A Calcium Sensor Discovered in Bluetongue Virus Nonstructural Protein 2 Is Critical for Virus Replication

After entering the host cells, viruses use cellular host factors to ensure a successful virus replication process. For replication in infected cells, members of the Reoviridae family form inclusion body-like structures known as viral inclusion bodies (VIB) or viral factories. Bluetongue virus (BTV) forms VIBs in infected cells through nonstructural protein 2 (NS2), a phosphoprotein. An important regulatory factor critical for VIB formation is phosphorylation of NS2. In our study, we discovered a characteristic calcium-binding EF-hand-like motif in NS2 and found that the calcium binding preferentially affects phosphorylation level of the NS2 and has a role in regulating VIB assembly.

Both proteins are lost during endocytosis, and the inner core is subsequently released into the cytoplasm. The BTV core consists of the remaining five proteins and the viral genome of 10 double-stranded RNA (dsRNA) segments. In addition to 7 structural proteins, 4 nonstructural proteins, NS1 to NS4, are synthesized during virus replication. Two of these are major nonstructural (NS) proteins, NS1 and NS2, which are synthesized during early infection, and each plays an essential role in virus replication. The third NS protein, NS3/NS3A, is a transmembrane protein and facilitates release of the newly assembled BTV (1)(2)(3). NS4 is newly identified, and its function is still not fully characterized and a fifth putative nonstructural protein (4)(5)(6). Previously, we reported that NS3 interacts with cellular exocytic pathway protein p11 (S100A10), a protein known to facilitate Ca 2ϩ uptake, suggesting indirect involvement of NS3 with Ca 2ϩ -related signaling pathways (1,7). Several enveloped and nonenveloped viruses employ their proteins (Tat, gp120, nef of HIV-1, HBx of hepatitis B virus [HBV], NSP4 of rotavirus, P7 of hepatitis C virus [HCV]) to modulate cellular Ca 2ϩ hemostasis for ensuring a successful viral life cycle (8,9). For example, rotavirus expresses membrane-localizing NSP4 protein that binds Ca 2ϩ and influences Ca 2ϩ homeostasis (9). BTV and rotavirus belong to the same family; however, unlike rotavirus NSP4, a Ca 2ϩ binding protein in BTV is yet unknown.
In this study, we used bioinformatics to identify whether any of the BTV proteins has a Ca 2ϩ binding motif. We identified EF-hand-like motif in NS2, which is the only viral encoded phosphoprotein and is essential for replication (10). The 357 amino acid (aa) long NS2 is the major component of viral inclusion bodies (VIBs), the sites for viral capsid assembly and genome packaging. The identified EF-hand-like motif in NS2 was comparable to those found in other member proteins of the EF-hand superfamily that are characteristically known for Ca 2ϩ binding. Using recombinant purified protein, together with biochemical and biophysical analysis, we demonstrated that Ca 2ϩ binding changes the secondary structural conformation of NS2. Moreover, our cryo-electron microscopy (CryoEM) analysis of NS2 oligomer in the presence of Ca 2ϩ exhibited a helical architecture. By site-specific targeted mutagenesis in the recombinant NS2 and in the replicating viral genome by reverse genetics, we identified the specific Ca 2ϩ binding site of NS2 and demonstrated its importance in NS2 phosphorylation level, formation of VIBs, and virus replication. Altogether, our results suggest that Ca 2ϩ sensing by NS2 influences NS2 phosphorylation and may be involved in the regulation of VIB assembly/disassembly, a process critical for virus replication and the release of newly assembled cores from VIBs (2,11).

RESULTS
Computer-based sequence analysis of BTV proteins reveals putative Ca 2؉ binding site in NS2. In order to identify putative Ca 2ϩ binding motif, we used the SMART motif search program for each of the 11 BTV proteins, including seven structural proteins (VP1 to VP7) and four NS proteins (12). In our linear sequence search, only NS2 exhibited signature residues (aa 200 to 300) of EF-hand-like motif, found in Ca 2ϩ binding proteins of the EF-hand superfamily (Fig. 1). In particular, presence of acidic amino acids Asp and Glu in the region aa 250 to 262 suggests the calcium binding potential of this segment of NS2 (Fig. 1A) (13)(14). However, we found that the relative positioning of signature residues of the EF-hand motif and Ca 2ϩ binding residues identified in NS2 is different than has been observed in a typical EF-hand containing calcium binding proteins (CaBP), thus making this putative motif less obvious. These acidic residues in NS2 are continuous rather than alternate as found in the case of standard EF-hand motifs (Fig. 1A). These clusters (aa 250 to 262) of Asp (D) and Glu (E) of NS2 resemble more closely the "Ca 2ϩ bowl" found in BK (big potassium) channels (15)(16)(17)(18)(19)(20)(21). Further, these Asp and Glu residues are highly conserved among different BTV serotypes, indicating that the putative Ca 2ϩ binding motif is likely to be important for BTV replication (Fig. 1B).
Demonstration of Ca 2؉ binding ability of NS2 oligomers and helix-to-coil transition in secondary structure. To validate our bioinformatics analysis, we ex-pressed NS2 in Escherichia coli in a nonphosphorylated form that was purified and analyzed by SDS-PAGE gel ( Fig. 2A). Purified NS2, free of any bound metals and pretreated with Chelex 100, was examined for Ca 2ϩ binding activity during metal titration experiment through circular dichroism (CD). The changes in intrinsic far-UV CD spectra were recorded, as the direct measure of Ca 2ϩ binding to purified NS2, without (apo) or with (holo) Ca 2ϩ at increasing concentrations of Ca 2ϩ from 5 M to 10 mM to calculate the dissociation constant, K d . NS2 showed a K d value of 53.9 (Ϯ8.4) M for Ca 2ϩ binding. In comparison, another divalent ion, Mg 2ϩ , when tested for NS2 binding in a parallel titration experiment with the same concentration range, showed a K d value of 2.48 (Ϯ 0.4) mM, suggesting a weaker binding of Mg 2ϩ compared to Ca 2ϩ ( Fig. 2B and C).
Also, NS2 formed stably intact oligomers and eluted at an elution volume on the size exclusion column corresponding to an estimated molecular weight of ϳ440 kDa, suggesting decamers (Fig. 2D). The stability of the oligomers was not dependent on Ca 2ϩ binding, as both apo NS2 and NS2-Ca 2ϩ eluted at the same elution volume. To investigate further the effect of Ca 2ϩ on NS2 secondary structure, which was largely alpha helical, we analyzed the far-UV CD spectrum at optimum Ca 2ϩ concentrations based on the prior titration experiments (Fig. 3A). NS2 showed dose-dependent changes in secondary structure elements in the presence of Ca 2ϩ (Fig. 3B).
The effect of Ca 2ϩ on NS2 spectra was diminished in the presence of 0.4 M EGTA, consistent with chelation of the Ca 2ϩ ion (Fig. 3A) (18). Importantly, at a higher molar concentration of Ca 2ϩ ([Ca 2ϩ ] ϭ 44.67 M), NS2 showed a very different CD spectrum (Fig. 3B) suggestive of a helix-to-coil transition in response to Ca 2ϩ binding, as shown by other Ca 2ϩ binding proteins from the EF-hand superfamily, for example, calmodulin (22). The change of helix to coil also prompted us to analyze in silico predicted secondary structure of NS2 near the Ca 2ϩ binding site. The computer program PSIPRED suggested helix, beta strands, and coils in NS2 protein in different regions; however, IUPred2, a specific program to predict intrinsic unfolded regions or coil, suggested unfolded regions are mainly located near and at the Ca 2ϩ binding site ( Fig. 4A and B) (23,24).
Identification of Ca 2؉ binding site of NS2. To determine which Asp and Glu residues in the predicted region of aa 250 to 262 are responsible for Ca 2ϩ binding activity, we generated a series of recombinant NS2 mutant proteins by site-specific mutagenesis targeting these three sites ( Fig. 1; Table 1). The amino acid substitu-  tions were introduced by replacing negatively charged residues Asp (D) and Glu (E) with Ala (A), a neutral amino acid, which would have minimal impact on overall protein structure, unlike deletion mutants. The NS2 alanine mutant proteins, such as DDDE 250 -253 AAAA (A1), ED 256 -257 AA (A2), and DDE 260 -262 AAA (A3), respectively, were then purified and analyzed by gel electrophoresis to determine that each mutant protein expressed was stable and equivalent to wild-type NS2 (wtNS2) (Fig. 4C). Further, size exclusion chromatography analysis of each protein showed that all three mutant proteins appeared equivalent to wtNS2, with an approximate molecular mass of ϳ440 kDa (data not shown). Prior to investigation of Ca 2ϩ binding activity of NS2 mutants, we compared their CD spectra with that of wtNS2 and calculated the estimated secondary structure elements using the BeStSel program (25) (Fig. 5A). There was no significant change in the percentage of helix, beta strands, or turns in mutants, indicating no major changes in the secondary structure elements (Fig. 5B). CD titration analysis of each NS2 mutant protein was then performed in the presence of Ca 2ϩ . The two mutants A2 and A3 showed the K d values (ϳ64 M and ϳ44 M, respectively) similar to that of wtNS2 (K d value of ϳ54 M), suggesting not much change in the Ca 2ϩ binding activities of mutants A2 and A3 ( Fig. 5C and D). However, the mutant A1 showed weak Ca 2ϩ binding as reflected from an increase in K d value to ϳ150 M ( Fig.  5C and D). Thus, the four alanine residue (aa 250 to 253) substitutions of Glu and Asp residues have affected Ca 2ϩ binding ability significantly, indicating that the three consecutive Glu and an Asp DDDE at aa 250 to 253 are important for Ca 2ϩ binding activity of NS2. Ca 2؉ binding enhances phosphorylation level of NS2 in vitro. Since the Ca 2ϩ binding site (aa 250 to 253) is in juxtaposition of serine residues (S249, S259), known for phosphorylation modification, it was more likely that Ca 2ϩ -mediated transition in the secondary structure elements could affect the level of NS2 phosphorylation. Therefore, we undertook an in vitro kinase assay for NS2 phosphorylation, a modification required for VIB assembly (2,11). Since casein kinase II alpha (CK2␣) is responsible for NS2 phosphorylation, we used CK2␣ and purified unmodified NS2 as substrate (11). Our data showed that, for the fixed ratio of substrate and kinase, a much higher signal of [␥-32 P]-labeled phosphate group transfer to NS2 was achieved in the presence of Ca 2ϩ ions than that in the presence of Mg 2ϩ ions ( Fig. 6A and B). Presence of the Mg 2ϩ ion showed a minimum basal level of phosphorylation of substrate NS2, determined at two different metal ion concentrations. Taken together, of these two ions, Ca 2ϩ binding specifically increases phosphorylation level of purified NS2 protein. Further, we did not observe any increase in the activity of CK2␣ for another substrate (e.g., glutathione S-transferase [GST]) in the presence of Ca 2ϩ ions (control; data not shown). To confirm further the specificity of Ca 2ϩ binding activity on NS2 phosphorylation, we assessed the three NS2 mutants A1, A2, and A3 as substrates for CK2␣ kinase assay in the presence of Ca 2ϩ (Fig. 6C). The NS2 mutants A2 and A3 did not show any significant change in the level of phosphorylation compared to that of wtNS2. In contrast, NS2 mutant A1 showed a marked decrease in the level of phosphorylation, suggesting a critical role of the calcium binding site on phosphorylation of the protein by the CK2␣ kinase (Fig. 6C).
Subunits in NS2 oligomers are arranged in helical fashion. Since NS2-Ca 2ϩ interaction changes the percentage of helix in NS2, we investigated whether this change has any destabilizing effect on the oligomeric state of NS2. To this end, purified NS2 in the presence of Ca 2ϩ was visualized by cryo-electron microscopy (CryoEM). In total, 159,361 particle images were selected from 2,712 CryoEM micrographs and subjected to image classification to obtain two-dimensional (2D) class averages. Particles with clear helical organizational features were observed in some 2D classes (Fig. 7). The pitch and outer diameter of those helical particles were measured to be 75 Å and 91 Å, respectively ( Fig. 7A and B), matching the parameters of the helical structure of NS2 N-terminal domain observed by X-ray crystallography (26). Modeling of the crystal structure into this CryoEM average suggests that a single turn of the NS2 helix is contributed by 10 monomers (Fig. 7C). The full-length NS2 oligomer showed a helical overall structure, and the N-terminal domain of NS2 contributes to the formation of the helical configuration. Two-dimensional classes with clear "ring-like" feature could be further assigned to helical NS2 oligomers on their front view ( Fig. 7B and D) since their outer diameters are the same as those of "helical particles" (Fig. 7A). Interestingly, clear Discovery of Ca 2ϩ Binding Protein in BTV Journal of Virology density could be observed inside the "ring," which corresponds to the center of the NS2 helical shape (Fig. 7B). Considering the C-terminal end of the N-terminal domain points toward the inside of the helical particle (K160; purple sphere in Fig. 7C), we propose that the C-terminal domains of NS2 are located inside of the helical structure formed by the N-terminal domains of individual subunits of the helical oligomers of NS2. Disruption of putative Ca 2؉ binding residues in NS2 affects virus replication. The above data demonstrated that NS2 possesses a specific Ca 2ϩ binding site. Such specific calcium binding activity is likely to influence virus fitness in infected cells.
To address this issue, we introduced site directed substitution mutations in the BTV genome and studied the effect using reverse genetics (RG) as described previously (27). A set of alanine substitution mutations on NS2 were designed ( Table 1). Three of these mutations in the encoding S8 segment were the same NS2 sites, A1 (DDDE 250 -253 AAAA), A2 (ED 256 -257 AA), or A3 (DDE 260 -262 AAA). In addition, two multisite mutations in S8, A1 ϩ A2 (DDDE 250 -253 AAAA ϩ ED 256 -257 AA) and A1 ϩ A2 ϩ A3 (DDDE 250 -253 AAAA ϩ ED 256 -257 AA ϩ DDE 260 -262 AAA) were created (Table 1) to assess if the other two sites (A2 and A3) have influence on Ca 2ϩ binding residues aa 250 to 253 during virus replication. For a negative control, we used an available NS2 phosphorylation mutant SAA, in which phosphorylated serine sites (S249 and S259) were previously substituted by alanine residues (SAA) that perturbed virus replication (2,11). When BSR cells were transfected with each mutant S8 together with 9 remaining RNA segments for virus recovery by reverse genetics, only A2 and A3 mutant viruses were recovered successfully but not A1 or the others that included the A1 mutation (A1 ϩ A2 and A1 ϩ A2 ϩ A3) (Fig. 8). Subsequently, each RNA cocktail was then used to transfect BS8 cells, which stably express wtNS2 protein (segment 8) to validate the RG experiment and viability of mutant viruses. In parallel, BSR cells were also transfected similarly for comparison. Cells were fixed 48 h posttransfection. The mutant virus A2 and A3 formed plaques both in BSR and BS8 cells with similar phenotypes of wt virus and titers (PFU/ml values of ϳ7 log 10 ), suggesting no apparent change due to these mutations (Fig. 8). In contrast, the mutant viruses A1, A1 ϩ A2, and A1 ϩ A2 ϩ A3 and the negative control SAA mutant virus showed typical plaque-forming phenotype only in the NS2 complementary BS8 cells. These data highlighted the critical role of Asp and Glu residues at aa 250 to 253 (site A1) and further validated the RG experiment of the mutant S8 that failed to recover in normal BSR cells (Fig. 8).
To investigate further the failure of virus recovery with the mutation at the A1 site, we assessed whether Asp and Glu (aa 250 to 253) residues identified for Ca 2ϩ sensing were critical for the NS2-triggered VIB formation, the sites of virus assembly. BSR cells were, therefore, infected with either wt virus or one of the three mutant viruses A1 (DDDE 250 -253 AAAA), A2 (ED 256 -257 AA), or A3 (DDE 260 -262 AAA) recovered from the NS2  (Fig. 9). The A1 mutant virus after infection in cell showed ϳ1.6-fold smaller VIBs than wt virusinfected cells (Fig. 9B). Taken together, these data suggest that Ca 2ϩ binding Asp and Glu residues in aa 250 to 253 have a role in VIB formation and virus replication, consistent with the in vitro Ca 2ϩ binding and kinase assay data ( Fig. 2 and 6). NS2 phosphorylation decreases following disruption of putative Ca 2؉ binding motifs in NS2. To obtain direct evidence on whether the defective Ca 2ϩ binding motif resulting in diminished VIBs was due to poor phosphorylation of NS2, BSR cells were infected with wt and the mutant viruses in addition to NS2 phosphorylation-negative mutant SAA following optimized protocol for BTV (11). BSR cells were infected with the viruses at a multiplicity of infection (MOI) of 1 and NS2 purified by immunoprecipitation (11). Pulldowns were confirmed by Western blotting and the gel stained with Pro-Q Diamond phosphoprotein gel stain followed by densitometry to determine the relative phosphorylation states. NS2 phosphorylation for mutants A2 and A3 was not significantly different from that of the wt; however, NS2 phosphorylation in mutant A1 was significantly reduced (ϳ70%). The SAA mutant virus showed no phosphorylation in BSR cells as previously reported (Fig. 10) (11). The decrease in phosphorylation in mutant A1 is consistent with the data obtained from in vitro phosphorylation experiment and poor Ca 2ϩ binding observed from CD ( Fig. 5 and 6). These data indicate that poor Ca 2ϩ binding, due to disruption of Asp and Glu amino acid residues in the Ca 2ϩ binding motif (aa 250 to 253), specifically interferes with NS2 phosphorylation in cells infected with Ca 2ϩ mutant virus.

DISCUSSION
Several viruses are known to synthesize Ca 2ϩ binding proteins containing EF-handlike motifs (28). Our in silico methods predicted the presence of a unique EF-hand-like calcium binding motif in NS2, resembling more closely a Ca 2ϩ bowl-like structure with clusters of Asp and Glu residues. The absence of a high-resolution structure of fulllength NS2 limits our understanding of the detailed structural features of the Ca 2ϩ bowl. Therefore, we have taken a genetic approach to validate the predicted Ca 2ϩ binding activity of this region. We tested alanine substitution of these residues by generating recombinant mutant NS2 proteins, A1, A2, and A3, targeting three sites of the putative Ca 2ϩ binding region. Our biochemical and biophysical experiments with these mutant proteins demonstrated that NS2 is indeed a Ca 2ϩ binding protein, and the binding of Ca 2ϩ to the negatively charged residues in the A1 site (aa 250 to 253) is highly specific. Our data also suggest that the unusual Ca 2ϩ bowl-like motif of NS2 acts as a Ca 2ϩ sensor. In addition, in the presence of a specific chelating agent, such as EGTA, the CD spectrum of NS2-Ca 2ϩ was partially reversed. Further, like several other CaBP, NS2 also demonstrated a contrasting change in the percentage of helix in NS2 (helix-to-coil) upon Ca 2ϩ binding in far-UV CD spectroscopic measurements. The Ca 2ϩ concentrations used in the far-UV CD experiments were to a 1:1 stoichiometry (approximately) with NS2, which is a decamer in purified form. More importantly, when we introduced the same mutations into the NS2-encoding gene of a replicating viral genome, the A1 mutation failed to recover infectious virus and plaque formation, reflecting the importance of the Ca 2ϩ sensing activity of NS2 for the production of infectious virus.
To investigate the mechanism behind failure of virus recovery, we tested the effect of calcium binding mutants on the phosphorylation state of NS2 protein both in vitro (recombinant NS2 protein) and in vivo (NS2 expressed by mutant virus particles in cells). In our in vitro kinase assay, the level of phosphorylation was increased in the presence of Ca 2ϩ , indicating that Ca 2ϩ -mediated changes in secondary structural elements in NS2 might have led to better access to the phosphorylation sites to the enzyme. The data from the kinase assay was consistent with the CD data on Ca 2ϩ -mediated changes in secondary structural elements (helix) in NS2. These Ca 2ϩ -induced changes in NS2 are significant, although they do not appear to destabilize its oligomeric state as revealed by CryoEM analysis and size exclusion chromatography. This suggests that it is the oligomeric NS2 that senses the Ca 2ϩ ions, possibly through coordination between the protomers, rather than individual monomeric subunits independently sensing Ca 2ϩ ions and forming a decamer. Interestingly, a ring-like shape was inferred for full-length NS2 based on negative-stain electron microscopy (29). In this study, CryoEM analysis of full-length NS2 shows that the subunits are arranged in helical configuration (10 subunits/pitch). In vivo, similar effects on phosphorylation state of NS2 in cells infected with mutant virus suggest a crucial role of NS2 in host's calcium homeostasis, which is possibly linked to the wider role of NS2 in VIB assembly/disassembly. Several viral proteins are reported to alter cellular calcium homeostasis to assist viral life cycle by modulating mainly membrane-associated Ca 2ϩ pumps. For instance, HBx of HBV modulate Ca 2ϩ pumps at the plasma membrane (30), viral proteins such as Vpr (HIV-1) modulate Ca 2ϩ pumps associated with mitochondrial membranes (31), and NSP4 of rotavirus (Reoviridae) do the same to Ca 2ϩ pumps at the endoplasmic reticulum (32). It is noteworthy that, although NS2 is largely cytosolic not localizing to any membrane, it demonstrates significant potential to manipulate the host's Ca 2ϩ signaling through characteristic calcium binding motif (EF-hand-like) during the assembly of VIBs, the viral assembly factories.
Specifically, different members of the Reoviridae family also form similar VIB-like structures (referred to as viral factories or viroplasms in rotavirus/reovirus), although each have their own unique features (33,34). One of the important regulatory factors critical for VIB/viroplasm assembly is phosphorylation of viral proteins, such as NSP2 of rotavirus and NS2 of BTV, which controls self-oligomerization of VIB-forming viral proteins (2,33). Thus, our study provides further insights on the mechanism behind the regulation of this phosphorylation of VIB-forming protein NS2 in BTV. We found that Ca 2ϩ binding preferentially affects phosphorylation level of the NS2, suggesting the broader relevance of NS2 in the context of modulating Ca 2ϩ -related signaling. During VIB assembly in the infected cells, phosphorylated NS2 is produced abundantly, and its level can be correlated with the abundance of CaBP calmodulin. These NS2 molecules play multiple roles, such as recruiting different RNA segments, self-oligomerization forming large inclusion bodies for replication, and viral genome packaging. The complex mechanism of assembly and disassembly of VIBs must be reversible and precisely regulated such that it is in sync with several other cellular processes, which is necessary to avoid any aggregate formation within the host cells during the assembly/disassembly. Ca 2ϩ signaling controls diverse cellular processes. Understanding the details of NS2-Ca 2ϩ interaction greatly expands our knowledge of VIB assembly and disassembly in the context of calcium homeostasis of the cell (35)(36)(37)(38)(39)(40)(41). Although the details of the complex mechanism of VIB assembly are not well understood, the importance of phosphorylation of NS2 in VIB assembly was clearly demonstrated in our earlier study where mutations of serine residues S249 and S259 (phosphorylation site) aborted VIB assembly (2). In light of our current data, we propose a model of NS2-mediated VIB assembly in which the Ca 2ϩ -sensing activity of NS2 is linked to its phosphorylation status (2,11), which in turn controls the VIB assembly required for virus replication and genome packaging.

MATERIALS AND METHODS
Construction of expression plasmids, protein purification, and gel filtration. Based on the SMART motif search results (12), the plasmids for NS2 mutants were generated by the QuikChange mutagenesis method (Agilent Technologies) from the expression plasmids pET 15b-NS2 for bacterial expression and pCAG NS2 for mammalian cell expression (27,42). Plasmids used for reverse genetics were pCAGGS BTV1 protein expression plasmids (pCAG-VP1, pCAG-VP3, pCAG-VP4, pCAG-VP6, pCAG-NS2) and T7 plasmids for BTV transcripts as previously described (27). Site-directed mutagenesis of BTV1 NS2 was performed in both T7 plasmid for the segment 8 (encoding NS2) template, pET15b-NS2, and pCAG-NS2 template using the following mutagenic primers (5=-3=): DDD 250-252 AAA (AGGTGAAGACTCT GAGTGCCGCTGCTGAACAAGGTGAGGATGC), ED 256-257 AA (CGATGATGAACAAGGTGCGGCTGCGAGTGACG ATGAAC), and DDE 260-262 AAA (CAAGGTGAGGATGCGAGTGCCGCTGCACACCCAAAAACTCATA). Obtained mutants were sequenced using internal NS2 primers in order to confirm the presence of the desired mutation(s). The wild-type NS2 and mutant proteins were expressed in the E. coli bacteria strain BL21(DE3) pLysS. The culture was grown at 37°C until optical density at 600 nm (OD 600 ) reached 0.5 to 0.6 and induced with 0.5 mM isopropyl-␤-D-thiogalactopyranoside (IPTG). The bacterial cultures were grown at 37°C for the next 4 h postinduction. Cells were then lysed for the protein purification. Ni-nitrilotriacetic acid (Ni-NTA) affinity purification was used to purify the wild-type and mutant proteins eluting in the presence of 250 mM imidazole (elution buffer, 20 mM Tris-HCl, 150 mM NaCl, pH 7.4, 250 mM imidazole). In order to remove traces of nucleic acid from the purified protein, samples were treated with Benzonase nuclease or micrococcal nuclease where necessary. For gel filtration, a Superdex 200 10/300 GL (GE/Cytiva) column was used in a running buffer of 20 mM HEPES, 100 mM NaCl, pH 7.4, at a flow rate of 0.2 ml/min. Circular dichroism. The circular dichroism (CD) spectra of wild-type apo NS2 (treated with Chelex 100 resin; Bio-Rad) were recorded in CD buffer (20 mM HEPES, 100 mM NaCl, pH 7.4) at room temperature (20°C). All buffers were pretreated with Chelex 100 resin to remove divalent metal ions contaminants. For calcium binding studies, the apo NS2 sample was titrated with Ca 2ϩ salts in increasing concentration. Far-UV CD spectrum data were collected from 260 to 195 nm with a 0.5-mm rectangular cell path length at 20°C on the Applied Photophysics Chirascan and Chirascan Plus spectrometers (Leatherhead, UK) attached to a Peltier unit (Quantum Northwest TC 125). The UV and CD spectra were smoothed (window factor of 4; Savitzky-Golay method) and analyzed using Origin v6 and APL Pro-Data Viewer v4.2.15. For comparative metal ion titration experiments, same preparation of the wtNS2 protein sample was aliquoted in two halves just before titration on CD instrument. Equal amounts of Ca 2ϩ and Mg 2ϩ were added to each tube, respectively, in the increasing concentrations to obtain same concentrations for two metal ions. The Ca 2ϩ or Mg 2ϩ concentrations were taken from the 5 M to 10 mM range. For K d measurement, CD 222 nm was plotted (y axis) for each Ca 2ϩ concentration (x axis), and the K d value was derived using GraphPad Prism choosing inbuilt one-site total function for nonlinear regression (curve fit). Likewise, similar concentrations of mutant proteins A1, A2, and A3 were used for titration, and a similar Ca 2ϩ concentration range (5 M to 10 mM) was studied.
In vitro kinase assay with [␥-32 P]ATP. For the kinase assay, substrate protein NS2 (6ϫHis tag) and kinase enzyme CK2 (GST tag) were expressed separately in E. coli cells and purified using nickel and glutathione Sepharose, respectively. The kinase reaction was conducted in a 50-l reaction mixture in 1ϫ reaction buffer (20 mM Tris, 100 mM NaCl, pH 7.4, 1 mM dithiothreitol [DTT]). The reaction tubes were prepared by adding two different concentrations of Mg 2ϩ or Ca 2ϩ while keeping the substrate (0.7 mg/ ml) and enzyme (0.4 mg/ml) concentrations unchanged. For other tubes, different mutants of NS2 were added with Ca 2ϩ . The reaction was started by adding 10 Ci of [␥-32 P]ATP (Pekin Elmer; 250 Ci) and 0.5 mM ATP in each tube at an interval of 15 s. Additional 1ϫ buffer was added to the tubes to adjust the dilution factor, making each tube contain up to 50 l of reaction volume. The reaction tubes were then incubated for 30 min at 37°C. After incubation, the reaction tube was boiled in SDS sample buffer and loaded on SDS-PAGE to run at 90 V for 3 h. The SDS-PAGE gel was then dried, exposed to film, and observed under imager.
CryoEM sample preparation, image collection, and data processing. For CryoEM, 2.5 l of NS2 in 20 mM Tris, 150 mM NaCl, pH 7.4, and supplemented with 0.1 mM CaCl 2 , was applied to thin continuous carbon film on lacey grid (Ted Pella) and blotted using a Vitrobot Mark IV (Thermo Fisher Scientific) with the standard Vitrobot filter paper at 4°C. The blotting time was set to 6 s, blotting force was set to 2, and drain time was set to 1 s. The grid was flash-frozen in liquid ethane and stored in liquid nitrogen. A total of 2,712 micrographs were collected on a Titan Krios 300 kV electron microscope (Thermo Fisher Scientific) equipped with a Gatan imaging filter (GIF) Quantum LS and a Gatan K2 Summit direct electron detector operated in super-resolution mode at magnification of 130K (calibrated pixel size of 0.535 Å on the sample level). The GIF slit width was set to 20 eV. The dose rate on the camera was set to ϳ6.5 electrons/pixel/s. The total exposure time of each movie was 8 s, which fractionated into 40 frames of images with 0.2-s exposure time for each frame. Dose-fractionated frames were 2ϫ binned (pixel size, 1.07 Å) and aligned for beam-induced drift correction using UCSF MotionCor2 (43). The defocus values of the micrographs were determined by CTFFIND4 (44) to be in the range of Ϫ1.0 m and Ϫ4.0 m. From a total of 2,712 micrographs, 463,691 particles were boxed out in 400 by 400 square pixels and 2ϫ binned to 200 by 200 square pixels (pixel size of 2.14 Å) to speed up further data processing with Relion 3.0 (45). After one round of 2D classification, 159,361 particles (34.4% of all particles) were selected and subjected to the second round of 2D classification. Represented 2D classes are selected and used for the measurement of pitch and diameter of spiral NS2 oligomers.
Cells, viruses, and reverse genetics. BSR cells (BHK-21 subclone) or BS8 (BSR cells stably expressing NS2/segment of BTV) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% (vol/vol) fetal bovine serum (FBS) at 35°C in 5% CO 2 . Media for BS8 was also supplemented with puromycin. Each mutant and wild-type virus were recovered (either from BSR or BS8 cells) by reverse genetics as previously described (27). Each recovered virus was plaque-purified, amplified, and titrated using plaque assay. For reverse genetics, synthetic single-stranded RNAs were prepared by runoff in vitro transcription from T7 PCR products using T7 RNA polymerase. Transcripts were prepared with antireverse cap analogue (ARCA) using mMESSAGE mMACHINE T7 Ultra kit (Ambion) as previously described (27). Reverse genetics was performed as previously described (27). Briefly, at day 1, 70 to 80% confluent BSR monolayers were transfected with pCAG-VP1, pCAG-VP3, pCAG-VP4, pCAG-VP6, and wild-type or mutated pCAG-NS2 (120 ng each) using EndoFectin (GeneCopoeia) according to the manufacturer's instructions and incubated at 35°C in 5% CO 2 overnight. At day 2, the cells were transfected with BTV1 exact copy RNA transcripts (S8 wild type or mutated) using EndoFectin (GeneCopoeia) overlaid with 1% agarose and incubated 3 days at 35°C in 5% CO 2 . Visible plaques were picked up and resuspended in 1% FBS containing medium, and/or cells were subsequently fixed with 10% formaldehyde and stained with crystal violet. Each recovered virus was plaque-purified, amplified, and harvested 3 days postinfection. Viruses were titrated using plaque assay.
Immunofluorescence and VIB analysis from microscopy data. BSR cells were grown on coverslips and infected at a multiplicity of infection (MOI) of 1 with NS2 wild-type or mutant-recovered viruses. Twenty-four hours postinfection, cells were fixed with 4% paraformaldehyde (Sigma-Aldrich) solution, permeabilized with 0.5% Triton X-100 (Sigma), blocked with 1% bovine serum albumin (BSA) (Sigma), and subsequently stained using a guinea pig anti-NS2 primary antibody (lab made) and an anti-guinea pig Alexa 488 coupled secondary antibody (Thermo Fisher Scientific). Nuclei were stained using Hoechst 33342 (Thermo Fisher Scientific). Images were acquired using an ϫ100 oil objective and a Zeiss Axiovert LSM 880 confocal microscope supplied with the Zen software. For each infection condition (wild-type virus versus A1, A2, or A3 mutant viruses) (Table 1), five fields were randomly selected and z-stacks (14 to 19 slices) were acquired (x, 1,912; y, 1,912; 12-bit). Each field contained in average of 17 infected cells, and the experiment was repeated three times independently. Maximum intensity projection of each z-stack was performed using Zen software, and obtained images were further processed using ImageJ software (v1.52a; https://imagej.nih.gov/ij/). Only particles (i.e., VIBs) with a size Ͼ0.5 m 2 were selected for particle analysis. The experiment was performed three times, and in total 1,600 VIBs were used for the