key: cord-0000650-k2io05hy authors: Belhouchet, Mourad; Mohd Jaafar, Fauziah; Firth, Andrew E.; Grimes, Jonathan M.; Mertens, Peter P. C.; Attoui, Houssam title: Detection of a Fourth Orbivirus Non-Structural Protein date: 2011-10-12 journal: PLoS One DOI: 10.1371/journal.pone.0025697 sha: 15f2b1915443fff0466f0dd89c7c0ab6761833b4 doc_id: 650 cord_uid: k2io05hy The genus Orbivirus includes both insect and tick-borne viruses. The orbivirus genome, composed of 10 segments of dsRNA, encodes 7 structural proteins (VP1–VP7) and 3 non-structural proteins (NS1–NS3). An open reading frame (ORF) that spans almost the entire length of genome segment-9 (Seg-9) encodes VP6 (the viral helicase). However, bioinformatic analysis recently identified an overlapping ORF (ORFX) in Seg-9. We show that ORFX encodes a new non-structural protein, identified here as NS4. Western blotting and confocal fluorescence microscopy, using antibodies raised against recombinant NS4 from Bluetongue virus (BTV, which is insect-borne), or Great Island virus (GIV, which is tick-borne), demonstrate that these proteins are synthesised in BTV or GIV infected mammalian cells, respectively. BTV NS4 is also expressed in Culicoides insect cells. NS4 forms aggregates throughout the cytoplasm as well as in the nucleus, consistent with identification of nuclear localisation signals within the NS4 sequence. Bioinformatic analyses indicate that NS4 contains coiled-coils, is related to proteins that bind nucleic acids, or are associated with membranes and shows similarities to nucleolar protein UTP20 (a processome subunit). Recombinant NS4 of GIV protects dsRNA from degradation by endoribonucleases of the RNAse III family, indicating that it interacts with dsRNA. However, BTV NS4, which is only half the putative size of the GIV NS4, did not protect dsRNA from RNAse III cleavage. NS4 of both GIV and BTV protect DNA from degradation by DNAse. NS4 was found to associate with lipid droplets in cells infected with BTV or GIV or transfected with a plasmid expressing NS4. The genus Orbivirus currently includes twenty two distinct virus species, with genomes composed of 10 segments of linear double stranded RNA (dsRNA), that are vectored by Culicoides midges, ticks, phlebotomine flies, anopheline or culicine mosquitoes. The three economically most important orbiviruses: Bluetongue virus (BTV) (the Orbivirus 'type-species') African horse sickness virus (AHSV) and Epizootic hemorrhagic disease virus (EHDV) are all transmitted by Culicoides biting-midges [1] . Several tick-borne orbiviruses can infect humans, including members of the Changuinola virus, Corriparta virus, Lebombo virus, Orungo virus and Great island virus (GIV) species. The coding assignments of the 10 BTV genome segments were initially determined in 1983 [2, 3, 4] . Seven distinct structural proteins (VP1 to VP7) and 3 distinct non-structural proteins (NS1, NS2 and NS3) were identified in orbivirus infected cells, or after in vitro translation of viral RNA. In most cases each genome segment encodes a single protein from a single open reading frame (ORF), expect segment 9 (Seg-9) and segment 10 (Seg-10), both of which encode two nearly identical proteins initiated from in-phase AUG codons close together near the upstream termini (VP6 and VP6a encoded by Seg-9, and NS3 and NS3a encoded by Seg-10) [2, 5] . However, in vitro translation of BTV RNA segments reproducibly generated a number of smaller translation products of unknown significance, that were usually dismissed as unimportant byproducts of translation [2] . The icosahedral orbivirus core-particle is constructed as two concentric protein shells, the sub-core layer which contain 120 copies/particle of the T2 protein (VP3 of BTV), and the coresurface layer composed of 780 copies/particle of the T13 protein (VP7 of BTV). VP1, VP4 and VP6 are minor enzymatic proteins that are packaged along with the ten genome segments within the central space of the virus core [6, 7] . The orbivirus outer-capsid layer is composed of two additional structural proteins (VP2 and VP5 of BTV), which mediate cell-attachment and penetration during initiation of infection. These outer-capsid proteins are more variable than the core proteins and most of the non-structural proteins, and the specificity of their reactions with neutralising antibodies determines the virus serotype (as exemplified by VP2 of BTV [8] ). The relative number and locations of the BTV structural proteins have been determined in biochemical and structural studies using cryo-electron microscopy and X-ray crystallography [7, 9, 10, 11, 12] . NS1 is the most abundant protein in BTV infected cells, forming tubules that may be involved in translocation of progeny virus particles to the cell membrane [13, 14] . BTV NS2 can be phosphorylated by ubiquitous cellular kinases and is an important matrix protein of the granular viral inclusion bodies (VIB) that form within the cytoplasm of infected cells. VIB represent the primary site of virus replication and assembly. The smallest of the BTV non-structural proteins that were previously identified, are membrane glycoproteins NS3 and NS3a, which are expressed in large amounts in insect cells, but not in mammalian cells. They are involved in the release of progeny virus particles from infected cells [15] . In some orbiviruses (e.g. AHSV) NS3/NS3a are highly variable and it has been suggested that they may be involved in determination of both vector competence and virulence [16] . BTV-Seg-9 encodes the minor core protein VP6, which is a helicase. Recent bioinformatic analyses have identified a new overlapping ORF in Seg-9 of both insect-borne and tick-borne orbiviruses, although the putative protein (identified here as NS4) varies in size between 10 kDa and 22.5 kDa [17, 18] . We report the synthesis and detection of NS4, in the cytoplasm and nuclei of cells infected with insect-borne and tick-borne orbiviruses (represented by BTV and GIV respectively). All animal immunisation work was conducted according to the recommendations in the Animals (Scientific procedures) Act of the Home Office of the UK and the Directive on the protection of Animals used for Experimental and other scientific purposes of the EU. The protocol was approved by the Ethics Committee of animal experiments at the Institute for Animal Health in the UK (Project license number 70/7060). All surgery was performed under sodium pentobarbital anaesthesia, and all efforts were made to minimize suffering. BHK-21 (American type cell culture collection) were grown at 37uC under 5% CO2 in Glasgow's minimum essential medium (GMEM), supplemented with 10% foetal bovine serum, 10% tryptose phosphate broth, penicillin G (100 IU/ml) and streptomycin (100 mg/ml). Culicoides sonorensis KC cells were grown at 28uC in Schneider's insect medium supplemented with 15% fetal bovine serum. Confluent monolayers of BHK-21 cells were infected with either BTV-8 (isolate NET2006/04) or Great Island virus (GIV) (isolate CAN1971/01) at a multiplicity of infection (MOI) of 0.1 pfu/cell. Infected cell cultures were incubated at 37uC for 72 hours until cell lysis began. The cells were then scraped into the supernatant and centrifuged at 3,000 g for 10 minutes. RNA was extracted from cell pellets using guanidinium isothiocyanate (RNA NOW reagent: Biogentex, Tx, USA) as described earlier [19] . KC cells were infected at an MOI of 0.1 pfu/cell and then incubated at 28uC for 7 days. Both BHK-21 and KC cell pellets were used in western blot analyses as described below. Viruses were purified from BHK-21 infected cells, as previously described using a discontinuous sucrose gradient [20] . Virus particles formed a blue opalescent band at the interface of the sucrose solutions. This was recovered and further purified by layering onto a continuous PercollH gradient as previously described [21] , using an SW41 rotor (100000 g, 1 hour, 4uC). The virus formed a blue band which was collected, diluted in 0.1 M Tris-HCl and pelleted at 10000 g for 1 hour. Bioinformatic analyses of the overlapping ORF in Seg-9 of BTV and GIV The hydrophobicity profile of different NS4 proteins was analysed using the Kyte and Doolittle hydrophobicity plot with a window size of 11 amino acids (aa) [22] . Sequence relatedness to proteins in public databases was assessed using the NCBI's BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi)) and the pfam software (http://pfam.sanger.ac.uk/search/sequence). Amino acid alignments of NS4 of various orbiviruses were generated using the Clustal X program [23] and pairwise aa identities calculated using the MEGA 4 package [24] . The presence of 'coiled-coils' was indicated by analyses using the program 'COILS' (http://www.ch. embnet.org/cgi-bin/COILS_form_parser) and the PredictProtein server (http://www.predictprotein.org). The presence of nuclear localisation signals were analysed by PredictNSL, implemented in the PredictProtein server, and the cNLS Mapper (http://nlsmapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi). Synonymous site conservation within the BTV VP6 coding sequence was analysed as described previously [25] . For this procedure, alignment columns in which the reference sequence (GenBank accession number: NC_006008) contained gap characters were removed so that the plots are in reference sequence coordinates. The RNA of BTV-8 or GIV was separated by 1% agarose gel electrophoresis. Seg-9 was cut from the gel using a clean scalpel blade, purified using RNaid kit (MP Biomedicals) and cDNA was synthesised using a single primer amplification technique as previously described [19] ). The ORFs in Seg-9 from BTV-8 (between nucleotides 182 and 415, accession number: AM498059) and GIV (between nucleotides 176 and 748: accession number HM543473) were PCR amplified using specific primers tailed with restriction enzyme sites shown in table 1. The pGEX-4T-2 vector and Seg-9 PCR products were doubledigested with EcoRI and NotI (BTV-8) or EcoRI and XhoI (GIV) enzymes (Invitrogen). The pCI-neo vector and Seg-9 PCR products were double-digested with EcoRI and NotI (BTV-8) or EcoRI and XbaI (GIV) enzymes. Digested products were gel purified using Genclean kit (Qbiogen). Corresponding vectors and PCR products were ligated overnight (O/N) at 16uC using T4 DNA ligase (Roche) to generate pGEX-BTVNS4, pGEX-GIVNS4, pCI-BTVNS4 or pCI-GIVNS4. These recombinant plasmids were used to transform XL1-Blue bacteria (Stratagene). Clones were recovered and grown in trypticase-soy-casein (TSC) medium containing 100 mg/ml ampicillin. The plamsids were subsequently purified using Qiaquick plasmid miniprep kit (Qiagen) and sequenced using the D-Rhodamine DNA sequencing kit and an ABI prism 377 sequence analyser (Perkin Elmer). Confirmed pGEX-BTVNS4 or pGEX-GIVNS4 plasmids were used to transform BL21 or C41 bacteria. A single colony of each plasmid was grown overnight (ON) in TSC/ampicillin, then used to seed 200 ml of fresh TSC/ampicillin. The bacteria were grown until OD600 0.5, then 0.5 mM IPTG was added for induction, for 4 hours at 37uC, or for 8 hours at 28uC. The bacterial cells were pelleted and processed using Bugbuster protein purification (Novagen) as previously described [26] . The soluble fraction of the fusion protein was purified by glutathione affinity chromatography using glutathione sepharose, as directed by the manufacturer (GE Healthcare). Proteins were analysed by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS-PAGE) using a 10% polyacrylamide separating gel (Miniprotean III) with a 3% stacking gel and stained with Coomassie brilliant blue, as described Table 2 . Percentage amino acid identity values between NS4 of BTV, EHDV, AHSV, GIV, PHSV and YUOV. Amino acid identity ranged from 5% to50%. The highest identity exists between BTV and EHDV (50%) followed by PHSV and YUOV-1 (30%). Amino acid identity in NS4 between the tick-borne and insect-borne viruses ranged between 5% and 18%. doi:10.1371/journal.pone.0025697.t002 previously [21] . The purified fusion protein was used to immunize rabbits (Harlan) with an initial injection, followed by 4 boosts at 2 weeks interval in the presence of Montanide ISA50 (Seppic) as an adjuvant. BTV-8 or GIV infected BHK-21 cells (5610 6 cells) and BTV-8 infected KC cells (5610 6 cells) were dissolved for 10 min at 100uC in 1 ml of sample denaturation buffer (160 mM Tris-HCl, 4 mM EDTA, 3.6% SDS, 60 mM DTT, 0.2% ß-mercaptoethanol, 0.8% methionine, 800 mM sucrose). A volume of 20 ml was analysed per well, by electrophoresis in a minigel (Miniprotean III tank -Bio-Rad). Purified and pelleted virus particles were also dissolved in sample buffer and analysed by SDS-PAGE, using a 4-20% gradient polyacrylamide gel. Resolved proteins were electro-blotted on 0.2 mm nitrocellulose membrane (Bio-Rad) using 20 mM Tris, 0.05% SDS, 150 mM glycine and 20% V/V isopropanol transfer buffer. Membranes were blocked with 5% skimmed milk, in Tris buffered saline (TBS: 25 mM Tris/HCl, 150 mM NaCl, 2 mM KCl, pH 7.4) and incubated over night with a dilution of 1/300 rabbit antisera. Membranes were washed three times with TBS-Tween-20 (TBS containing 0.05% Tween-20) and further incubated with monoclonal, anti-rabbit, peroxydase conjugate (Sigma), diluted at 1/750 in 5% skimmed milk. After 2 hours the membrane was washed three times with TBS-Tween-20 and developed using 4-chloronaphthol (Sigma) in presence of hydrogen peroxide. Logarithmically growing BHK-21 cells were infected with BTV-8 (1pfu/cell) for 24 hours then harvested and washed once with PBS. Nuclear extracts were prepared from 2.5610 7 cells using the NE-PER nuclear and cytoplasmic extraction reagent kit (Pierce), as directed by the manufacturer. The nuclear extract was mixed volume to volume with sample denaturation buffer and analysed by SDS-PAGE using a 4-20% gradient polyacrylamide gel. Resolved proteins were electro-blotted on 0.2 mm nitrocellulose membrane as described above, blocked with 5% skimmed milk, in TBS-Tween-20 and incubated over night with a dilution of 1/300 anti-BTV-8 NS4 rabbit antiserum. The membranes were washed three times with TBS-Tween-20 and further incubated with monoclonal, anti-rabbit, peroxydase conjugate (Sigma), diluted at 1/750 in 5% skimmed milk. After 2 hours the membranes were washed three times with TBS-Tween-20 then incubated with Lumilight plus (Roche) chemiluminescent detection reagent, as described by the manufacturer. X-Omat radiographic films (Kodak) were exposed for 10 minutes to membranes then developed as described by the manufacturer. BHK-21 cells were grown on coverslips placed at the bottom of a 24 well plates. 50% confluent cells were infected with 0.1 pfu/ cell of BTV-8 or GIV, incubated at 37uC for 4 hours or 24 to 72 hours, then fixed in 4% paraformaldehyde and processed for immuno-fluorescence. Briefly, rabbit antisera raised against NS4 of BTV-8 or GIV and a mouse anti-alpha tubulin antibody were both diluted 1/500 in PBS containing 0.5% bovine serum albumin (PBS-A) and applied to the fixed cell. After 1 hour incubation at room temperature (RT), slides were washed in PBS, then incubated with Alexa Fluor 488 conjugated anti-rabbit IgG (Invitrogen) and Alexa Fluor 568 conjugated anti-mouse, both diluted 1/250 in PBS. After labelling with primary and secondary antibodies, the cells were stained with DAPI (1:10,000) for 15 BHK-21 cells grown in 24 well plates (75% confluence), were transfected in triplicate, with pCI-BTVNS4 or pCI-GIVNS4 (4 mg/well) using Fugene-6 (Roche). At 48 hours post-transfection, the cells were fixed in 4% paraformaldehyde and processed for immuno-fluorescence, using anti-NS4 antibodies, as described above. Identification of NS4 in cells transfected with pCI-BTVNS4, using anti-BTV-8 immune serum from infected mice BHK-21 cells were transfected with pCI-BTVNS4 using Fugene-6. At 48 hours post-transfection, the cells were dissolved in sample denaturation buffer as described above. Cell lysates were analysed by SDS-PAGE/Western blot, using an immune serum (diluted 1/50, in 5% skimmed milk) from mice infected with BTV-8. Non-transfected cells were used as control. Nucleic acid protection assays dsRNA binding proteins can compete with Dicer (an endoribonuclease of the RNAse III family), reducing its ability to cleave long dsRNAs into 21 bp-long 'interfering' RNAs [27] . A dsRNA ladder (New England Biolabs) with sizes ranging from 500 to 21 bp was used as a template for Dicer cleavage. A competition assay with Dicer (Mobitech), was carried in the presence of 150 ng of expressed BTV-8 or GIV NS4, in a final volume of 20 ml containing 8 ml of the Dicer reaction, 2 mg of dsRNA, 1 mM ATP, 2.5 mM MgCl2. The reaction was incubated at room temperature for 20 minutes, followed by addition of a 1.5 U of Dicer, then incubated for a further 6 hours at 37uC. The reaction products were analysed by 3% agarose gel electrophoresis. DNAse I is an endodeoxyribonuclease that can degrade dsDNA into 59 phosphorylated tetranucleotides [28] . A dsDNA ladder (Promega) with sizes ranging from 2645 to 36 bp provides a target for DNAse I cleavage. Competition assays between DNAse I (Roche) and BTV-8 or GIV NS4, were carried out in a final volume of 20 ml, containing 2 ml of 10X DNAse I buffer and 2 mg of dsDNA. The reaction was incubated at room temperature for 20 minutes, followed by addition of a 2 U of DNAse I, then incubated for a further 30 minutes at 37uC. The completed reaction was heated at 99uC for 1 minute to inactivate the DNase and the reaction products were analysed by 2% agarose gel electrophoresis. The outer capsid protein VP9 of Banna virus (BAV, genus Sedaornavirus, family Reoviridae) expressed in E.coli [29] was used as a control in both RNAse and DNAse assays. A colorimetric assay was developed to detect interactions between NS4 and dsRNA. Synthetic dsRNA was prepared with the 59-end of one strand linked to biotin via a 15-atom mixed polarity tetraethylene glycol spacer (59-Biotine TEG). This design allows the dsRNA to be captured at the bottom of a well of 96 well plate coated with streptavidin, while keeping the dsRNA free as a target for NS4 binding. The sequence of the +ve strand is: 59-Biotine-TEG-UGGAAGCGGCUGGCAAUUAAUUUUGGU-GUC-39 and that of the negative strand is 59-GACAC-CAAAAUUAAUUGCCAGCCGCUUCCA-39. Increasing concentrations (from 1 to 640 ng) of the dsRNA in PBS were added to separate wells of a streptavidin-coated 96 well plate (Pierce) and allowed to bind at room temperature for 2 hours. The wells were washed three times with TBS-Tween-20, then two hundred microlitres of a 5% solution of bovine serum albumin (BSA) in PBS, was added in each well, to block non-specific sites. After washing 3 times with TBS-Tween-20 a fixed amount (150 ng) of either BTV or GIV NS4 in binding buffer (20 mM Tris-HCl pH 7.5, 50 mM KCl, 2 mM MgCl2, 2 mM MnCl2 and 5% glycerol) was added per well, prior to incubation for 30 minutes at 25uC. After the wells had been washed 3 times with TBS-Tween-20, rabbit anti-BTV or anti-GIV NS4 sera was diluted 1/250 in 5% BSA and 100 ml was added to each well, then the plates were incubated at 25uC for 2 hours. After washing three times, 100 ml of peroxydase conjugated antirabbit antibody was added (diluted 1/750 in 5% BSA) to each well. The plates were incubated at 25uC for 2 hours, then washed 3 times with TBS-Tween-20. One hundred microliters of SureBlue TMB 1-component microwell peroxidase substrate (tetramethyl benzidine from KPL) was added per well, then incubated for 30 minutes at 25uC. The reaction was stopped by adding 100 ml of 1 M HCl and the plate was read at OD 450 nm. Wells not containing dsRNA/NS4, were included as negative controls. Wells from which the dsRNA was omitted, but in which NS4 (BTV or GIV) alone was incubated were also included as controls. The program MLOGD models and compares sequence evolution in single-coding and dual-coding sequences. It has previously been used to identify a second ORF, in a different but overlapping reading frame from that encoding the viral helicase (VP6 of BTV), within Seg-9 of the insect-borne orbiviruses [17, 18, 30] . This ORF was also identified in tick-borne orbiviruses [18] . The length of the putative translation product is highly variable, even between closely related Orbivirus species. In BTV and EHDV it is approximately 10 kDa, in Peruvian horsesickness virus (PHSV) and Yunnan orbivirus (YUOV) it is approximately 13.5 kDa, while in AHSV it is approximately 17 kDa, and in GIV it is approximately 22.5 kDa (twice as long as in BTV). These NS4 sequences contain a high proportion of charged residues, with basic R+K (arginine + lysine) content ranging from 13% to 22%, while acidic E+D (glutamic + aspartic acids) content ranges from 12% to 22%. Each NS4 protein contains 4-5 histidine residues, with the exception of the BTV protein, which contains none. The levels of pairwise nucleotide conservation at synonymous sites within aligned sequences of the VP6 ORF, were used to assess the functional importance of the NS4-ORF. Complete or near-complete VP6-encoding sequences from BTV showed strikingly enhanced conservation in the region corresponding to the NS4 ORF (figure 1), supporting and extending previous computational analyses [17] . Enhanced conservation was also apparent at the 59 end of the VP6 coding sequence, indicating that this region (like the terminal noncoding regions of orbiviruses) is likely to contain functionally important elements. Amino acid identity between NS4 of the different Orbivirus species compared ranged from 5% to 50%. Highest identity was detected between BTV and EHDV (50%), followed by PHSV and YUOV-1 (30%). Amino acid identity in NS4 between the tickborne and insect-borne viruses, ranged between 5% to 18% (table 2). Local blast analyses using BLAST-P or TBLAST-N identified significant matches (as defined by the E value in BLAST) between NS4 proteins encoded by other orbiviruses. Analysis of NS4 protein sequences using the pfam program, which uses the hidden Markov model (HMM) based profiles to identify or predict protein functionalities [31, 32] , revealed strong similarities to certain conserved functional motifs. AHSV NS4 exhibits strong relatedness over almost its entire length with DUF domains that have helical structures known to be involved in nucleic acid binding and/or modification [33] . Previous analysis of GIV NS4 identified a 72 amino acid fragment (aa 82 to 153) with 39% similarity to dsRNA-binding domains of similar length (approximately 68aa) in other reovirus proteins [ [18] or other dsRNA binding proteins [34] . BTV NS4 (77aa long) also exhibits relatedness (over aa 14-54) to a DUF domain, belonging to the MetJ/Arc repressor superfamily [35] , which has a ribbon-ribbonhelix-helix DNA-binding motif, with the beta-ribbon located in and recognising the major groove of operator DNA. BTV NS4 shows strong relatedness to fzo-mitofusin protein, a putative transmembrane GTPase. The fzo protein has a coiled-coil structure and mediates mitochondrial fusion [36] . Another protein family with a coiled-coil structure, which also shows a strong match with BTV NS4, is EMP24_GP25L. Members of this family have been implicated in transporting 'cargo' from the endoplasmic reticulum (ER) and are related to the previously described GOLD domain [37] , which is always found combined with lipid-or membrane-association domains. Sequence analyses indicate that PHSV NS4 (111 aa long) contains a coiled-coil domain between aa 75 and 111, YUOV NS4 (113 aa long) contains two coiled-coils domains between aa 5 to 45 and 75 to 105, and AHSV NS4 (143 aa long) contains coiled-coil domains between aa 5 to 85 and aa 110-140. The BTV NS4 (77 aa long) appears to contain only a single coiled coil structure, between aa 27 and 77. Two overlapping potential nuclear localisation signals' (NLS) were identified in the aa sequence of PHSV NS4 ( . Although all of these NLS were monopartite, the GIV NS4 was found to contain a bipartite NLS (position 113-141: RKRGLEFLLLPLHEYVTHCAKEDIR-IYES). The prediction cut-off scores for all these NLS as defined by PredictNSL and cNLS ranged from 4 to 8, indicating dual nuclear/cytoplasmic localisations of a given protein [38] . The aa region 55 to 129 of GIV NS4 showed 29% identity (55% similarity) to aa 1823 to 1890 of UTP20 (a component of the nucleolus). NS4 of BTV and GIV were successfully cloned into pGEX-4T-2 and expressed in C41 at 28uC, as partially soluble proteins fused to GST (figure 2). The soluble fraction was used in competition assays with DNase I or endoribonucleases belonging to the RNAse III family and in binding assays with dsRNA. In contrast when these proteins were expressed in BL-21 they were totally insoluble and formed inclusion bodies. The inclusion bodies fraction was purified using bugbuster reagent, solubilised and used for immunization of rabbits. Western blots analyses, using rabbit antisera raised against recombinant BTV-8 NS4, showed that NS4 is expressed in BTV-8 infected Culicoides KC cells and BHK-21 ( figure 3, figure 4) . The antiserum identified a single protein band with an apparent molecular weight of approximately 12 kDa in BTV-8 infected cells, which is close to the molecular weight calculated for NS4, from the sequence of Seg-9 (,10 kDa). The anti-BTV NS4 antiserum is therefore specific to NS4 and does not cross react with other viral proteins. Western blot analysis using non-infected BHK-21 cells, showed that anti-BTV NS4 rabbit antiserum does not cross react with cellular proteins. A similar analysis, using antisera raised against recombinant GIV NS4, identified a protein of approximately 20 kDa in GIV infected cells (figure 5), corresponding to the theoretical molecular weight of NS4 deduced from the sequence of GIV Seg-9. The anti-GIV NS4 antiserum is therefore specific to GIV NS4 and does not cross react with other viral proteins. Western blot analysis using non-infected BHK-21 cells, showed that anti-GIV NS4 rabbit antiserum does not cross react with cellular proteins. Western blot analyses of purified BTV virus particles showed no reaction with anti-BTV NS4 antibodies, indicating that NS4 is 'non-structural' ( figure 6A, 6B ). Figure 7 shows infected and non-infected BHK-21 cells probed with anti-BTV-8 VP2 antibodies raised in mice. Figure 8 shows infected and non-infected BHK-21 cells probed with anti-BTV8 immune serum from experimentally infected mice. NS4 was identified in the nuclear fraction of BTV-infected BHK-21 cells harvested at 24 hours post-infection by western blot. Rabbit anti-BTV NS4 immune serum, identified the same band in the nuclear extract that was previously identified in infected cell lysates (figure 9). No band was identified in non-infected nuclear extracts. NS4 was detected as early as 4 hours post-infection, mainly in the cytoplasm of BHK-21 cells infected with BTV-8 or GIV ( figure 10A, 10B) . At 24 hours post-infection, NS4 formed small aggregates throughout the cytoplasm and nucleus, suggesting that it makes specific interactions with itself and/or other infected cell components ( figure 11A, 11B) . This is consistent with bioinformatic analyses which identified nuclear localisation signals in NS4 of GIV and BTV (as well as YUOV, PHSV, EHDV, AHSV). Although not all cells are morphologically intact at 72 hours postinfection, with GIV or BTV-8, at this stage NS4 was present in the cell membrane ( figure 12A, 12B) . This is consistent with bioinformatic analysis showing similarities between NS4 and membrane-associated proteins. Another set of cells, which were collected at 36 hours PI contained cells at different stages of infection. Those at an advanced stage of infection had depolymerised and depleted tubulin (figure 13A). No immuno-fluorescence signal was detected when non-infected cells were labelled using anti-NS4 antibodies (figure 13B, indicating that the anti-NS4 antiserum does not cross react with cellular proteins. Further analyses with confocal microscopy identified nucleolar fluorescence using anti-NS4 antibodies in cells infected with either BTV or GIV. Localisation of the NS4 to the nucleoli was visible by confocal fluorescence as well as by overlaying the fluorescence signal onto cells imaged by differential interference contrast microscopy ( figure 14A, 14B) . Localisation of NS4 to the nucleoli was confirmed using anti-fibrillarin antibodies, giving a fluorescence signal that was super-imposable on that of NS4 in the nucleoli ( figure 15 ). BHK-21 cells transfected with pCI-BTVNS4 or pCI-GIVNS4 resulted in expression of NS4 in both the cytoplasm and nucleus ( figure 16A, 16B) . NS4 was also detected in the nucleoli (figure 16B). Expressed NS4 was abundant in the cytoplasm where it formed aggregates similar to those found in infected cells. In many cells NS4 formed spherical bodies with 0.7 and 1 mm in diameter ( figure 16A and 16B) . Similar spherical bodies were occasionally also observed in cells infected with BTV-8 or GIV (figure 17A). Staining with the lipid stain oil-red-O, showed that these spherical bodies are associations between NS4 and lipid droplets ( figure 17B and figure 18A, 18B) . These bodies were identified in BTV-8 infected cells ( figure 17B and figure 18B) , where the oil-red-O stains lipids in the centre of the droplet while NS4 surrounds the lipid droplet. Figure 18A shows cells transfected with pCI-GIVNS4 stained with oil-red-O. Figure 19 shows non-infected cells stained with oil-red-O, where lipid droplets stain with red only. Similar data were recently reported for rotaviruses, where VP2, VP6 or NSP5 were found to associate with lipid droplets [39] . The mice immune serum from an animal infected with BTV-8 identified a protein in cells transfected with pCI-BTVNS4 expressing BTV-8 NS4. The protein band had the same size as that identified by the anti-BTV NS4 rabbit immune serum in BTV-infected cells. No band was identified in non-transfected cells (figure 20). Incubation of a dsRNA ladder (500-21 bp) with Dicer led to cleavage of long dsRNAs, generating 21 bp-long RNAs. Incubation of the dsRNA ladder with BTV NS4 or GIV NS4 alone did not alter dsRNA integrity. dsRNA preincubated with NS4 of GIV was protected against Dicer cleavage, consistent with previous findings regarding the presence of a dsRNA-binding domain. However, BTV NS4 did not protect dsRNA against Dicer and dsRNA was still processed into 21 bp long fragments, as analysed by agarose gel electrophoresis ( figure 21) . Incubation of dsRNA with BAV outer capsid protein VP9 (as a control) did not affect Incubation of a dsDNA ladder (2645-36 bp) with DNAse I led to degradation, while incubation with BTV or GIV NS4 only did not affect dsDNA integrity. However, dsDNA pre-incubated with either BTV or GIV NS4 was at least partially protected against In the colorimetric assay to detect NS4-dsRNA binding, the wells devoid of dsRNA were all negative, with a very low background (values close to zero). Negative control wells, containing only dsRNA, also had OD values close to zero (figure 23). Wells containing biotinilated dsRNA and BTV NS4 were also negative, indicating that BTV NS4 does not bind dsRNA. However, the wells containing dsRNA and GIV NS4, had increasing OD values with an almost linear relationship between the fixed NS4 concentration (150 ng/well) and the increasing dsRNA concentration, reaching a plateau at 320 ng of dsRNA/well (figure 23). This confirms the existence of a dsRNAbinding domain in GIV NS4, which is absent from BTV NS4. Within the family Reoviridae, genome segments encoding more than one protein, from distinct, ORFs have been previously reported for the aquareoviruses, fijiviruses, orthoreoviruses, rotaviruses, phytoreoviruses and oryzaviruses [1] . Genome segments of the orthoreoviruses, phytoreoviruses, oryzaviruses and rotaviruses can be bi-or tri-cistronic with overlapping ORFs. Those in the phytoreoviruses (Seg-9 and Seg-12), orthoreoviruses (segment S1) and rotaviruses (Seg-11) were also found to be expressed in infected cell cultures [40, 41, 42, 43, 44] . Translation of overlapping ORFs from reovirus genome segments has usually been shown to be dependent on leaky scanning [41, 43, 44] although scanning-independent ribosome shunting has also been described [42, 45] . An overlapping ORF in Seg-9, designated as ORFx, was recently identified by bioinformatic analysis in both insect-borne and tick-borne orbiviruses [17, 18] . ORFx appeared to encode a protein with the potential to bind dsRNA that was tentatively named as VP6db [18] . However, in line with previous orbivirus protein nomenclature, we have renamed these proteins, based on their theoretical size and absence from the virion, as non-structural protein 4 (NS4). GIV NS4 was previously shown to contain significant aa sequence matches with dsRNA-binding domains [18] . Further analyses of their amino acid sequences indicate that NS4 may be structured as 'coiled-coils' and that BTV NS4 exhibits significant relatedness (as identified by the pfam programme) with nucleic acid binding proteins that also have coiled-coils or helical structures and are associated with ER or cell membranes. It was suggested that translation of ORFx may be initiated via 'leaky ribosome-scanning' [17] , although the presence of additional AUG codons between the VP6 initiation codon and the presumed NS4 initiation codon in some orbiviruses (including BTV) suggests that additional mechanisms for bypassing intervening AUG codons may be operating [17] . An A-rich polypurine tract is present upstream of the NS4 ORF in all of the sequenced Orbivirus species, except the tick borne St. Croix River virus (SCRV, [46] . The SCRV NS4-ORF (nt 101-379) is interrupted by a stop codon at position 217. Although hydrophobicity profiles of putative NS4 proteins from BTV, AHSV, PHSV, YUOV and GIV as analysed using the Kyte and Doolittle algorithm [22] are somewhat variable, overall they show broadly similar patterns of conserved domains, indicating that the NS4 proteins are generally hydrophilic ( figure 24 ). BTV-8 infected KC or BHK-21 cells and GIV infected BHK-21 cells all contain NS4, as revealed by western blot analyses and confocal microscopy, confirming the existence of a new and previously un-described protein, encoded by ORFx of orbivirus Seg-9. NS4 has not previously been detected in purified BTV virus or core particles [20] . Western blot analyses of purified BTV-8 particles confirmed that the protein is 'non-structural'. Bioinformatic analyses indicate that NS4 contains coiled-coils and is structurally related to other mammalian proteins, with helical or coiled-coil regions. These analyses also suggest that the NS4 may be functionally related to proteins involved in nucleic acid binding, or associated with lipids and membranes. Nuclear localisation signals were predicted in NS4 of PHSV, YUOV, EHDV, BTV, AHSV and GIV. All these proteins are rich in arginine and lysine residues that are essential for NLS [47] . Double-stranded RNA-binding proteins (DRBPs) do not recognize specific nucleotide sequences but interact primarily with A-form double helix RNAs, which differ from the typical dsDNA B-form helix in that the minor groove is shallow and broad while the major groove is narrow and deep. This conformation allows DRBPs to bind non-specifically to dsRNAs. Indeed, the lack of nucleotide binding recognition suggests that target specificity may generally be governed through interactions with other proteins, since many DRBPs bind strongly but nonspecifically to any dsRNA structure in vitro. GIV NS4 can protect dsRNA from degradation by RNAse III endoribonucleases, confirming previous sequence analyses indicating the presence of a dsRNA-binding domains [18] . BTV NS4 which is half the theoretical size of its counterpart in GIV, lacks dsRNA binding domains and did not protect dsRNA from Dicer or RNAse III. NS4 of GIV and BTV both failed to protect ssRNA or ssDNA from degradation by RNAse A, or nuclease S1 respectively (data not shown). However, NS4 of both GIV and BTV did protect dsDNA from degradation by DNAse I, indicating an ability to bind dsDNA. Fluorescent confocal microscopy confirmed that NS4 is expressed in both BTV and GIV infected cells, and starts to accumulate in the cytoplasm and nucleus (as fine aggregates) as early as 4 hours post-infection. However, at 72 hours postinfection NS4 was associated with the cell membrane. This is consistent with analyses suggesting similarities between NS4 and ER-lipid-or membrane-associated proteins [37] . Cells infected with BTV or GIV or transfected with plasmids expressing NS4, showed interaction of NS4 with lipid droplets within the cytoplasm. This is consistent with bioinformatic analysis that identified similarities between NS4 and lipid-associated domains. Similar data were recently reported for VP2, VP6 and NSP5 of rotavirus [39] . . Colorimetric assay to detect interactions of NS4 with dsRNA. The graph shows colorimetric OD readings plotted against concentrations of dsRNA. Increasing concentrations (from 1 to 640 ng) of a biotinylated dsRNA were bound to wells coated with streptavidin. BTV NS4 or GIV NS4 were added to the wells in triplicate. Wells not containing dsRNA/NS4 were included as negative controls. Wells from which dsRNA was omitted, but in which NS4 (BTV or GIV) alone was incubated were also included as controls. Only wells containing the dsRNA to which GIV NS4 was added reacted with anti-GIV antibodies as indicated by increasing OD readings. The readings were almost linear (reaching a plateau at 320 ng of dsRNA) indicating that dsRNA acted as a target for binding of GIV NS4. doi:10.1371/journal.pone.0025697.g023 Viruses can interact with components of the nucleolus [48, 49] and viral proteins can co-localise with proteins such as nucleolin, B23 and fibrillarin (components of the nucleolus). The use of antifibrillarin antibodies identified NS4 in the nucleoli of cells harvested at 24 hours post-infection. Although NS4 was detected in the nucleoli late in infection, anti-fibrillarin antibodies failed to detect fibrillarin. This may reflect BTV induced apoptosis, leading to nuclear condensation and DNA fragmentation, blebbing of the plasma membrane and shrinkage [50, 51] , and/or host cell shut-off [4] . Similar findings were reported in rotavirus (another member of the family Reoviridae, genus Rotavirus) where NSP2 protein was found to cause depolymerisation of tubulin [52] . As part of their replication strategy, viruses can use nucleolar components to favour viral transcription and translation, or alter the cell cycle [48, 49] . Western blot analysis indentified NS4 in the nuclear fraction of BTV infected cells, while immunofluorescence confocal microscopy co-localised NS4 to the nucleolus. GIV-NS4 showed sequence similarity to UTP20, a small subunit processome component and a component of the nucleolus. UTP20 is part of the U3 small nucleolar RNA (snoRNA) protein complex (U3 snoRNP) and is involved in 18S rRNA processing [53] . Whether NS4 interferes with the processing of the 18s rRNA remains to be clarified in future work. The ability of GIV NS4 to protect dsRNA from cleavage by endoribonucleases of the RNAse III family and its ability to bind dsRNA agree with sequence analyses that indicated the presence of a dsRNA binding domain in GIV NS4 [18] . The inability of BTV NS4 to protect dsRNA from cleavage by endoribonucleases of the RNAse III family and its inability to bind dsRNA in a platebased colorimetric assay are in agreement with sequence analyses that failed to detect a dsRNA binding domain in its aa sequence [18] . Other reoviruses dsRNA-binding proteins include Sigma 3 of mammalian orthoreovirus (found in both the cytoplasm and nucleus), and pns10 of rice dwarf virus [54, 55] . SCRV, which persistently infects tick cells but does not grow in mammalian cells, appears to have a non-functional NS4 ORF that is interrupted by a stop codon. These observations suggest that NS4 expression could play a role in productive infection of mammalian cells. The data presented here show that the orbivirus genome encodes four distinct non-structural proteins (NS1-NS4). NS1 and NS3 play an important role in orbivirus exit mechanisms from infected cells [15] . BTV infects mammalian cells, usually resulting in a lytic infection, while infection of KC cells derived from the BTV vector Culicoides sonorensis, become persistently infected with little or no evidence of cell lysis [56, 57] . Previous work showed that intracellular expression of an NS1 specific antibody fragment (scFv) destabilised the formation of NS1 tubules in BTV infected cells [14] . As a consequence, cells became persistently infected and viruses exited by budding instead of via cell lysis. Although BTV NS3 is effectively expressed in insect cells [15] , it is much less abundant in mammalian cells [4] . It was suggested previously [14] that the relative levels of NS1 to NS3 synthesised during infection dictate the fate of cellular pathogenesis as of whether the virus exit occurs by lysis or budding. The rapid accumulation of NS4 in the cytoplasm as early as 4 hours post-infection suggests that this protein plays an early role in the virus replication cycle. At 72 hours post-infection NS4 was absent from the nucleus which could be the consequence of changes affecting the nucleus and the integrity of the nuclear membrane. The presence of the NS4 in the plasma membrane late in infection suggests that it may play a role, alongside NS1 and NS3, in virus exit. Further co-localisation studies will be carried out to assess NS4 interactions with other viral or cellular protein components. Virus Taxonomy Eighth Report of the International Committee on Taxonomy of Viruses Comparison of type 1 bluetongue virus protein synthesis in vivo and in vitro Identification of bluetongue virus type 17 genome segments coding for polypeptides associated with virus neutralization and intergroup reactivity Assignment of the genome segments of bluetongue virus type 1 to the proteins which they encode Sequence of genome segment 9 of bluetongue virus (serotype 1, South Africa) and expression analysis demonstrating that different forms of VP6 are derived from initiation of protein synthesis at two distinct sites The highly ordered double-stranded RNA genome of bluetongue virus revealed by crystallography Translocation portals for the substrates and products of a viral transcription complex: the bluetongue virus core Analysis and phylogenetic comparisons of full-length VP2 genes of the 24 bluetongue virus serotypes Structure of bluetongue virus particles by cryoelectron microscopy Structure of Broadhaven virus by cryoelectron microscopy: correlation of structural and antigenic properties of Broadhaven virus and bluetongue virus outer capsid proteins The atomic structure of the bluetongue virus core Structural studies of orbivirus particles Characterization of the tubules associated with the replication of three different orbiviruses Role of an arbovirus nonstructural protein in cellular pathogenesis and virus release Release of bluetongue virus-like particles from insect cells is mediated by BTV nonstructural protein NS3/NS3A Development of a mouse model system, coding assignments and identification of the genome segments controlling virulence of African horse sickness virus serotypes 3 and 8 Bioinformatic analysis suggests that the Orbivirus VP6 cistron encodes an overlapping gene Complete sequence of Great Island virus and comparison with the T2 and outercapsid proteins of Kemerovo, Lipovnik and Tribec viruses (genus Orbivirus, family Reoviridae) Strategies for the sequence determination of viral dsRNA genomes Purification and properties of virus particles, infectious subviral particles, and cores of bluetongue virus serotypes 1 and 4 Structural organization of an encephalitic human isolate of Banna virus (genus Seadornavirus, family Reoviridae) A simple method for displaying the hydropathic character of a protein The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools MEGA: Molecular Evolutionary Genetics Analysis software for microcomputers A conserved predicted pseudoknot in the NS2A-encoding sequence of West Nile and Japanese encephalitis flaviviruses suggests NS1' may derive from ribosomal frameshifting Recombinant VP7-based enzyme-linked immunosorbent assay for detection of immunoglobulin G antibodies to Colorado tick fever virus Largescale production of dsRNA and siRNA pools for RNA interference utilizing bacteriophage phi6 RNA-dependent RNA polymerase The specificity of five DNAases as studied by the analysis of 59-terminal doublets The structure and function of the outer coat protein VP9 of Banna virus Detecting overlapping coding sequences in virus genomes Pfam: multiple sequence alignments and HMM-profiles of protein domains Hidden Markov model approach for identifying the modular framework of the protein backbone Exploration of uncharted regions of the protein universe A conserved doublestranded RNA-binding domain. roc Natl Acad Sci Crystal structure of omega transcriptional repressor encoded by Streptococcus pyogenes plasmid pSM19035 at 1.5 A resolution Developmentally regulated mitochondrial fusion mediated by a conserved, novel, predicted GTPase The GOLD domain, a novel protein module involved in Golgi function and secretion Six classes of nuclear localization signals specific to different binding grooves of importin alpha Rotaviruses Associate with Cellular Lipid Droplet Components To Replicate in Viroplasms, and Compounds Disrupting or Blocking Lipid Droplets Inhibit Viroplasm Formation and Viral Replication Characterization of the NSP6 protein product of rotavirus gene 11 Polycistronic (tri-or bicistronic) phytoreoviral segments translatable in both plant and insect cells Leaky scanning and scanning-independent ribosome migration on the tricistronic S1 mRNA of avian reovirus Biosynthesis of reovirus-specified polypeptides. The s1 mRNA synthesized in vivo is structurally and functionally indistinguishable from in vitro-synthesized s1 mRNA and encodes two polypeptides, sigma 1a and sigma 1bNS Analysis of ribosome pausing during translation of reovirus S1 and S4 mRNAs in virus-infected and vector-transfected cells Facilitated leaky scanning and atypical ribosome shunting direct downstream translation initiation on the tricistronic S1 mRNA of avian reovirus Complete sequence characterization of the genome of the St Croix River virus, a new orbivirus isolated from cells of Ixodes scapularis Expanding the definition of the classical bipartite nuclear localization signal Interaction of the coronavirus nucleoprotein with nucleolar antigens and the host cell The Nucleolus-a gateway to viral infection? Bluetongue virus outer capsid proteins are sufficient to trigger apoptosis in mammalian cells Bluetongue virus induces apoptosis in cultured mammalian cells by both caspase-dependent extrinsic and intrinsic apoptotic pathways Sequestration of free tubulin molecules by the viral protein NSP2 induces microtubule depolymerization during rotavirus infection Human 1A6/DRIM, the homolog of yeast Utp20, functions in the 18S rRNA processing Multiple functions of Rice dwarf phytoreovirus Pns10 in suppressing systemic RNA silencing Regulated, stable expression and nuclear presence of reovirus double-stranded RNA-binding protein sigma3 in HeLa cells Susceptibilities of 14 cell lines to bluetongue virus infection Cell lines from Culicoides variipennis (Diptera: Ceratopogonidae) support replication of bluetongue virus