GP64 and two reporters (green fluorescent protein [GFP] and β-glucuronidase [β-Gluc]). The cell surface levels of GP64s and infectious virus production were determined at a narrow time period after transfection (24 h p.t.) (Fig. 8B, panel a). Also, the GFP fluorescence and β-Gluc activities were measured (at 24 h p.t.) to ensure that transfection efficiencies and initiation of infection by the different bacmids were equivalent. At 24 h p.t., the percentages of GFP-positive cells in transfections with the same amount of DNA (4 μg/well) for the different bacmids (V^WT-GFP, V^E390A-GFP, V^G391A-GFP, and V^N381/K389A-GFP) ranged from 30.1% to 32.5%, and the activities of β-Gluc in cell lysates from the different bacmid-transfected cells were similar (data not shown), suggesting that transfection efficiencies were equivalent for all bacmids and that, in each case, the bacmid initiated an infection that progressed into the late phase of the infection cycle. Since the cell surface levels of E390A, G391, and N381/K389A constructs were significantly lower than that of WT GP64 in plasmid-transfected cells or in virus-infected cells (Fig. 2 and 7), we generated a range of decreasing GP64 cell surface levels by transfecting Sf9 cells with decreasing quantities of the bacmid expressing WT GP64 (Fig. 8B, panel b). In bacmid-transfected cells, the cell surface levels of E390A, G391A, and N381/K389A constructs were reduced by approximately 62.3% to 78.4%, and similar reductions were observed from transfection with 0.5 μg or 1.0 μg of the bacmid expressing WT GP64 (Fig. 8B, panel b). Infectious BV production was then measured from the same transfected cells (Fig. 8B, panel c). Comparisons of virus production in cells expressing the modified GP64s with that from WT GP64-expressing cells (in which the cell surface levels of GP64 were similar; see dashed lines in Fig. 8B, panels b and c) showed that there was no significant reduction of infectious BV production in cells expressing the E390A, G391A, or N381/K389A construct compared with the level of BV production from cells expressing a similar level of WT GP64 (Fig. 8B, panel c). Together, these results suggest that the E390A, G391A, and N381/K389A constructs affect efficient virus entry but not egress of progeny BVs.

DISCUSSION

During the entry of enveloped viruses, fusion of the viral envelope and cellular membrane mediated by viral fusion proteins is required for delivery of the viral genome into host cells (1, 3, 4). Based on the postfusion structure that has been determined, GP64 is grouped with herpesvirus gB, rhabdovirus G, and thogotovirus Gp proteins as class III viral fusion proteins (18, 22, 23, 26). In its postfusion structure, GP64 is composed of five domains (18). Domain IV is located at the membrane-distal end of the molecule and coordinates with the top end of domain III to form a hydrophobic pocket (Fig. 1). Prior studies showed that the disulfide bond C382-C402 in domain IV of GP64 is essential for the protein’s stability (21). In the current study, we extended the observation by using site-directed mutagenesis to demonstrate that the intramolecular contacts (N381-N385, N381-K389, N381-Q401, N381-I403, N384-Y388, N385-W393, K389-W393, and D404-S406), the Y388-X-E390-G391-R392-W393 motif, and the C-terminal N407 in domain IV of AcMNPV GP64 are critical for low-pH-triggered conformational change and fusion pore expansion during membrane fusion.

In alanine-scanning mutagenesis, the single replacement of N407 or disruption of the D404-S406 contact had no significant effect on the cell surface expression of the modified GP64s but resulted in a dramatic reduction of fusion activities. Dye transfer assays indicated that these alanine substitution constructs could efficiently induce fusion pore formation but appeared to be defective in fusion pore expansion (Fig. 4). D404, S406, and N407 reside at the C terminus of domain IV and connect with the loop region of domain V (Fig. 1). Prior studies indicated that the interaction of the loop region of domain V with the central coiled-coil in domain III of herpes simplex virus 1 (HSV-1) gB promotes the gB structural transition to a postfusion conformation (38). In response to a low-pH trigger, the D404A, D404/S406A, and N407A constructs showed dramatically less efficient transitions from prefusion to postfusion states (as detected by AcV1 binding) (Fig. 5), suggesting that the D404-S406 contact and N407 may play a role in the low-pH-triggered conformational change of GP64.

Similarly, the contact network formed by N381, N385, K389, Q401, and I403 (N381-N385 and N381-K389 in loop 1, and N381-Q401 and N381-I403 between loop 1 and loop 2) in the middle region of domain IV was also identified as playing a role in the refolding or stability of GP64. Double-alanine replacement of N381 and Q401 or N381 and I403 had a dramatic effect on the fusion activities of the modified GP64s. Both constructs (N381/Q401A and N381/I403A) underwent the low-pH-triggered conformation change and induced pore formation but seemed to be defective in fusion pore expansion (Fig. 4). In domain IV, the N381-Q401 and N381-I403 contacts surround the intramolecular disulfide bond C382-C402 (Fig. 1D), which is essential for GP64 folding and stability (21). In the well-established hemifusion-pore membrane fusion model, conformational change triggered by low pH or other mechanisms promotes the insertion of fusion peptides or fusion loops of viral fusion proteins into a target membrane. Then, the fusion proteins fold back toward themselves and form energy-stable postfusion structures to drive fusion pore expansion (3, 4). The negative effect of disrupting one of the two contacts N381-Q401 and N381-I403 on GP64-mediated fusion may be interpreted as the substitution attenuating or destabilizing the disulfide bond C382-C402, which in turn affects the modified protein’s refolding or stability. For another two constructs, the N381/N385A and N381/K389A constructs, the mutations resulted in a membrane fusion defect at the initial stage (hemifusion). Low-pH treatments appear to alter the prefusion structure of these modified GP64s (Fig. 5A), suggesting that the N381-N385 and N381-K389 contacts play a role in the transition from the prefusion to the postfusion state. Surprisingly, the virus expressing N381/K389A was able to replicate, although it was very inefficient in producing infectious BVs. In infected cells, the N381/K389A construct induced low syncytium formation. It seems that, in infected cells, the N381/K389A mutant may undergo partially proper folding and induce fusion pore formation but affect pore expansion. The impact of N381/K389A on virus entry but not on virus egress further suggests the role of N381-K389 in membrane fusion.

In addition to forming the necessary N381-N385 and N381-K389 contacts in loop 1, residues N385 and K389 also form contacts with W393, which resides at the bottom of loop 2 in domain IV (Fig. 1D). Disruption of these contacts (W393A, N385/W393A, or K389/W393A) resulted in low expression of the modified GP64 at the cell surface and defective membrane fusion. Modelled-structure analyses indicated that disrupting the N385-W393 or K389-W393 contact resulted in the top region of loop 2 being shifted to the center of the molecule by about 2.5 Å and a shift of the backbone and side chains of E390-G391-R392, which reside at the base of loop 1, toward the outside of the molecule by about 2 Å (data not shown), suggesting that the N385-W393 and K389-W393 contacts are critical for maintaining the conformation of domain IV. Similar to the negative effect observed for these constructs, disruption of the N384-Y388 contact at the top region of loop 1 (Y388A or N384/Y388A) or the single replacement of E390, G391, or R392 also caused a dramatic reduction of the cell surface level of the modified GP64 and defective membrane fusion at the initial stage (hemifusion) (Fig. 2 and 4). For all these constructs, the low-pH trigger seems to alter the prefusion conformation (Fig. 5A), suggesting that the N384-Y388, N385-W393, and K389-W393 contacts and E390, G391, and R391 also play a role in the conformational change of GP64.

Similar to the replication observed for virus expressing N381/K389A, virus expressing E390A or G391A also replicated inefficiently. The defect in virus replication seems to result from the reduced efficiency of virus entry (Fig. 8A). Although these two constructs also underwent low-pH-triggered conformational change in virus-infected cells, the significant reduction of AcV1-binding activities under low-pH conditions suggests they have a role in the conformational change of GP64. In addition, in the postfusion structure of GP64, E390-G391-R392 is located at the tip of the lid structure of the hydrophobic pocket and forms the turn between loop 1 and loop 2 (Fig. 1C and D). Previous studies have demonstrated that an arginine-glycine-aspartic acid (RGD) motif in some viral proteins binds integrin to promote virus entry or activate the integrin signal pathway to facilitate the intracellular trafficking of viral proteins (39). Intriguingly, replacement of E390 with aspartic acid (E390D) had no effect on the expression and fusion activity of the modified GP64 and virus infection, whereas replacement of R392 with histidine or lysine (R392H or R392K) abolished the cell surface expression and fusion activity of GP64 (data not shown). The receptor-binding domain of GP64 has been mapped within the N-terminal 160 aa (12). Therefore, the conserved EGR motif of GP64 (DGR in thogotovirus Gp) may be unrelated to the plasma membrane binding of GP64 but may mediate the interaction of GP64 with some cellular factors (such as integrin) to promote the trafficking of GP64 to the cell surface or proper folding.

Among class III viral fusion proteins, VSV G is the only one for which high-resolution structures of prefusion and postfusion states have been solved (24, 25). In the low-pH-triggered conformational change of VSV G, the prefusion-to-postfusion structural transition proceeds through monomeric stages of G with the domains mostly retaining their structures (26, 27). A similar conformational change may be adopted by herpesvirus gB, according to a recently determined low-resolution prefusion structure of HCMV gB (32). In the prefusion structures of VSV G and HCMV gB, domain IV is located at the lateral side of the molecules and proximal to the middle or upper regions of domain I, which contains fusion loops. However, in postfusion structures, domain IV in both proteins is rearranged to the top end of the molecule and distal from domain I, which resides at the bottom of the molecule (24, 25, 32, 40). Structural comparison and phylogenetic analyses suggest that the class III viral fusion proteins have evolved from a common ancestor (23). In addition, although class III viral fusion proteins from different viral families exhibit distinct characteristics in activation and interaction with the target membrane, they share similar molecular architecture in the postfusion state (23, 26). Intriguingly, the negative effect of mutations in domain IV on membrane fusion (i.e., defects in hemifusion and fusion pore expansion) in the current study is similar to that observed in prior studies for mutations in fusion loops of AcMNPV GP64 (17, 28). Based on the prefusion structure of VSV G (25), we generated a predicted prefusion structure of GP64 using the I-TASSER approach (Fig. 9A) (33) and proposed a model of roles of domain IV in the prefusion-to-postfusion conformational change of GP64 (Fig. 9B). In this model, we hypothesize that under neutral-pH conditions, the cell surface- or viral-envelope-localized GP64 may adopt a compact prefusion conformation, in which domain IV may be proximal to domain I. Upon a low-pH trigger, domain I extends outside and exposes fusion loops that interact with the target membrane. Following the formation of the long central coiled-coil in domain III (and perhaps the interaction of domains III, IV, and V) and the extension of domain V, domain IV is pushed to the top end of the molecule. These structural rearrangements bring the two membranes close to fusion (Fig. 9B).

FIG 9.

Proposed roles of domain IV in membrane fusion and low-pH-triggered conformational change of AcMNPV GP64. (A) The predicted prefusion structure of AcMNPV GP64 (domains I to IV) that was generated by the I-TASSER approach using the prefusion structure of VSV G (PDB ID 5I2S) (25) as the template (left), the postfusion structure of GP64 (middle), and the critical residues and contacts within domain IV that are involved in GP64 conformation change and membrane fusion (right). Domains I to V (DI to DV), PTM (pretransmembrane domain), and TM (transmembrane domain) are colored in green, red, blue, cyan, magenta, orange, and olive, respectively. FL1, fusion loop 1; FL2, fusion loop 2. Residues and contacts are indicated as magenta spheres and sticks, respectively. (B) Roles of domain IV in prefusion-to-postfusion conformation change of GP64. In the prefusion state (neutral-pH conditions), GP64 is anchored on the viral envelope (or the cell surface) in a compact form. Upon low-pH trigger, the structure of GP64 opens to expose fusion loops that reside at the tip of domain I (step 1). Following the extension of domain V and the formation of the long α-helix in domain III, fusion loops target the cellular membrane and interact with it (step 2). Refolding of GP64 that may be driven by the formation of long coiled-coils in the center of domain III and the interaction of domains III, IV, and V promotes the close proximity of two membranes and yields the outer membrane leaflet merger (hemifusion, step 3a) and the inner membrane leaflet merger (fusion pore formation and expansion, step 3b). The potential negative effects of mutations in domain IV on membrane fusion and conformational change of GP64 are indicated.

In conclusion, we examined the functional roles of domain IV in baculovirus GP64-mediated membrane fusion. Each of the intramolecular contacts and residues within domain IV was analyzed for its effects on GP64-mediated membrane fusion and virus infection. The intramolecular contacts formed by the conserved residues N381, N384, N385, Y388, W393, Q401, I403, D404, and S406, the Y388-X-E390-G391-R392-W393 motif that made the turn between loop 1 and loop 2, and residue N407 at the C terminus of domain IV were identified as important for low-pH-triggered conformational change and/or fusion pore expansion of GP64. These data will be more definitively fit into the structure transition when the prefusion conformation of GP64 becomes available, and they highlight the refolding mechanism of class III viral fusion proteins during membrane fusion.

MATERIALS AND METHODS

Cells and transfections.

Spodoptera frugiperda Sf9 cells were grown in TNMFH medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum (Gibco) and maintained at 27°C. Cells in 6-well plates (1 × 10⁶ cells per well) or 12-well plates (2 × 10⁵ cells per well) were transfected with plasmid or bacmid DNA using a CaPO₄ precipitation method (16).

Mutagenesis and construction of plasmids and bacmids.

Site-directed mutagenesis was performed with overlap PCR. The C-terminal 882-bp fragments of AcMNPV gp64 encoding alanine substitution mutations were amplified using pBiepA (36) as the template. PCR products were digested with NotI and EcoRI and then inserted into pBiepA to generate plasmid Pie1-XpBlue (X represents a specific mutation of GP64). To generate AcMNPV bacmids encoding the substitution constructs of GP64, the promoter of AcMNPV gp64 was amplified by PCR using the bacmid DNA and then digested with SacI and XbaI and inserted into GFPpFB (41) to generate the plasmid Pgp64-GFPpFB. Then, the fragment encoding WT or modified GP64 that was isolated from pBiepA or Pie1-XpBlue with XbaI and EcoRI was separately inserted into Pgp64-GFPpFB to generate the pFastbac plasmid Pgp64-XpFB. To generate AcMNPV bacmids expressing certain constructs of GP64 (WT, E390A, G391A, or N381/K389A) and GFP, a cassette containing the AcMNPV ie1 promoter, gfp, and the poly(A) sequence of AcMNPV gp64 were isolated from GFPpBlue (42) with KpnI and HindIII and inserted into Pgp64-XpFB to generate the pFastbac plasmid Pgp64-X-Pie1-GFPpFB. The resulting pFastbac constructs were each cloned into the polyhedrin gene locus of an AcMNPV gp64-null bacmid (19) by Tn7-mediated transposition (43). All constructs were verified by DNA sequencing. The plasmid and bacmid DNAs used for transfection were isolated with the HiPure midiprep kit (Invitrogen).

cELISA.

The cell surface-localized GP64 proteins were detected using cELISA as previously described (36). Briefly, Sf9 cells in 12-well plates were transfected with plasmid expressing WT GP64 or modified GP64 or infected with virus expressing WT or modified GP64 proteins. At 36 h p.t. or 48 h p.i., the cells were fixed with 0.5% glutaraldehyde and the relative levels of cell surface-localized GP64 were detected by using anti-GP64 MAb AcV5 (Santa Cruz Biotechnology), goat anti-mouse IgG conjugated with β-galactosidase (SouthernBiotech), and the substrate chlorophenol red-β-d-galactopyranoside (CPRG; Roche Life Science). To determine the low-pH-triggered conformational change of the cell surface-localized GP64 proteins, cELISA was performed with anti-GP64 MAb AcV1 (Santa Cruz Biotechnology). For this assay, transfected or infected Sf9 cells (at 36 h p.t. or 48 h p.i.) were incubated in PBS adjusted to different pH values (4.5 to 7.0) for 20 min and then fixed with 2% paraformaldehyde for 30 min. The following steps in cELISA were similar to those used for MAb AcV5.

Immunofluorescence analysis.

Sf9 cells in 12-well plates were transfected with 2 μg of plasmid expressing WT or modified GP64. At 36 h p.t., the cells were fixed with 2% paraformaldehyde for 30 min. The cell surface-localized GP64 proteins were visualized by indirect immunofluorescence using MAb AcV1 and Alexa Fluor 488-conjugated goat anti-mouse IgG (Thermo Fisher Scientific) as described previously (36).

Fusion assay.

The fusion activity for modified GP64 proteins was measured as previously described (36). Briefly, Sf9 cells in 12-well plates were transfected with plasmids (2 μg DNA for each plasmid expressing the modified GP64 or 0.005 to 2 μg DNA for plasmid expressing WT GP64) or infected with virus encoding WT or modified GP64 (multiplicity of infection [MOI] = 5). At 36 h p.t. or 48 h p.i., the cells were incubated in PBS at pH 5.0 for 3 min. After washing once with PBS at pH 7.4, the cells were maintained in TNMFH medium at 27°C for 4 h. Then, the transfected or infected cells were fixed with methanol and stained with 0.1% eosin Y and 0.1% methylene blue. The number of nuclei found in syncytia that contained at least five nuclei was scored. The ratio of nuclei in syncytial masses to those in a field was calculated and then normalized to the ratio for WT GP64 that was localized to the cell surface at equivalent levels.

Hemifusion and pore formation assay.

The hemifusion and pore formation assay was performed as previously described with minor modifications (36). Briefly, rabbit red blood cells (RBCs) were collected by centrifugation of the fresh blood at 500 × g, 4°C for 5 min. After washing twice with PBS (pH 7.4), the cells were dually labeled with R18 and calcein-AM (Invitrogen). At 36 h p.t. or 48 h p.i., the transfected or infected Sf9 cells were washed twice with PBS (pH 7.4) and then incubated with the labeled RBCs for 20 min. After removing the unbound RBCs, the cells were washed three times with PBS (pH 7.4). Then, the cells were incubated in PBS at pH 5.0 for 3 min. After washing twice with PBS (pH 7.4), the cells were maintained in TNMFH medium at 27°C for 20 min. Then, the transfer of fluorescence between RBCs and Sf9 cells was photographed using epifluorescence microscopy (Nikon Eclipse Ti). Five fields were randomly selected to score for dye transfer. The efficiency of hemifusion or pore formation was estimated by the ratio between Sf9 cells to which R18 or calcein-AM, respectively, was transferred and RBC-bound Sf9 cells.

Transfection-infection assay.

Sf9 cells in 6-well plates were transfected with 4 μg of each of the recombinant AcMNPV bacmids encoding WT or modified GP64. At 96 h p.t., the supernatants were collected and clarified by centrifugation (3,000 × g, 10 min) and then used to infect a new monolayer of Sf9 cells. At 96 h p.i., the supernatants were harvested and the infectious virus titers were determined by a 50% tissue culture infection dose (TCID₅₀) assay.

Virus growth curve analysis.

Sf9 cells in 12-well plates were infected with virus expressing WT or modified GP64 at an MOI of 5 or 0.1 for 1 h. After removing the viral inoculum, the cells were washed once with TNMFH medium and then incubated at 27°C. At different time points (24 to 120 h) p.i., the cell culture supernatant was collected and infectious virus titers were determined by TCID₅₀ assays.

BV purification.

BVs were purified as previously described with minor modifications (41). Briefly, Sf9 cells were infected with virus expressing WT or modified GP64 (E390A, G391A, or N381/K389A) with an MOI of 1. At 96 h p.i., infected-cell supernatants were centrifuged at 28,000 rpm, 4°C for 90 min through a 25% (wt/vol) sucrose cushion. Pellets were resuspended and overlaid onto a 30% to 55% (wt/vol) continuous sucrose gradient and centrifuged at 28,000 rpm, 4°C for 90 min. Then, the virus fraction was diluted with TNMFH medium and centrifuged again under the same conditions. Virus pellets were resuspended in TNMFH medium, and the potential contamination was removed by filtration. Virus titers were determined by TCID₅₀ assays.

BV binding and internalization assay.

Sf9 cells in 12-well plates were prechilled at 4°C for 30 min and then incubated with purified virus expressing WT or modified GP64 (E390A, G391A, or N381/K389A) (MOI = 5) at 4°C for 1 h. After removing the virus inoculum, cells were washed twice with cold TNMFH medium. Then, one set of the cells was collected and total DNA was extracted at 4°C using the DNeasy blood and tissue kit (Qiagen). Another set of the cells was incubated at 27°C for 1 h to allow virus internalization. Then, the cells were treated with citrate buffer (40 mM sodium citrate, 135 mM NaCl, and 10 mM KCl, pH 3.1) for 1 min to inactivate and remove noninternalized viruses as described previously (15), and then total DNA was extracted. Viral genomic DNA in total DNA extracts isolated from two sets of the infected cells was measured by qPCR (CFX96 Touch real-time PCR system; Bio-Rad). Each PCR mixture contained 10 μl SYBR premix Ex Taq II (TaKaRa), 2.5 μM each primer (ODV-e56F, 5′-GATCTTCCTGCGGGCCAAACACT-3′, and ODV-e56R, 5′-AACAAGACCGCGCCTATCAACAAA-3′), and 1 ng of the DNA. A standard curve was generated by using plasmid ODV-e56pGEM as previously described (42).

Virus egress assay.

Sf9 cells in 6-well plates were transfected with 4 μg of each of the recombinant AcMNPV bacmids expressing modified GP64 (E390A, G391A, or N381/K389A) or 0.5 to 4 μg of bacmid expressing WT GP64. At 24 h p.t., one set of the cells was used to score GFP fluorescence-positive cells under an epifluorescence microscope and evaluate the transfection efficiency. Another set of the cells was solubilized with 0.5% NP-40 in PBS (pH 7.4), and the β-Gluc activity was measured using the substrate 4-nitrophenyl β-d-glucuronide (PNPG; Sigma-Aldrich). The cell supernatants in a third set of the transfected cells were collected, infectious virus titers were determined by TCID₅₀ assays, the cells were fixed with 0.5% glutaraldehyde, and cell surface levels of GP64s were assessed by cELISA with MAb AcV5.

Western blotting.

Transfected or infected Sf9 cells were lysed with Triton X-100 buffer (150 mM sodium chloride, 0.1% Triton X-100, 50 mM Tris, pH 8.0) containing protease inhibitor cocktail (Roche Life Science). The extracted proteins were separated by 6% or 10% SDS-polyacrylamide gels under reducing and nonreducing conditions. After transferring to a polyvinylidene difluoride (PVDF) membrane (Millipore), the blots were blocked in a 4% milk TBST (10 mM Tris pH 8.0, 150 mM sodium chloride, 0.05% Tween 20) solution. GP64 proteins were detected with MAb AcV5 and alkaline phosphatase-conjugated goat anti-mouse IgG (Promega). Immunoblots were visualized using nitro-blue-tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) (Promega).