ABSTRACT
A total of 2,691 mosquitoes representing 17 species was collected from eight locations in southwest Cameroon and screened for pathogenic viruses. Ten isolates of a novel reovirus (genus Dinovernavirus) were detected by culturing mosquito pools on Aedes albopictus (C6/36) cell cultures. A virus that caused overt cytopathic effects was isolated, but it did not infect vertebrate cells or produce detectable disease in infant mice after intracerebral inoculation. The virus, tentatively designated Fako virus (FAKV), represents the first 9-segment, double-stranded RNA (dsRNA) virus to be isolated in nature. FAKV appears to have a broad mosquito host range, and its detection in male specimens suggests mosquito-to-mosquito transmission in nature. The structure of the T=1 FAKV virion, determined to subnanometer resolution by cryoelectron microscopy (cryo-EM), showed only four proteins per icosahedral asymmetric unit: a dimer of the major capsid protein, one turret protein, and one clamp protein. While all other turreted reoviruses of known structures have at least two copies of the clamp protein per asymmetric unit, FAKV's clamp protein bound at only one conformer of the major capsid protein. The FAKV capsid architecture and genome organization represent the most simplified reovirus described to date, and phylogenetic analysis suggests that it arose from a more complex ancestor by serial loss-of-function events.
IMPORTANCE We describe the detection, genetic, phenotypic, and structural characteristics of a novel Dinovernavirus species isolated from mosquitoes collected in Cameroon. The virus, tentatively designated Fako virus (FAKV), is related to both single-shelled and partially double-shelled viruses. The only other described virus in this genus was isolated from cultured mosquito cells. It was previously unclear whether the phenotypic characteristics of that virus were reflective of this genus in nature or were altered during serial passaging in the chronically infected cell line. FAKV is a naturally occurring single-shelled reovirus with a unique virion architecture that lacks several key structural elements thought to stabilize a single-shelled reovirus virion, suggesting what may be the minimal number of proteins needed to form a viable reovirus particle. FAKV evolved from more complex ancestors by losing a genome segment and several virion proteins.
INTRODUCTION
Human and livestock diseases are commonly caused by viruses transmitted from enzootic hosts by arthropod vectors (1). Mosquito surveillance studies have been key to understanding the prevalence, distribution, and outbreak potential of the majority of these pathogens, as well as for the detection of novel pathogenic and nonpathogenic agents.
The family Reoviridae consists of double-stranded RNA (dsRNA) viruses with 9 to 12 genome segments. Reoviruses have both a wide geographic distribution and wide host range, having been isolated from fungi, plants, insects, ticks, arachnids, fish, marine protists, crustaceans, mammals, and birds (2). They are divided into two subfamilies, Spinareovirinae and Sedoreovirinae, which are classified based on their core structures. Spinareovirinae viruses, such as the recently emerged human pathogen Melaka virus and the mammalian orthoreovirus type 3, are called “turreted” reoviruses, because they encode spikes (protrusions) on the surface of the virion. The subfamily Sedoreovirinae contains the “nonturreted” viruses with multiple smooth shells, which include human and other animal pathogens, such as rotaviruses and orbiviruses (3, 4). In addition to this structural classification, amino acid identities of >30% in the polymerase sequence and the presence and sequence of conserved terminal nucleotide (nt) motifs have been used as criteria for the classification of genera within this family (2, 5–9).
The genus Dinovernavirus (subfamily Spinareovirinae) contains a single isolate designated Aedes pseudoscutellaris reovirus (APRV) that was isolated from the AP61 mosquito cell line (8), which was established in 1974 from A. pseudoscutellaris larvae from Fiji (10). Phylogenetic analysis of APRV sequences indicates that it represents a distinct genus within the family and that its closest genetic and structural relationships are with reoviruses in the genera Cypovirus, Fijivirus, and Oryzavirus (8).
APRV is the only reovirus known to contain a 9-segment dsRNA genome and, with the exception of cypoviruses (CPVs), is the only other reovirus to have a single-shelled virion. The innermost shell contains two interlocking peanut-shaped monomers of the major capsid protein, which are chemically but not structurally identical. Therefore, this T=1 layer has sometimes been described as “T=2” to emphasize the conformational nonequivalence of the two chemically identical monomers per asymmetric unit. Most reoviruses contain at least part of a second shell, comprised mainly of trimers arranged in a T=13l lattice. In contrast to cypoviruses that form polyhedra/occlusion bodies, APRV is nonoccluded (8). The structure of the cypovirus cytoplasmic polyhedrosis virus is the highest-resolution single-particle cryoelectron microscopy (cryo-EM) structure solved to date (11). The single-shell capsid of cypoviruses is also thought to have an unusual stability (12, 13), which has been explained by the inclusion of the small protrusion (SP) domain in the major capsid protein, the N-terminal anchor of the major capsid protein, and the presence of “clamp” (also called “cement”) proteins, which are homologs of aquareovirus VP6 and orthoreovirus σ2 (11, 14).
Here, we describe the first Dinovernavirus (family Reoviridae) to be detected in nature, which was isolated during a surveillance study in the Fako division of Cameroon in 2010 and tentatively designated Fako virus (FAKV). FAKV is related to both single-shelled and partially double-shelled viruses (3) and has minimal genetic composition. These characteristics suggest that its virion structure could shed light on the evolutionary history and structural diversity of this family. A subnanometer three-dimensional (3D) structure of the FAKV virion determined by cryo-EM and single-particle reconstruction revealed a simplified virion architecture with only four polypeptides per asymmetric unit. We therefore hypothesize that FAKV evolved from a more complex ancestor by serial loss-of-function events within the subfamily. We also describe for the first time the genetic, ecological, and phenotypic characteristics of a naturally circulating dinovernavirus, as well as show evidence of mosquito-to-mosquito transmission in nature.
MATERIALS AND METHODS
Mosquito collection and virus isolation.Between 15 July and 8 August 2010, mosquitoes were collected at seven locations in Cameroon by using a combination of landing collections and CDC light traps (John Hock, Gainesville, FL) baited with dry ice. After collection, mosquitoes were sorted by sex, date, and location of collection into conspecific pools ranging in size from one to 50 individuals. Pools were frozen at −80°C until assayed for virus on cultures of Vero and C6/36 cells as described previously (15, 16).
Virus purification and transmission electron microscopy.Aedes albopictus (C6/36) cells were grown to 80% confluence in eight 150-cm2 culture flasks and infected with FAKV at a multiplicity of infection (MOI) of ∼0.1 PFU/cell. Three days postinfection, cell culture supernatants were collected and virus was precipitated using 7% polyethylene glycol (PEG) 8000 and 2.3% NaCl (wt/vol). Pellets containing virus were resuspended in 4 ml of TEN (10 mM Tris-HCl, pH 7.5; 1 mM EDTA, pH 7.8; and 0.1 M NaCl) buffer, pH 7.8, and purified on a 20% to 70% continuous sucrose gradient by centrifugation at 210,000 × g for 1.5 h using an SW41Ti rotor (Beckman, Indianapolis, IN). The virus band was collected and applied to a 100-kDa Amicon filter (Millipore, Billerica, MA). Sucrose was then removed by 10 washes with TEN. Purified virus was harvested in 100 μl of TEN with or without magnesium (10 mM) and stored at 4°C. Transmission electron microscopy of thin sections from infected C6/36 cells and negatively stained carbon-coated copper grids containing purified virions was performed as previously described (16).
Virus propagation and RNA extraction.Virus was subsequently inoculated onto an 80% confluent C6/36 cell monolayer in a 150-cm2 culture flask. The culture medium was collected 6 days postinfection and centrifuged at 1,860 × g to pellet cell debris. The supernatant was collected and PEG was added to 7% (wt/vol), after which the mixture was incubated at 4°C for 24 h prior to centrifugation at 3,100 × g for 40 min. The supernatant was removed and the virus pellet resuspended in 250 μl of TEN buffer by repeated pipetting. Viral RNA was extracted from purified virions using TRIzol LS (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions.
Sequencing.The complete genome sequences for 4 isolates were determined using the Illumina HiSeq1000 platform (Illumina, Inc., San Diego, CA) as previously described (17). Intrahost genomic diversity was assessed using VarScan v2 with a 3% threshold (18).
Sequence analysis.Nucleotide and open reading frame (ORF) sequences were submitted to BLASTx and BLASTp to identify sequences that share homology with FAKV. Protein families were identified based on BLASTp analysis. Nucleotide and amino acid sequences for the ORFs within all 9 segments were determined with EnzymeX using a 200-nt threshold for each ORF (Aalsmeer, Netherlands).
Phylogenetic analysis.Previously published reovirus RNA-dependent RNA polymerase (RdRp) amino acid sequences (n = 29) were downloaded from GenBank and aligned with the FAKV RdRp (ORF detected within segment 2) translated sequence using the ClustalW program and then manually adjusted in Se-Al (http://tree.bio.ed.ac.uk). Sequence identities were calculated using GENEIOUS v5.6 (Auckland, New Zealand). A neighbor-joining (NJ) phylogenetic tree was constructed using PAUP* version 4.0b (Sinauer Associates, Inc., Sunderland, MA, USA), with Mimoreovirus acting as an outgroup. Bootstrapping was used to assess the robustness of tree topologies using 1,000 replicate NJ trees.
Given the large sequence divergence within the subfamily Spinareovirinae and to differentiate ancestral from derived clades, a Bayesian coalescent analysis was performed using BEAST (19). BEAST utilizes a Bayesian Markov chain Monte Carlo (MCMC) approach to jointly estimate trait diffusion patterns from sampled sequences, while concurrently taking into account phylogenetic uncertainty arising from both the sequence data and trait diffusion processes. The statistical support for each trait under study is therefore obtained by implementing the analyses over all plausible trees, thereby removing the bias of using a parsimony approach to assess trait evolution over a single phylogeny that may not be truly accurate. All GenBank reference RdRp sequences for the Spinareovirinae were analyzed along with FAKV, assuming a Yule speciation process and an uncorrelated lognormal molecular clock (20). Tracer version 1.5 (http://tree.bio.ed.ac.uk/software/tracer/) was used to monitor stationarity and efficient topology mixing, as diagnosed using effective sample size (ESS) statistics. TreeAnnotator version 1.8.0 (http://beast.bio.ed.ac.uk) was used to summarize the posterior tree distribution and FigTree version 1.3.1 (http://beast.bio.ed.ac.uk) to visualize the annotated maximum clade credibility (MCC) tree.
Virus quantification.A plaque assay based on A. albopictus (C7/10) cells was used for quantification of FAKV as previously described (21), except that fixed cells were stained with 2% crystal violet.
Cryoelectron microscopy.Purified virions were applied to 200-mesh Quantifoil grids (2-μm hole size) and blotted inside a Vitrobot Mark IV (FEI Company) for 3.5 s at 6°C. From virions purified in TEN buffer only, 1,385 particles at a pixel size of 2.78 Å/pixel were selected from data imaged on a JEM3200FSC 300-kV cryoelectron microscope (JEOL Ltd., Tokyo, Japan) and a DE-20 camera (Direct Electron, LP, San Diego, CA) at 16 frames per second. Motion correction and damage compensation were performed on a per-particle basis to combine information from the 32 frames per specimen area (22, 23). From virions purified in TEN buffer with supplemental magnesium, 2,464 full particles (intact capsid and genome) were manually boxed with EMAN2 (24) from data imaged on a JEM2010F electron cryomicroscope (JEOL Ltd., Tokyo, Japan) and a USA4000 charge-coupled device (CCD; Gatan, Pleasanton, CA) at 3.62 Å/pixel.
Cryo-EM density map and interpretation.For 3D reconstruction, data derived from virus purified with and without Mg2+ were divided into halves, and each was refined independently. EMAN2.1 was used for automated boxing, contrast transfer function correction, and generation of data-derived initial models. 2D alignment and 3D reconstruction were performed with EMAN2.1 (24) in the case of full particles and with multipath simulated annealing (MPSA) (25) in the case of empty particles. The Fourier shell correlation (FSC) between the structures from independent halves was used to measure the resolution of the combined map (26). Visualization of the density map was performed in UCSF Chimera (27). All filtration operations were performed in EMAN2. To measure the volume of regions of the asymmetric unit, semiautomatic segmentation of the capsid was performed using Segger (28).
To determine the location of secondary structural elements and clarify protein boundaries in the map, subvolumes were extracted, analyzed, and repositioned in the original map. For each protein in the asymmetric unit, a subvolume extending 20 Å out from the location of its closest structurally characterized homolog from CPV was extracted. Gorgon was then used to automatically skeletonize density and to manually identify secondary structural elements in the extracted subvolume (29, 30).
Homology models for all three proteins were constructed using Modeler v9.8 (31, 32). Where secondary structural elements identified in Gorgon corresponded to secondary structural elements in the homology model, the path between these elements was traced manually in Chimera. Regions of uncertainty were not traced or segmented; these include the termini of the major capsid proteins. Finally, the partially traced paths were used to define the borders of polypeptides in the original map.
Structural protein characterization.Proteins extracted from purified virions were subjected to polyacrylamide gel electrophoresis under denaturing and reducing conditions on a 4% to 20% gradient gel (Bio-Rad, Hercules, CA). Gels were stained with Coomassie blue or with the SilverQuest silver staining kit (Life Technologies, Carlsbad, CA). Three bands were excised from the silver-stained gel, and in-gel trypsin digestion was performed. Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) was performed, and peptides were identified by searching the NCBI database or all predicted FAKV ORFs with Mascot (Matrix Science, Boston, MA).
In vitro culture of FAKV.Cell lines derived from African green monkey kidney (Vero), baby hamster kidney (BHK-S), and mosquitoes (A. albopictus; C6/36 and C7/10) were inoculated with FAKV at a multiplicity of infection of 10 PFU/cell (titers based on C7/10 cell plaque assays), using cell culture supernatant from infected C6/36 cells as the inoculum. Mosquito cultures were incubated at 28°C and observed for cytopathic effects (CPE) for 7 days; mammalian cells were maintained at 37°C and observed for 14 days. Dulbecco's minimal essential medium (DMEM) containing 5% fetal bovine serum (FBS), 1% penicillin-streptomycin, and 1% tryptose phosphate broth was used to maintain mosquito cells postinfection, and DMEM containing 2.5% FBS and 1% penicillin-streptomycin was used to maintain mammalian cells. Culture fluid from each of the mammalian cell lines was then serially blind-passaged three times, using undiluted supernatant from the previous passage as the inoculum. Cells were observed for CPE, and the third-passage supernatant was screened for FAKV RNA by reverse transcription (RT)-PCR using primers based on the VP9 segment (available from authors on request). This method detected FAKV RNA at a titer of 0.01 PFU/ml. One-step RT-PCRs were performed with the Titan one-step RT-PCR kit (Roche Diagnostics, Indianapolis, IN).
Inoculation of mice.Culture medium from C6/36 cells infected with FAKV was used to inoculate intracerebrally 10 2-day-old CD1 mice (∼105 PFU/ml). Mice were observed for signs of illness for 14 days. Animal experiments were approved by the UTMB Institutional Animal Care and Use Committee (IACUC), protocol number 9505045.
RESULTS
Mosquito collection and FAKV isolation.A total of 138 pools containing 2,691 mosquitoes representing 17 species (see Table S1 in the supplemental material) were collected and screened for arboviruses by CPE assays on Vero and C6/36 cells. Ten cytopathic viruses were isolated via screening on C6/36 cells (Table 1), but no CPE were observed on Vero cells. Locations and GPS coordinates in decimal degrees where the 10 isolates were made include Ekona (4.12521, 9.19569), Malende (4.20305, 9.25555), and Muea (4.10224, 9.18398). Cytopathic effects were observed 4 days postinfection in both C6/36 and C7/10 cells (Fig. 1), which are different clones of an A. albopictus cell line. These viruses could not be identified using standard immunofluorescence techniques against commonly detected, medically important orthobunyaviruses, flaviviruses, alphaviruses, rhabdoviruses, and orbiviruses. To assess whether the observed CPE were a result of virus infection, transmission electron microscopy of ultrathin sections of infected C6/36 cell monolayers was performed and revealed the presence of a single-shelled, turreted virus that was similar in size and structure to turreted reoviruses such as cypoviruses, dinovernaviruses, and oryzaviruses. Virus particles clustered primarily in the host cell cytoplasm and inside cytoplasmic vacuoles (see Fig. S1A and B in the supplemental material). Negative-stain imaging suggested that FAKV virions were approximately 450 to 500 Å in diameter and were icosahedral in shape, with 6 of the 12 large turrets visible in projection images (see Fig. S1C).
Viruses isolated in this study
Phase-contrast micrograph showing the cytopathic effects of CSW77 in representative cell lines: (A) negative control for C7/10; (B) C7/10 cells infected with CSW77. Note the significant syncytium formation and cell adhesion postinfection. (C) Typical plaques observed by CSW77 infection on C7/10 cells.
Analysis of FAKV nucleotide sequences.Illumina sequencing performed on purified viral RNA revealed the presence of a nine-segment dsRNA genome totaling 23,170 nt. Approximately 41% of the reads in the prototype CSW77 sample mapped to the viral genome, resulting in about 5.78 million read pairs mapped out of 14.1 million total in the sample. The percentage of reads mapping to the viral genome in the other 3 strains ranged from 3.5% to 20.5%. Host and viral sequences were also extracted from the resultant sequencing reads in an attempt to determine the presence of any additional viral sequences, but none was detected in any of the 4 samples tested. For prototype strain CSW77, BLASTx searches on translated amino acid sequences indicated the presence of a Dinovernavirus (family Reoviridae) tentatively designated Fako virus (FAKV), which belongs to a recently described genus containing a single, insect-specific virus known as Aedes pseudoscutellaris reovirus (APRV) (8). Based on the RdRp, FAKV shared the highest amino acid sequence identity, at 89.7%, with APRV, followed by the cypoviruses with ∼20% similarity (see Table S2 in the supplemental material). Interestingly, submission of FAKV sequences to UniProt Metagenomic and Environmental Sequences (UniMES) resulted in no matches, suggesting this virus has not been detected during previous metagenomic studies. All 10 isolates of FAKV shared >99% sequence identity across 850 nt of the segment 9 sequenced, which suggests that they belong to a single replicating lineage. This was further supported by complete genome sequencing of 4 additional isolates.
Table 2 shows the genome organization of the prototype FAKV strain, CSW77. Segment names 1 through 9 were designated based on similarity in sequence identity to APRV. The lengths of the various segments were very similar between FAKV and APRV, with the majority of the segments differing by only 1 to 3 nucleotides, except for segments 4 (28-nt difference) and 9 (7-nt difference). The terminal sequences of FAKV were conserved, beginning with AGU at the 5′ end and ending with AGU at the 3′ end (Table 2). This pattern is conserved among the dinovernaviruses, cypoviruses, and fijiviruses.
Lengths of Fako virus genome segments, predicted proteins, and conserved terminal sequences at the 5′ and 3′ endsa
We evaluated the intrahost genomic diversity of FAKV to determine if it increases its genetic diversity by incorporating large numbers of single-nucleotide polymorphisms (SNPs), insertions, or deletions. We performed this analysis by realigning all Illumina sequencing reads against the viral genome. The results indicated only a few SNPs (see Table S4 in the supplemental material), suggesting limited intrahost sequence diversity.
Phylogenetic analysis of FAKV.A neighbor-joining (NJ) phylogeny based on RdRp amino acid sequences of 29 representative members of the family Reoviridae and prototype strain CSW77 provided further support (i.e., 100% bootstrap) for FAKV's assignment to the genus Dinovernavirus (Fig. 2). This phylogenetic position is also supported by the inferred Bayesian MCC phylogeny based on representative Spinareovirinae RdRp sequences (Fig. 3). Table S3 in the supplemental material shows the species used in the phylogenetic analyses. There was no evidence of reassortment, as all FAKV's segments were more similar to the corresponding segments of APRV (71 to 85% nucleotide identity) than they were to segments of other reoviruses.
Neighbor-joining phylogeny of representative Reovirus sequences based on an amino acid alignment of the RdRp protein. Nodes are labeled with bootstrap values greater than or equal to 95%. FAKV is highlighted in bold. The scale bar indicates percent amino acid divergence. Members of the subfamily Spinareovirinae are highlighted in blue, and members of the subfamily Sedoreovirinae are highlighted in red. The various genera are also indicated in bold. Tip labels include the ICTV abbreviations. Sequences used for phylogenetic analyses are shown in Table S4 in the supplemental material.
Bayesian MCC tree for the subfamily Spinareovirinae based on 1,232 amino acids of the RdRp protein. Terminal branches of the tree are colored according to the trimer positions of the taxon at the tip. Internal branches are colored according to the most probable (modal) trimer positions of their parental nodes. All internal nodes had posterior probabilities (clade credibilities) of ≥0.95. Clamp proteins, their positions, and their probabilities of occurrence at the relevant nodes are indicated in bold. The number of genomic segments each genus contains is highlighted in color. The scale bar represents nucleotide substitutions per site.
FAKV phenotypic characteristics.FAKV was detected in 3 mosquito species within the genus Aedes and at least one species within the genus Eretmapodites (Table 1); it was not detected in any of the other 4 mosquito genera tested, including known Culex and Anopheles arboviral vectors. FAKV was detected in male specimens of both A. albopictus and Eretmapodites dracaenae, suggesting vertical or venereal transmission in nature. Whether plants, vertebrates, or other invertebrates may play a role in FAKV's transmission requires further testing.
Unlike APRV, the prototype virus of the Dinovernavirus genus, FAKV, caused cytopathic effects in mosquito cells; but like APRV, it appears to lack the ability to replicate in mammalian cells. Cytopathic effects were apparent by day 5 postinfection in both A. albopictus cell lines tested. Figure 1 shows the typical CPE (delayed replication and cell aggregation) and plaques created by FAKV infection in C7/10 cells. Interestingly, FAKV lost the ability to cause CPE in C6/36 cells upon serial passaging (>5 passages) in that line. Sequencing these multiply passaged strains might lead to the identification of mutations associated with reduced cytopathogenicity. These results suggest that the failure of APRV to cause CPE reflects adaptation to replication in cell culture due to serial passaging in a cell line that was first established in 1974. Another difference observed between APRV and FAKV was that APRV lost infectivity after freezing at −20°C or −80°C, but FAKV remained infectious and showed no decline in titer following such treatment (stored for up to 2 years at −80°C). Treatment with various concentrations of NP-40 detergent did not abolish FAKV infectivity (data not shown).
Lack of FAKV replication in vertebrate cells.FAKV did not produce CPE and was not detected by RT-PCR after serial, blind passages in either of the mammalian cell lines tested. Furthermore, none of the 10 intracerebrally inoculated, 2-day-old mice showed signs of illness or disease over the 14-day observation period. Our preliminary characterization of FAKV as an insect-specific reovirus supports previous observations with APRV and the insect-specific nature of this genus.
3D map of genome-free FAKV capsids.Cryo-EM micrographs of FAKV exhibited particles with a diameter of roughly 580 Å, not including the protruding spikes (Fig. 4A). Both genome-associated and genome-free particles were observed, with genome-free capsids predominating, especially in viral preparations purified in the presence of EDTA. However, adding 50 mM EDTA to purified samples prepared without EDTA did not change the ratio of genome-associated to genome-free particles after 1- or 24-h incubation at 4°C (data not shown). This suggests that preparation conditions other than cation concentration contributed to this ratio of empty to full particles. The structure of the FAKV genome-free capsid was determined to a 7.8-Å resolution (Fig. 4B) by single-particle cryoelectron microscopy using icosahedral averaging and was deposited in the electron microscopy databank (EMD-6001). The resolution was estimated by independently reconstructing two halves of the data set (33), known as the current gold standard (see Fig. S2 in the supplemental material). α-Helices were well resolved in all polypeptides of the capsid, but the individual strands within β-sheets were not differentiated (Fig. 5).
The 7.8-Å structure of Fako virus determined by cryoelectron microscopy. (A) A typical micrograph collected with the DE-20 camera illustrates the preponderance of empty particles in preparations purified in TEN buffer. Particle boxes were extracted from each subframe, and particles were reconstituted by summing aligned, damage-corrected boxes. (B) The structure of the genome-free capsid was solved using the multipathway simulated annealing package and is colored radially. The inset shows a cutaway revealing the empty capsid interior and is labeled with radial measurements of the capsid outer surface, the turret peak, and the capsid inner surface. The structure of the FAKV full capsid was solved to 17 Å (C) to confirm that the absence of clamp near the 3-fold axis was not an artifact of genome release and agrees with the empty capsid structure (B). The inset shows a cutaway revealing ordered layers of dsRNA within the virion.
Protein composition of FAKV virions. Three segmented asymmetric units of FAKV (A) are colored by chain with different tints of the same color, representing symmetry-related copies of a protein. There are four proteins per asymmetric unit: a turret (purple) at the 5-fold axis, a dimer of the major capsid protein (blue and yellow), and a clamp protein (red) that binds to the major capsid protein surface. In contrast, the previously published structure of CPV (B) (11) and all other turreted reoviruses contain one clamp for each major capsid protein (red and green), and some have an additional clamp spanning the 2-fold axis. The turret protein (C to E), clamp protein (F to H), and major capsid protein (I to K) are contrasted between CPV and FAKV. The FAKV density map of each protein (C, F, I; colored as described above) was segmented from the 7.8-Å empty capsid reconstruction. A CPV map was obtained from the EM Data Bank (EMDB5256) (11) and filtered to similar resolution (D, G, J; gray). The FAKVα-helices (light green) and β-sheets (dark green) were compared to the α-helices (blue) and β-strands (cyan) of the CPV atomic structure (PDB 3IZX). While most secondary structural elements in the turret (E) and major capsid protein (K) are conserved, the clamp protein varies greatly around a conserved helix-barrel core (H).
FAKV was determined to be a single-shelled reovirus with gross morphology similar to that of Bombyx mori CPV or to rice ragged stunt virus (RRSV) (see Fig. S3 in the supplemental material) (3, 34, 35). Each of the 60 asymmetric units in FAKV can be segmented into 4 density regions using Chimera and Gorgon, which are annotated in different colors (Fig. 5A). Similar results were obtained using Segger segmentation software (see Fig. S4). Two of these regions (cyan and yellow) formed an icosahedral shell with inner and outer radii of approximately 250 and 290 Å, respectively (Fig. 4B). When these densities were computationally extracted and aligned to each other, they agreed to 7.9 Å at an FSC of 0.5, which is close to the resolution of the map. This structure is consistent with the organization of all reoviruses, whereby each asymmetric unit contains two copies of the major capsid protein in similar conformations for a total of 120 monomers per capsid. Among the 26 α-helices and 8 β-sheets identified with Gorgon, all α-helices and 5 β-sheets of this major capsid protein were also seen in the major capsid protein of CPV, with which the predicted open reading frame of FAKV segment 3 shares 24% amino acid identity (Fig. 5K).
The FAKV shell was decorated with 12 turrets protruding 360 Å from the center of the particle, each with a radius of 75 Å. A protruding density occupying about 50,000 to 60,000 Å3 occurred once per asymmetric unit adjacent to the turret (Fig. 4B and 5A and F; see also Fig. S3 in the supplemental material). This region was occupied by the clamp (also termed “cement”) protein in cypovirus (large protrusion protein [LPP]) (12) (Fig. 5B; see also Fig. S3C), aquareovirus (VP6) (14), and orthoreovirus (σ2) (4). When Gorgon was used to identify secondary structural elements in this density (Fig. 5H), locations of 9 of 12 α-helix and 1 of 2 β-sheets matched to those in the CPV LPP clamp protein (Fig. 5G) with which it shared 23% amino acid identity. In contrast, fewer than half of its helices and no β-sheets were shared with the aquareovirus VP6 clamp protein, with which it shared no sequence homology detectable by BLASTp (36).
The organization of major capsid protein's helix-rich lobe is highly conserved among reoviruses (37), making it a useful benchmark to validate the reconstruction and segmentation steps used in our cryo-EM studies of the genome-free FAKV virion. In contrast to the usually structurally divergent clamp protein, comparison of the segmented α-helices in the major capsid protein helix-rich lobe to those of its closest homolog revealed no significant differences (Fig. 5C to E).
We observe a density beneath the turret that likely represents the polymerase complex (see Fig. S3E in the supplemental material). This component was blurred; no helices were resolved. When a nonicosahedral feature is icosahedrally averaged, this effect is typically observed (38). A similar density can be seen in the structure of rice dwarf virus (25, 39).
Identification of FAKV proteins and assignment to open reading frames.Coomassie-stained SDS-PAGE gels of proteins extracted from purified virions revealed four bands, whose pattern was remarkably similar to that observed for purified CPV virions (35). Fresh virions were electrophoresed and silver stained to excise bands for mass spectrometry, and intense bands with estimated molecular masses of roughly 136, 120, 61, and 40 kDa were observed, as well as genome RNA segments and very faint protein bands that may represent trace contaminants or degradation products (Fig. 6A).
Electrophoresis of FAKV proteins. (A) Purified FAKV virions were analyzed by SDS-PAGE and silver staining. Decreasing 10-fold dilutions in water of the same FAKV preparation are indicated by a triangle beneath lanes 1 to 3. (B) Putative open reading frames identified in the nine-segment genome of FAKV are colored to match their location in the virion. (C) Protein chains of one asymmetric unit of FAKV were segmented using Chimera and Gorgon. The folds and positions of these chains were similar to the turret, major capsid, or clamp of other Spinareovirinae. Open reading frames and protein chains are colored goldenrod and teal for two copies of the major capsid protein, purple for the turret protein, and red for the clamp protein. Lines indicate assignment of an SDS-PAGE band to a structural ORF.
The band of about 136 kDa was much more intense than any other and therefore likely corresponds to the major capsid protein, because in all other reoviruses this protein is present at 120 copies per virion, more than any other protein in the virus inner shell. We observed two copies of the major capsid protein per asymmetric unit in the capsid structure (Fig. 5A). Mass spectrometry revealed the dominant component of this band to be FAKV segment 3, with a predicted mass of 136 kDa, which also exhibits sequence homology to the major capsid proteins of other reoviruses.
The RdRp is the most conserved protein among reoviruses, and we assigned FAKV segment 2 to the RdRp based on its homology to other reovirus RdRp proteins. The RdRp should be present at 12 or fewer copies per virion, and its predicted molecular mass is 143 kDa. An independent band corresponding to RdRp was not seen. The RdRp is present in a relatively low copy number per virion. Also, the thick band of the 136-kDa major capsid protein should migrate at a similar rate and could obscure an RdRp band. Consistent with this interpretation, two FAKV proteins exhibited statistically significant protein scores in the mass spectrometric analysis of the bright band around 136 kDa: the segment 3 major capsid protein described above and FAKV segment 2's RdRp. Fewer tryptic peptides were observed for segment 2 (107 peptides matching to the segment 2 ORF compared to 1,706 peptides matching to the segment 3 ORF), consistent with its lower copy number per virion.
The band of about 120 kDa, second in intensity, likely corresponds to the turret protein, which we assigned by homology to FAKV segment 5 with a predicted mass of 121 kDa. One copy of the turret was present per asymmetric unit in our cryo-EM capsid structure.
The 61-kDa protein was assigned by mass spectrometry to FAKV segment 6, with a predicted molecular mass of 62 kDa. The only significant BLAST hits were cypovirus proteins homologous to CPV protein V4. In CPV, this protein is present in virions at substoichiometric concentrations (35), but its function is unknown and it has never been located in particles. The leucine zipper motif previously predicted in V4 (40) is disrupted in FAKV.
The 40-kDa band, which had higher intensity than the 61-kDa band, was assigned by mass spectrometry to FAKV segment 7, whose predicted molecular mass is 40 kDa. The only detectable homologous sequence was APRV segment 7, which was previously annotated as a putative nonstructural protein with no known homologs (8). The cryo-EM structure of FAKV contained three unique density regions per asymmetric unit, corresponding to the major capsid protein (segment 3), the turret protein (segment 5), and the clamp protein. These three unique density regions should correspond to the three most abundant proteins among purified virions. This suggests that segment 7 is the FAKV clamp protein.
In support of the assignment of segment 7 as the FAKV clamp, we measured the volume of the clamp directly and by comparing the clamp volume to the capsid volume. Assuming a specific volume of 0.73 ml/g (41), the density of the putative clamp protein region would be occupied by roughly 40 to 50 kDa of protein. Local contrast variations were normalized in EMAN2.1, and the ratio of clamp protein volume to total capsid volume was measured as 9.74%. Since an asymmetric unit consists of one turret of 121 kDa and two major capsid proteins at 136 kDa each, the clamp mass would be roughly 42 kDa. Therefore, of all the bands observed by electrophoresis of purified virions, the abundant band corresponding to segment 7 matches the measured clamp volume better than any other band.
In Oryzavirus, a 43-kDa clamp protein is produced from a 67-kDa precursor (42), a process which also occurs in CPV (40). However, both the volume measurement by cryo-EM and densitometry of electrophoresed purified virions yielded estimates of the clamp mass that are consistent with the full mass of the segment 7 open reading frame.
In summary, four polypeptides were identified and assigned to ORFs in the FAKV asymmetric unit (Fig. 5A): the turret protein, two subunits of the major capsid protein dimer, and a clamp protein adjacent to the turret structure. The RdRp and a protein of unknown function were also present at less than one copy per asymmetric unit. Figure 6B indicates the putative functions of the various predicted proteins based on our study and comparisons with APRV and other reoviruses. Protein lengths and masses were generally comparable; however, several other possible proteins were detected in alternative reading frames. Three of the smaller open reading frames detected within segments 1, 3, and 4 did not show detectable sequence homology to other reovirus proteins, and the functions of these proteins remain unknown. It is unclear whether these proteins may truly exist or be expressed in nature.
Absence of the clamp protein near the 3-fold axis.Unique among the Spinareovirinae, FAKV lacked the second clamp protein near the icosahedral 3-fold symmetry axis (see Fig. S3 in the supplemental material). To confirm that the unexpected absence of the clamp protein was not an artifact of empty capsid preparation, we reconstructed full particles purified in TEN buffer and magnesium to a 17-Å resolution (Fig. 4C). The Fourier shell correlation between masked maps of the empty and full capsids was 0.5 at 20 Å (see Fig. S2 in the supplemental material), which was similar to the resolution of the full-capsid map. This suggests that the maps are largely identical in the resolution range at which a clamp protein was observable close to the turret near the 5-fold axis in both maps (Fig. 5A; see also Fig. S3). We note that the ordered shells of dsRNA were interrupted by the presence of a density beneath each turret, corresponding to the putative RdRp also observed in the empty capsids.
Ancestral trait reconstruction within the subfamily Spinareovirinae.We performed a Bayesian coalescent analysis on the RdRp sequences of 22 representative reovirus reference sequences within the subfamily Spinareovirinae using BEAST (19), incorporating ancestral trait reconstructions (using a discrete diffusion model) to trace the evolutionary events that led to the emergence of FAKV structural and genetic characteristics. The tree topology was identical to that of the rooted NJ phylogeny, despite differences in representative sequences and phylogenetic method employed.
Our analysis predicted that the ancestor of the Spinareovirinae had 600 outer shell trimers forming a partial T=13l outer layer and two clamps per asymmetric unit (Fig. 3). The five quasiequivalent positions in a T=13 shell are named by the letters P through T (see Fig. S3E in the supplemental material); only the P position nearest the 5-fold vertex was unoccupied in the ancestor of turreted reovirues. The MCMC phylogeny also showed that cypoviruses and dinovernaviruses evolved from a 10-segment ancestor, with no trimer proteins existing per asymmetric subunit, and with clamp proteins at both the 3- and 5-fold positions (Fig. 3). Dinovernaviruses subsequently lost a genomic segment and the 3-fold clamp from this ancestral virus.
DISCUSSION
FAKV phenotypic characteristics.The isolation of FAKV from male mosquitoes (which do not take blood meals) and the lack of replication in vertebrate systems both suggest vertical rather than horizontal transmission. In particular, intracerebral inoculation of infant mice is typically a highly permissive in vivo environment for mammalian arboviruses. The 10 FAKV isolates detected from mosquitoes of different species and genders were very similar in sequence, suggesting that they are members of a single evolving lineage. The isolation of a single replicating FAKV lineage from several mosquito genera also suggests some form of horizontal transmission, possibly through simultaneous feeding on carbohydrate sources. If the virus underwent long-term coevolution with individual mosquito species, we would expect multiple lineages and considerably more genetic diversity among the FAKV sequences determined. FAKV likely occupies a distinct ecological niche from APRV, which has not been observed in the wild, and whose probable host A. pseudoscutellaris is a Pacific Island mosquito that exists under different climatic conditions (43). Taken together with the sequence divergence in the RdRp gene alignment (24%), we therefore propose that all 10 isolates described herein belong to a novel species, which we name Fako virus, and that FAKV and APRV are distinct member species of the genus Dinovernavirus.
The roles of dinovernaviruses in nature are still unknown. It would be interesting to determine if infection with FAKV precludes, interferes with, or enhances infection of arboviruses or other viruses within the subfamily Spinareovirinae. It is also important to determine if mosquitoes infected with these viruses undergo any pathological or physiologic changes. The A. albopictus cells used here lack an intact RNA interference response (44), so the cytopathic effects observed in cell culture may not be observed in vivo or in the natural vectors identified in this study. Intrathoracic inoculation of Aedes aegypti mosquitoes with FAKV had no visible effect on these mosquitoes or their survivability.
Both the in vitro and in vivo systems used in this study represent highly permissive environments for the replication and detection of mammalian arboviruses. The inability to detect FAKV or FAKV genomic RNA from these systems provides strong support for the insect-specific nature of this genus. However, further studies using a more diverse array of vertebrate cell cultures, in vivo models, or perhaps more sensitive techniques, such as real-time PCRs, are necessary to truly preclude the possibility of FAKV replication in vertebrate cells. Also of importance is the question of whether or not FAKV has the ability to infect plants. This would add another level of complexity within the FAKV transmission cycle. Mosquitoes do feed on the nectar of plants as a primary sugar source; hence, the opportunity exists for transmission between these hosts, as does the possibility for horizontal transmission between mosquitoes.
Stability of the T=1 capsid.FAKV, like CPV, is a true single-shelled virus; it lacks trimers at any of the T=13l positions. The mechanism of capsid stability in CPV has been previously investigated, because this has hitherto been the only reovirus for which an inner-shell-only structure is stable (13, 45). Among the proposed factors conferring this stability include (i) the insertion of the SP domain (12), (ii) the N-terminal anchor, which extends the interface between the two major capsid chains (11), and (iii) the two clamp proteins (4). While visualization of the N-terminal anchor is not possible at this resolution, FAKV virions are functional and stable without an SP domain insertion (Fig. 4C) and half the usual complement of clamp proteins. Interestingly, our results show that several of the structural features previously thought to account for the stability and maintenance of a single-shelled reovirus are in fact dispensable. Fewer proteins than previously proposed are necessary for its stability. Further work is necessary to investigate the specific biophysical mechanisms that confer the stability and determine the degree of stability of the FAKV capsid.
Evolution of the family Reoviridae.FAKV clusters together with APRV (8) in the genus Dinovernavirus that contains 9-segment turreted reoviruses. Because serial passages not only cause mutations but may also result in major genetic changes, such as genome segmentation, it was previously unclear if APRV originated as a 10-segment virus and lost a genomic segment during cell culture passaging (46). Previous work has also shown that some segmented viruses may have an accessory segment that is not necessary for further replication of the virus (47). However, our data now confirm the existence of an extant, fully functional 9-segment reovirus, the first nine-segment dsRNA genome virus to be observed in nature. The T=1 inner layer consists of two interlocking peanut-shaped monomers. Each monomer has a helix-rich lobe and a mixed α/β lobe. This fold and organization are conserved within the T=1 Totiviridae and Cystoviridae as well (48, 49). Thus, Totiviridae, Cystoviridae, and Reoviridae are 3 families sometimes nicknamed “T=2,” because they have 2 copies of the major capsid protein per asymmetric unit. Currently, it is unclear whether Reoviridae and/or Cystoviridae is the product of a horizontal gene transfer event combining a T=1 Totiviridae inner capsid shell with a T=13l Birnaviridae outer shell (50) or whether the Totiviridae evolved by deletion of the T=13l layer from a more complex reovirus or cystovirus ancestor (48). Within the reovirus family, results of the Bayesian analyses suggest that FAKV evolved from a double-shelled ancestor by loss of the outer T=13l coat of trimers. Its sister genus Cypovirus is also single-shelled, but it carries polyhedrin, a trimeric protein whose central β-sandwich is suspiciously similar to the fold observed in the second-shell trimers of more basal reoviruses (51). Polyhedrin forms intracellular nanocrystals, which protect virions from harsh conditions in occlusions. FAKV is nonoccluded and lacks any homolog to these trimeric proteins. The most parsimonious explanation is that the T=13l capsid trimer lost its structural function in an ancestor of single-shelled reoviruses, and then the gene was neofunctionalized (into polyhedrin) in cypoviruses but deleted in dinovernaviruses.
Unique among all turreted reoviruses, the FAKV clamp protein (corresponding to aquareovirus VP6 and orthoreovirus σ2) binds only one site per asymmetric unit. In all structurally characterized turreted reoviruses, the clamp protein binds at least two different sites, each interacting predominantly with one of the two major capsid proteins in an asymmetric unit: a 5-fold proximal and 3-fold proximal clamp. Orthoreoviruses contain a third clamp binding site near the 2-fold symmetry axis. These interactions are similar but not identical, as the clamp can also interact with the turret or with dissimilarly oriented domains (3, 4, 11, 14). No protein density was observed that would sterically hinder the clamp from binding at its usual site. Because this clamp is present near the 5-fold axis at one copy per asymmetric unit, it must be properly expressed and folded. We therefore conclude that subtle changes to the binding interface, the clamp, or both are probably responsible for the evolution of this synapomorphy in FAKV. Determining the structure of the FAKV capsid at atomic resolution will permit valuable comparisons to the structure of its closest relative, allowing for a better understanding of how the clamp usually tolerates differences in the binding site arising from the nonequivalent positions occupied by the capsid proteins on which it sits.
In summary, FAKV arose from a more complex ancestor by serial loss-of-function events, including the following: the loss of the second shell, possibly through an Oryzavirus-like intermediate having only three of the usual 10 copies of the outer shell protein (3); the loss of one copy per asymmetric unit of the clamp protein; and the loss of one genomic segment from a 10-segment ancestor (Fig. 3). It is an interesting finding that the single-shelled reoviruses are those that are restricted to insect hosts.
Conclusions.We report the first detection and isolation of a 9-segment reovirus from nature and present the first structural studies of a member of this genus. The complete genome sequences presented here should be helpful in the development of detection methods to study the distribution of this genus, as well as aid in further understanding the evolutionary history of the family Reoviridae. With only four polypeptides per asymmetric unit (one turret, two major capsid proteins, and one clamp) and 9 segments, FAKV illustrates a minimal but functional reovirus and demonstrates the dispensability of several key structural features. The detection and characterization of FAKV increases our knowledge of the known geographic distribution, genomic characteristics, genetic diversity, and host range of another group of insect-specific viruses as well as the family Reoviridae.
ACKNOWLEDGMENTS
This work was supported by NIH grant P41GM103832 to W.C., NIH contract HHSN272201000040I/HHSN27200004/D04 to R.B.T. and N.V., Robert Welch Foundation grant Q1242 to W.C., a fellowship from the John S. Dunn Foundation to E.B.F., and the Robert E. Shope International Fellowship in Infectious Diseases from the American Society of Tropical Medicine and Hygiene to A.J.A. A.J.A. is supported by the James W. McLaughlin endowment fund. J.T.K. is supported by the Nanobiology Interdisciplinary Graduate Training Program (NIBIB T32EB009379) through the Gulf Coast Consortia.
We thank Matthew Baker and Andrew Debevec for helpful suggestions, Jill Thompson at the UTMB Recombinant DNA core for assisting with the next-generation sequencing, Xuemei Luo for performing mass spectrometry experiments, and Naoyuki Miyazaki for furnishing the map of rice ragged stunt virus.
Funding agencies had no role in study design, data collection and analysis, the decision to publish, or preparation of the manuscript.
FOOTNOTES
- Received 4 August 2014.
- Accepted 20 October 2014.
- Accepted manuscript posted online 29 October 2014.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JVI.02264-14.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.