Structural Characterization of Per Os Infectivity Factor 5 (PIF5) Reveals the Essential Role of Intramolecular Interactions in Baculoviral Oral Infectivity

ABSTRACT

Baculoviruses initiate oral infection in the highly alkaline midgut of insects via a group of envelope proteins called per os infectivity factors (PIFs). To date, no high-resolution structural information has been reported for any PIF. Here, we present the crystal structure of the PIF5 ectodomain (PIF5e) from Autographa californica multiple nucleopolyhedrovirus (AcMNPV) at a 2.2-Å resolution. It revealed an open cavity between the N-terminal E1 domain and the C-terminal E2 domain and a cysteine-rich region with three pairs of disulfide bonds in the E2 domain. Multiple conserved intramolecular interactions within PIF5 are essential for maintaining its tertiary structure. Two conserved arginines (Arg8 and Arg74) play critical roles in E1-E2 interactions, and mutagenesis analysis supported their crucial role in oral infection. Importantly, the reduction in the oral infectivity of the Arg8, Arg74, or cysteine mutant viruses was related to the proteolytic cleavage of PIF5 by the endogenous protease embedded in occlusion bodies during alkaline treatment. This suggested that the structural stability of PIF5 under physiological conditions in the insect midgut is critical for baculoviral oral infectivity.

IMPORTANCE Per os infection mediated by PIFs is the highly complex mechanism by which baculoviruses initiate infection in insects. Previous studies revealed that multiple PIF proteins form a large PIF complex on the envelope of virions, while PIF5 functions independently of the PIF complex. Here, we report the crystal structure of AcMNPV PIF5e, which, to our knowledge, is the first atomic structure reported for a PIF protein. The structure revealed the precise locations of three previously proposed disulfide bonds and other conserved intramolecular interactions, which are important for the structural stability of PIF5 and are also essential for oral infectivity. These findings advance our understanding of the molecular mechanism of baculovirus oral infection under alkaline conditions.

INTRODUCTION

Baculoviruses are arthropod-specific large DNA viruses that have been isolated from insect species belonging to the orders Lepidoptera, Hymenoptera, and Diptera (1). In nature, baculovirus infection is initiated by occlusion-derived viruses (ODVs), which are embedded in occlusion bodies (OBs). After being ingested by insect larvae, OBs disintegrate under alkaline conditions within the larval midgut and release ODVs into the gut lumen (2). ODVs then initiate primary infection in midgut epithelial cells, and this process is designated per os infection (oral infection). Budded viruses (BVs) produced by ODV-infected midgut epithelial cells systemically spread to different tissues (also referred to as secondary infection) within the larval body.

Successful primary infection is mediated by a group of ODV-specific envelope proteins known as per os infectivity factors (PIFs) (3, 4). Autographa californica multiple nucleopolyhedrovirus (AcMNPV) is the most well-characterized baculovirus, and our current knowledge regarding PIFs is derived largely from AcMNPV. To date, 10 different PIFs (PIF0 to -9) that reside in the AcMNPV ODV envelope have been reported to be essential for oral infectivity (5). These PIFs are highly conserved among the baculovirus genomes sequenced to date (3–5). The deletion of any individual pif gene leads to a complete loss of oral infectivity but does not affect the production of BVs, except in the case of pif8 (ac83), which also contains a cis-acting element essential for nucleocapsid assembly (6, 7). Previous studies showed that PIF1, -2, and -3 assemble into a stable core complex of ~230 kDa, while other PIFs (except PIF5) associate with this complex to form a full complex of ~500 kDa (5).

In contrast to other PIFs, PIF5 appeared to be absent from the PIF complex, suggesting that PIF5 may function independently of the PIF complex during primary infection (4). AcMNPV PIF5 was originally designated ODV-E56 because it was identified as an envelope protein of ODVs and had an experimental molecular weight of ~56 kDa (8). The deletion of odv-e56 in AcMNPV significantly compromised the oral infectivity of ODVs, which established the encoded protein as PIF5 (9). ODVs obtain their envelope within the nucleus of infected cells, and PIF5 was found to be enriched in baculovirus-induced intranuclear microvesicles (8). Purified PIF5 was reported to interact with several proteins in brush border membrane vesicles prepared from the larval midgut, but PIF5 did not impact the binding or fusion of ODVs to midgut epithelial cells (9). We recently reported that PIF5 is a downstream substrate of the baculovirus sulfhydryl oxidase P33, and the six cysteine residues in AcMNPV PIF5 are all involved in the formation of disulfide bonds, which are essential for the function of PIF5 during ODV oral infection (10).

In this study, we report the crystal structure of the PIF5 ectodomain at a 2.2-Å resolution. Based on structure analysis, recombinant viruses containing point mutations in predicted key amino acids of PIF5 were constructed, and their oral infectivity was tested. Our results showed that multiple intramolecular interactions of PIF5 are essential for its structure and function.

RESULTS

Crystallization of the soluble ectodomain of PIF5.

Sequence analysis of PIF5 (376 amino acids in length) with the Phobius algorithm indicated that the protein possesses a putative transmembrane domain (TM) close to its C terminus (4) (Fig. 1A). For downstream crystallization, we first designed a soluble construct of the PIF5 ectodomain (termed PIF5e here) by truncating C-terminal residues 322 to 376. PIF5e was highly expressed in Schneider 2 (S2) Drosophila melanogaster cells, with an N-terminal Drosophila BiP signal peptide and a C-terminal V5/hexahistidine tag. Size exclusion chromatography suggested that purified PIF5e is monomeric in solution. Mass spectrometry showed that the protein has a molecular mass of 40.9 kDa. Using a microseeding procedure (see Materials and Methods), purified PIF5e crystallized at neutral pH in the space group P2₁2₁2 with a single molecule in the asymmetric unit, and the optimal crystals were diffracted to a 2.2-Å resolution (Table 1). Initial phasing was solved from crystals of selenium methionine (SeMet)-labeled PIF5e by the single-wavelength anomalous diffraction (SAD) method.

FIG 1.

Structural overview of PIF5e. (A) Domain organization of PIF5. The E1, E2, and putative transmembrane (TM) domains are shown in cyan, gray, and orange, respectively. The first residue of each domain and the cysteine residues are indicated below. The connectivity of the disulfide bonds in the cysteine-rich region (CRR) is shown. (B) The PIF5e structure depicted in a ribbon representation. The N-terminal E1 domain, the CRR, and the α8-helix are shown in cyan, gold, and magenta, respectively. The disordered loop is shown as a dashed line. The cysteine residues are represented as green sticks. (C) Diagram showing the topology of secondary structural elements within PIF5e. The cyan arrows denote β-strands. The orange and red cylinders denote α-helices and 3₁₀-helices, respectively. (D) Electrostatic potential at the surface of PIF5e at different pH values. The structure is shown as a solvent-accessible surface colored by electrostatic potential ranging from −10 kT/e (red, negative charge) to +10 kT/e (blue, positive charge).

TABLE 1.

Data collection and refinement statistics for PIF5e and the SeMet derivative^a

Parameter	Value
	PIF5e native	PIF5e SeMet
Data collection statistics
Space group	P21212	P21212
Cell dimensions
a, b, c (Å)	68.173, 92.360, 52.646	68.102, 91.592, 53.237
α, β, γ (°)	90, 90, 90	90, 90, 90
Wavelength (Å)	0.9789	0.9792
Resolution range (Å)	46.18–2.20 (2.28–2.20)	46.03–2.60 (2.69–2.60)
Rmerge	0.139 (0.455)	0.102 (0.635)
I/σI	12.9 (5.1)	21.4 (3.8)
Completeness (%)	99.8 (98.4)	99.9 (99.3)
Redundancy	12.8 (12.9)	12.9 (12.5)
Refinement statistics
No. of reflections	17,358 (1,673)
Rwork/Rfree (%)	19.9/24.4
Ramachandran plot favored regions (%)	98.3
Ramachandran plot outliers (%)	0
No. of atoms
Protein	2,359
Water	91
Avg B-factors (Å2)
Protein	30.44
Water	28.99
RMSD
Bond lengths (Å)	0.007
Bond angles (°)	0.78

^aStatistics for the highest-resolution shell are shown in parentheses. RMSD, root mean square deviation.

Structural overview of PIF5e.

Most residues of PIF5e were clearly defined in the density map, except for the C-terminal region (residues 306 to 321) and a surface loop (residues 215 to 220) (Fig. 1B). Additional density resulted from 6 residues from the expression vector pMT/BiP/V5-His (Pro-Trp-Pro-Gly-Val-Pro) at the N terminus. A search of the protein structure database using the DALI server (11) revealed that no proteins shared a highly similar overall architecture with PIF5e.

PIF5e can be generally divided into two structural domains: the N-terminal domain (residues 1 to 104) and the C-terminal domain (residues 105 to 305) (Fig. 1B). For simplicity, we named the N- and C-terminal domains of the PIF5 ectodomain E1 and E2, respectively. E1 contains seven α-helices (α1 to α7), one β-sheet (β1 to β3), and one 3₁₀-helix (η1), whereas E2 contains seven α-helices (α8 to α14), two β-sheets (β4 to β5 and β6 to β8), and three 3₁₀-helices (η2 to η4) (Fig. 1C). The cysteine-rich region (CRR) (residues 203 to 281) is localized in E2 (Fig. 1A and B). The disordered region (residues 215 to 220) resides in the CRR (Fig. 1B).

PIF5e has a total solvent-accessible surface area (SASA) of 14,164 Ų (using a probe radius of 1.4 Å). Basic residues (Arg, His, and Lys) account for 11% (1,498 Ų), while acidic residues (Asp and Glu) contribute 14% (1,990 Ų), resulting in a slightly anionic surface at neutral pH (Fig. 1D). However, when the pH was raised to 11, the surface charge of PIF5e was highly negative, especially at the E1-E2 interface (Fig. 1D). To evaluate the stability of PIF5e at alkaline pH, selected PIF5e crystals were soaked for 10 min in a buffer of pH 11 (100 mM N-cyclohexyl-3-aminopropanesulfonic acid [CAPS] and 25% polyethylene glycol 3350 [PEG 3350]). The crystals were able to diffract to a 4.0-Å resolution in the space group P2₁2₁2 and had cell dimensions (a = 68.017 Å, b = 91.414 Å, and c = 53.861 Å) similar to those of the crystals at neutral pH (Table 1), suggesting that the overall architecture of PIF5e might be relatively stable during the pH transition.

In addition, weak electron densities (not included in the final Protein Data Bank [PDB] model) have been observed in the direct vicinity of Asn183 (Fig. 2A), indicating that PIF5e may be a glycoprotein carrying N-linked glycans at Asn183. Deglycosylation of purified PIF5e with glycopeptidase F (GPF) showed that N-glycosylation does occur in the S2-expressed recombinant protein (Fig. 2B).

FIG 2.

Purified PIF5e is glycosylated. (A) Additional electron densities connected with Asn183 denoted by a dotted red ellipse in the 2mFo-DFc map at the 1.0-RMSD contour level. (B) PIF5e was digested without (−) or with (+) glycopeptidase F under denaturing conditions. Molecular mass marker lanes (M) and sizes (in kilodaltons) of individual markers are indicated on the left. PIF5e-specific bands are indicated by solid (for purified PIF5e) or empty (for deglycosylated PIF5e) arrows.

Strong intramolecular interactions detected between E1 and E2 and within the CRR.

The E1-E2 interface results in a wide-open cavity (9 Å deep) (Fig. 3A and B). Helix α8, the longest helix in PIF5e, runs from the bottom to the mouth of this cavity (Fig. 1B). The cavity is mainly lined by hydrophilic amino acids (Arg71, Asp77, His111, Gln114, Asp257, Ser259, Cys281, Glu283, and Asp289). Asp293 and nearby residues from E2 appear to cluster to form a negative region at the bottom of the cavity (Fig. 3B). Several residues at the E1-E2 interface are responsible for interdomain interactions (Fig. 3A), in which Arg8 and Arg74 from E1 are involved in the strongest interactions. The side chain of Arg8 forms hydrogen bonds with the main-chain carboxyl oxygen atoms of the nearby Leu273, Pro274, and Gln277 residues from E2 (Fig. 3C). Arg74, in proximity to Arg8, interacts with Asp107 via hydrogen bonding and salt bridge interactions (Fig. 3C and Table 2). Arg74 also forms hydrogen-bonding interactions with the backbone nitrogen atom of Phe108 and the carboxyl oxygen atom of Ala275. In addition, 5 residues from E1 (Arg51, Ser67, Arg100, Arg101, and Asn104) are also involved in the E1-E2 interactions (Arg51-NH2-to-Pro262-O distance; 3.40 Å; Arg51-NH1-to-Asp263-OD1 distance, 3.11 Å; Ser67-OG-to-Pro258-O distance, 3.35 Å; Arg100-O-to-Val105-N distance, 2.90 Å; Arg101-NH1-to-Glu198-OE1 distance, 3.05 Å; Asn104-OD1-to-Ala196-N distance, 2.78 Å; Asn104-OD1-to-Gly197-N distance, 3.15 Å).

FIG 3.

E1-E2 interactions. (A) Residues at the E1-E2 interface. Residues from E1 are represented as light cyan spheres. Residues from E2 are represented as purple spheres. (B) Slice through the surface of PIF5e to show negative (Asp293) and positive (Arg8) residue clusters in the cavity. (C) Close-up view of the cavity region highlighting conserved E1-E2 interactions. Arg8 and Arg74 are from E1, which is shown as a cyan ribbon. Atoms involved in hydrogen-bonding interactions are connected with dashed lines, and the atom-to-atom distance is labeled (in angstroms). For clarity, partial stick models are shown for those residues whose main-chain O or N atoms are involved in interactions. Colors for the stick models are as follows: nitrogen is shown in blue, and oxygen is shown in red.

TABLE 2.

Summary of salt bridges and N-O bridges in PIF5e^a

Positive residue	SASA of positive residue (Å2)b	Negative residue	SASA of negative residue (Å2)	COM (Å)c	1st atom pair			2nd atom pair
					Atom 1	Atom 2	Distance (Å)	Atom 1	Atom 2	Distance (Å)
Lys12	41.62	Asp23	23.05	4.90	NZ	OD2	3.95
Arg26	34.56	Asp23	23.05	4.44	NH1	OD1	3.35	NH2	OD1	3.89
Arg51	19.96	Asp263	46.10	4.63	NH1	OD1	3.11	NH2	OD1	3.51
Arg71	47.55	Asp68	40.90	4.61	NH2	OD1	2.76
Arg74	0.00	Asp107	14.07	3.71	NH2	OD1	3.31	NH2	OD2	3.49
Arg82	59.80	Asp90	57.44	4.18	NH2	OD1	2.71	NH2	OD2	3.35
Arg101	64.09	Glu198	71.52	5.03	NH1	OE1	3.05
Lys113	23.37	Asp293	2.14	3.77	NZ	OD1	2.73
Lys121	18.37	Asp289	12.20	3.15	NZ	OD1	2.60	NZ	OD2	3.67
Arg130	7.23	Asp213	19.59	4.71	NH2	OD2	2.68
Lys153	2.62	Asp177	1.80	3.45	NZ	OD2	2.81
Lys194	39.71	Glu201	63.35	3.58	NZ	OE1	3.49	NZ	OE2	3.88
Arg208	2.47	Asp227	2.40	3.58	NH2	OD1	3.45	NH2	OD2	2.71
Arg254	24.54	Glu249	3.79	3.62	NH2	OE1	3.53	NH2	OE2	2.76

^aGeometrically, a salt bridge is defined by two criteria: (i) the distance between the nitrogen atom of the Arg, Lys, or His side chain and the carboxyl oxygen atom of the Asp or Glu side chain is less than 4.0 Å, and (ii) the distance between the centers of the charged groups in oppositely charged residues is less than 4.0 Å (34). In N-O bridges, the first criterion must be satisfied, but for the second criterion, the center distance is greater than 4.0 Å. There are 14 residue pairs in PIF5e satisfying the first criterion.

^bThe atoms included in the calculation of the solvent-accessible surface area (SASA) are side chain nitrogen atoms or carboxyl oxygen atoms of Arg (NH1 and NH2), Lys (NZ), Glu (OE1 and OE2), and Asp (OD1 and OD2).

^cThe atoms included in the center of mass (COM) calculation were the charged regions of Arg (NE, NH1, NH2, and CZ), Lys (NZ), Glu (OE1, OE2, and CD), and Asp (OD1, OD2, and CG) (13).

Salt bridges and nitrogen-oxygen (N-O) bridges are two categories of electrostatic interactions that mainly contribute to the stability of proteins (12). We examined the electrostatic interactions between positively charged residues (Arg, Lys, or His) and negatively charged residues (Asp or Glu) in proximity. There are seven salt bridges (Arg74–Asp107, Lys113–Asp293, Lys121–Asp289, Lys153–Asp177, Lys194–Glu201, Arg208–Asp227, and Glu249–Arg254) and seven N-O bridges (Lys12–Asp23, Asp23–Arg26, Arg51–Asp263, Asp68–Arg71, Arg82–Asp90, Arg101–Glu198, and Arg130–Asp213) in the PIF5e structure (Table 2). Most of the nitrogen or oxygen atoms involved in electrostatic interactions had a SASA greater than zero (except Arg74) (Table 2), suggesting that most residues forming the salt or N-O bridges are presented on the surface of PIF5e (Fig. 4A and B). Usually, the salt bridges in proteins are formed between residues that are close in sequence to stabilize the local structure (13); however, most residue pairs involved in salt bridges from PIF5e were well separated, suggesting that their role in PIF5e is to stabilize the overall architecture.

FIG 4.

Electrostatic interactions in PIF5e. (A and B) Detailed views of the salt bridges and N-O bridges in PIF5e. The solvent-accessible surface of PIF5e is shown in two orientations. Positively and negatively charged residues are shown in blue and red, respectively. (C) Conserved salt bridges in the cysteine-rich region in stereo view. Disulfide bonds are indicated by green sticks. The β7- and β8-strands are shown in gold.

Three disulfide bonds (Cys203–Cys224, Cys210–Cys241, and Cys253–Cys281) are distributed throughout the CRR, which exhibits less secondary structure than the rest of PIF5e (Fig. 4C). Among the 6 cysteine residues, only 2 resided in the secondary structure (Cys203 in β7 and Cys281 in β8). The sulfur atoms of cysteine residues involved in disulfide bonds were largely buried in the protein, with only sulfur atoms of Cys203 (9.24 Ų) and Cys241 (3.19 Ų) being solvent accessible. Interestingly, two salt bridges, Arg208–Asp227 and Glu249–Arg254, were also situated in the CRR, which further stabilizes the structure (Fig. 4C).

As a core gene, pif5 is conserved among all sequenced genomes of baculoviruses. The family Baculoviridae contains four genera, namely, Alphabaculovirus, Betabaculovirus, Gammabaculovirus, and Deltabaculovirus (14, 15). Amino acid sequences from representative species were aligned for conservation analysis of intramolecular interactions (Fig. 5). Arg8 and Arg74 were strictly conserved residues whose side chains were involved in E1-E2 interactions, as shown in Fig. 3C. The salt bridge of Arg74–Asp107 was also conserved. In the CRR, two disulfide bonds (Cys210–Cys241 and Cys253–Cys281) and two salt bridges (Arg208–Asp227 and Glu249–Arg254) were conserved. The strict conservation of these residues and the intramolecular interactions suggested that the structural integrity of PIF5 may be important for its function. The flexible loop (residues 211 to 226) in the CRR is highly variable among baculoviruses, but its relative position in PIF5e is expected to be conserved since the intramolecular interactions mentioned above lock the flexible region in a rigid frame.

FIG 5.

Structure-based sequence alignment of baculovirus PIF5. The five representative species are from Alphabaculovirus group I (AcMNPV), Alphabaculovirus group II (Helicoverpa armigera single nucleopolyhedrovirus [HearNPV]) (GenBank accession number NP_075084), Betabaculovirus (Cydia pomonella granulovirus [CpGV]) (accession number NP_148802), Gammabaculovirus (Neodiprion abietis nucleopolyhedrovirus [NeabNPV]) (accession number YP_667864), and Deltabaculovirus (Culex nigripalpus nucleopolyhedrovirus [CuniNPV]) (accession number NP_203406). Note that since no structural information is available for residues 306 to 376 (AcMNPV), the corresponding C-terminal sequences are not included in the alignment. Secondary structural elements are indicated above the sequences, as follows: β-strands are marked as arrows, α-helices and 3₁₀-helices are depicted as coils, and β-turns are denoted by TT. The green numbers indicate three disulfide bond pairings. Three conserved salt bridges are denoted by S1, S2, and S3.

Mutagenesis studies revealed the essential role of Arg8 and Arg74 in oral infectivity.

To evaluate the functional roles of intramolecular interactions in PIF5e, in addition to our six previously constructed mutant viruses containing single-site cysteine mutations (C203S, C210S, C224S, C241S, C253S, and C281S) (10), six new pif5 mutants (R8A, R8E, R74A, E198A, N104A, and Ha-loop) were constructed to examine the impacts on BV production and in vivo oral infectivity (Fig. 6A and B). All five single-point mutations (R8A, R8E, R74A, E198A, and N104A) were at sites related to E1-E2 interactions. Arg8 and Arg74 are conserved residues in the pocket between the E1 and E2 domains, which likely play important roles in the E1 and E2 interactions, as mentioned above, while the less conserved Glu198 (which forms an N-O bridge with Arg101) (Table 2) and Asn104 (which forms two hydrogen bonds with the backbone N atoms of Ala196 and Gly197, respectively) residues are distributed on the protein surface. The Ha-loop mutant was constructed by replacing the flexible loop (RQDPNLNQSDVTI, containing the disordered fragment NLNQS) between Cys210 and Cys224 with the corresponding sequence (VLDVNNLNDVNV) of Helicoverpa armigera single nucleopolyhedrovirus (HearNPV) (Fig. 5) to test the flexibility of the loop region in the CRR.

FIG 6.

Generation of PIF5 mutations and their impact on oral infection. (A) Construction of recombinant bacmids. The pif5 mutant bacmids were generated by Tn7-mediated transposition based on a constructed pif5 knockout bacmid. (B) Mutation sites on the PIF5e structure. The target amino acids and the loop region are highlighted in red. (C) Transfection and infection assays of pif5 mutants. Sf9 cells transfected or infected with different mutants were observed by fluorescence microscopy for EGFP expression at 24 and 96 h posttransfection (p.t.) or 72 h postinfection (p.i.). WT, wild type. (D) Bioassay of PIF5 mutant OBs. The infectivity rates are presented as means ± standard deviations (SDs) from two independent tests. P values were determined by Student’s t tests. ***, P < 0.001; ns, no significant difference.

The recombinant bacmids containing the enhanced green fluorescent protein (EGFP) reporter gene were transfected into Sf9 cells to assess the effects of the six mutations on BV production. As shown in Fig. 6C, the cells transfected with the wild-type bacmid and all of the bacmids harboring mutations exhibited similar increases in fluorescence intensity from 24 h to 96 h posttransfection (p.t.). Subsequently, an infection assay showed successful infection by all mutants at 72 h postinfection (p.i.), indicating that these mutations have no obvious impact on BV production, which is in agreement with the role of PIF5 as an oral infection factor.

The OBs of mutant recombinant viruses were purified from larvae injected with the recombinant BVs and were used for an oral infection assay. The oral infectivity of the R8A, R8E, and R74A mutant viruses was profoundly impaired compared with those of the wild-type and pif5-repaired viruses, and there was no substantial difference in infectivity between these mutants and the Δpif5 control group (Fig. 6D). Meanwhile, the OBs of the recombinant viruses harboring N104A, E198A, and Ha-loop mutations retained oral infectivity. The disordered loop in the CRR can be functionally replaced by that of HearNPV, suggesting the flexibility of the loop sequence for PIF5 function. Collectively, these results indicate that Arg8 and Arg74 of PIF5 are functionally important for oral infection.

PIF5 mutants lacking oral infectivity underwent proteolytic degradation during ODV release from OBs.

To understand why the mutations affected oral infectivity, we checked whether the mutated PIF5 proteins were correctly packaged into ODVs by Western blotting. To inactivate the endogenous proteinase, OBs were preincubated at 80°C for 1 h before ODVs were released using alkaline buffer. Except for the pif5 knockout virus, PIF5 was packaged into ODVs successfully for all mutant viruses, as shown in Fig. 7A. Unexpectedly, Western blot analysis of ODVs derived from unpreheated OBs showed that PIF5 in the mutant viruses that lost oral infectivity (R8A, R8E, and R74A) underwent proteolytic degradation, resulting in protein products of less than 40 kDa, while the wild-type virus and other mutant viruses exhibiting normal oral infectivity possessed intact PIF5 (Fig. 7B). We also conducted similar experiments using previously constructed cysteine mutants (C203S, C210S, C224S, C241S, C253S, and C281S), all of which lost oral infectivity (10). These results showed that the PIF5 in the cysteine mutants was packed into ODVs although for C224S and C241S at a significantly reduced amount (Fig. 7C). Furthermore, the mutated PIF5 proteins underwent different levels of proteolytic degradation when OBs were not preheated (Fig. 7D). Collectively, these results suggested that intramolecular interactions are critical for the stability of PIF5 against proteolytic degradation under alkaline conditions and, consequently, for oral infectivity.

FIG 7.

PIF5 stability in the ODVs of PIF5 mutant virus. For each mutant virus, purified ODVs from OBs with (A and C) or without (B and D) heat inactivation (80°C for 1 h) were treated with SDS-PAGE loading buffer. Samples were resolved by 10% SDS-PAGE, followed by Western blotting with the indicated primary antibodies against PIF5 or VP39. M, standard molecular markers. Bands corresponding to intact PIF5 are indicated by an arrow.

DISCUSSION

The PIFs are a unique group of viral membrane proteins mediating the entry of baculoviral ODVs into insect midgut cells. PIF5 has been found to operate independently from the PIF complex. In this report, we present the crystal structure of PIF5e, which is structurally stabilized by multiple conserved intramolecular interactions (disulfide bonds, hydrogen bonds, and salt bridges). Our mutagenesis studies showed that many of these interactions are essential for effective oral infectivity. Importantly, the loss of oral infectivity appears to correlate with PIF5 degradation by endogenous proteases in OBs during alkaline treatment.

OBs of several baculoviruses, including AcMNPV, have been shown to contain a host-derived protease, which is suggested to aid in the disassociation of OBs and/or the release of ODVs under alkaline conditions (16–18). Here, we showed that wild-type PIF5 remained intact during alkaline treatment, whereas PIF5 derived from the mutants that lost oral infectivity was degraded if the endogenous protease in the OBs was not heat inactivated (Fig. 7), suggesting that the structural stability of PIF5 is essential for its oral infectivity. Apart from the endogenous protease, there are many other proteases in the insect gut (19), some of which may be used for processing PIFs. For example, PIF0 (P74) undergoes two sequential proteolytic digestion steps, the first involving the endogenous protease in OBs and the second involving the trypsin-like protease in the alkaline host midgut (20). However, our structural and functional results revealed that the proteolytic cleavage of PIF5 proteins under alkaline conditions has a negative impact on the oral infectivity of baculoviruses.

PIF5 degradation in mutant viruses produced small fragments of various lengths (most ~40 kDa), with the exception of the C224S and C241S mutants, in which PIF5 appeared to be fully degraded (Fig. 7). Given that the N-terminal E1 domain has a predicted molecular weight of ~12 kDa and that PIF5 is anchored onto the ODV membrane through the TM domain near the C terminus, we speculated that most of the E1 domain had been removed and that some sites near the E1/E2 boundary were degradation hot spots during alkaline treatment (Fig. 5). As shown by the PIF5e structure, Arg8 and Arg74 play important roles in E1-E2 interactions. When E1 lacks the stabilizing force of the E1-E2 interactions, sensitive sites may be exposed for degradation. Although cysteines are not directly involved in E1-E2 interactions, it is worth noting that Arg8 and Arg74 form hydrogen bonds with the main-chain O or N atoms of the residues in the CRR (Fig. 3C). Structural disruption of the CRR caused by cysteine mutations might also reduce the interactions of the CRR with Arg8 and Arg74, and thus, E1 may also be destabilized in such cases.

The crystal structure of PIF5e enabled us to map the precise pairing patterns of cysteine residues (Cys203–Cys224, Cys210–Cys241, and Cys253–Cys281) in the CRR (Fig. 4C) and amended our previous prediction about disulfide bonds (10). Interestingly, two conserved arginine-related salt bridges (Arg208–Asp227 and Glu249–Arg254) also stabilize the CRR. Our data suggested that conserved intramolecular interactions in the CRR are critical for the stabilization of the baculoviral PIF5 protein under alkaline conditions. It has been reported that most PIF proteins are rich in cysteines, with some being highly conserved (3, 4) and important for PIF complex formation and oral infectivity (6, 21). It will be interesting to investigate whether the cysteines in other PIFs have functions similar to those of PIF5 in future studies.

PIF5 was originally named ODV-E56 because it migrated in SDS-PAGE gels at a molecular weight (~56 kDa) higher than the expected molecular weight (~40.9 kDa) (8). Truncated PIF5e expressed in S2 cells exhibited an experimental molecular weight of ~41 kDa (by mass spectrometry), which was higher than its theoretical molecular weight (~39.5 kDa). Our results showed that S2-expressed PIF5e is glycosylated at Asn183 (Fig. 2), the only relatively conserved N-glycosylation site in PIF5, as predicted previously (22). In baculovirus-infected cells, PIFs are expressed in the cytoplasm, sequentially transported to the nuclear envelope and intranuclear microvesicles, and finally incorporated into the ODV envelope (5, 23). Posttranslational modifications of PIF5 in the infected cell may be significantly different from those of secretory glycoproteins (as in the case of S2 expression), which may result in the glycosylation of PIF5 being undetected on the ODV membrane (8). It would be interesting to investigate the posttranslational modifications of PIF5 and their roles in protein function.

Many enveloped viruses, such as influenza, dengue, and chikungunya viruses, release their genomes into host cells via virus-cell membrane fusion triggered by the low pH in the endosome (24, 25). In these viruses, fusion proteins on the viral surface mediate a membrane fusion reaction through dramatic, acid-dependent conformational changes. Histidine residues in fusion proteins have been proposed as the pH sensors for membrane fusion (24, 25) since the pK_a for the histidine side chain (6.45) is in the acidic range (26). In contrast, AcMNPV PIF proteins provide unique targets for membrane fusion in the alkaline midgut (pH 10 to 11) of a lepidopteran larva (19), and the amino acid residues with ionizable side chains, especially those with a pK_a value of between 10 and 11, may play important roles during viral fusion in alkaline microenvironments. Based on determinations from wild-type proteins, lysine (average pK_a = 10.68) and tyrosine (average pK_a = 10.98) have pK_a values in this alkaline range (26). However, there are no conserved lysine or tyrosine residues in baculovirus PIF5 (Fig. 5). In addition, PIF5e crystals were resistant to alkaline buffer (pH 11), indicating that PIF5 may not be a protein that is sensitive to the neutral-to-alkaline pH transition. Thus, PIF5 is unlikely to be a metastable fusion protein that undergoes significant conformational changes at alkaline pH.

In conclusion, our crystal structure of PIF5e provides some initial insights into the molecular architecture of this membrane protein that may help it resist harsh alkaline/proteinase conditions. In the future, further structural characterization of the PIF complex or individual PIF proteins will be required to understand how these molecules act in concert to mediate the entry of ODVs into insect midgut cells.

MATERIALS AND METHODS

Expression and purification of the ectodomain of PIF5 in S2 cells.

A nucleotide fragment encoding PIF5 residues 1 to 321 from the AcMNPV genome (GenBank accession number NC_001623) was subcloned into the Drosophila expression vector pMT/BiP/V5-His containing a C-terminal hexahistidine tag (Invitrogen). The resultant expression vector and the selection vector pCoHygro carrying the hygromycin B resistance gene in a 19:1 ratio were cotransfected into S2 cells grown in ESF 921 insect cell culture medium (Expression Systems) with Cellfectin II reagent (Gibco) according to the manufacturer’s protocol. Next, 300 μg/mL of hygromycin B (Vetec) was added to the medium for selection after 24 h of incubation at 27°C. A stable-expression cell line was then selected within 3 weeks. The expression of the ectodomain of PIF5 (PIF5e) was induced by 0.5 mM CuSO₄ at a cell density of 0.6 × 10⁷ cells/mL for 6 days. Hexahistidine-tagged PIF5e secreted from the culture supernatants was collected by removing cells and debris via centrifugation at 4,000 × g and buffer exchange into 50 mM Tris-HCl (pH 8.0)–150 mM NaCl. PIF5e was then purified by Ni affinity (Ni-agarose resin from Cwbio) and size exclusion chromatography (HiLoad 16/600 Superdex 200 pg; GE Healthcare Life Sciences). Deglycosylation assays of PIF5e with glycopeptidase F (TaKaRa) under denaturing conditions were performed according to the manufacturer’s instructions.

To prepare SeMet-labeled PIF5e, S2 cells expressing PIF5e were grown in ESF 921 cell culture medium to 0.6 × 10⁷ cells/mL. The cells were then rinsed twice in ESF 921 Delta series methionine-deficient cell culture medium (Expression Systems) and cultured in this medium for 24 h. Protein expression of the SeMet derivative of PIF5e was induced by the addition of 0.5 mM CuSO₄ after the addition of l-SeMet (Anatrace) to the medium to a final concentration of 150 μg/mL for 3 h. After 6 days of induction, SeMet-labeled PIF5e was collected and purified in the same manner as described above for native PIF5e.

Crystallization, structure determination, and refinement.

Purified PIF5e in 50 mM Tris-HCl (pH 8.0)–150 mM NaCl was subjected to buffer exchange into 200 mM sodium acetate (pH 5.6) and then concentrated to ~10 mg/mL prior to crystallization. Crystals of native PIF5e were grown using the hanging-drop vapor diffusion method and the microseeding technique at 14°C. Drops were prepared by mixing 1 μL of PIF5e and 1 μL of a precipitant solution containing 0.1 M Bicine (pH 8.0) and 25% (wt/vol) PEG 3350 to a final pH of ~7.0. After 21 days, small crystals from the droplets were crushed, and the crystal seeds were diluted over a series of concentrations (1:10, 1:100, and 1:1,000) in the precipitant solution. The seeds at different concentrations were then added in preequilibrated drops. Diffractable crystals were obtained at ~18 days. Crystals of SeMet-labeled PIF5e were prepared in a similar way, using crushed crystals of native PIF5e as microseeds.

Crystals were briefly soaked in a solution containing 0.1 M Bicine (pH 8.0), 25% (wt/vol) PEG 3350, and 25% (vol/vol) cryoprotectant glycerol before cryocooling in liquid nitrogen. X-ray diffraction data were collected on the BL17U1 and BL19U1 beamlines of the Shanghai Synchrotron Radiation Facility (SSRF), using an Eiger X 16M and a Pilatus3 6M detector, respectively. The data were then processed using XDS (27), as summarized in Table 1.

The crystal structure of PIF5e was solved by single-wavelength anomalous dispersion (SAD) phasing using the anomalous data collected from the SeMet crystals. The model was manually built in Coot (28) and iteratively refined by phenix.refine (29). The final refined model of PIF5e was validated using the MOLPROBITY server (30), which reported 98.34% of the residues in the Ramachandran-favored regions and 1.66% in the allowed regions. Structural calculations were performed using EDPDB (31), and structural illustrations were generated using PyMOL (PyMOL molecular graphics system, version 1.8; Schrödinger, LLC). The addition of hydrogen atoms and assignment of atomic charges and radii at different pHs were carried out by the PDB2PQR server (http://server.poissonboltzmann.org/pdb2pqr), using the PROPKA algorithm with the PARSE force field (32). Continuum electrostatic calculations were performed with APBS (32), and the resulting electrostatic potential of PIF5e was mapped onto its surface with PyMOL.

Generation of pif5 mutant recombinant viruses.

The recombinant pif5 knockout virus AcBacΔpif5, the pif5-repaired virus AcBacΔpif5-pif5, and the cysteine mutants had been constructed previously in our laboratory (10). AcBac-Syn (33) was used as the wild-type control. The construction of pif5 mutant recombinant viruses was performed as described previously (Fig. 6A) (10). Briefly, the pif5 promoter was fused to the single-point or Ha-loop-substituted pif5 gene constructed by overlapping PCR using the primers listed in Table 3. These fusion fragments were ligated into the BamHI and EcoRI sites of the pFastBacDual plasmid (Invitrogen) to generate mutation donor vectors, which were transformed into Escherichia coli DH10B cells containing bacmid AcBacΔpif5 and the helper plasmid. The mutant recombinant bacmids were selected by resistance screening and confirmed by PCR.

TABLE 3.

Primers used in this study

Primer	Sequence (5′–3′)a
Ppif5-F (BamHI)	gcgGGATCCCAAATGCGCCTTCTCGCCCAG
pif5:S-R (EcoRI)	gcgGAATTCTTAGCTGTCCATGTGCTGGCGTTCGAATTTAGCAGCAGCGGTTTCTTTTCGAGGGGCCGTTGTTGG
R8A-F′	GTTTTTTTTCAAATCTTGCCGCAGTCAATAAATTGTATC
R8A-R′	GATACAATTTATTGACTGCGGCAAGATTTGAAAAAAAAC
R8E-F′	GTTTTTTTTCAAATCTTGAGGCAGTCAATAAATTGTATC
R8E-R′	GATACAATTTATTGACTGCCTCAAGATTTGAAAAAAAAC
R74A-F′	CATAAACAGAATTACTGCTAACAATGATGTCCCC
R74A-R′	GGGGACATCATTGTTAGCAGTAATTCTGTTTATG
N104A-F′	GCGTGGACGCCGTGCCAGAC
N104A-R′	GTCTGGCACGGCGTCCACGC
E198A-F′	GACGCCGGAGCAATCGCCGAGGC
E198A-R′	GCCTCGGCGATTGCTCCGGCGTC
Ha-loop-F′	GTTCTGGACGTGAACAATTTGAACGATGTTAACGTATGCCCATTCGACCCCTTGTTG
Ha-loop-R′	TACGTTAACATCGTTCAAATTGTTCACGTCCAGAACACAAGTGCGCTGCAATAACAAGC

^aThe underlined nucleotides indicate restriction enzyme sites. For efficient digestion of the PCR products, extra bases (gcg) were added to the 5′ end of the restriction sites.

To generate the recombinant viruses, the bacmids were transfected into Sf9 cells using Cellfectin (Invitrogen) according to the manufacturer’s protocol. Sf9 cells were cultured at 27°C in Grace’s insect medium supplemented with 10% fetal bovine serum. BV supernatants were harvested at 96 h p.t., and viral titers were estimated by a 50% tissue culture infective dose (TCID₅₀) endpoint-dilution assay. To test the infectivity of recombinant viruses, Sf9 cells were infected with each recombinant virus at a multiplicity of infection (MOI) of 1 TCID₅₀/cell. The cells were observed with an Evos fluorescence microscope (Thermo Fisher Scientific) at different time points after transfection or infection.

Per os infection assay.

To produce viral OBs, each viral suspension containing BVs (2.5 × 10⁵ TCID₅₀/larva) was injected intrahemocoelically into fourth-instar Spodoptera frugiperda larvae. At 6 days p.i., OBs were purified from the liquefied larval cadavers by differential centrifugation and counted using a hemocytometer. Fourth-instar Spodoptera exigua larvae were used for droplet feeding with a total volume of 1 to 2 μL of the purified OBs at a dose of 1 × 10⁹ OBs/mL. At 48 h p.i., the infection state of the larva body was observed with a fluorescence stereomicroscope (SZX16; Olympus), and the infectivity rate was calculated. The experiments were performed in duplicate, and 24 larvae were used in each group for the feeding test.

Western blot analysis of PIF5 proteins in ODVs from preheated or unpreheated OBs.

PIF5 stability in ODVs was determined by Western blotting. To inactivate the endogenous protease, OBs were heated at 80°C for 1 h. ODVs contained in 1 × 10⁸ OBs (with or without preheating) were purified according to previously described procedures, with slight modifications (20). OBs were dissolved in an alkaline buffer (0.1 M Na₂CO₃, 0.15 M NaCl, 10 mM EDTA [pH 11.0]) at room temperature for 5 min, followed by neutralization with a 1/10 volume of Tris-HCl (500 mM; pH 7.5). After removing nondissolved debris by centrifugation at 2,400 × g for 5 min at 4°C, the ODV-containing supernatant was collected and centrifuged at 20,000 × g for 30 min at 4°C. Next, the ODV pellets were boiled with SDS-PAGE loading buffer. Prepared samples were analyzed by Western blotting with anti-PIF5 monoclonal antibody (mAb) (23D7 or 24B2) (1 : 5,000 dilution) or anti-VP39 rabbit serum (5) (1 : 5,000 dilution) as the primary antibody, followed by incubation with horseradish peroxidase (HRP)-conjugated secondary antibody against rabbit or mouse IgG (Sigma-Aldrich). Signals were detected with the SuperSignal West Pico chemiluminescent substrate (Pierce). Anti-PIF5 mAbs (23D7 and 24B2) are reported here for the first time and were generated by injecting mice with purified prokaryotically expressed PIF5e and isolating the mAbs after three rounds of screening of mouse hybridoma cells. The mAbs 23D7 and 24B2 recognize PIF5 epitopes of amino acids 241 to 250 and 101 to 110, respectively.

Data availability.

The atomic coordinates of the three-dimensional structure of PIF5e have been deposited at the Protein Data Bank (PDB) with the accession number 7X77.