Edited by Michael G. Rossmann, Purdue University, West Lafayette, IN, and approved December 7, 2016 (received for review October 6, 2016)
Here, we present a structural analysis of the genome delivery of slow bee paralysis virus (SBPV) that can cause lethal infections of honeybees and bumblebees. The possibility of blocking virus genome delivery would provide a tool to prevent the spread of this viral pathogen. We describe the three-dimensional structures of SBPV particles in a low-pH buffer, which imitates the conditions that the virus is likely to encounter after cell entry. The low pH induces a reduction in the contacts between capsid proteins and a formation of pores within the capsid that may serve as channels for the genome release. Our work provides a structural characterization of iflavirus genome release.
Viruses from the family Iflaviridae are insect pathogens. Many of them, including slow bee paralysis virus (SBPV), cause lethal diseases in honeybees and bumblebees, resulting in agricultural losses. Iflaviruses have nonenveloped icosahedral virions containing single-stranded RNA genomes. However, their genome release mechanism is unknown. Here, we show that low pH promotes SBPV genome release, indicating that the virus may use endosomes to enter host cells. We used cryo-EM to study a heterogeneous population of SBPV virions at pH 5.5. We determined the structures of SBPV particles before and after genome release to resolutions of 3.3 and 3.4 Å, respectively. The capsids of SBPV virions in low pH are not expanded. Thus, SBPV does not appear to form “altered” particles with pores in their capsids before genome release, as is the case in many related picornaviruses. The egress of the genome from SBPV virions is associated with a loss of interpentamer contacts mediated by N-terminal arms of VP2 capsid proteins, which result in the expansion of the capsid. Pores that are 7 Å in diameter form around icosahedral threefold symmetry axes. We speculate that they serve as channels for the genome release. Our findings provide an atomic-level characterization of the genome release mechanism of iflaviruses.
The family Iflaviridae includes arthropod pathogens, some of which infect economically important insects such as western honeybees (Apis mellifera), bumblebees, and silkworms (1). The iflavirus slow bee paralysis virus (SBPV) was identified in 1974 in the United Kingdom as a causative agent of honeybee colony mortality (2, 3). The related iflaviruses deformed wing virus and varroa destructor virus are found worldwide and, in combination with the ectoparasitic mite Varroa destructor, cause collapses of honeybee colonies (4). Despite the efficient transmission of SBPV by Varroa destructor (5), it only rarely causes disease in honeybees. However, it is common in bumblebees and solitary bees, whereas honeybees are probably an accidental secondary host (6). Honeybees are vital for agricultural productivity (7) and for maintaining the biodiversity of wild flowering plants (8). Furthermore, bumblebees and solitary bees are also important pollinators of specific commercial crops (3).
The family Iflaviridae belongs, together with Dicistroviridae and Picornaviridae, to the order Picornavirales of small nonenveloped viruses. The viruses from the family Iflaviridae have icosahedral capsids containing positive-sense ssRNA genomes about 9,500 nt long (9). These genomes encode a single polyprotein that is cotranslationally and posttranslationally cleaved into functional subunits. The capsid proteins originating from one polyprotein form a protomer—the basic building block of the capsid. The entire capsid consists of 60 such protomers, each of them composed of capsid proteins VP1–3, arranged in 12 pentamers (10). VP1 subunits are clustered around fivefold axes, and VP2 and VP3 form hetero-hexamers around icosahedral threefold axes. Unlike in the related picornaviruses and dicistroviruses, capsid protein VP3 of SBPV contains a C-terminal extension that folds into a globular protruding (P) domain positioned at the virion surface (10⇓⇓–13). At neutral pH, the P domains form “crowns” on the virion surface around each fivefold axis of the virus. However, the P domains can shift 36 Å toward the threefold axes in a different solution composition. The P domain consists of a central twisted antiparallel β-sheet surrounded by shorter β-sheets and α-helix. The P domain contains an Asp–His–Ser catalytic triad that is, together with the four surrounding residues, conserved among iflaviruses. These residues may participate in receptor binding or provide the protease, lipase, or esterase activity required for the entry of the virus into the host cell (10).
Virus capsids have to be stable and compact to protect the virus RNA genome in the extracellular environment. However, the capsids also need to allow genome release at the appropriate moment after the contact of the virus with its host cell. Whereas there is limited information about the cell entry of iflaviruses, the related picornaviruses that infect vertebrates have been extensively studied as models for nonenveloped virus genome delivery. Picornaviruses usually enter the host cells by receptor-mediated endocytosis (14). The acidic environment of the endosomes triggers the uncoating of some picornaviruses, including minor group rhinoviruses (15, 16). In contrast, major group rhinoviruses, which use receptors from the Ig superfamily, are stable at low pH and their genome release requires receptor binding (17, 18). Regardless of the trigger mechanism, enterovirus genome release is preceded by the formation of an uncoating intermediate called the altered (A) particle that has an expanded capsid with pores, N termini of VP1 subunits exposed at the particle surface, and released VP4 subunits (19⇓⇓⇓⇓–24). Amphipathic sequences from the N termini of VP1 and myristoylated VP4 subunits interact with endosome membranes and enable the delivery of enterovirus genomes into the cell cytoplasm (25⇓–27). It has been previously proposed that the enterovirus genome leaves the capsid via a pore located at a twofold icosahedral axis (19⇓–21). The resulting empty capsids are conventionally referred to as B particles (19). Acidic pH in the endosomes also induces the genome release of the aphthoviruses foot and mouth disease virus (FMDV) and equine rhinitis A virus (ERAV) (28⇓–30). However, aphthovirus genome release is not connected to capsid expansion or the formation of stable openings in the capsid. Furthermore, FMDV and ERAV empty capsids rapidly dissociate into pentamers after the genome release (28⇓–30).
Genome release has also been studied for a triatoma virus (TrV), a member of the insect virus family Dicistroviridae. Cryo-EM reconstructions of genome-containing and empty TrV particles determined to resolutions of 15–22 Å demonstrated that the TrV empty particles are very similar to the native virions. After the genome release, the empty TrV capsids rapidly disassemble into pentamers (13).
Delivery of the genome into the cell cytoplasm constitutes a critical step in the life cycle of positive-sense ssRNA viruses. However, molecular details underlying the uncoating of iflaviruses are currently unknown. To describe the genome release mechanism of SBPV, we used cryo-EM to determine the structures of SBPV particles in a low-pH buffer mimicking conditions that the virus is likely to encounter during cell entry. By comparing the structures of full and empty particles, we determined that iflaviruses use an uncoating mechanism distinct from that of picornaviruses and dicistroviruses.
The exposure of SBPV to pH 5.5 or lower induces a gradual release of the RNA genome from the virions (Fig. 1). Similar effects were previously described for the iflavirus sacbrood virus and several picornaviruses (31⇓–33). The induction of genome release under acidic conditions implies that SBPV uses endosomes for cell entry (14). Alternatively, SBPV virions might be endocytosed, escape the endosomes, and subsequently uncoat in the cytoplasm. The nonphysiological pH 3.5 induced precipitation of SBPV particles that subsequently could not be observed in the cryo-EM images (Fig. 1A). Only 30–50% of SBPV virions released their genomes in the in vitro low-pH conditions (Fig. 1B). It is possible that the binding of SBPV to currently unknown receptors in vivo might increase the efficiency of the genome release. The effect of the low pH on SBPV genome release was further verified by the observation that SBPV virions in a buffer with pH 5.5 release their genomes at 52.5 °C, whereas those in buffer with neutral pH at 54.2 °C (Fig. 1C). Thus, the stability of SBPV virions is similar to that of previously studied vertebrate picornaviruses (19, 22, 34).
Low pH triggers genome release of SBPV. (A) Cryo-electron micrographs of SBPV virions show increasing number of empty particles with decreasing pH. (B) Graph showing dependence of the fraction of empty SBPV particles on pH of the buffer. (C) Stability of SBPV virions at different pH values. SBPV virions were mixed with Sybr Green dye II and heated to indicated temperatures (x axis). The fluorescence signal increases as the dye binds to RNA that is released from thermally destabilized particles. The dashed line represents SBPV at neutral pH, and the full line represents SBPV at pH 5.5. Error bars indicate SDs of the measurements (n = 3). Please Materials and Methods for details.
The exposure of SBPV to pH 5.5 for 18 h resulted in a mixed population of full virions and empty particles (Fig. 1A). Reconstructions of the two forms were determined independently to resolutions of 3.3 and 3.4 Å, respectively (Fig. 2 and Table S1). The quality of the cryo-EM map was sufficient to enable the virion structure to be built, except for residues 1–12, 181–193, and 246–266 of VP1; 1–35, 191–201, and 261 of VP2; and 1, 73–75, 218–221, and 267–431 of VP3. The structure of the low-pH genome-containing virion is nearly identical to that of the neutral pH SBPV virion, which was characterized previously by X-ray crystallography (Fig. 2 C and D) (10). The rmsd of the corresponding Cα atoms of the two structures is 0.67 Å. The major capsid proteins VP1–3 have jellyroll β-sandwich folds with β-strands named according to the picornavirus convention B to I (Fig. 2H, Fig. S1, and Movie S1) (35). The two antiparallel β-sheets forming the β-sandwich fold contain the strands BIDG and CHEF. The loops are named according to the secondary structure elements they connect. The differences between the native and low-pH virion structures are restricted to the loops of the capsid proteins exposed at the particle surface, some of which are not resolved in the cryo-EM map of the low-pH virus (Fig. 2H). Subunit VP3 of SBPV has a C-terminal extension that forms a globular P domain (Figs. 2 A, B, and H, and 3). It was shown previously that the P domains can occupy different locations at the virion surface (10). In agreement with one of the SBPV structures determined previously by X-ray crystallography, the P domains in the cryo-EM reconstruction are located close to the fivefold axes and form “crowns” at the virion surface (Fig. 2 A–E). However, the volumes of the cryo-EM electron density map corresponding to the P domains lack high-resolution details, most likely because the P domains can attach to different positions at the capsid surface. This results in a smearing of the high-resolution details of the map (Fig. 2 D and E), because the reconstruction process depends on an averaging of the images of many particles.
Cryo-EM structures of full virions and empty SBPV particles at pH 5.5. Structures of SBPV at pH 5.5 before (A) and after genome release (B). The solvent-accessible surfaces are rainbow-colored based on their distance from the particle center. The positions of P domains that are not well resolved in the cryo-EM maps are indicated as gray transparent surfaces. Central slices of electron density maps are shown for SBPV virions at pH 6.5 (C), virions at pH 5.5 (D), and empty particle at pH 5.5 (E). White indicates high values of electron density. The positions of selected icosahedral symmetry axes are labeled. F and G show representative electron densities of SBPV virion and empty particle at resolutions of 3.3 and 3.4 Å, respectively. The maps are contoured at 2.0 σ. (H) Superposition of cartoon representations of icosahedral asymmetric units of native SBPV virion with VP1 shown in blue, VP2 in green, and VP3 in red. The structure of the P domain is shown in semitransparent gray. The structure of SBPV virion at pH 5.5 is shown in magenta, and that of the empty particle at pH 5.5 is in gray. Parts of the capsid proteins with the largest structural differences among the three structures are highlighted in the native virion in cyan.
Movements of capsid proteins associated with SBPV genome release. Capsid protein subunits of the empty SBPV particle are shown in blue, green, and red for VP1, VP2, and VP3, respectively. The P domain of VP3 from the empty capsid is not shown. Relative positions of the capsid proteins in the native SBPV virion are shown in gray. Shifts between the positions of the individual capsid proteins subunits in the two structures are indicated. The position of an icosahedral fivefold symmetry axis is shown as a dashed line.
Cryo-EM reconstruction and structure refinement statistics of SBPV virion and empty particle at pH 5.5
Stereoview of superposition of cartoon representations of icosahedral asymmetric units of native SBPV virion with VP1 shown in blue, VP2 in green, and VP3 in red. The structure of the P domain is shown in semitransparent gray. The structure of SBPV virion at pH 5.5 is shown in magenta, and that of the empty particle at pH 5.5 is in gray. The parts of the capsid proteins with the largest structural differences among the three structures are highlighted in the native virion in cyan.
Three-dimensional classification of images of full SBPV virions at pH 5.5 did not identify a subclass of particles with expanded capsids. Therefore, unlike in many picornaviruses (16, 24), the exposure of SBPV virions to pH 5.5 did not induce the transition of the virions to the A form.
The GH loop of SBPV subunit VP2, which harbors an RGD motif, is disordered in the low-pH virion structure (Fig. 2H). The RGD sequence enables proteins to bind to integrins (36), which are used as receptors by several viruses (37, 38). We previously speculated that SBPV may use integrins to infect honeybee cells (10). The low-pH–induced flexibility of the GH loop might enable the virus to detach from the integrin receptor after it enters the endosome. This could be required for subsequent interaction of the virion with the membrane of the endosome and delivery of the virus genome into the cell cytoplasm. Other parts of capsid proteins that contain residues that are not resolved in the low-pH virion structure are the BC, GH, and HI loops of VP3 and the N terminus and GH loop of VP1 (Fig. 2H).
Although structures of the SBPV capsids at pH 6.5, determined by X-ray crystallography (10), and 5.5, determined by cryo-EM, are similar, there are differences in the distribution of the genomic RNA inside the particles (Fig. 2 C and D). The genome of SBPV is a 9,500-nt-long ssRNA molecule that cannot entirely obey the icosahedral symmetry of the capsid. However, both X-ray crystallography and single-particle reconstruction used for SBPV structure determination use the icosahedral symmetry of the capsid that provides the dominant signal in the experimental data. Therefore, both the cryo-EM and crystallographic electron density maps contain information about the icosahedrally symmetrized distribution of the genome. In the neutral-pH SBPV structure, the RNA is distributed uniformly within the capsid cavity (Fig. 2C). It was previously speculated that the disordered N-terminal parts of major capsid proteins VP1–3 of picornaviruses, which carry positively charged residues, are in direct contact with the genome (20). However, the N termini of the capsid proteins are fully determined in the crystal structure of SBPV (10), whereas the details of the RNA structure are not even partially resolved (Fig. 2C). Therefore, the SBPV genome appears to be folded in the lumen of the capsid without specific interactions with the surrounding capsid proteins. In contrast, in the low-pH SBPV virion structure, the RNA density is accumulated in a spherical shell with a radius of 110–90 Å and in a central part of the virion within a sphere with a radius of 80 Å (Fig. 2D). These two regions are separated by a spherical shell 90–80 Å in diameter with a lower RNA density. The SBPV RNA does not appear to form specific contacts with the capsid, similar to the previously studied A particles of enteroviruses (22, 39). No particles in the process of RNA release were observed in the cryo-micrographs of low-pH SBPV. Therefore, we speculate that the genome release from SBPV virion is stochastic and rapid with only short-lived genome release intermediates. The conformational changes of the capsid associated with the genome release (Fig. 3), described in detail below, result in alterations in the distribution of charge on the inside of the capsid (Fig. 4). Whereas the areas around the icosahedral threefold axes on the inside of the capsid are strongly negatively charged in the genome-containing virions, they become neutral in the empty particle (Fig. 4). The changes in the distribution of the RNA induced by the low pH and the alteration of charge distribution within the capsid might facilitate the release of the genome from the SBPV virion.
Comparison of charge distribution on the inside of SBPV virion and empty particle. Comparison of electrostatic potential distribution on the inside of the native virion (A) and empty particle (B). Three pentamers of capsid protein protomers are displayed. The borders of a selected icosahedral asymmetric unit are highlighted with a black triangle. The Insets show details of the electrostatic potential distribution around threefold symmetry axes.
The genome release of SBPV results in the formation of empty particles that are expanded compared with the native virions (Figs. 2 A and B, and 3). The radius of the particle, measured as the distance of the center of mass of a protomer from the particle center, changes from 133 Å in the native virus to 136 Å in the empty low-pH particle. Accordingly, the volume of the capsid increases from 6.4 × 106 to 7.2 × 106 Å3. The expansion of the capsid is achieved by movements of the capsid proteins VP1, VP2, and VP3 3.7, 2.6, and 3.1 Å away from the particle center, respectively (Fig. 3). In addition, the subunits VP1, VP2, and VP3 rotate 3.1°, 1.7°, and 1.9°, respectively. These shifts and rotations are approximately one-half of those previously reported for the conversion of enteroviruses to A particles (19, 20). The rotations together with small conformational changes to residues located at the periphery of the pentamers allow the capsid proteins to remain in contact after the expansion of the capsid.
The expansion of the SBPV capsid is connected to a loss of structure of the first 35 residues from the N terminus of VP2 (Fig. 5 A and B). The strands β1 and β2 of VP2, which in the native virion mediate contacts between pentamers of capsid protein protomers by extending the β-sheet CHEF of VP3 from the neighboring pentamer, are not resolved in the empty SBPV particle (Fig. 5 A and C). Furthermore, residues 218–221 of VP3 that form part of the interpentamer interface in the native virus are not structured in the empty particle (Fig. 5B). The conformational changes of the SBPV capsid linked to the genome release result in a reduction of the interpentamer contacts from 2,800 to 1,200 Å2, which corresponds to a 68% decrease in the buried surface area (Fig. S2). The loss of the structure of the N-terminal arm of VP2 accounts for the removal of 800 Å2 of the buried surface area of the interface. In contrast, the contacts within the protomer and within the pentamer are only reduced by 25 and 14%, respectively (Fig. S2). The reduction in the interface areas cannot be directly converted to binding energies, but the restricted interpentamer contacts indicate that the expanded SBPV particles are less stable than the native virions.
Changes to the SBPV capsid structure associated with genome release. Cartoon representation of capsid of SBPV virion at neutral pH (A) and empty particle at pH 5.5 (B) viewed from the particle inside. VP1 subunits are shown in blue, VP2 in green, and VP3 in red. Selected subunits are shown in bright colors. Positions of icosahedral twofold and threefold axes are indicated by ovals and triangles, respectively. Dashed boxes outline areas displayed in higher details in indicated panels. (C) Interaction of β2 of VP2 with βF of VP3 from neighboring pentamer in SBPV virions. Hydrogen bonds are shown as dashed black lines. The interaction is not present in the empty SBPV particle. (D) Interaction of N terminus of VP1 with β3 of VP2 present only in native and low-pH SBPV virions. Comparison of interactions of residues around threefold symmetry axis of the capsid in native SBPV (E) and empty low-pH particle (F). Electron density map is contoured at 1.5 σ. Comparison of interface between pentamers of capsid protein protomers in SBPV virion (G) and empty particle (H). Interactions between α3 helices of VP2 subunits related by icosahedral twofold axis in SBPV virion (I) and empty particle (J).
Buried surface areas of interfaces within SBPV virions at different pH levels and of empty particle. A list of buried surface areas is shown in A. Individual subunits are labeled according to their relative positions shown in B. (B) Capsid surface representation of SBPV virion with subunits VP1, VP2, and VP3 shown in blue, green, and red, respectively. Icosahedral asymmetric units considered for buried surface calculations are labeled with letters.
The movements of the capsid proteins away from the particle center and reduction in the interpentamer interfaces result in the formation of pores around icosahedral threefold axes of the SBPV capsid (Figs. 4 and 5 E and F). Besides the subunit movements, the pores are enlarged by conformational changes of residues 126–129 from the DE loop and 221–224 from the HI loop of VP2 and 136–141 from the DE loop and 223–225 from the HI loop of VP3 that are located around the threefold axis (Fig. 5 E and F). Upon particle expansion, the loops move away from the threefold axis, giving rise to a 7-Å-diameter pore (Fig. 5F). The size of the pore is not sufficient to allow the release of ssRNA, which would require an aperture about 10 Å in diameter. However, the loops of capsid proteins located around the threefold axis are more flexible (average temperature factor, 70 Å2) than the rest of the capsid (average temperature factors, 50 Å2). Therefore, the loops at the border of the channels located at threefold axes might fold back and expand the pore as necessary for the release of the genome. Two additional small openings 1–2 Å in width are formed at the interface between the pentamers (Fig. 5H). In contrast, there are no pores in genome-containing SBPV virions both under neutral- and low-pH conditions (Fig. 5G). Because of their small size, the pores at the interface between the pentamers are unlikely to be channels for RNA egress. In addition, in some parts of the capsid the pores might be obscured by the N termini of VP2 subunits, which are not visible in the averaged cryo-EM reconstructions of the empty SBPV particles.
The strand β3 that extends the β-sheet BIDG of the VP2 subunit is not structured in the empty particle, whereas it interacts with Glu-2 and Arg-3 from the N terminus of VP1 in the virions (Fig. 5D). The loss of the structure of the VP2 β3 strand might contribute to alterations in the structure of the N terminus of VP1 as discussed below.
Several surface loops are not resolved in the cryo-EM electron density map of the empty particles, similar to the structure of full SBPV virions at pH 5.5 (Fig. 2H). However, the GH loop of VP3 (residues 185–210) has a different conformation in the empty particle than in the native virion, whereas it is partially disordered in the low-pH virion structure (Fig. 2H). The neutral-pH structure of the loop is stabilized by interactions with the C terminus and EF loop of VP1 and the GH loop from VP2 subunits from the neighboring icosahedral asymmetric unit. Residues 198–205 of the loop are disordered in the low-pH virion structure (Fig. 2H). However, in the empty particle, the GH loop is structured and interacts with the CD loop and α2 from the same VP3 subunit and with DE and FG loops of VP1 from the neighboring icosahedral asymmetric unit. Therefore, the GH loop might stabilize interactions within the pentamer of capsid protein protomers. Similar to the low-pH SBPV virion structure, the high-resolution details are not resolved in the P-domain region of the cryo-EM map of the empty particle (Fig. 2 D and E).
During cell entry, nonenveloped viruses have to breach a biological membrane to deliver their genomes to the cell cytoplasm. Previously, it has been shown that both myristoylated and unmodified VP4 minor capsid proteins (which are 60–80 residues long) of several picornaviruses and dicistroviruses can induce the lysis of liposomes (26, 40, 41). In contrast, the VP4 subunits of SBPV have been predicted to be only 20 residues long, and lack the myristoylation signal sequence (42). In silico analysis predicts that residues 4–21 of VP4 of SBPV form an amphipathic α-helix in which the polar and hydrophobic residues are segregated to the opposite sides (Fig. S3A) (43, 44). Helices with this type of charge distribution have been shown to interact with membranes (45). Nevertheless, electron density corresponding to the VP4 subunits could not be identified in the SBPV virion structures (10). Furthermore, mass spectrometry analysis [liquid chromatography–tandem mass spectrometry (LC-MS/MS)] shows that the SBPV virions contain at least some subunits in which VP4 was not cleaved from the N terminus of VP3 (Fig. S4). Therefore, it is not clear whether the VP4 peptides contribute to the delivery of the SBPV genome across the biological membrane.
Interaction of SBPV virions with liposome membranes. (A) Helical wheel representation of predicted structure of SBPV VP4 subunit that forms a putative amphipathic helix with hydrophobic residues clustered on one side and polar residues on the opposite face. (B) Effect of SBPV particles on the integrity of liposomes. Liposomes filled with the self-quenching fluorescent dye calcein were incubated with SBPV virions in buffers with pH 7.0 or 5.5, and the increase in the fluorescence signal caused by the release of calcein from the liposomes was measured. The heated virus samples were preincubated at 63 °C for 5 min to induce the release of VP4 subunits from the capsids. The 10 mM Triton X-100 detergent was used as a positive control to induce the liposome disruption. The error bars represent SD of the mean calculated from triplicates. Please see SI Materials and Methods for details.
SBPV virions contain uncleaved VP4/VP3 subunits. (A) SDS polyacrylamide gel of purified SBPV virions showing bands of capsid proteins VP1, VP2, and VP3. The presence of the VP4 peptide at the N terminus of VP3 would add only 2.1 kDa to the molecular mass of VP3 and may not be distinguished in the gel. (B) Amino acid sequence of P1 polypeptide of SBPV. The borders of the predicted cleavage sites are indicated with a colored background. N-terminal amino acids of subunits are highlighted with a brown background, whereas C-terminal residues are highlighted in magenta. Amino acids that are parts of the peptides that were detected in the Mass Spec analysis are shown in the following colors: VP1 in blue, VP2 in green, and VP3 and VP4 in red. The predicted sequence of VP4 is highlighted with a yellow background. Peptides “C” and “D” that contain VP4 residues and were identified by the LC-MS/MS mass spectrometry are highlighted with black rectangles. These two peptides were identified in the band corresponding to VP3 indicated by an asterisk in A. (C and D) Mass spectra of peptides from the putative VP4 sequences.
We suggested previously that the P domains of SBPV and several other iflaviruses contain a putative catalytic triad Asp300–His283–Ser284 that might have esterase, protease, glycosidase, or lipase activity (10). However, in this study, the membrane lytic activity of SBPV could not be detected under neutral- or low-pH conditions on liposomes composed of phosphatidyl-choline, phosphatidyl-ethanolamine, lysophosphatidyl-choline, sphingomyelin, phosphatidyl-serine, and phosphatidyl-inositol (Fig. S3B). Thus, we can now exclude the possibility that the putative active site in the P domain cleaves the abovementioned lipids.
Conformational changes of capsids associated with genome release were previously studied for viruses from the families Picornaviridae and Dicistroviridae (13, 19⇓–21, 30, 39). There are functionally important differences between the two families as well as with the genome release mechanism of SBPV described here. The uncoating of enteroviruses from the family Picornaviridae is preceded by the formation of expanded A particles (19⇓–21, 23). The A particles contain two types of pores located at twofold and between twofold and fivefold symmetry axes of the capsids, have N termini of VP1 subunits exposed at the virion surface, and spontaneously release VP4 subunits (46, 47). The formation of enterovirus A particles is characterized by the movement of α3 helixes from the VP2 subunits away from the icosahedral twofold axis, which results in the formation of a 9 × 20-Å pore (22, 39). In contrast, the α3 helices of VP2 subunits remain tightly associated in the SBPV empty capsid (Fig. 5 I and J). The buried surface area between the two SBPV helices α3 is 800 Å2 in the native virus and 600 Å2 in the empty particle.
The N-terminal regions of VP1 subunits of enteroviruses contain sequences, which were proposed to form amphipathic α-helices that disrupt endosome membranes, and together with VP4 subunits enable translocation of the virus genome to the cytoplasm (26, 27, 41). Twelve residues from the N terminus of SBPV VP1 become disordered upon genome release (Fig. 5 A and B). However, the disordered peptide is not sufficiently long to reach the surface of the capsid or to interact with a lipid bilayer (48). In contrast, the disordered N termini of VP1 subunits of enteroviruses are about 60 residues long (19, 20). Moreover, a pore at the base of the canyon, which was shown to be the site of the externalization of the N termini of VP1 subunits in coxsackievirus A16 (19, 22), is not present in the empty particle of SBPV. Therefore, the structure of the SBPV empty particle together with the negative results of the liposome lysis experiment described above indicate that the N terminus of SBPV VP1 is unlikely to interact with membranes. The genome release of TrV from the family Dicistroviridae does not involve structural changes to the capsid before the genome release (13). However, the empty capsids of TrV are compact and do not contain any pores that might serve as channels for genome release.
The conformational changes of iflavirus capsid associated with genome release are distinct from those of both picornaviruses and dicistroviruses. Specific features of SBPV genome release are the loss of structure of the N termini of VP2 subunits and the formation of pores around the threefold icosahedral axes of symmetry of the capsid. Unlike in enteroviruses, the contacts between VP2 capsid proteins close to the icosahedral twofold axes are retained in the empty SBPV particle. SBPV virions do not form A particles, and the genome appears to be released through pores located at the threefold axes.
The propagation of SBPV in honeybee pupae and subsequent purification were carried out as described previously (10). Images were recorded in a FEI Titan Krios electron microscope operated at 300 kV with an FEI Falcon II camera. The particles were separated into two half-datasets for all of the subsequent reconstruction steps to follow the “gold-standard” procedure for resolution determination (49). Three-dimensional refinement was carried out using the 3dautorefine procedure of RELION (50).
The propagation of SBPV in honeybee pupae and subsequent purification were carried out as described previously (10).
Aliquots of 15 μL of SBPV at a concentration of 2 mg/mL in 30 mM Hepes, pH 7.2, and 150 mM NaCl were dialyzed into buffers containing 50 mM NaCl and 30 mM sodium acetate at a pH of 6.5, 5.5, 5.0, 4.5, and 3.5 overnight at 4 °C using dialysis buttons (Hampton Research). The dialyzed solutions were clarified by centrifugation at 10,000 × g at 4 °C for 10 min to remove precipitated virus particles. The resulting virus solution was applied to Quantifoil R2/1, mesh 200, copper holey carbon grids and vitrified in liquid ethane using a Vitrorobot mark IV plunge freezing device (FEI) with the settings blot force 0 and blotting time of 2 s. Grids were imaged using a F20 FEI Tecnai transmission electron microscope.
Virions at a concentration of 0.02 mg/mL either in 0.25 M Hepes, pH 7.5, 0.1 M NaCl, or in 0.25 M sodium acetate, pH 5.5, 0.1 M NaCl buffer, were incubated with SYBR Green II (Thermo Fisher Scientific), diluted 3,000 times from the stock solution according to the manufacturer’s instructions. The mixture was heated from 25 to 95 °C in 1 °C increments with a 2-min incubation time at each temperature in a real-time PCR instrument (Roche; LightCycler 480). The fluorescence signal increases either as the dye interacts with RNA that is released from the thermally destabilized particles, or the dye might be able to enter the particles. Readings were normalized by subtracting the fluorescence signal observed at 37 °C from the readings observed at all other temperatures. All measurements were carried out in triplicate. The thermal stability of the virus was estimated as the temperature corresponding to an increase in the fluorescence to 50% of the maximal value obtained when all virions were thermally denatured.
A solution of SBPV at a concentration of 2 mg/mL was dialyzed into 30 mM sodium acetate, pH 5.5, 50 mM NaCl buffer. A volume of 3.5 μL of the solution was applied to a holey carbon grid (Quantifoil R2/1, mesh 300; Quantifoil Micro Tools) and vitrified by plunging into liquid ethane using an FEI Vitrobot Mark IV. Grids with the vitrified sample were transferred to a FEI Titan Krios electron microscope operated at 300 kV, aligned for parallel illumination in nanoprobe mode. The column of the microscope was kept at −196 °C. Images were recorded with an FEI Falcon II direct electron detection camera under low-dose conditions (21 e−/Å2) with underfocus values ranging from 1.0 to 3.0 µm at a nominal magnification of 75,000, resulting in a pixel size of 1.07 Å. Each image was recorded in movie mode with 0.47 s of total acquisition time and saved as seven separate movie frames. In total, 5,325 micrographs were acquired. The frames from each exposure were aligned to compensate for drift and beam-induced motion using the program SPIDER (51).
Particles of SBPV (512 × 512 pixels) were extracted from the micrographs using the program e2boxer.py from the package EMAN2 (52), resulting in 8,668 images of the full virions and 13,633 images of empty particles. Contrast transfer function parameters from each micrograph were estimated using the program ctffind4 (53). The dataset was homogenized by several rounds of 2D classification in RELION (50). The resulting datasets of 8,437 and 12,049 images of full and empty particles, respectively, were subjected to 3D classification using RELION (50), which used the low-pass–filtered (60 Å) structure of the native SBPV virion determined by X-ray crystallography with removed P domains as a starting model. Initial 3D classification of full particles into four classes resulted in datasets containing 3,146, 2,448, 1,566, and 1,277 particles. For the empty particle, the datasets contained 1,699, 2,064, 3,406, and 4,880 images. Classes 1 and 3 from the full-particle classification exchanged particles, and classes 2, 3, and 4 from the empty-particle classification exchanged particles. Therefore, for the final 3D refinement, particles from these classes were pooled, which yielded 4,712 particles for the full-virion reconstruction and 10,350 particles for the empty-particle reconstruction. Subsequently, the particles were separated into two half-datasets for all of the subsequent reconstruction steps to follow the “gold-standard” procedure for resolution determination (49). Final 3D refinement was carried out using the 3dautorefine procedure of RELION (50). The resulting map was masked with a threshold mask and B factor sharpened (−98 Å2 for the full particle and −105 Å2 for the empty particle).
The initial model, derived from the structure of the native SBPV virion (PDB ID code 5J96) (10), was rigid-body fitted into the B factor-sharpened cryo-EM maps using University of California, San Francisco (UCSF) Chimera (54) and subjected to manual rebuilding using the program Coot, and coordinate and B-factor refinement using the program Phenix (55, 56).
The lipids phosphatidyl-choline, phosphatidyl-ethanolamine, lysophosphatidyl-choline, sphingomyelin, phosphatidyl-serine, and phosphatidyl-inositol (purchased from Avanti Polar Lipids) were dissolved in chloroform and mixed in molar ratios of (43:23:13:9:6:6). Chloroform was evaporated for 2 h in a rotary evaporator set to 200 mbar, 10 × g, and 37 °C. The dried lipid film was resuspended in a buffer consisting of 10 mM Hepes, pH 7.0, 100 mM KCl, and 100 mM calcein (self-quenching fluorescent dye). Liposomes were prepared by multiple passages of the rehydrated lipid–buffer mixture through a membrane with an average pore diameter of 100 nm using a mini extruder (Avanti Lipids). The nonencapsulated calcein dye was removed by five rounds of ultracentrifugation for 20 min at 20,000 × g in 4 °C using a Sorvall table top MTX-150 ultracentrifuge. For calcein release, reactions were mixed in a 96-well plate containing 1 mg/mL purified SBPV native virus; purified SBPV viruses heated for 15 min at 56 °C were incubated with 50 μM lipid either in 100 mM NaCl, and 10 mM Hepes, pH 7.0, or in 100 mM NaCl, Na-acetate, pH 5.0. Triton X-100 detergent was used at a concentration of 0.1% (vol/vol). Calcein release from liposomes was monitored as the increase in fluorescence in a FLUOstar Omega fluorescence plate reader (BMG LABTECH) with an excitation wavelength of 485 nm and emission wavelength of 525 nm. The measurements were done in triplicates.
The volumes of the particles were calculated using the programs Mama and Voidoo from the software package Uppsala Software Factory (57). The average radii of the virus particles were calculated using the program Moleman2 from Uppsala Software Factory (57). Figures were generated using the programs UCSF Chimera (54) and PyMOL (The PyMOL Molecular Graphics System, version 1.7.4; Schrödinger). The potential of VP4 to form an amphipathic α-helix was identified using the Heliquest server (43).
Selected protein bands were manually excised from a gel, and after destaining and washing procedures, each band was incubated with trypsin (sequencing grade; Promega) without reduction and alkylation of cysteine residues. The digestion was performed for 2 h at 40 °C on a Thermomixer (Eppendorf). Digested peptides were extracted from gels using 50% (vol/vol) acetonitrile solution with 2.5% (vol/vol) formic acid and concentrated in SpeedVac concentrator (Thermo Fisher Scientific). LC-MS/MS analyses of peptide mixture were done using RSLCnano system (Thermo Fisher Scientific) on-line connected to Impact II Ultra-High Resolution Qq-Time-of-Flight mass spectrometer (Bruker). Before LC separation, tryptic digests were on-line concentrated on trap column. The peptides were separated using Acclaim Pepmap100 C18 column (3-µm particles, 75 μm × 500 mm; Thermo Fisher Scientific; 300 nL/min) by the following gradient program [mobile phase A: 0.1% FA in water; mobile phase B: 0.1% FA in 80% (vol/vol) acetonitrile]: the gradient elution started at 1% of mobile phase B and increased to 30% during the first 35 min, then 56% of mobile phase B in the 50th min and finally increased linearly to 90% in the next 5 min and remained at this state for the final 10 min. MS and MS/MS spectra were acquired in a data-dependent strategy with 3-s-long cycle time. Mass range was set to 150–2,200 m/z and precursors were selected from 300 to 2,000 m/z. Exported MS/MS spectra were searched with in-house Mascot (Matrix Science; version 2.4.1) against the National Center for Biotechnology Information database (no taxonomy restriction) and local database supplied with the expected sequence. Mass tolerances of peptides and MS/MS fragments for MS/MS ion searches were 10 ppm and 0.1 Da, respectively. Oxidation of methionine and propionylamidation of cysteine as optional modifications, one enzyme miscleavage, and correction for one 13C atom were set for all searches. Peptides with statistically significant peptide score (P < 0.05) were considered. Manual MS/MS spectra assignment validation was done.
We thank the Core Facility (CF) Cryo-Electron Microscopy and Tomography and CF Proteomics supported by the Czech Infrastructure for Integrative Structural Biology research infrastructure (LM2015043 funded by Ministry of Education, Youth and Sport of the Czech Republic) for their assistance in obtaining the scientific data presented in this paper. This research was carried out under the project Central European Institute of Technology 2020 (LQ1601). Access to computing and storage facilities owned by parties and projects contributing to the National Grid Infrastructure MetaCentrum, provided under the program “Projects of Large Infrastructure for Research, Development, and Innovations” (LM2010005), is greatly appreciated. Computational resources were provided by the IT4Innovations Centre of Excellence project (CZ.1.05/1.1.00/02.0070; LM2011033). The research leading to these results received funding from the European Research Council under the European Union's Seventh Framework Program (FP/2007-2013)/ERC through Grant 355855 and from European Molecular Biology Organization installation Grant 3041 (to P.P.).
Author contributions: P.P. designed research; S.K. and A.P. performed research; J.d.M. contributed new reagents/analytic tools; S.K., T.F., and P.P. analyzed data; and S.K. and P.P. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: Cryo-EM electron density maps of the low-pH SBPV virion and empty particle have been deposited in the Electron Microscopy Data Bank, https://www.ebi.ac.uk/pdbe/emdb/ (accession numbers EMD-4063 and EMD-4064), and the fitted coordinates have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 5LK7 and 5LK8, respectively).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1616562114/-/DCSupplemental.
Freely available online through the PNAS open access option.
