Unraveling the Entry Mechanism of Baculoviruses and Its Evolutionary Implications

Abstract

The entry of baculovirus budded virus into host cells is mediated by two distinct types of envelope fusion proteins (EFPs), GP64 and F protein. Phylogenetic analysis suggested that F proteins were ancestral baculovirus EFPs, whereas GP64 was acquired by progenitor group I alphabaculovirus more recently and may have stimulated the formation of the group I lineage. This study was designed to experimentally recapitulate a possible major step in the evolution of baculoviruses. We demonstrated that the infectivity of an F-null group II alphabaculovirus (Helicoverpa armigera nucleopolyhedrovirus [HearNPV]) can be functionally rescued by coinsertion of GP64 along with the nonfusogenic F^def (furin site mutated HaF) from HearNPV. Interestingly, HearNPV enters cells by endocytosis and, less efficiently, by direct membrane fusion at low pH. However, this recombinant HearNPV coexpressing F^def and GP64 mimicked group I virus not only in its EFP composition but also in its abilities to enter host cells via low-pH-triggered direct fusion pathway. Neutralization assays indicated that the nonfusogenic F proteins contribute mainly to binding to susceptible cells, while GP64 contributes to fusion. Coinsertion of GP64 with an F-like protein (Ac23) from group I virus led to efficient rescue of an F-null group II virus. In summary, these recombinant viruses and their entry modes are considered to resemble an evolutionary event of the acquisition of GP64 by an ancestral group I virus and subsequent adaptive inactivation of the original F protein. The study described here provides the first experimental evidence to support the hypothesis of the evolution of baculovirus EFPs.

INTRODUCTION

Baculoviruses (Baculoviridae) are a family of rod-shaped viruses with circular, supercoiled, double-stranded DNA genomes ranging in size from 80 to 180 kb. They are pathogens specific to insects, particularly to Lepidoptera, Hymenoptera, and Diptera (1). Baculoviruses are used in a number of applications, including bioinsecticides, eukaryotic expression vectors, and potential gene therapy vectors (2). These applications have contributed significantly to fundamental studies on these viruses.

Baculoviruses are highly diverse, but they have significant gene and genome conservation and several features that make them an ideal model to study the evolution of DNA viruses (3, 4). The family Baculoviridae contains four genera: Alphabaculovirus, Betabaculovirus, Gammabaculovirus, and Deltabaculovirus (5). The genome content and gene composition of gamma- and deltabaculoviruses are quite divergent from those of Alphabaculovirus and Betabaculovirus, implying that they were probably derived from much more ancient lineages (6). Alphabaculovirus can be further subdivided into group I and group II viruses based on molecular evolutionary analyses of polyhedrin and other genes (5, 7). Group II viruses show more variation than viruses of group I, which suggests an earlier lineage than group I (8).

The entry of enveloped viruses into host cells is mediated by envelope fusion proteins (EFPs) (9). The progeny budded viruses (BVs) appear to have at least two distinct types of EFPs, GP64 (in group I alphabaculoviruses) and F proteins (in group II alpha-, beta- and deltabaculoviruses). Although GP64 and F proteins are believed to have similar biological functions in virus entry and egress, they possess distantly related structures and modes of action (10–16). BVs were generally considered to enter host cells through clathrin-mediated endocytosis (17, 18). However, the GP64-containing baculovirus Autographa californica multiple nucleopolyhedrovirus (AcMNPV) was demonstrated to efficiently infect insect cells and to transduce mammalian cells via direct fusion with the plasma membrane, albeit at low pH (19).

F proteins are more diverse, with amino acid sequence identities as low as 20%, and are more widespread within the baculovirus family (14) than GP64 proteins. The latter are closely related (>74% amino acid identity) and are confined to and may have led to the evolution of group I NPVs. Despite their encoding GP64, an F-protein homologue (F-like protein) was identified in all group I NPVs. It has been reported that the AcMNPV GP64 can be pseudotyped with F proteins of group II alphabaculoviruses and betabaculoviruses (20–23). In contrast, it is difficult for GP64 to substitute the function of F protein. We once obtained gp64-pseudotyped F-null Helicoverpa armigera nucleopolyhedrovirus (HearNPV) vHaBacΔF-gp64 through many transfection attempts. However, the production of infectious BVs of vHaBacΔF-gp64 was ∼10³- to 10⁴-fold less than that of the native virus, and most of them had abnormal morphology (24). An attempt to replace the F protein with GP64 in Spodoptera exigua MNPV (SeMNPV) was also not successful (25). This implies evolutionary and functional divergence between GP64 and F proteins. The fact that F-like proteins have been retained during the evolution of group I alphabaculovirus genomes suggests that they might perform a necessary function(s) in the viral life cycle. A major difference between F and F-like proteins (Ac23 and Op21) is that the latter are unable to mediate low-pH-dependent virus-cell fusion due to the lack of a furin cleavage site (R-X-K/R-R) (26, 27). Although the F-like protein in group I viruses does not appear to be essential for virus replication and its precise function has not been determined, it was demonstrated that deletion of the F-like protein Ac23 from AcMNPV delays mortality of infected insects and decreases the in vitro infectivity of F protein-pseudotyped AcMNPV (27, 28).

Phylogenetic analyses suggest that F proteins might represent the ancestral baculovirus EFPs, while GP64s may have been acquired relatively recently (14). GP64 appears to have been the major contributor to the radiation of a distinct lineage (29, 30). Since a lineage of the Orthomyxoviridae, the thogotoviruses, also employ a homologue of GP64 as their EFP (31), it is possible that gp64 entered both viruses via recombination with a hitherto-unidentified host- or symbiotic-microorganism-derived gp64 homologue (29). Furthermore, sequence similarity between baculovirus F proteins and the env protein of a group of insect retroviruses, the errantiviruses, suggests the possibility that errantiviruses may have originated from a noninfectious insect gypsy-like retrotransposon that incorporated F protein from a baculovirus (29).

The presence of two different types of envelope fusion proteins and the apparent recent incorporation of gp64 that may stimulate the radiation of a distinct baculovirus lineage suggest that baculoviruses may provide novel insight into fusion protein-mediated virus entry and evolution. To further address the functional differences between the two types of EFPs and their evolutionary relationship, we constructed a series of recombinant viruses with different EFP combinations in an F-null HearNPV backbone. Specifically, we tested whether the infectivity of an F-null group II alphabaculovirus, HearNPV, can be functionally rescued by coinsertion of GP64 with a nonfusogenic F protein (mutated furin cleavage site, F^def) or with the remnant F-like protein (Ac23) of group I AcMNPV. Interestingly, we found that this recombinant pseudotyped group II virus (F^def plus GP64) resembled a group I virus in its EFP composition and entry mechanism. Neutralization and competition assays were performed to assess the contribution of a nonfusogenic F protein and GP64 to binding and to fusion of permissive cells. These experiments are believed to have experimentally revealed a significant step in the evolution of baculoviruses.

MATERIALS AND METHODS

Cell culture and viruses.

HzAM1 and Sf9 cells were cultured at 27°C in Grace's insect medium (pH 6.0; Gibco-BRL) supplemented with 10% fetal bovine serum (FBS). The F-null HaBacmid (HaBacΔF) (32), gp64-null AcBacmid (AcBacΔgp64) (28), and two control viruses, vHaBacΔF-HaF (32) (Fig. 1A) and vHaBac-gp64 (33) (Fig. 2A), were previously constructed at our laboratory.

FIG 1.

Entry pathway of group II baculovirus. (A) Schematic representation of F-repaired HearNPV HaBacΔF-HaF. (B) Transfection-infection assay. HzAM1 cells were transfected with HaBacΔF-HaF or F-null HaBacΔF. At 6 days p.t., supernatants from the transfected cells were used to infect healthy HzAM1 cells. Fluorescent images were taken at 3 days p.t. or p.i. (C) Low-pH-triggered infection assays with inhibition of the endocytosis pathway. HzAM1 cells were incubated with different MOIs of vHaBacΔF-HaF at 4°C. The unbound viruses were washed off, and virus-cell fusion was triggered by exposure of cells to pH 4.8 for 5 min. The endocytosis pathway was blocked with NH₄Cl. Normal infections without inhibitor and low-pH treatment were set as the endocytosis control. Infections were observed at 24 h p.i., and the percentages of infected cells were determined by flow cytometric assay. The experiments were repeated twice. Bars represent standard errors of the means. **, P < 0.01. (D) Infection of HzAM1 cells conducted with inhibitors of cytoskeleton molecules. HzAM1 cells were infected with HearNPV at an MOI of 5. Normal infection without chemical (NH₄Cl, nocodazole [Noc; microtubule-depolymerizing drug], or cytochalasin D [CD; actin polymerization inhibitor]) or low-pH treatment was set as a control. The functions of microtubules and myosin-like proteins were blocked by nocodazole (20 μM) or cytochalasin D (2 mg/ml) in the endocytosis pathway or direct fusion pathway (low-pH + NH₄Cl). The percentages of infected cells were quantified by flow cytometry at 24 h p.i. Data were normalized to the control value. (E) Time course analysis of IE1 transcription in HzAM1 cells infected by HearNPV (MOI = 5) through the endocytosis or direct fusion pathway. The cell samples were collected at 0, 30, 60, 90, and 120 min p.i. DNase I-treated total RNAs were used for RT-PCR analysis of IE1 transcription. 18S rRNA was used as an internal control. H₂O, PCR using double-distilled water as the template; N, RT-PCR using total RNA of uninfected HzAM1 cells; D, PCR using total RNA after DNase I treatment; P, PCR using HearNPV bacmid DNA as the template.

FIG 2.

Construction and characterization of a receptor binding domain and fusion domain complementary baculovirus. (A) Schematic representation of recombinant HearNPV bacmids HaBacΔF-gp64 (AcMNPV gp64-pesudotyped virus), HaBacΔF-HaF^def (furin cleavage site-mutated HaF-pseudotyped virus), and HaBacΔF-HaF^def-gp64 (F-null HearNPV, which coexpresses HaF^def and gp64). (B) Transfection-infection assays of the indicated recombinant bacmids. (C) Western blot analyses of recombinant viruses. BV proteins of vHaBacΔF-HaF, vHaBacΔF-HaF^def-gp64, and vHaBac-gp64 were probed with anti-HaF₁ (top), anti-GP64 (middle), and anti-VP80 (BV nucleocapsid protein [bottom]) antibodies, respectively. M, molecular mass markers. (D) One-step growth curves of vHaBacΔF-HaF, vHaBacΔF-HaF^def-gp64, and vHaBac-gp64. The results are from 3 replicate experiments. Bars represent standard errors of the means. TCID₅₀, 50% tissue culture infective dose.

Construction of recombinant bacmids.

Recombinant bacmids were generated according to the Bac-to-Bac manual (Invitrogen). AcMNPV gp64 was amplified by PCR from AcMNPV (bMON14272) bacmid DNA with primers gp64-1 (5′-AAGCTTGCCTCAATGCTACTAGTAAATC-3′) and gp64-2 (5′-AAGCTTGTGAGTTCAAGTCTCGCC-3′). Site-directed mutagenesis of arginine to lysine in the HaF furin cleavage site (Ha^def) was performed similarly, using primer pair HaF-1 (5′-AAGCTTATGGTTGCGATAAAAAGTAGTATG-3′) and HaF-2 (5′-AACGAAGTTCAATCCAATGTTTTTTTTGTTGCGACTCGAGAATGTTG-3′) (lysine underlined) and primer pair HaF-3 (5′-AACCAACATTCTCGAGTCGCAACAAAAAAAACATTGGATTGAACTTCGTT-3′) (lysine underlined) and HaF-4 (5′-GGATCCAAGCTTCGTAGGGATTTGCCGTCG-3′) for the first round of PCR and external primer pair Ha-1 and Ha-4 for the second round. The Ac23 gene was amplified by PCR with primers Ac23-1 (5′-GGATCCATGTTGGCTTGCAAATTCAGTCA-3′) and Ac23-2 (5′-GAATTCTTATTTTATTCTTTCTATAATCATAGGGTACAAC-3′) using AcMNPV bacmid DNA as a template. All PCR products were cloned into the pGEM-T Easy vector (Promega) for sequencing. Then, the envelope genes were digested with EcoRI and inserted into pFB-Op166 (32), generating transfer vectors pFB-Op166-gp64, pFB-Op166-Ha^def, and pFB-Op166-Ac23. The Op166-gp64 cassette was removed from pFB-Op166-gp64 by SacI digestion and cloned into the pFastBacDual vector (Invitrogen) to create pFBD-gp64. The Op166-HaF^def and Op166-Ac23 cassettes were excised from pFB-Op166-HaF^def and pFB-Op166-Ac23 by KpnI digestion and inserted into pFBD-gp64, generating the pFBD-HaF^def-gp64 and pFBD-Ac23-gp64 transfer vectors. Transfer vectors were transposed into the attTn7 integration site of HaBacΔF (Bac-to-Bac manual; Invitrogen). Recombinant bacmids were selected by gentamicin resistance and identified by PCR with M13 primers.

Transfection and infection.

HzAM1 cells (5 × 10⁵) were transfected with each recombinant bacmid DNA (1 μg) according to the Bac-to-Bac manual. At 6 days posttransfection (p.t.), 1-ml aliquots of supernatants were taken and infections were performed as described previously (32). Transfection-infection results were monitored by fluorescence microscopy at 72 h postinfection (p.i.) or p.t.

Western blotting.

To confirm the correct expression and incorporation of envelope proteins into the recombinant viruses, BVs were harvested by ultracentrifugation and Western blot analyses were performed with polyclonal antibodies against HaF₁, Ac23, or VP80 and anti-GP64 monoclonal antibody (AcV5), as described previously (28, 32). The final signals were detected after treatment with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate (SABC, China).

Assay of viral growth characteristic.

One-step growth curves were performed to analyze the growth properties of recombinant viruses as described previously (32). Statistical analysis was performed with one-way analysis of variance (ANOVA) in GLM (SPSS Inc.).

Virus entry via the low-pH-triggered direct fusion pathway.

Virus entry via direct fusion with the plasma membrane was performed as described previously (19). Briefly, cells were pretreated with Grace's medium containing 40 mM NH₄Cl at 28°C for 30 min and chilled to 4°C for an additional 30 min. The cells were incubated with recombinant viruses at a multiplicity of infection (MOI) of 2.5, 5, or 10 in the NH₄Cl-containing medium at 4°C for 1 h. Unbound virions were removed, and the virus-cell fusion was triggered with acidic Grace's medium (pH 4.8) in the presence of NH₄Cl at 28°C for 5 min. Subsequently, the medium was removed and the cells were further incubated with NH₄Cl-containing medium (pH 6.0) at 28°C for 4 h. The cells were washed twice with fresh Grace's medium and cultured in normal Grace's insect medium supplemented with 10% FBS for 20 h before fluorescence microscopy or flow cytometric analyses. Cells infected without chemical treatment under normal pH were set up as controls. For the cytoskeleton molecule inhibition experiment, cytochalasin D (2 mg/ml) and nocodazole (20 μM) were used to inhibit the functions of actin and microtubules, respectively, during normal infection and direct fusion triggered by low pH. The experiments were repeated twice, and standard deviations were calculated by Microsoft Excel 2003. The data were transformed logarithmically and statistically analyzed with one-way ANOVA in GLM (SPSS Inc.).

Viral entry kinetics.

The HearNPV entry kinetics of normal infection in comparison to those of low-pH-mediated infection was determined by reverse transcription-PCR (RT-PCR) based on the transcription of the ie1 gene, similar to that previously described (19). The total RNA was isolated with TRIzol (Invitrogen) at 0, 30, 60, 90, and 120 min after the binding procedure. The synthesis of the first-strand cDNA was performed using Moloney murine leukemia virus (MMLV) reverse transcriptase (Promega) and oligo(dT) primer (5′-CTGATCTAGAGGTACCGGATCCTTTTTTTTTTTTTTT-3′) according to the manufacturer's instructions. The PCR amplification of IE1 cDNA was performed using the IE1-specific primers IE1-F (5′-CCGTATCGCTGCGTCGCGGTAGCGGA-3′) and IE1-R (5′-GTTCGGACGGGTCTCGTGGCTGCGG-3′). The internal control, 18S rRNA, was amplified by 18srRNA-F (5′-GACGCGCGTGTTATAACGTA-3′) and 18srRNA-R (5′-GACGCGAGAACACATAACGA-3′).

Syncytium formation assays.

HzAM1 cells were infected with vHaBacΔF-HaF^def-gp64, vHaBacΔF-Ac23-gp64, vHaBacΔF-HaF, or vHaBac-gp64 at an MOI of 1. At 48 h p.i., cells were briefly washed with 1 ml of Grace's insect medium (pH 6.0) and treated with low-pH Grace's insect medium (pH 4.8) for 5 min at room temperature. Cells were then incubated with 2 ml of pH 6.0 Grace's insect medium containing 10% FBS, and the multinuclear cells were detected with a fluorescence microscope at 24 h posttreatment. Cells were fixed and stained with Hoechst 33258, and the average number of nuclei in 1 syncytium was calculated using 10 randomly chosen syncytia. The experiments were done in triplicate.

Neutralization assays.

HzAM1 cells (2 × 10⁵ cells per well) were inoculated into 24-well tissue culture plates. vHaBacΔF-HaF, vHaBac-gp64, and vHaBacΔF-HaF^def-gp64 (MOI = 5) were incubated with increasing amounts of anti-HaF₂ antiserum or neutralizing antibody (AcV1) against GP64 for 1 h at room temperature, and the virus-antibody mixtures were added to 24-well tissue culture plates to allow for virus attachment. After 1 h, the mixtures were replaced with Grace's insect medium containing 10% FBS. At 24 h p.i., the infection rates were quantified with a Beckman Coulter (EPICS XL) flow cytometer.

RESULTS

Group II alphabaculovirus enters cells via the direct fusion pathway only at a low efficiency.

Initially, an F-repaired vHaBacΔF virus (vHaBacΔF-HaF) (Fig. 1A and B) was used to characterize the entry mechanism of a group II alphabaculovirus. In the presence of endocytosis inhibitor NH₄Cl, infection of HzAM1 cells by vHaBacΔF-HaF was significantly blocked, suggesting that endocytosis was utilized by group II viruses (Fig. 1C). When cell-bound virions were activated by exposure to low pH, the infection rate of vHaBacΔF-HaF, via presumably the direct fusion pathway, was partially recovered, to ∼30% that of the parental virus (MOI = 5) (P < 0.01) (Fig. 1C). This differs from group I (GP64) viruses, which recover 100% when activated by low pH in the presence of NH₄Cl (19), indicating that F protein is not efficient to mediate entry of HearNPV into HzAM1 cells via the direct fusion pathway. When cytochalasin D and nocodazole were used to inhibit actin- and microtubule-dependent transport, we observed that HearNPV nucleocapsid movement to the nucleus was actin dependent but microtubule independent (Fig. 1D). Time course analysis of immediate-early gene (ie1) transcription suggested that HearNPV utilizes both pathways for cell entry at nearly the same rate (Fig. 1E). These results suggested that the postentry events of nucleocapsid trafficking of HearNPV via the two pathways are of similar modes and kinetics.

To determine whether F can mediate efficiently fusion between the viral envelope and the plasma membrane and enter other cells, we constructed an HaF-pseudotyped gp64-null AcMNPV (AcBacΔgp64-HaF). Interestingly, AcBacΔgp64-HaF was able to infect Sf9 cells by direct fusion almost as efficiently as internalization by endocytosis (data not shown). However, the same virus entered HzAM1 cells by direct fusion at low levels (data not shown), suggesting that different cellular factors may also take part in the baculovirus entry pathway.

Expression of F^def and GP64 together, but not individually, is essential to rescue the infectivity of an F-null group II alphabaculovirus.

We tested whether the low direct fusion activity of F proteins compared to GP64 might be a reason for the inefficient fusion of HearNPV to HzAM1 cells. It has been shown that F proteins can readily replace GP64 in AcMNPV (20, 22), but it was extremely difficult to obtain infectious F-null HearNPV pseudotyped with gp64 (HaBacΔF-gp64) (Fig. 2A and B). Although we have once randomly obtained an infectious vHaBacΔF-gp64 through about a dozen transfection attempts, it quickly started to lose infectivity (10³- to 10⁴-fold reduction in BV titer) and became unstable during storage (24). This instability is likely to be due to the inefficient budding process as evidenced by abnormal virion morphology (24). It is possible that a specific interaction of the cytoplasmic tail domain (CTD) of F with group II BV proteins may occur during virus egress. We tested this hypothesis by deleting the 7-amino-acid (aa) CTD from GP64 and replacing it with the 48-aa CTD of HaF. However, even though the chimera GP64_△CTD^HaFCTD was correctly expressed and mediated efficient infection of gp64-null AcMNPV, it was inadequate to produce infectious pseudotyped HearNPV (data not shown).

Furin cleavage is essential for the activation of F to a fusogenic form (15). To further investigate the functions of F protein other than membrane fusion, the HaF furin cleavage site was mutated from RNKR to RNKK, generating a fusion-disabled HaF (HaF^def). This mutation in HaF furin cleavage site rendered the recombinant HaBacΔF-HaF^def virus noninfectious (Fig. 2A and B). Although neither GP64 nor HaF^def alone can efficiently replace F protein, a recombinant F-null HearNPV bacmid coexpressing the defective envelope protein HaF^def and GP64 produced infectious progeny viruses (HaBacΔF-HaF^def-gp64 [Fig. 2A and B]). Western blotting was performed to detect the expression, cleavage, and incorporation of envelope proteins into recombinant BVs (Fig. 2C). As expected, due to the mutated furin cleavage site, anti-HaF₁ antiserum detected only a band (∼80 kDa) of an uncleaved HaF₀ in vHaBacΔF-HaF^def-gp64, while for vHaBacΔF-HaF and vHaBac-gp64, most of the HaF was processed into the F₁ subunit, with a minor band of uncleaved F₀ remaining. The anti-GP64 AcV5 antibody detected the expected band of ∼64 kDa in viruses containing gp64 (vHaBacΔF-HaF^def-gp64 and vHaBac-gp64) but not in vHaBacΔF-HaF. One-step growth curves further demonstrated that production of infectious BVs by vHaBacΔF-HaF^def-gp64 was comparable to that of the control vHaBacΔF-HaF virus (P > 0.05) (Fig. 2D). These results suggested that coincorporation of HaF^def and GP64 leads to efficient entry and budding of F-null HearNPV, with the fusogenic function being provided by GP64.

Group II virus coexpressing fusion-defective F^def and GP64 enters cells via the direct fusion pathway as efficiently as a group I virus.

The entry pathway of vHaBacΔF-HaF^def-gp64 was subsequently studied, and it was found that the virus infected HzAM1 cells via the direct fusion pathway quite efficiently. The rates of infection through this pathway were even significantly higher than those via endocytosis at three tested MOIs (P < 0.01) (Fig. 3A). The result implied that GP64 can mediate efficient fusion of virus with the plasma membrane of HzAM1 cells and fully rescued the defective entry of group II HaBacΔF-HaF^def by direct fusion. These properties endowed vHaBacΔF-HaF^def-gp64 with the characteristics of group I.

FIG 3.

Entry pathways and fusion activities of recombinant pseudotyped baculoviruses. Results of low-pH-triggered infection assays with inhibition of the endocytosis pathway of vHaBacΔF-HaF^def-gp64 (A) and vHaBac-gp64 (B) are shown. The experiments were done in triplicate. Bars represent standard errors of the means. *, P < 0.05; **, P < 0.01. (C) Syncytium formation assay. HzAM1 cells were infected with individual recombinant viruses at an MOI of 5 or left uninfected (mock). At 48 h p.i., cells were incubated with Grace's insect medium at pH 6.0 (top) or 4.8 (bottom) for 5 min and then examined for syncytium formation after 24 h. Multinuclear syncytia are indicated by arrows (left). For each set of virus-infected cells, 10 syncytia were randomly chosen and the average numbers of nuclei per syncytium were scored (right). The experiments were repeated twice.

Similar to vHaBacΔF-HaF^def-gp64, a pseudotyped virus, vHaBac-gp64, coexpressing a native HaF and GP64 also produced progeny viruses efficiently in HzAM1 cells (Fig. 2B and D). However, unlike with vHaBacΔF-HaF^def-gp64, the infection rate of vHaBac-gp64 via direct fusion ranged from 24% to 29%, depending on the MOI, compared to that via endocytosis (P < 0.01) (Fig. 3B). This is similar to the situation with wild-type (wt) HearNPV (F). The low-pH-dependent membrane fusion assay shows that the fusogenicity of vHaBac-gp64 (F plus GP64) is lower than that of vHaBacΔF-HaF^def-gp64, which uses GP64 as the sole fusion protein (Fig. 3C). We therefore propose that the presence of a second fusogenic protein, F, may interfere with the fusion of GP64, and if this is right, it provides a possible explanation for the evolution of a fusion-defective F-like protein in group I alphabaculovirus.

F or F^def proteins are crucial for cellular receptor attachment, and GP64 is crucial for fusion.

The respective roles of GP64 and HaF proteins during recombinant virus entry were investigated by neutralization assays. As shown in Fig. 4A, anti-HaF₂ subunit antiserum (putative receptor binding domain of HaF) efficiently blocked the infection of all three recombinant viruses (vHaBacΔF-HaF, vHaBacΔF-HaF^def-gp64, and vHaBac-gp64) in a dose-dependent manner, whereas preimmune rabbit serum lacked neutralizing activity (data not shown). The largest amounts (1 μl) of anti-HaF₂ antiserum completely inhibited infection by vHaBacΔF-HaF but failed to completely neutralize infection by vHaBac-gp64 or vHaBacΔF-HaF^def-gp64 (Fig. 4A). These results reveal that in these constructs, HaF and HaF^def proteins appear to be the major receptor binding proteins, while GP64 may also contribute to this process in a manner yet to be determined. These results also demonstrated that improper interaction of GP64 with a cellular receptor can limit the infectivity of GP64-pseudotyped F-null HearNPV.

FIG 4.

The roles of GP64 and F protein during virus entry by neutralization assays. vHaBacΔF-HaF (white bars), vHaBacΔF-gp64 (red bars), and vHaBacΔF-HaF^def-gp64 (green bars) were incubated with increasing amounts of either anti-HaF₂ (A) or anti-GP64 AcV1 (B) antibodies. The infection rate was quantified by flow cytometric analysis. (C) Entry model of three types of recombinant viruses: vHaBacΔF-HaF (left), vHaBac-gp64 (middle), and vHaBacΔF-HaF^def-gp64/vHaBacΔF-Ac23-gp64 (right). EE, early endosome; LE, late endosome; NPC, nuclear pore complex.

Anti-GP64 monoclonal antibody (AcV1) recognizes a prefusion conformational epitope and does not inhibit virus attachment to cells (34). As GP64 serves as the only active fusion protein in vHaBacΔF-HaF^def-gp64, the AcV1 antibody blocked infection in a dose-dependent manner, as expected (>98% of the virus was neutralized with 25 μl of antibody) (Fig. 4B). However, similar to the case with the control virus vHaBacΔF-HaF, the infection of vHaBac-gp64 was not affected by AcV1 (Fig. 4B), indicating that the F protein alone can mediate fusion, irrespective of the contribution from GP64.

We propose entry patterns for these recombinant viruses in Fig. 4C. The infection process of vHaBacΔF-HaF is initiated by specific interaction of F protein with an HzAM1 cellular receptor (Fig. 4C, left). Virions are internalized via clathrin-mediated endocytosis. As for vHaBac-gp64 (Fig. 4C, middle), the attachment event is mainly mediated by F protein, while in the late endosome (LE), both GP64 and native F protein are able to trigger the fusion. As indicated on the right side Fig. 4C, the recombinant vHaBacΔF-HaF^def-gp64 virus contains two types of envelope proteins, HaF^def (nonfusogenic receptor binding protein) and GP64. HaF^def binds to the cell surface receptors and mediates viral entry via endocytosis. Under the LE low-pH condition, only GP64 mediates the fusion process. The entry pattern of vHaBacΔF-HaF^def-gp64 is similar to that of paramyxoviruses, which have separate proteins for attachment and fusion (35).

Function of group I F-like protein (Ac23) in the context of an F-null group II virus pseudotyped with gp64.

Unlike the functional F proteins of group II alpha- and betabaculoviruses, a group I F-like protein (Ac23) is an attenuated F homologue lacking fusogenic activity and is dispensable in cell-to-cell spread of AcMNPV (27, 28). It was proposed that Ac23 along with GP64 may be receptor binding in AcMNPV (36). To test the function of Ac23, we constructed an F-null HearNPV bacmid pseudotyped with both Ac23 and GP64 (HaBacΔF-Ac23-gp64) or with Ac23 alone (HaBacΔF-Ac23) as a control (Fig. 5A). Ac23 alone could not replace F protein in HzAM1 cells (Fig. 5B), but when it was coexpressed with GP64, infectious progeny viruses were produced, as evidenced by the spread of the infection among cells (Fig. 5B). Western blotting confirmed the correct expression and incorporation of Ac23 and GP64 in vHaBacΔF-Ac23-gp64 (Fig. 5C). Syncytium formation assay revealed that similar to vHaBacΔF-HaF^def-gp64, vHaBacΔF-Ac23-gp64 (F^like plus GP64) has enhanced fusogenicity compared to that of vHaBac-gp64 (F plus GP64) (Fig. 3C). However, the titer of vHaBacΔF-Ac23-gp64 was about 400-fold lower than that of vHaBacΔF-HaF at the late stage of infection (Fig. 5D). The decreased infectivity of vHaBacΔF-gp64-Ac23 may be due to the fact that Ac23 is an F protein heterogeneous to that of HearNPV (shares ∼15% amino acid identity with HaF) and does not function efficiently for HzAM1 cells.

FIG 5.

Construction and characterization of an F-null group II baculvovirus expressing group I F-like protein and GP64. (A) Schematic representation of recombinant HearNPV bacmids HaBacΔF-Ac23 and HaBacΔF-Ac23-gp64. (B) Transfection-infection assays of recombinant bacmids in HzAM1 cells. (C) Western blot analysis of recombinant viruses. BV proteins of vHaBacΔF-Ac23-gp64 and vHaBacΔF-HaF were probed with anti-GP64, anti-Ac23, anti-HaF₁, and anti-VP80 antibodies, respectively. M, molecular mass markers. (D) Infectious BV production assay of vHaBacΔF-HaF or vHaBacΔF-Ac23-gp64 at an MOI of 1. The results are from 3 replicate experiments. Bars represent standard errors of the means.

Compared with a randomly obtained vHaBacΔF-gp64 (24), vHaBacΔF-Ac23-gp64 is readily propagated in each transfection trial. Further viral growth curve and binding assays revealed increased production of infectious BVs and an enhanced cell binding capability of vHaBacΔF-Ac23-gp64 compared to those of vHaBacΔF-gp64 (data not shown). Therefore, Ac23 contributes to the infection of a gp64-pseudotyped F-null HearNPV.

DISCUSSION

In the current study, we experimentally simulated a possible major incident in baculovirus evolution by constructing a series of recombinant baculoviruses to examine the incorporation of gp64 that may have led to a major new lineage of baculoviruses, group I of the Alphabaculovirus genus. Our results support the hypothesis that the acquisition of gp64 allowed a baculovirus to use an alternate protein to fuse with cell membranes and to adapt and replicate in new insect hosts, which may have propelled the radiation of group I alphabaculovirus.

The potential roles of GP64 and F proteins in respective recombinant baculoviruses were investigated by neutralization tests. Anti-GP64 AcV1 antibody efficiently neutralized vHaBacΔF-HaF^def-gp64, indicating that GP64 functions as a fusion protein. Antiserum against the HaF₂ subunit efficiently blocked the infection of all three recombinant viruses (vHaBacΔF-HaF, vHaBacΔF-HaF^def-gp64, and vHaBac-gp64), irrespective of the presence of GP64, suggesting that F is the major receptor binding protein. In members of the Paramyxoviridae, such as Sendai virus and respiratory syncytial virus, the entry process is mediated by separate attachment and fusion proteins (37). As for baculoviruses, retroviruses, influenza viruses, and many other viruses, receptor binding and membrane fusion events are executed by one viral glycoprotein. However, this study mimicked the entry mechanism of paramyxovirus and demonstrated that the receptor binding and fusion process of baculovirus infection can be mediated by two separate proteins.

It has been documented that viruses containing uncleaved F₀ precursor are noninfectious; they bind to target cells, while the viral genome does not enter (38). In HIV infection, uncleaved GP160 is able to bind CD4 and therefore tether the virus to the target cell (39). These characteristics provide theoretical support that HaF^def and F-like proteins retain binding activity. However, with larger amounts (5 μl) of anti-HaF₂ antibody, the infection of vHaBacΔF-HaF could be completely neutralized, but vHaBacΔF-HaF^def-gp64 and vHaBac-gp64 retained a low level of infection (data not shown). This suggests that GP64 may play a minor role in receptor attachment in vHaBacΔF-HaF^def-gp64 and vHaBac-gp64 (Fig. 4C). So far, the insect cellular receptor of baculoviruses remains a mystery. GP64 and F protein were suggested to exploit distinct receptors in Sf21 and Tn 5B1-4 cells, whereas our previous report and this study implied that GP64 may utilize the same cell surface molecule with F protein to gain entry into HzAM1 cells (24, 40, 41). Due to the fact that GP64 can bind to HzAM1 cells, albeit only at low levels, when F protein was neutralized in vHaBacΔF-HaF^def-gp64 (Fig. 4A), we suspect that this cellular molecule may not be a high-affinity receptor for GP64 and instead may be a molecule such as a phospholipid or proteoglycan (42).

It was previously suggested that Ac23 functions in the AcMNPV binding process (36). An enhanced green fluorescent protein (EGFP)-marked AcMNPV binds to and enters HzAM1 cells but fails to produce infectious progeny, and the promoter of a late gene (polyhedrin) does not operate (data not shown). Therefore, it is quite logical to suggest the presence of a potential cellular receptor(s) for AcMNPV (GP64 and/or Ac23) on the surface of HzAM1 cells. HearNPV competed with vHaBacΔF-HaF and vHaBacΔF-Ac23-gp64, suggesting that the same receptor is utilized by these viruses (data not shown). In addition, vHaBacΔF-Ac23-gp64 showed higher production of infectious BVs and cell binding capability than did vHaBacΔF-gp64 (24) (data not shown). Thus, we suggest that the F-like protein Ac23 of vHaBacΔF-Ac23-gp64 may contribute to receptor binding in place of HaF^def in vHaBacΔF-HaF^def-gp64 (Fig. 4C).

Interestingly, by exploring the entry pathways of these pseudotyped viruses, we found that vHaBacΔF-HaF^def-gp64 resembled group I virus in its ability to enter host cells via the direct fusion pathway under experimental conditions. This functional simulation process further implies a possible correlation between EFP composition/function and the virus entry mechanism (Fig. 6A). Our data also suggested that the baculovirus entry pathway could be shifted from endocytosis to fusion by acid treatment, depending upon the interaction of specific viral envelope glycoprotein with host cells.

FIG 6.

Entry and evolution model of baculovirus. (A) Schematic representation of the relationship between functional simulation process and the entry mechanism of recombinant pesudotyped viruses. The more ancestral group II baculoviruses (termed “starting virus”) use F protein as the receptor binding and fusion protein and enter cells via endocytosis (En), and the capture of gp64 allowed intermediate virus 1 to use a new protein for fusion. After a long time of adaptation, GP64 took over the fusogenicity of F, with this F^def only executing its receptor binding function. This gave the intermediate virus 2 a new pathway (direct fusion [DF]) for entry. The incorporation of GP64 as the viral binding protein could lead to evolution of F^def lacking its binding function and of GP64 as the sole binding and fusion protein of group I alphabaculoviruses; this may promote the evolution of this new branch and allow them to enter and replicate in new hosts. (B) Suggested model of baculovirus evolution based on EFPs. The ancestral baculovirus utilized the archetypal F protein as the EFP, while GP64 was randomly captured by a certain ancestral baculovirus and coexisted with ancestral F protein on the surface of BV, represented by vHaBac-gp64 in this study. Over a long course of adaptive evolution, GP64 took over the fusogenicity function of F protein, which allowed for the accumulation of mutations in the F protein furin cleavage site, typified by the recombinant viruses vHaBacΔF-HaF^def-gp64 and vHaBacΔF-Ac23-gp64. Endowed with a new protein, this branch of baculoviruses eventually evolved into group I alphabaculoviruses.

In summary, this study is believed to have experimentally revealed a possible major event in the course of baculovirus evolution. As proposed in Fig. 6B, by constructing a series of EFP-modified or recombinant pseudotyped baculoviruses, we revealed the consequence of the incorporation of gp64, which appears to have led to the formation of a new lineage, group I Alphabaculovirus. We also demonstrated that (i) inefficient binding of GP64 with HzAM1 cells can limit GP64 pseudotyping of F-null HearNPV; (ii) baculovirus binding and fusion events are executed by two heterologous proteins, much like the entry mechanism of paramyxoviruses; and (iii) a native F-like protein in the alphabaculoviruses might play an important role in the group II alphabaculovirus infection process. The successful reconstruction of these recombinant viruses not only reenacted possible evolutionary events of baculovirus but also provided an explanation for the presence of a fusion-defective F-like protein in group I alphabaculovirus. In addition, the EFP pseudotyping-based baculovirus entry system developed in this study is applicable in functional studies on other virus glycoproteins.