Isolation, Identification, and Characterization of Novel Arenaviruses, the Etiological Agents of Boid Inclusion Body Disease

Animals.

The study was performed on 35 captive boid snakes (Boa constrictor [30 snakes], Corallus annulatus [2 snakes], Corallus hortulanus [2 snakes], and Python reticulatus [1 snake]) of variable ages and weights (Table 1). The animals were captive snakes from Germany, the United Kingdom, and Costa Rica that had been submitted for diagnostic purposes by their owners to the Institut für Veterinär-Pathologie, University of Giessen, Germany, the Division of Veterinary Pathology, School of Veterinary Science, University of Liverpool, United Kingdom, or the Departmento de Patologia, Escuela de Medicina Veterinaria, University of Heredia, Costa Rica, between 2000 and 2012. The snakes were killed according to a schedule 1 procedure, and a full diagnostic postmortem examination was performed in order to confirm or exclude BIBD. Tissue samples from the dead animals were subjected to the different tests with owners' consent. For these diagnostically motivated necropsies, no ethical permission was required in any of the universities involved. Animals that were submitted alive were euthanized with exposure to CO₂ for 15 min, followed by decapitation. From these animals, blood was collected and stored at −70°C, and a blood smear was prepared. From animals that were euthanized by the submitting veterinarian, a blood smear had been prepared prior to death.

Establishment of BIBD-positive and -negative permanent primary boid cell lines.

Four histologically confirmed BIBD-positive and two histologically BIBD-negative juvenile B. constrictor snakes (age, 14 days to 4 months; weight, 51 to 68 g) from three different breeders were used for the establishment of BIBD-positive and -negative boid tissue cultures. These animals had been submitted alive by their owners to the Institut für Veterinär-Pathologie, University of Giessen, Germany. Immediately after euthanasia, sterile samples of brain, heart, kidney, liver, and bone marrow were retained and subjected to tissue culture. Subsequently, a full postmortem examination was performed, and samples of a range of organs were processed for histological examinations. The organ material for culturing was washed three times in sterile phosphate-buffered saline (PBS), trimmed into blocks (>1 mm), and digested in 10× trypsin three times. Supernatants were centrifuged (500 × g at room temperature [RT]) for 5 min, and cells were suspended in 5 ml of HEPES buffered cell culture medium with 10% fetal bovine serum (FBS; Biochrom), inactivated at 56°C for 30 min in sterile cell culture dishes 5 cm in diameter, and incubated at 30°C. One liter of cell culture medium was prepared containing 873.5 ml of basal medium Eagle, (1× BME; Biochrom, Berlin, Germany) with 100 ml of tryptose phosphate broth (TPB [Difco, Sigma-Aldrich, Germany]; 29.5 g solubilized in 1 liter of aqua bidest, autoclaved at 121°C for 21 min), 15 ml of HEPES buffer (1 M; Biochrom), 10 ml of l-glutamine (200 mM l-glutamine; Biochrom), 1 ml of gentamicin (10 mg/ml; Biochrom), and 0.5 ml of nystatin (100,000 IU/ml Nystatin Lederle; Valeant Pharmaceuticals, Eschborn, Germany), pH 7.2 to 7.3. During the first 6 days a 50% medium exchange was performed at 8-h intervals, followed by a full medium exchange every fourth day. After 14 days, cultures with proliferating and adherent cells were trypsinized and transferred into 25-cm² tissue culture flasks and incubated at 30°C. The cells were screened for the development and persistence of the characteristic IB by collecting an aliquot of cells and performing a light microscopy examination on formalin-fixed, paraffin-embedded cell pellets and by transmission electron microscopy (TEM) on glutaraldehyde (GA)-fixed and processed pellets after each second or third passage. Cell lines originating from the histologically BIBD-positive snakes were defined as BIBD positive (development of IB), whereas control cell lines (naive cultures) from histologically BIBD-negative snakes were defined as BIBD negative (no IB formation after several passages).

In vitro infection experiments and confirmation of BIBD in tissue cultures.

To demonstrate the causative relationship between the as yet unidentified infectious agent and BIBD, supernatants from BIBD-positive heart, kidney, and bone marrow cultures were filtered (0.45-μm-pore-size syringe filter) and added (1 ml each) to BIBD-negative tissue cultures of the same or different organ origin. After 24 h of incubation, medium was exchanged, and the cultures were maintained for 8 to 14 days. Identical cell cultures inoculated with BME (mock infected) served as negative controls.

In an attempt to confirm that the causative agent of BIBD can infect naive boid cultures, a control kidney cell line was inoculated with blood, serum, homogenized full blood, and homogenized liver tissue (1 ml each) from one juvenile, histologically BIBD-negative Python reticulatus snake and seven histologically BIBD-positive and one-negative B. constrictor snakes (Table 1, snakes 5 to 11, 28, and 35). A further identical attempt was undertaken with blood samples from seven adult B. constrictor snakes (Table 1, snakes 36 to 42) that were submitted for intra vitam diagnosis of BIBD by cytological examination. After 24 h of adsorption at 30°C, the cultures were washed three times with medium and maintained for 24 days at 30°C. During this time period, they were passaged twice, at 9 and 18 days postinoculation (dpi). A cell pellet was prepared for histological examination with each passage. The supernatants were collected and subjected to reverse transcription-PCR (RT-PCR) for arenaviruses and immunoblotting.

Histological and cytological examinations of snake tissues and cultured cells.

Tissue samples from the central nervous, respiratory, gastrointestinal, urogenital, lymphatic, and endocrine systems of all snakes were fixed in 10% formalin and paraffin wax embedded. Consecutive sections (3 to 5 μm) stained with hematoxylin-eosin (HE) and Giemsa served to confirm or exclude BIBD, based on the presence or absence of the typical intracytoplasmic IB in a wide range of cells (2, 7–9). Blood smears were stained with May-Grünwald Giemsa and examined for the presence of IB in circulating blood cells.

Cultured cells were detached by trypsinization and pelleted by centrifugation (for 3 min at 2,850 × g at 8°C). The supernatant was collected and stored at −20°C for future use. Cell pellets for histological and immunohistochemical (IHC) investigations were fixed in 2.5% paraformaldehyde (PFA) in 0.2 M phosphate-buffered saline (PBS; pH 7.4) for 24 h at 5°C and then paraffin wax embedded. Sections (3 to 5 μm) were prepared, stained with HE, and examined for the presence of IB or used for the immunohistochemical demonstration of arenavirus antigen within cells.

TEM of tissues and cultured cells.

Fresh brain, liver, pancreas, and kidney samples of selected BIBD-positive snakes were fixed in 5% glutaraldehyde (GA), buffered in 0.2 M cacodylic acid buffer, pH 7.3, for 24 h, and embedded in epoxy resin. Toluidine blue-stained semithin (1.5 μm) sections were used to select areas for the preparation of ultrathin (75 nm) sections that were contrasted with lead citrate and uranyl acetate and examined with a JEOL JEM1400 transmission electron microscope at 80 kV. The ultrastructure of the IB in cells in tissues of BIBD-positive snakes served as a reference for IB in tissue cultures.

Cell pellets (see above) for TEM were fixed in 1.5% GA, buffered in 0.2 M cacodylic acid buffer, pH 7.3, for 12 h at 5°C, and processed and examined as described above.

Isolation of IB (68-kDa protein) from infected cells and peptide mass fingerprinting.

The IB characteristic for BIBD have previously been shown to consist of an unidentified 68-kDa protein (7). We therefore aimed to isolate this protein from our BIBD-positive boid cell lines and identify it using mass spectrometry. BIBD-positive (snake 1) and -negative (snake 26) boid cells were washed three times with PBS and lysed by addition of 500 μl of lysis buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100, and EDTA-free protease inhibitor cocktail [Roche]). After thorough mixing and centrifugation for 5 min at 10,000 × g at 4°C, the supernatant was collected. The remaining cell debris pellet was washed three times with 1 ml of lysis buffer prior to addition of Laemmli sample buffer (LSB). The isolated protein (68-kDa band in SDS-PAGE gel) found in BIBD-infected cell lysates was excised from the gel and was in-gel trypsinized according to standard protocols (20). The trypsinized peptides were purified by C₁₈ ZipTip pipette tips (Millipore), and the peptide masses were recorded using matrix-assisted laser desorption ionization–two-stage time of flight mass spectrometry (MALDI-TOF/TOF) (Bruker Daltonics).

Ultracentrifugation, purification, and concentration of viruses.

Supernatants, collected from BIBD-positive permanent cell lines (originating from snakes 1 and 2) and from naive boid cells (snake 26) infected with supernatants of BIBD-positive cells from snakes 1 and 2 or with tissue extracts from one BIBD-positive snake (snake 5), were precleared from cell debris by centrifugation (30 min at 6,000 × g at room temperature [RT]). The precleared supernatant was transferred to UltraClear tubes. Subsequently, a cushion of 30% sucrose in TEN (50 mM Tris, 1 mM EDTA, 150 mM NaCl) was placed under the supernatant, using a sterile needle and syringe, followed by ultracentrifugation for 2 h at 27,000 rpm and 5°C, using an SW41 (smaller supernatant volumes; 1 ml of sucrose) or an SW28 (larger supernatant volumes; 3 ml of sucrose) rotor. The resulting pellets were resuspended in PBS for further analysis.

For detailed morphological studies, nucleic acid isolation, and the production of the polyclonal antisera, one virus isolate (UHV; originating from snake 1) was used and further purified by layering the concentrated virus on a 0 to 70% (in TEN) sucrose density gradient, followed by ultracentrifugation (SW41 rotor; 2.5 h at 40,000 rpm at 10°C). The fractions were collected by puncturing the tube bottom.

Isolation and analysis of nucleic acids.

RNA was isolated from infected and mock-infected cell samples (Vero E6 and boa cells) with Trisure isolation reagent, according to the manufacturer's protocol. The nucleic acids from virus-containing density gradient fractions were isolated using a High Pure Viral Nucleic Acid Kit (Roche) according to the manufacturer's instructions.

Isolated viral nucleic acids were analyzed using standard protocols and formaldehyde agarose gel electrophoresis. RNase-free DNase I (Thermo Scientific) and DNase-free RNase A (Thermo Scientific) treatments of viral nucleic acids (containing also carrier RNA from a Nucleic Acid Kit [Roche]) were performed by incubating the nucleic acid in the appropriate buffer with 1 unit of the respective enzyme 30 min at 37°C).

Sequencing.

Sequencing of the viral genome (snake 1) was initiated by 454 sequencing. The viral RNA template was derived from viruses purified by ultracentrifugation in a sucrose density gradient (the same material as that for nucleic acid analysis). RNA was reverse transcribed with random hexamers and a pan-arenavirus terminal primer (5′-CGCACMGDGGATCCTAGGC-3′) using Expand reverse transcriptase (Roche). The template was further amplified by PCR with the same terminal primer using Taq polymerase (Thermo Scientific) and Taq Extender PCR additive (Agilent Technologies). PCR products were purified with a PCR purification kit (Qiagen) and subjected to pyrosequencing by the 454 method (Roche) at the DNA sequencing and genomics core facility, Institute of Biotechnology, University of Helsinki, Finland. The whole-genome sequence was completed by conventional RT-PCR with primers designed based initially on the sequences retrieved by 454 and later on the sequences retrieved by Sanger sequencing. Full-length S and L segments were amplified using RevertAid Premium reverse transcriptase (Thermo Scientific) in the RT step with terminal primer optimized for BIBD arenaviruses (5′-GCACCGTGGATCCTAGGC-3′), followed by PCR with the same primer using Taq polymerase (Thermo Scientific) with Taq Extender. These PCR products were cloned into pGEM-T vector (Promega) before sequencing. Alternatively, a 5-kb fraction of the L segment was amplified using primers LF1 (5′-AAACTGAGATCAACCATTCACGAACTGA-3′) and LR1 (5′-AACCCAGTTACCACTGAACTTTGGAAAGC-3′) and directly sequenced after agarose gel purification using a gel extraction kit (Qiagen).

Phylogenetic analyses.

Sequences were handled with BioEdit (21), and alignments were built with ClustalX (22). Phylogenetic trees were estimated using the Bayesian approach implemented in the program BEAST (23), convergence of parameters was assessed using TRACER (23), and each run was continued until the effective sampling size of all parameters was greater than 200.

Antibodies.

Polyclonal rabbit anti-Machupo virus serum, human anti-Bolivian hemorrhagic fever serum, rabbit anti-JUNV serum, and a pool of monoclonal anti-Junin virus antibodies had been obtained earlier from the CDC, Atlanta, GA. The monoclonal antibody (MAb) 1.1.3 against LCMV nucleoprotein was a kind gift from Michael Buchmeier, University of California, Irvine, CA.

For the production of rabbit antiserum against UHV, sucrose gradient fractions containing the virus were pooled, and the buffer was exchanged to PBS, using a 100-kDa-cutoff centrifugal filter device (Millipore). The virus suspension was inactivated by Triton X-100 (final concentration of 0.01%) and used at 1.2 mg/ml total protein to immunize two rabbits (animals 21923 and 21924) according to the following protocol: initial injection (200 μg/animal), followed by boosters (100 μg/animal) on days 7 and 14, first bleeding on day 28 accompanied by booster (200 μg/animal), booster on day 42, and final bleeding on day 49. Immunizations were performed by BioGenes GmbH (Berlin, Germany). The rabbit antisera were tested by titration against virus lysate using enzyme-linked immunosorbent assays (ELISAs). Briefly, 96-well Polysorp plates (Nunc) were coated with purified virus (1:5,000 in 0.1 M NaHCO₃, pH 9.3; overnight at 4°C). Antisera were titrated according to standard ELISA protocols; the detection was based on horseradish peroxidase (HRP)-conjugated secondary antibody and colorimetric detection, utilizing 3,3′,5,5′-tetramethylbenzidine (TMB; Sigma-Aldrich) substrate solution. The antiserum (21923) that gave higher enzyme immunoassay (EIA) values than the 1:100 dilution of the preimmunization serum up to a higher dilution (1:12,800) was subsequently used. The IgG fraction was purified using a HiTrap Protein G column (GE Healthcare) operated by BioCAD Vision Workstation (PerSeptive Biosystems) according to standard protocols. The nonspecific background to boid cells was adsorbed from the IgG fraction by passing the fraction several times through a CnBr-Sepharose (GE Healthcare)-coupled lysate of naive boid cells according to the manufacturer's recommendations.

For immunofluorescence, Alexa Fluor 555-conjugated goat anti-rabbit (dilution, 1:1,000; Invitrogen) and Alexa Fluor 488-conjugated goat anti-rabbit (dilution, 1:1,000; Invitrogen) were used as conjugates. For immunoblotting, IRDye800CW-labeled donkey anti-rabbit (dilution, 1:10,000; Li-Cor Biosciences), IRDye700CW-labeled donkey anti-human (dilution, 1:5,000; Rockland Immunochemicals), and IRDye800CW-labeled sheep anti-mouse (dilution, 1:10,000; Rockland Immunochemicals) were used as conjugates.

SDS-PAGE and immunoblotting.

SDS-PAGE separations were done according to standard protocols on 8 to 12% gels. For immunoblotting, the proteins were transferred onto nitrocellulose filter (Whatman) by wet blotting. The membrane was incubated in blocking buffer (2 to 3% skimmed milk in TEN–0.05% Tween 20 [TEN-T]) on an orbital shaker (30 min at RT), followed by incubation with primary antibodies or serum, diluted 1:200 to 1:2,000 in blocking buffer (overnight at 4°C). After three 5-min washes, the respective secondary antibody (see dilutions above) in blocking buffer was added, and the membrane was incubated on an orbital shaker (1 h at RT). After three washes with TEN-T and one wash with PBS, the results were recorded using an Odyssey Infrared Imaging System (Li-Cor).

Cell samples were prepared for SDS-PAGE and/or immunoblotting by three washes with sterile PBS, followed by lysis (50 mM Tris, 150 mM NaCl, 1% Triton X-100, complete EDTA-free protease inhibitor cocktail [Roche]) and determination of the protein concentration, using a bicinchoninic acid (BCA) protein assay kit (Pierce, Thermo Scientific) according to the manufacturer's protocol. The cell lysates were diluted in LSB to obtain a final protein concentration of 0.5 to 1.0 μg/μl and then boiled and stored at −70°C. Typically, 10 to 20 μg of total protein/lane was used for immunoblotting. Concentrated virus samples were diluted in LSB, boiled, separated by SDS-PAGE, and transferred onto nitrocellulose for immunoblotting.

SDS-PAGE samples of whole blood collected from one P. reticulatus and eight B. constrictor snakes (snakes 5 to 11, 28, and 35) (Table 1) were prepared by mixing 5 μl of blood with 25 μl of LSB, followed by 5 min of boiling. For the analysis of liver samples of the same snakes, a liver sample was weighed, and LSB was added to obtain 50 μg/μl (wet liver/LSB). For immunoblotting, 2.5 μl of whole blood or 250 μg of liver homogenate was separated by SDS-PAGE. The immunoblotting was done with rabbit anti-UHV (1:2,000) as described above.

Infection of naive boid cultures with virus isolated from BIBD-positive boid cultures.

Virus purified by ultracentrifugation from supernatants of boid kidney cells infected in vitro with BIBD-positive tissue culture supernatants (snakes 1 and 2) or organ material (snake 5) was used to infect control boid kidney cell cultures (snake 26). After virus inoculation, cells were maintained for 8 days at 28 to 30°C. Subsequently, supernatants were collected for virus purification (for RT-PCR) and immunoblotting, and cell pellets were prepared for histological examination (detection of typical IB) and immunohistochemistry for arenavirus antigen.

Immunohistochemistry.

Viral antigen was demonstrated in situ by immunohistochemistry in the tissues of all 35 boid snakes (Table. 1). The streptavidin peroxidase method was applied, using an UltraVision anti-rabbit HRP detection system (Thermo Scientific). Briefly, sections (3 to 5 μm) were prepared from fixed, paraffin-embedded tissues. After deparaffinization, sections underwent antigen retrieval (incubation in citrate buffer [pH 6.0] in a microwave oven), blocking of endogenous peroxidase activity (incubation in 3% H₂O₂ in PBS for 10 min at RT), and blocking of nonspecific staining. They were then incubated for 15 to 18 h at 4°C with the primary antibody (negatively purified IgG fraction of rabbit 21923 anti-arenavirus antiserum [rabbit anti-UHV]; 5 μg/ml in TBS-Tween). This step was followed by incubation with a goat anti-rabbit antibody and the streptavidin peroxidase according to the manufacturer's protocol. The reaction was visualized with diaminobenzidine tetrahydrochloride (DAB), followed by counterstaining with hematoxylin. Consecutive sections incubated with a nonreactive rabbit polyclonal antibody instead of the specific primary antibody served as negative controls.

Antibody detection by competitive ELISA.

A competitive ELISA was established to demonstrate arenavirus antibodies in the blood of boid snakes. Polysorp (Nunc) plates were coated with inactivated virus (1:5,000 in 0.1 M NaHCO₃, pH 9.6) and blocked as described above. Plasma collected from one P. reticulatus and eight B. constrictor snakes (snakes 5 to 11, 28, and 35) (Table 1) was mixed at dilutions of 1:25 to 1:3,600 with rabbit anti-UHV (1:8,000 in PBS-Tween [PBS-T]), and the assay was performed as described above (see the paragraph “Antibodies”). The ability of snake plasma to interfere with the binding of rabbit antibodies was calculated by comparing the values obtained in the presence of plasma (average of 1:25 to 1:100 dilutions) to values measured in the presence of FBS. The percentage of inhibition was calculated according to the following equation: 100 × (1 − absorbance_{plasma+antibody}/absorbance_FBS+antibody).

Immunofluorescence.

The number of fluorescent foci in virus suspensions was determined as follows: boid kidney cells (from snake 26) or Vero E6 cells in a 75-cm² flask were detached with 3 ml of trypsin-EDTA and suspended in 10 ml of HEPES-minimal essential medium (MEM). Aliquots of cell suspension (∼10⁶ cells/ml; 30 μl/well) were dispensed on 10-well diagnostic slides (Marienfeld). Virus suspension (10 μl) was added, and a 10-fold dilution series (on slides with duplicates) was prepared. Slides were incubated at 30°C for 48 to 72 h, washed three times with PBS, air dried, and fixed for 10 min with ice-cold acetone. Subsequently, slides were incubated with preadsorbed rabbit anti-UHV (1:500 in PBS; 35 μl/well) for 45 min at 37°C, washed three times with PBS, air dried, and incubated with Alexa Fluor 555-conjugated goat anti-rabbit (1:1,000 in PBS) for 45 min at 37°C. Slides were washed three times with PBS and twice with Milli-Q water, air dried, and cover slipped. Fluorescent foci were counted in wells with a suitable dilution.

Alternatively, Vero E6 or boid kidney cells grown on coverslips (Marienfeld) on 24-well plates (Cellstar; Greiner Bio-One) at 30°C were infected with UHV by the addition of virus (1:25 to 1:100) to conditioned cell culture medium. After 3 to 7 days of incubation at 30°C, cells were washed three times with PBS and fixed with 4% PFA for 15 min at RT. Cells were then permeabilized with 0.3% Triton X-100 in PBS containing 3% bovine serum albumin (BSA). Coverslips were incubated with preadsorbed rabbit anti-UHV (1:500 in PBS with 2 to 3% Top-Block [Fluka]) at RT for 1 to 2 h, washed three times with PBS, and incubated with Alexa Fluor 555-conjugated goat anti-rabbit (1:1,000 in PBS with 2 to 3% Top-Block) for 45 min at RT. Subsequently, coverslips were washed three to five times with PBS and once with Milli-Q water, incubated for ∼1 min with Hoechst 33342 (1 μg/ml in Milli-Q water), washed with Milli-Q water, air dried, and mounted on slides for microscopic evaluation.

Diagnostic RT-PCR protocols.

An RT-PCR protocol was established to study liver, plasma, and blood samples as well as tissue culture supernatants of virus isolations from BIBD-positive snakes. RNA was isolated from liver samples using Trisure isolation reagent according to the manufacturer's protocol. RNA from blood and plasma samples of snakes and from supernatants of virus-infected cell cultures was extracted by a GeneJET Viral DNA and RNA Purification Kit (Thermo Scientific). The infectious agent was initially confirmed to be an arenavirus by a pan-arenavirus RT-PCR (24). Subsequently, an RT-PCR specific for BIBD-arenaviruses was designed with primers LF2 (5′-GGTTGGTTTCCAAGTGTAAACGACAGA-3′) and LR2 (5′-ACTAAATGCCTTCTGACCTTCACC-3′). RevertAid Premium reverse transcriptase and PhusionFlash polymerase (Thermo Scientific) were used according to the manufacturer's instructions, and PCR products were directly sequenced.

TEM and three-dimensional image processing of arenavirus isolate UHV.

UHV (either pelleted by ultracentrifugation through a sucrose cushion or purified by density gradient fractionation) was transferred into 25 mM Tris–150 mM NaCl, pH 8.0, and further concentrated, using a 100-kDa centrifugal ultrafiltration device (Millipore). The virus suspension was negatively stained with uranyl acetate, and images of the viruses were recorded using a JEOL 1400 transmission electron microscope at 80 kV.

The density gradient virus preparation was also used for cryo-electron microscopy (cryo-EM). The purified virus was mixed with 10-nm colloidal gold and vitrified on Protochips C-flat CF-22-4C holey carbon-coated copper grids as previously described (25). Cryo-EM was conducted under low-dose conditions in a FEI Tecnai F20 microscope operated at 200 kV, using a Gatan 914 cryo-holder maintained at −180°C. Cryo-EM images were acquired with a Gatan Ultrascan 4000 charge-coupled-device (CCD) camera, using a nominal magnification of ×68,000. Cryo-electron tomography data sets were collected, using the SerialEM software (26), to acquire 11 tilt series (±60° at 2° increments) at nominal underfocus values of 3, 4, or 6 μm and a nominal magnification of ×39,400, resulting in a sampling of 0.38 nm per pixel.

Tomographic reconstructions were calculated using the program package IMOD (27, 28). The raw data were binned, resulting in a final sampling of 0.77 nm per pixel. Weighted back-projection was used for tomographic reconstructions unless stated otherwise. The simultaneous iterative reconstruction technique (SIRT) was used for tomographic reconstruction of a broken virion (see Fig. 5G). Subtomographic averaging was used to improve the reconstruction of the spikes that were evident on the virus surface. The coordinates of the virion spikes were manually defined from 31 low-pass-filtered type 1 (see Fig. 5B) virion tomograms, using the IMOD program 3dmod. A total of 2,463 tomographic subvolumes (64 by 64 by 64 pixels in size) were extracted, each containing a virion spike. Iterative subvolume alignment and averaging in PEET (29) was used to calculate an averaged spike structure. Three-fold averaging, contrast transfer function correction, modulation transfer function filtering, and low-pass filtering were implemented, and the final spike reconstruction was calculated. The resolution was estimated from the Fourier shell correlation (0.5 cutoff) in the program package BSOFT (30, 31) comparing two reconstructions calculated from two separate data sets that were aligned and averaged independently of each other. Data were visualized with IMOD and UCSD Chimera (32). The mass of the spike at two standard deviations above the mean was estimated in EMAN (33).

Infection of Vero E6 cells with concentrated virus.

The principal susceptibility of mammalian cells to BIBD arenaviruses was studied using Vero E6 cells (from African green monkey kidney) maintained in minimal essential medium (MEM) supplemented with 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin; 1 M HEPES, pH 7.4, was added to a final concentration of 25 mM (HEPES-MEM) to facilitate cultivation without CO₂. Supernatant of infected boid cells (snake 1) and UHV concentrated by pelleting through a sucrose cushion were used for infection. The virus was allowed to adsorb for 1 h at RT, after which fresh medium was added. Cells were maintained at 30°C without CO₂, the medium was exchanged after 10 days, and the supernatant was collected at 16 days.

Accession numbers.

The S- and L-segment sequences of UHV have been submitted to GenBank under accession numbers KF297880 and KF297881, and partial L-segment sequences of other strains have been submitted under accession numbers KF564796 to KF564805. The spike reconstruction has been deposited in the Electron Microscopy Data Bank (EMDB) with the accession code EMD-2424.

Postmortem diagnosis and morphological characterization of BIBD in boid snakes.

The histopathological examination of a range of organs from 35 constrictor snakes that had undergone a diagnostic postmortem examination to assess the BIBD status confirmed BIBD in 23 individuals (19 B. constrictor, 2 C. annulatus, and 2 C. hortulanus snakes). In six B. constrictor individuals, an inconclusive histopathological BIBD result was obtained, and six animals were BIBD negative (Table 1). The diagnosis was based on the presence of the typical intracytoplasmic eosinophilic to amphophilic IB in cells (Fig. 1). IB ranged from 5 to 25 μm in diameter and were found in practically all cell types, with the highest prevalence in the central nervous system (CNS), kidneys, pancreas, and liver. Cells with IB did not show any other changes and, in particular, no evidence of degeneration or death (i.e., apoptosis or necrosis). Ultrastructurally, IB were represented by accumulations of finely granular, electron-dense material in the cytoplasm, often located in proximity to the nuclear membrane (Fig. 1M to P). In blood smears from the 23 histologically confirmed BIBD cases, typical IB were found in red blood cells (Fig. 1E, inset) of 17 individuals; in the remaining six snakes, the blood cells did not exhibit IB. The histologically BIBD-negative cases also yielded negative results on the blood smears. The detailed results are provided in Table 1.

Permanent BIBD-positive and -negative boid cell lines.

Permanent cultures were successfully generated from the bone marrow, heart, and kidney of four histologically BIBD-positive (snakes 1 to 4) and two histologically BIBD-negative (snakes 26 and 27) juvenile B. constrictor snakes. The cultures have now each been passaged more than 30 times and can be considered permanent cell lines. Regardless of tissue origin and infection state, cultured cells vary in size (20 to 40 μm) but generally exhibit a round to spindle shape with a round to ellipsoid nucleus and partly vacuolated cytoplasm (Fig. 2A to C).

Light microscopically, cell pellets from the tissue cultures of the BIBD-positive B. constrictor snakes (snakes 1 to 4) exhibited characteristic, eosinophilic, often perinuclear, single or multiple intracytoplasmic IB with diameters of 1 to 5 μm. In early passages, IB were observed in approximately 5% of cells. In higher passages (≥30), the percentage of cells with IB decreased to below 2%. The ultrastructural features of the IB were identical to those seen in the tissues of BIBD-positive snakes (data not shown). The cell lines from BIBD-negative B. constrictor snakes did not develop any IB, as confirmed by histology and immunohistology (Fig. 2D and E).

Infection of naive boid cells with supernatants from BIBD-positive tissue cultures and organ material from BIBD-positive snakes.

To demonstrate that BIBD-positive tissue cultures produce an infectious agent, we inoculated BIBD-negative (naive) tissue cultures (heart, bone marrow, and kidney) with filtered (0.45-μm pore size) supernatants from BIBD-positive tissue cultures. All transmission attempts, regardless of the donor or recipient culture organ origin, successfully reproduced BIBD in the naive cells, as confirmed by the histological demonstration of the characteristic IB in the cell pellets, without signs of any cytopathic effects. The majority of infected cells showed single or multiple, 3- to 4-μm-diameter cytoplasmic inclusions at 6 days postinoculation (dpi). By 12 dpi, the size of the IB had increased to 5 to 10 μm in diameter. The ultrastructural features of the IB were identical to those seen in the tissues of BIBD-positive snakes (Fig. 1Q and R).

To demonstrate that BIBD can be directly transmitted from infected tissues to cultured boid cells, we inoculated naive boid kidney cells from snake 26 with extracts from blood, serum, blood cells, and liver of seven BIBD-positive and two BIBD-negative snakes (Table 1, snakes 5 to 11, 28, and 35) as well as with blood from six B. constrictor snakes that tested BIBD positive in a blood smear and one that had a BIBD-negative blood smear (Table 1). Regardless of the type of inoculum, material from BIBD-positive animals induced the characteristic IB in a large number of cells (up to 25%) in culture when examined at 6 to 8 dpi. In the positive cultures, the number of cells with IB declined to 3 to 4% during the following two passages. However, when supernatants of these later passages were used to inoculate other naive boid kidney cell cultures, IB developed within the same time scale and to the same extent as after the initial transmission attempt. The successful in vitro transmission of BIBD by 0.45-μm-pore-size-filtered BIBD-positive tissue culture supernatant and organ material from BIBD-positive snakes to naive boid cell cultures suggested the causative agent to be a virus, prion, or bacterium.

Enrichment of the 68-kDa protein from infected cells.

The distinctive 68-kDa protein that has been shown to comprise the IB (7) was demonstrated as a component of BIBD-positive (infected) but not of BIBD-negative boa cell lysates that were analyzed by SDS-PAGE (Fig. 3A). The protein was further characterized by peptide mass fingerprinting and laser-induced fragmentation and was later found to match the predicted amino acid sequence of the UHV NP (see the paragraph “Sequencing of the UHV genome” below).

Isolation of viruses from supernatants of cell cultures established from BIBD-positive snakes.

To identify the causative agent of BIBD, precleared supernatants of the permanent bone marrow cell line of a BIBD-positive B. constrictor (snake 1) and the permanent kidney cell line of a BIBD-negative B. constrictor (snake 26) were ultracentrifuged (Table 1). Material pelleted through a sucrose density cushion was solubilized and analyzed by SDS-PAGE. The pellet generated from BIBD-positive cell culture medium contained a double band at ∼45 kDa, a 68-kDa band, and an ∼250-kDa band not seen in the control culture pellet (Fig. 3B). The ultrastructural examination of the negatively stained pelleted supernatant of the BIBD-positive culture showed roughly spherical membranous particles that ranged from 150 to 200 nm in diameter and were sometimes decorated by spikes (Fig. 3C).

Since the pelleted supernatant of infected cells contained a large amount of virus-like particles, we prepared a large batch of infected cell culture medium and fractionated the pelleted material in a density gradient by ultracentrifugation. The SDS-PAGE gel from supernatant of the permanent BIBD-positive cell culture showed further enrichment of the 68-kDa protein which was not present in the control cell line (Fig. 3D). The ultrastructural analysis of these fractions (negative staining) demonstrated enrichment of the membranous particles (Fig. 3E), which were not found in the supernatants of the control cell line (data not shown).

Isolation and characterization of viral nucleic acids.

To determine whether the ultrastructurally identified particles were indeed virions, we isolated nucleic acids from the particle-containing density gradient fractions. The nucleic acids were separated by agarose gel electrophoresis and visualized by ethidium bromide. The medium of infected cells contained nucleic acids that appeared as two distinct bands with approximate sizes of 3,500 and 7,000 bp (Fig. 4A). Treatment of the nucleic acids with DNase I and RNase A resulted in degradation by RNase A only, indicating that the observed particles contain two segments of RNA (Fig. 4B). Isolation of the nucleic acids on an RNA gel demonstrated that the size of the S segment is below 4,000 bp (Fig. 4C), whereas the size of the L segment is below 8,000 bp (Fig. 4A and C). Based on the identification of a bisegmented RNA genome and on its size, we assumed that we had isolated an arenavirus. This was confirmed with RT-PCR targeting a highly conserved region in the RNA-dependent RNA-polymerase of arenaviruses (24); direct sequencing of the amplicon found the virus to be distant from rodent-borne arenaviruses. The isolated virus was designated University of Helsinki virus (UHV).

Structure of UHV.

The structure of purified, highly concentrated UHV from density gradient ultracentrifugation was studied using cryo-EM. The virions appeared as pleomorphic, typically spherical or slightly ellipsoid, with diameters in the range of 100 to 200 nm (Fig. 5A). Based on evident structural differences, we separated the particles into two groups: type 1 we called virions (Fig. 5B) as they were filled with granular density and had spikes on the surface and an additional secondary density layer under the membrane. Type 2 virus-like particles (Fig. 5C) were similar in size but had no spikes or only a few spikes and no clearly defined secondary density layer under the membrane but also had a granular interior. The presence of spikes and the second density layer were most apparent in subtomographic averages aligned from the membranes (Fig. 5B and C, insets). The secondary layer was approximately 5 nm distant from the membrane and was assigned as the Z layer through comparison with other arenaviruses (34).

Inspection of electron tomograms of the viruses showed that the spikes on the virions are distributed over the whole surface of the membrane (typically 11 to 15 nm apart), yet no local or long-range order was detected. The spikes extended ∼10 nm from the outer leaflet of the bilayer, with a distinctive head and a short stalk region. Subvolume alignment and averaging 1,847 (75% of the data set) spikes from virions revealed a stalk of approximately 1 nm in length and 4 nm wide, with a trimeric head region forming a cup-shaped structure of approximately 11 nm wide and 9 nm high; thus, imposing 3-fold symmetry on the structure spike improved the final resolution to 3.2 nm at a Fourier shell correlation cutoff of 0.5 (30) (Fig. 5E) but did not change the overall features compared to the spike on which symmetry was not imposed (Fig. 5D). The cup consisted of three slightly tilted finger-like protrusions that form a chiral structure. The estimated mass of the spike was ∼150 kDa at two standard deviations above the mean density.

Despite careful inspection of the interior of the virions in the tomograms, no continuous structures were detected. However, when we inspected tomograms of broken virions (Fig. 5G), the loosening of the tightly packed material in the virions revealed a “beads-on-a-string”-like structure. The beads were ∼3 nm in diameter at ∼6-nm intervals connected by a narrow string-like density attributable to the nucleic acid and associated proteins.

Sequencing of the UHV genome.

Full-genome sequencing and genetic characterization of nucleic acids isolated from virions purified by density gradient ultracentrifugation were initiated using 454 sequencing. This approach yielded almost exclusively viral sequences that were assembled into seven contigs of 210 to 746 nucleotides (nt) in length, covering approximately 40% of the genome. The full genome of UHV, excluding the terminal sequences, was obtained by primer-walking through the genome. The S segment was 3,405 nt long and contained two open reading frames (ORFs) in ambisense orientation: one coding for a putative nucleoprotein (NP) and the other coding for a putative glycoprotein precursor (GPC). The genome coding strategy of UHV (and arenaviruses in general) is shown in Fig. 6B, and the sizes of protein products are given in Table 2. The NP was confirmed to be the 68-kDa protein shown in Fig. 3B by mass fingerprinting with the following peptides identified: RPPPYAPR, LRRPPYAPR, IKIVIDEER, SLFEDLFYK, KPVFVDIEGPPR, KEDTMEMEVIDPR, LRDLNEDVAGMSMGVAK, LSGFELETGMSSEDHAK, LSLESDSSSNEMDEYIK, ILTPFQGDNQQIFVDIK, and DLANQFGTSPAVTITLMMMR. The inter-ORF region was 260 nt long and had no identity to rodent-borne arenaviruses but low identity to Golden Gate virus (GGV) and very low identity to California Academy of Sciences virus (CASV). The amino acid identities of the NP and the GPC were highest to GGV, whereas in comparison to rodent-borne arenaviruses, only the NP was alignable (Table 3). The GPC region of amino acids (aa) 234 to 423, presumably corresponding to UHV glycoprotein 2 (GP2), were more conserved than the region corresponding to glycoprotein 1 (Table 3). In a BLASTP search, the putative GP2 sequence retrieved a hit for the conserved domain superfamily “Ebola-like_HR1-HR2” present in filoviruses and retroviruses. This region represents a common structural pattern among class I viral fusion proteins (35).

The L segment was 6,930 nt long and contained two putative ORFs in ambisense orientation: a 2,072-aa-long RNA-dependent RNA-polymerase (RdRp) and a 116-aa-long Z protein. The inter-ORF region was 160 nt long, predicted to form a hairpin loop structure with similarity only to GGV and CASV. Amino acid identity of the putative RdRp was highest to GGV and lowest to rodent-borne arenaviruses (Table 3; Fig. 6A). No identical regions were found between the UHV Z protein and the Z proteins of rodent-borne arenaviruses, while the amino acid identity of UHV and GGV was remarkably high (95%). The overall high sequence divergence of UHV from all known sequences suggests that the characterized isolate represents a novel arenavirus species.

A phylogenetic tree based on RdRp amino acid sequences shows that UHV falls into the same group as the previously reported CASV and GGV (Fig. 6A); a similar topology was obtained using NP sequences (data not shown). These snake-borne arenaviruses are clearly distinct from both NWA and OWA clades of rodent-borne viruses. Nevertheless, in a phylogenetic tree based on conserved RdRp motifs, they share ancestry with the other members of the Arenaviridae family (Fig. 6C).

Production and characterization of a polyclonal antibody against UHV and detection of viral proteins in purified virus, cells, and tissues.

To study whether the novel virus or its proteins are found in the tissues of BIBD-positive snakes, we prepared a polyclonal rabbit antiserum against UHV by immunizing two rabbits with density gradient-purified virus particles (Fig. 7A) inactivated with 0.1% Triton X-100. The antigen preparation (Fig. 7B) was shown to contain mainly components that are presumably of viral origin, a doublet band at ∼45 kDa, an ∼68-kDa band, and an ∼250-kDa band, representing the viral GPs, NP, and RdRp, respectively.

The rabbit antiserum detected the virus in ELISA and recognized the 68-kDa protein in infected BIBD-positive cell lines in immunoblots. The antisera recognized several bands from the virus preparation used as the antigen (Fig. 7C), not all of which are likely to represent viral proteins. In addition, we also tested the ability of rabbit anti-Junin, rabbit anti-Machupo, mouse anti-LCMV, and human anti-BHV antibodies to cross-react with the proteins of UHV by immunoblotting. The antibodies against LCMV and MACV both reacted weakly with the NP of UHV from the concentrated virus preparation (Fig. 7D). Furthermore, the anti-UHV rabbit serum recognized the 68-kDa protein enriched from cell lysates of BIBD-positive boid kidney cells (Fig. 3A and 7E). Also human anti-BHF and rabbit anti-MACV sera recognized this protein (NP of UHV) of viral origin (Fig. 7F).

The negatively purified IgG fraction of the rabbit anti-UHV antiserum and PFA-fixed, paraffin-embedded cell pellets of the cell line from which UHV had been isolated (snake 1) served to establish an immunohistochemistry protocol for the confirmation of arenavirus infection in boid cells and tissues. This stained the IB within the cytoplasm of cells, whereas no reaction was seen in pellets of BIBD-negative cell lines (Fig. 2).

Infection of permanent BIBD-negative boid tissue cultures with purified arenavirus isolates.

To fulfill Koch's postulates in vitro, we used the viruses purified by density gradient ultracentrifugation from tissue culture supernatants of snakes 1, 2, and 5 to infect a naive boid kidney cell line (snake 26). The results were comparable to those of the initial transmission experiments. Within 6 to 8 days, the formation of the characteristic intracytoplasmic IB was observed; distribution, size, and number of IB during the first passages (passages 0 [P0] to P2) were also similar (Fig. 2F, H, and J). Immunohistochemistry with rabbit anti-UHV stained the IB within the cytoplasm of cells (Fig. 2G, I, and K) with a staining pattern identical to that observed in the original permanent cell line from which UHV was isolated. Neither the formation of IB nor an IHC reaction was seen in pellets of BIBD-negative cell lines (Fig. 2D and 7E).

Arenaviruses are consistently detected in and isolated from BIBD-positive but not BIBD-negative animals.

(i) Retrospective confirmation of arenavirus infection in BIBD-positive snakes by immunohistochemistry.

The immunohistochemistry protocol for the specific detection of snake-borne arenavirus proteins from UHV was applied to tissues of snakes included in the present study (histologically BIBD positive, questionable, or negative) (Table 1). All 23 histologically BIBD-positive constrictor snakes (19 B. constrictor, 2 C. annulatus, and 2 C. hortulanus) showed a clear positive reaction (Fig. 1E, G, I, and K) but with variable IB staining patterns. While smaller to middle-sized IB exhibited moderate to strong staining, larger IB showed a weak reaction in the periphery. All animals that had been tested BIBD negative by histology, i.e., 5 B. constrictor snakes and one P. reticulatus snake (snakes 26 to 30 and 35) (Table 1) also yielded negative immunohistochemistry results. Immunohistochemistry confirmed arenavirus infection (BIBD) in two animals with questionable histology results (snakes 20 and 21) and excluded the infection in the remaining four histologically questionable animals (snakes 31 to 34) (Table 1).

(ii) Isolation of snake arenaviruses, RT-PCR, and partial genome sequences from BIBD-positive snakes.

After the discovery of arenaviruses in BIBD-positive snakes, we obtained a further cohort of seven histologically BIBD-positive and two BIBD-negative boa constrictors and one BIBD-negative P. reticulatus snake (snakes 5 to 11, 26, 28, and 35) (Table 1) from which we collected tissue samples and blood immediately after death. Confirmation of arenavirus infection was attempted by virus isolation in the permanent naive boid kidney cells, RT-PCR and sequencing, immunoblotting, and immunohistochemistry. In all histologically BIBD-positive snakes (snakes 5 to 11) (Table 1), IB in tissues exhibited a positive immunohistochemical reaction, and the RT-PCR performed on fresh blood, plasma, and/or liver samples yielded a positive result. Apart from snake 11, the snakes also yielded a virus isolate in naive boid kidney cells which produced immunohistochemically positive IB in cultured cells, and material pelleted from the supernatant by ultracentrifugation was positive in immunoblots (data not shown), and positive by RT-PCR. The three histologically BIBD-negative animals (snakes 26, 28, and 35) were negative in all tests.

In addition, seven prospective blood samples from snakes with suspected BIBD that had been submitted for diagnostic purposes (snakes 36 to 42) (Table 1) were examined. Six of these yielded a virus isolate in naive boid kidney cells that produced immunohistochemically positive IB in cultured cells and was RT-PCR positive; one (snake 42) was entirely negative (Table 1).

Partial L-segment sequences were obtained from 10 of the 13 RT-PCR-positive snakes. These showed a high diversity within the BIBD-associated arenaviruses. At least three different phylogroups were evident among the isolated arenaviruses (Fig. 6C). Snake 11, which was positive for arenavirus antigen in immunohistochemistry and immunoblotting, tested negative by RT-PCR. Interestingly, the immunoblot showed a viral NP that was different from that of UHV in size (Fig. 8A, #8).

(iii) Detection of arenavirus in infected tissues by immunoblotting.

To determine whether the presence of viral antigens could be detected directly in tissue or blood of infected animals, we subjected a small amount of whole blood (2.5 μl) and liver tissue (250 μg of wet weight) of nine animals (snakes 5 to 11, 28, and 35) to SDS-PAGE and immunoblotting with rabbit anti-UHV. The results indicate that NP can be detected from both blood and liver (Fig. 8A). In the seven positive animals (snakes 5 to 11, all B. constrictor), the antigen was abundant in the liver but could also be detected in whole blood. Interestingly, the sizes of the NPs, the major antigens, appeared to vary between samples: in snakes 8 and 9 (Fig. 8A, liver samples 1 and 2), the NPs were approximately the same size as the NP in UHV (68 kDa), whereas the NPs were slightly larger in snakes 5 to 7 (Fig. 8A, liver samples 4 to 6); in snakes 10 and 11 the NPs were clearly the largest (Fig. 8A, liver samples 7 and 8). The observed variation in the sizes of the NPs mirrors the three phylogroups of viruses present in the group of snakes. Six of these were confirmed as BIBD positive by histology, immunohistochemistry, and cell culture inoculation (snakes 5 to 10), whereas snake 11 was the histologically positive animal that had been negative by immunohistochemistry, cell culture inoculation, and RT-PCR (see above). Snakes 28 (B. constrictor) and 35 (P. reticulatus) were negative in the immunoblotting and had yielded negative results in all other tests, confirming that they were BIBD negative.

Altogether, we investigated a total of 29 snakes with a definite BIBD diagnosis based on histology (n = 23) or cytology (n = 6) and a further 13 snakes in which histology excluded BIBD or yielded an inconclusive result (Table 1). In animals with BIBD, arenavirus infection was demonstrated by IHC and/or virus isolation, RT-PCR, and Western blotting, but it was not found in any of the seven snakes in which BIBD was excluded with certainty. This association of BIBD with arenavirus infection in the same individual was highly statistically significant (Fischer's exact test, P = 1.198 × 10⁻⁷), even if the six individuals with inconclusive BIBD status, of which two were found to be arenavirus positive and four were negative, were considered within either the BIBD-positive category (P = 1,223 × 10⁻⁵) or BIBD-negative category (P = 1,822 × 10⁻⁸).

Arenavirus antibodies in the plasma of snakes.

There is evidence that BIBD is an immunosuppressive disease (8). We therefore wanted to investigate whether BIBD-positive snakes had mounted an antibody response against the snake arenaviruses. Since secondary antibodies against immunoglobulins of boid snakes are not available, we established an indirect ELISA in which the test snake plasma is mixed with rabbit anti-UHV at various dilutions, and inhibition of binding to the plate coated with UHV is measured. Plasma of all examined snakes (one P. reticulatus snake [snake 35] and eight B. constrictor snakes [snakes 5 to 11 and 28]) (Table 1) inhibited the binding of rabbit anti-UHV (Fig. 8B). The background was set to 10 to 15% since the plasma of snake 35 (P. reticulatus; BIBD negative) (Table 1) inhibited at roughly this level. BIBD is considered lethal for pythons, and thus we assume that snake 35 did not have antibodies. Since the blood and liver samples of all BIBD-positive snakes were positive for arenavirus antigen (Fig. 8A), it seems likely that the observed inhibition is due to antibodies rather than viral antigens.

Productive infection of Vero E6 cells with concentrated UHV and its adaptation.

Vero E6 cells are commonly utilized to propagate arenaviruses. However, Stenglein and coauthors reported that the novel arenaviruses (CAS and GGV) they found in boid snakes with BIBD did not infect Vero cells (17). In our study, the incubation of Vero E6 cells with supernatants of a BIBD-positive tissue culture (UHV) induced 3- to 8-μm, round to ellipsoid intracytoplasmic, eosinophilic IB in the cells, which were positive by immunohistochemistry for arenavirus antigen (Fig. 9A). These were apparent at 12 to 16 dpi in the majority of cells in the initial culture and the first passages (P0 to P1). Compared to uninfected controls, infected cell cultures showed mildly reduced density and growth rates but no obvious cytopathic effects. In addition to immunohistochemistry, we used immunofluorescence and immunoblotting to confirm the presence of arenavirus antigen in the Vero E6 cultures (Fig. 9A, B, and D). The immunofluorescence signals in the Vero E cells were similar to those in infected boid cells (compare Fig. 9B to C). Furthermore, the increased amount of NP in Vero E6 cells infected with supernatant from infected Vero E6 cells indicates that UHV could be adapted to grow in Vero E6 cells (Fig. 9D).

Our approach and main findings.

We began our search for the causative agent of BIBD a decade ago, taking a more classical approach based on the establishment of cell cultures from diseased and nondiseased boids. We were able to propagate the disease in boid cell cultures originating from various BIBD-positive tissues since within a few days they developed IB similar to those in affected tissues (36). The supernatant of these cells reproduced the IB phenotype in naive boid cell cultures also after filtration and ultimately served for the isolation of a novel virus. Purification and further structural and genetic characterization identified a novel arenavirus(es) as the etiological agent(s) of BIBD. The results of our study provide novel information on host cell spectrum, genetic variation, antigenicity, pathology, and disease association of arenaviruses, expand our knowledge on the evolution of arenaviruses, and show that arenaviruses have a wider host spectrum than previously known.

Here, we report the isolation of novel arenaviruses from boid snakes with BIBD. One isolate, UHV, was thoroughly characterized and represents an addition to the group of BIBD-associated arenaviruses. UHV was purified from cultured cells of a BIBD-positive B. constrictor and sequenced by combining traditional and NGS techniques. Purification of the virus by ultracentrifugation and fractionation yielded highly pure viral RNA ideal for NGS, thereby minimizing the presence of nonviral sequences that represent a common challenge in NGS. Rabbit antiserum raised against purified and inactivated UHV served to confirm that UHV or related arenaviruses are present in all tissues of BIBD-affected snakes that exhibit IB. Some snakes with BIBD have also developed antibodies against UHV, as shown by a competitive ELISA using purified UHV and the rabbit antiserum.

Structural characterization of UHV.

We studied the ultrastructure of UHV and, facilitated by the high yields of purified UHV, have provided the first three-dimensional cryo-electron tomography structure of the BIBD arenavirus and more generally of an arenavirus spike complex. The latter is a trimer, as its mass in the reconstruction corresponds to that of approximately three copies of the 45-kDa glycoprotein precursor, identified from the amino acid sequence analysis, that undergoes maturation by cleavage. Our sequence analysis indicates that the glycoprotein precursor is matured by cleavage. GP2 is most likely a type I fusion protein, like the fusion proteins seen in influenza-, paramyxo-, filo-, and retroviruses (37). Hence, we assume that the spike complexes are metastable structures that, when triggered by low pH, undergo conformational changes to reveal the fusion peptide to allow fusion with the host cell. The spikes penetrate the membrane to interact with an underlying layer of multiple copies of the Z protein. The Z protein layer thus provides an interaction interface to link the beads-on-a-string nucleic acid-nucleoprotein-polymerase complexes to the sites of the budding virus, driven by the glycoprotein-Z protein interactions (38). This agrees with the identification of a RING domain in the UHV Z protein and the fact that the arenavirus Z protein can homopolymerize (39). In contrast to a previous two-dimensional study of New World Pichinde and Tacaribe arenaviruses and Old World LCMV, the nucleoprotein was not found to form a paracrystalline layer under the Z layer (34).

Genetic properties of the novel arenaviruses.

The identification of novel arenaviruses in snakes (17) opens a completely new perspective on this virus family that has so far been considered exclusively rodent borne, with the exception of TACV, which has been isolated from bats (14). Sequence comparison has shown that the “snake-borne” arenaviruses are genetically very distant from arenaviruses isolated from rodents or human patients with viral hemorrhagic fevers, and phylogenetic analysis clearly shows that they belong to a unique, monophyletic taxonomic unit. The International Committee on Taxonomy of Viruses (ICTV) states that arenaviruses are defined as unique species based on a combination of characteristics such as the host species, geographic distribution, disease associations, antigenic cross-reactivity, and significant sequence divergence at the amino acid level (40). Although many of the characteristics still await determination, the sequence divergence alone indicates that UHV represents a novel arenavirus species. Four distinct lineages of BIBD/snake-associated viruses can be distinguished among the currently available sequences. However, this number is likely to increase as it was evident from our samples alone that there are further arenaviruses that cause BIBD; we recovered virus isolates with distinct NP antigen patterns that are not detectable by current RT-PCR primers. Furthermore, phylogenetic analysis showed that all “snake-borne” arenaviruses are monophyletic and comprise their own taxonomic clade in the Arenaviridae family. Due to this, it is evident that a third serogroup or another taxonomic entity, comprising BIBD-associated arenaviruses, should be established in the near future. We propose this putative new taxonomic entity to be called boid inclusion body disease-associated arenaviruses (BIBDAviruses, or BIBDAV).

Evolution of arenaviruses.

The arenaviruses that were recently discovered in boid snakes may also initiate a shift in the current understanding of arenavirus evolution. So far, it is thought to have proceeded alongside the evolution of rodents. One hypothesis is that the new arenaviruses were originally introduced to snakes through ingestion of arenavirus-infected prey hosts, which could be rodents, soricomorphs, or bats. However, BIBDAV differ considerably from each other and drastically from rodent-borne arenaviruses. It is possible that the initial transmission to snakes occurred early in the evolution of the snakes, leading to coevolution of virus and host. This would be supported by the observation that the CASV isolated from an annulated tree boa is distinct from the viruses (GGV, CVV, and UHV) isolated from boa constrictors. Of these, CASV is the most ancestral one. Alternatively, there might have been multiple cross-species introductions from an unknown host reservoir(s) during the evolution of boid snakes or, later, to captive snakes. Currently, the prevalence of BIBD in wild snake populations as well as the origin of the snake viruses remains unknown. Since we have demonstrated that UHV readily infects mammalian cells, it is also possible that the snake viruses might have their origins in mammalian species. Recent analysis of NWA has suggested that arenaviruses can cross the species barrier much more readily than previously thought and that diversification would have been driven by geographic proximity rather than relatedness of the arenavirus hosts (41). The same study also showed that different arenavirus lineages have different constraints for entering a new host, and structural studies suggest that cell entry mechanisms determine the potential for host spread (42). The GP1 region of UHV shows very little homology to GP1s of classical arenaviruses. In rodent-borne arenaviruses, GP1 contains several sites that are important for receptor binding. The fact that GP1 sequences of different arenaviruses align poorly demonstrates the high divergence of this region (14). On the other hand, the arenaviral GP2 contains a conserved domain involved in membrane fusion, and a homologous domain is found in both filoviruses and retroviruses. Based on this homology, Stenglein and coauthors suggested that recombination has occurred between filoviruses and arenaviruses (17). The chimeric origin of the GPC is evident for one of the NWA lineages (14), indicating the recombination potential of this region. However, the striking structural conservation among class I viral fusion proteins, including those of arenaviruses and filoviruses, is well known (35, 43), and a common RNA virus ancestor has been proposed for the two viral families. Therefore, recombination is not essential to explain the pattern of homology within the S segment.

Association of arenaviruses with BIBD: pathogenesis and diagnostic aspects.

The morphological hallmarks of BIBD are the amorphous, electron-dense large IB in various cell types of the nervous, respiratory, urogenital, gastrointestinal, and lymphatic systems, the bone marrow, and circulating blood of constrictor snakes. We were able to reproduce the BIBD phenotype in vitro within 6 to 8 days after inoculation of naive boid cell cultures with both tissue extracts from diseased snakes and supernatants of BIBD-positive cell cultures. Furthermore, we repeatedly isolated arenavirus from these reinfected naive cells, thus fulfilling Koch's postulates in vitro. We confirmed the IB to contain a 68-kDa protein, as previously shown (7), which we identified as the arenaviral NP protein (in our case UHV NP) by immunoblotting and, subsequently, by peptide mass fingerprinting. Furthermore, tissue samples of all diseased animals contained arenaviral proteins, as demonstrated by immunohistochemistry and/or immunoblotting, and almost all diseased snakes (14/15) were positive by RT-PCR with a set of primers designed based on the sequences of UHV, GGV, and CASV (17). These primers will obviously need to be modified when new strains are detected. Indeed, the only RT-PCR-negative BIBD-positive snake demonstrated detectable arenavirus NP antigen, though of remarkably different size, suggesting yet another BIBDAV variant. With the set of tools that we have developed for the diagnosis of arenavirus infection and thereby BIBD (virus isolation in boid cells, RT-PCR, immunohistochemistry, and immunoblotting), we were able to (i) identify or exclude the disease postmortem/retrospectively in a cohort of 35 histologically BIBD-positive, -questionable, or -negative snakes and (ii) diagnose the infection from the blood of 17 animals by both cell culture inoculation and/or immunoblotting. The second approach should allow testing of living animals and thereby the screening of, e.g., quarantined animals intended for international transport. The association between the presence of BIBDAV RNA or antigen and the morphological diagnosis of BIBD was statistically highly significant, which further confirms the arenavirus-BIBD association. Notably, not all virus-positive cases had tested positive by blood cytology or histology, the current golden standard for the diagnosis of BIBD. Also, one B. constrictor that was clearly positive in blood cytology, histology, immunohistology, tissue culture virus isolation, and Western blotting was negative by RT-PCR, suggesting that the diagnostic methods for BIBD need to be refined further.

Conclusions.

In summary, this report on novel snake-borne arenaviruses provides compelling new information on these new viruses and their association with BIBD. Our results will allow new molecular virology studies on these viruses and their associated diseases, both at cell and host levels, to gain new insights into arenaviruses and their pathogenesis. The novel information obtained on the morphology and surface structures of arenaviruses suggest that the snake-borne viruses can be used as efficiently cultivable (likely low-biohazard) model viruses for the Arenaviridae. The presented data will also help to establish diagnostic, treatment, and vaccine approaches and will foster the understanding of epidemiological patterns and the prevention of transmission as well as the search for the reservoir of the viruses. Furthermore, the new knowledge of the arenaviral diversity will likely help to find yet new arenaviruses and arenavirus groups in other than boid ophidian families, e.g., colubrids and viperids, and to narrow the currently huge gap between the BIBDAV group and the classical mammalian arenaviruses.