Trafficking of Bluetongue Virus Visualized by Recovery of Tetracysteine-Tagged Virion Particles

ABSTRACT

Bluetongue virus (BTV), a member of the Orbivirus genus in the Reoviridae family, is a double-capsid insect-borne virus enclosing a genome of 10 double-stranded RNA segments. Like those of other members of the family, BTV virions are nonenveloped particles containing two architecturally complex capsids. The two proteins of the outer capsid, VP2 and VP5, are involved in BTV entry and in the delivery of the transcriptionally active core to the cell cytoplasm. Although the importance of the endocytic pathway in BTV entry has been reported, detailed analyses of entry and the role of each protein in virus trafficking have not been possible due to the lack of availability of a tagged virus. Here, for the first time, we report on the successful manipulation of a segmented genome of a nonenveloped capsid virus by the introduction of tags that were subsequently fluorescently visualized in infected cells. The genetically engineered fluorescent BTV particles were observed to enter live cells immediately after virus adsorption. Further, we showed the separation of VP2 from VP5 during virus entry and confirmed that while VP2 is shed from virions in early endosomes, virus particles still consisting of VP5 were trafficked sequentially from early to late endosomes. Since BTV infects both mammalian and insect cells, the generation of tagged viruses will allow visualization of the trafficking of BTV farther downstream in different host cells. In addition, the tagging technology has potential for transferable application to other nonenveloped complex viruses.

IMPORTANCE Live-virus trafficking in host cells has been highly informative on the interactions between virus and host cells. Although the insertion of fluorescent markers into viral genomes has made it possible to study the trafficking of enveloped viruses, the physical constraints of architecturally complex capsid viruses have imposed practical limitations. In this study, we have successfully genetically engineered the segmented RNA genome of bluetongue virus (BTV), a complex nonenveloped virus belonging to the Reoviridae family. The resulting fluorescent virus particles could be visualized in virus entry studies of both live and fixed cells. This is the first time a structurally complex capsid virus has been successfully genetically manipulated to generate virus particles that could be visualized in infected cells.

INTRODUCTION

Bluetongue virus (BTV), the prototype of the genus Orbivirus within the family Reoviridae, is a nonenveloped, architecturally complex virus. BTV has 26 distinct serotypes and is endemic in most parts of the world, often resulting in high morbidity and mortality in ruminants. The BTV genome, comprising 10 double-stranded RNA (dsRNA) segments (S1 to S10), encodes 7 structural proteins (VP1 to VP7) and 4 nonstructural proteins (NS1 to NS4) (1, 2). In the virus particles, the structural proteins are organized in two capsids—an outer capsid of VP2 and VP5 and an inner capsid, or “core,” of VP7 and VP3 that encloses the viral transcription complex (VP1, VP4, and VP6)—in addition to the viral genome (1, 3). Three-dimensional structural studies of virions by cryoelectron microscopy has revealed that VP2 is arranged as trimers on the virion surface, protruding as spike-like structures from the surfaces of the virus particles (4, 5). VP2 alone is responsible for viral hemagglutination activity, viral serotype specificity, and the attachment of virions to the host cell (5–9). The second outer capsid protein, VP5, is also arranged in trimers but is less exposed than VP2 and is globular (5). Structurally, VP5 resembles the fusion proteins of enveloped viruses, and its external surface consists of amphipathic α-helical regions that have been suggested to play an active role in the penetration of endosomal membranes to release BTV cores into the cytoplasm (5, 10, 11).

Although studies have elucidated the roles of cellular factors in BTV entry, it is still not clear how the virus particles are trafficked into cells. In contrast to studies of other viruses, such as poliovirus and vaccinia virus (12–14), a major difficulty in previous studies on the entry process of BTV and its interactions with various cellular components has been the lack of real-time live-cell imaging. For this purpose, fluorescence visualization of single virus particles or viral proteins in fixed and live cells provides a valuable means of studying the interactions between viral and cellular proteins not only during virus entry but also during the trafficking, assembly, and release of the virus (15, 16). Since the BTV genome consists of 10 segmented dsRNA molecules, each approximately 0.8 to 3.9 kb, the capacity of each segment to accommodate foreign genes is limited (17). Hence, as an alternate strategy, we used the biarsenical-tetracysteine (TC) technology, which involves the use of small TC tags with a CCPGCC motif, which can be inserted into a protein without the risk of disrupting the overall structure of the targeted protein (18–20). The tagged proteins are specifically recognized by membrane-permeant biarsenical dyes that fluoresce when bound to the cysteine pairs in the TC motif. In addition, the differential labeling of the tagged proteins with two fluorescent biarsenical dyes, FlAsH (green) and ReAsH (red) (18, 19), makes this technology a powerful tool for the real-time visualization of nascent protein synthesis and trafficking in cells. To date, this technology has been used successfully in enveloped viruses (21–24) but not for any complex capsid virus such as BTV.

Although the functions of VP2 and VP5 have been documented (5–9, 11, 25–27), the detailed mechanisms of the roles of both proteins in virus trafficking have not been studied in depth. In addition, while a 7-Å structure of VP2 has been reported (5), the lack of a VP2 structure at an atomic level has made it difficult to design strategies for the insertion of foreign tags that will not disrupt the secondary structure of the protein. Thus, this study was designed to investigate whether biarsenical-TC tagging technology could be utilized for BTV as a means for investigating the trafficking of virus particles during virus entry. In this study, we have successfully utilized biochemical methods, sequence comparison data, and a BTV reverse genetics system (28) to insert tags into VP2 that do not disrupt the structure-function relationship of the virus particles. To our knowledge, this is the first report on the successful tagging of a structural protein for any nonenveloped virus. This resulted in clarification of the BTV entry pathway in mammalian cells and showed, for the first time, that the two outer capsid proteins are separated from each other during the early stages of virus entry. Our study not only allows for further investigation of VP2 protein trafficking in live cells infected with the mutant virus but also opens up possibilities for tagging other BTV proteins and other orbiviruses.

MATERIALS AND METHODS

Cell lines, viruses, and bacteria.

BSR and HeLa cells were maintained as described previously (29). Wild-type (WT) BTV serotype 1 (BTV1) (South African strain) and TC-tagged BTV1 stocks were propagated, and their titers were determined in BSR cells (28). Recombinant baculoviruses expressing BTV10 with S-tagged VP2 or with His-tagged VP2 fragments were propagated in Spodoptera frugiperda cells (Sf9) as described previously (29).

Antibodies and reagents.

All antibodies used against BTV proteins were generated in our laboratory. Purified recombinant VP2, VP5, and NS2 expressed from baculoviruses were used to generate monospecific polyclonal antibodies in rabbits as described previously (30). Antibodies against EEA1 and CD63 were obtained from Abcam. Fluorescently labeled secondary antibodies, Alexa Fluor 488, Alexa Fluor 546, and the biarsenical dye FlAsH were obtained from Invitrogen. Ammonium chloride, trypsin, and Dynasore were obtained from Sigma. Cells were treated with 30 μM ammonium chloride or 80 nM Dynasore for 30 min prior to infection (31). For Dynasore-treated cells, the drug was maintained in the medium during the 30 min of virus incubation.

Plasmids and site-directed mutagenesis.

Site-directed mutagenesis was performed to insert TC tags into the BTV1 S2 sequence (32). Briefly, two complementary primers were used to insert the nucleotide sequence (TGTTGTCCCGGGTGTTGT) encoding the TC tags into a pUCBTV1T7S2 (28) template. The following primers were used for site-directed mutagenesis: VP2TC94_F (5′-CGGTTGTTGAAAGTACGAGATGTTGTCCCGGGTGTTGTCACAAGAGTTTCCATACGAA-3′), VP2TC94_R (5′-TTCGTATGGAAACTCTTGTGACAACACCCGGGACAACATCTCGTACTTTCAACAACCG-3′), VP2TC352_F (5′-CGATACTTTTAATTGTTGTCCCGGGTGTTGTACACGAGTGTGGTGGTCGAAC-3′), VP2TC352_R (5′-ACAACACCCGGGACAACAATTAAAAGTATCGGAGGCTG-3′), VP2TC420_F (5′-TTGACTTT GTCGCGGAACCTTGTTGTCCCGGGTGTTGTGGGATTAAAATTGTTCATTG-3′), and VP2TC420_R (5′-CAATGAACAATTTTAATCCCACAACACCCGGGACAACAAGCGACAAAGTCAA-3′).

Recovery of tagged viruses.

The T7 BTV capped (BTV1 S1, BTV1 S3 to S10) and uncapped (BTV1 S2, BTV1 S2 with a TC tag) transcripts were generated as described previously (28). The mutant BTV particles were recovered by following a method described previously (33). Genomic dsRNA from cells infected with control or mutant BTV was analyzed as described previously (28).

Virus growth kinetics.

Monolayers of BSR cells were synchronously infected with either wild-type or mutant BTV1 at a multiplicity of infection (MOI) of 1, and plaque assays were carried out as described previously (29) at 0, 24, and 48 h postinfection (p.i.). Western blotting (WB) was undertaken to monitor the expression of BTV proteins VP2, VP5, and NS2, while the cellular protein tubulin was used as the loading control. Each blotting experiment was repeated three times, and the amount of protein expressed was quantitated by ImageJ. The means and standard errors of the virus titers, as well as the intensities of the viral and cellular protein bands, were calculated (SigmaPlot 2000; Systat Software, Inc.), and the P values were determined by Excel (Microsoft).

Fluorescence and confocal microscopic analysis of TC-tagged proteins and viruses.

Live- and fixed-cell analyses of tagged-virus trafficking were undertaken by synchronously infecting HeLa cells at MOIs of 50 and 10, respectively. BSR cells infected with BTV1-VP2TC1 or BTV1-VP2TC2 (BTV1 with a TC tag insertion between amino acid positions 94 and 95 or 352 and 353) were processed for biarsenical labeling with FlAsH at different times p.i. according to the manufacturer's recommendations. Both live and fixed HeLa cells were imaged by confocal microscopy with a Zeiss LSM 510 microscope. Cells infected for live-cell imaging were washed with Opti-MEM I reduced serum medium (Invitrogen) and were stained with FlAsH solution (2 μM) for 30 min at 4°C. After the cells were washed with 2,3-dimercapto-1-propanol (BAL), live images were captured every 16.7 s by confocal microscopy with a 488-nm laser and appropriate fluorescein filters on a prewarmed stage that was maintained at 37°C. Subsequently, the images were compiled into a movie at 8 images per second in ImageJ, and a cartoon was created using Adobe Photoshop Elements software, version 8.0.

Fixed-cell analysis was carried out as described previously (29). Images were obtained using LSM 510 Image Browser software and were processed using Adobe Photoshop Elements software, version 8.0. (Microsoft). Each set of fixed-cell experiments was repeated at least three times to generate either localization or colocalization data that could be quantified. Colocalization was judged by the appearance of yellow spots formed by the merging of red and green signals generated by the florescent tags attached to the secondary antibodies. The means and standard errors of the percentages of localization or colocalization were calculated (SigmaPlot 2000; Systat Software, Inc.), and the P values were determined by Excel (Microsoft).

Recombinant expression of the amino-terminal fragment of VP2.

For the production of a recombinant baculovirus expressing BTV10-VP2₉₄, 94 amino acids from the amino-terminal end of BTV10 VP2 were inserted into the baculovirus expression vector pAcYM1, and a His tag was introduced upstream of the start codon of VP2 (30). In addition, for bacterial expression, the first 94 N-terminal residues of BTV1 were inserted into pRSETA (Invitrogen). Each construct was verified by sequencing. Recombinant baculoviruses expressing His-tagged BTV10-VP2₉₄ were produced, plaque purified, and propagated using standard baculovirus recombination procedures (34). His-tagged BTV1-VP2₉₄ was expressed in Escherichia coli BL21 as described previously (35). The soluble and insoluble fractions from bacterial and insect cells were prepared and analyzed by Western blotting (36).

Digestion of purified virus particles and VP2 with trypsin and mass spectrophotometry analysis.

Purified virus particles and VP2 were incubated with increasing concentrations of trypsin (1 ng, 10 ng, 100 ng) for 30 min at 37°C and were resolved by SDS-12% PAGE. Bands representing the 100-kDa VP2 protein fragment and the complete 110-kDa protein were excised from the gel and subjected to in-gel digestion with trypsin. Subsequently, the products were analyzed with a liquid chromatography-tandem mass spectrometer (QToF Micro; Waters Corp., Milford, MA) as described previously (37).

RESULTS

Identification of putative exposed regions in VP2.

VP2, the host attachment protein, is responsible for virus entry. Hence, strategies were adopted to minimize the potential impediment to virus infectivity that may be caused by the fusion of the TC tag to VP2. Since, in the absence of an atomic structure, the locations of flexible loop-linker regions that could be utilized for the insertion of TC tag were still not clear, two strategies were adopted to identify such exposed loop regions in VP2. First, analysis of VP2 amino acid sequences (ExPASy) of two different BTV serotypes (BTV1 and BTV10) revealed the presence of putative trypsin cleavage sites that were common to the two VP2 sequences (data not shown). Since trypsin could access only sites that were exposed on the surface of the protein and not those present internally, it was hypothesized that these putative cleavage sites could be the exposed loop-linker regions in VP2 and thus could potentially be utilized for the insertion of TC tags.

Consequently, purified virus particles (BTV1) and VP2 protein (of BTV10) were digested with increasing concentrations of trypsin for 30 min at 37°C to identify the presence of potential enzyme cleavage sites in VP2 (Fig. 1). Although digestion of purified virus particles (Fig. 1A, left) and VP2 protein (Fig. 1A, right) with 100 ng of trypsin showed the presence of protein bands with smaller sizes on an SDS-PAGE gel, the patterns of the digested products were different for virus particles and VP2 protein (Fig. 1A, compare left and right panels). Briefly, digestion of purified virus particles yielded a very faint fragment of 110 kDa and two smaller fragments measuring 40 kDa and 10 kDa (Fig. 1A, left). In contrast, digestion of purified VP2 protein under the same conditions resulted in two fragments of 100 kDa and 10 kDa (Fig. 1A, right). The difference in the digestion pattern between purified virus particles and VP2 alone can be attributed to the possibility that VP2 in virions adopts conformations that are more susceptible to proteolysis. Furthermore, the smaller digested products of VP2 in both purified virus particles (Fig. 1A, left) and VP2 protein (Fig. 1A, right) were detectable only with higher concentrations of trypsin (100 ng), not with lower concentrations (10 or 1 ng). Control virus particles incubated at 37°C for 30 min in the absence of trypsin showed no breakdown products. Since the presence of the 10-kDa protein band was noted in both digested virus particles and purified VP2 protein, the larger fragment (100-kDa product) from digested VP2 was excised from the SDS-PAGE gel (Fig. 1A, right, asterisk). After purification from the SDS-PAGE gel, smaller peptide fragments were generated by trypsin digestion of the 100-kDa purified product, and the peptides were further analyzed by mass spectrometry. The undigested 110-kDa purified protein, treated in the same way, served as a control. Evaluation of the peptide fragments generated from both the 100-kDa digested product and 110-kDa undigested VP2 by a Swiss-Prot-based Mascot search confirmed that these peptides were derived from BTV10 VP2 (Table 1). Peptides generated from the 100-kDa fragment mapped perfectly to the BTV10 VP2 sequence after amino acid position number 94, suggesting that the 100-kDa protein fragment lacked the first 94 amino acids (Fig. 2A), in accord with the release of a 10-kDa fragment as a result of trypsin digestion of purified VP2. In contrast, peptides generated from undigested full-length VP2 (110 kDa) mapped to the entire length of the protein (Fig. 2B).

FIG 1.

Identification of putative exposed regions in VP2 by biochemical and sequence analysis. (A) Digestion of purified intact BTV1 (left) and recombinant BTV10 VP2 protein (right) with increasing concentrations of trypsin (1 ng to 100 ng). The control consists of undigested BTV1 and BTV10 VP2. The digested products were run on an SDS-PAGE gel and were stained with Coomassie blue stain. Concentrations of trypsin, molecular masses, and virus proteins are given at the top, left and right, respectively. The arrows indicate digested products, and the asterisk indicates products that were sent for mass spectrometry analysis. (B) Expression of 10-kDa BTV1 (top) and BTV10 (bottom) VP2 as soluble (sol) and insoluble (ins) fractions in E. coli and Sf9 cells, respectively. The lysates were fractionated, run on an SDS-PAGE gel, and analyzed by Western blotting. Molecular masses and virus proteins are given on the left and right, respectively.

TABLE 1.

Peptides generated in the mass spectrophotometry analysis of VP2 segments

VP2 fragment	Peptide
aa 95-956 (100-kDa protein)	SEQIVNK
	IKDEWK
	MDIQPLK
	VHSMGITR
	SFFNFIR
	DEMALSTR
	EAFPNFLK
	TNPYPCLR
	GVEFMTDTK
	FQFKLDEK
	MLDVQEEPK
	MIQGGWDQER
	IGYRGQPYER
	CNHTNLDLLR
	VALIILSNYER
	WSLRPEYGQR
	LIPVEVGAYANR
	SEQIVNKYYYSR
	WAIDDKMDIQPLKV
	MPGHVFGNDELMTK
	TVCPHSGGTFYTFR
	IPDEIRTEIAELNR
	EHETYMHPAVNDVFR
	SIVLIIVGDDKLEPQIR
	TSAESLEYALGPYYDLR
	EHETYMHPAVNDVFRR
	MLDVQEEPKDEMALSTR
	LNLFDTNLAVGDEIIHWR
	SANADTIYYDYYPLENGAK
	QSAVFEHMAQQDDFSTLTDYTK
aa 1-956 (110-kDa protein)	GEIATWK
	SEQIVNK
	LLDITLR
	NDGVVVPR
	ILSAIGRK
	VHSMGITR
	YYYSRR
	GKIPDEIR
	RHVLEMK
	SFFNFIR
	VGGSATDDGR
	DEMALSTR
	ADDPWSNR
	EAFPNFLK
	ERPLEDNK
	YKEMFDR
	TNPYPCLR
	GVEFMTDTK
	FQFKLDEK
	MLDVQEEPK
	MIQGGWDQER
	WINSPMFNAR
	CNHTNLDLLR
	IPESDMIDVPR
	WSLRPEYGQR
	LIPVEVGAYANR
	VDDEGGKHNLIK
	LQLFGDTLSLGQR
	MIQGGWDQERFK
	SEQIVNKYYYSR
	WAIDDKMDIQPLK
	MPGHVFGNDELMTK
	TVCPHSGGTFYTFR
	ERPLEDNKYVFAR
	FIRTDQEHVNIFK
	ETTHDDGYICVSQK
	IPDEIRTEIAELNR
	QWSIPMILFDQVIR
	EHETYMHPAVNDVFRR
	SLTTTEMFHILQGAAYALK
	LNLFDTNLAVGDEIIHWR
	VTLDNHCSVNHQLFNCIVK
	SANADTIYYDYYPLENGAKR
	TYELVAHSERENMSESYQVGTQR

FIG 2.

Peptide mapping. (A) Peptide fragments generated by digestion of the 100-kDa fragment of BTV10 VP2 were mapped against the amino acid sequence of full-length BTV10 VP2. (B) For the control, peptides generated by digestion of full-length BTV10 VP2 were mapped against BTV10 VP2. The peptides are shaded.

In order to analyze whether the 10-kDa VP2 fragment was an autonomous domain that could be expressed as a stable soluble product, the amino-terminal fragments of two different BTV serotypes, BTV1 (Fig. 1B, top) and BTV10 (Fig. 1B, bottom) were expressed as His-tagged fusion proteins (BTV1-VP2₉₄ and BTV10-VP2₉₄) in either a bacterial (E. coli) (Fig. 1B, top) or a eukaryotic (baculovirus) (Fig. 1B, bottom) expression system. The E. coli and Sf9 cells expressing the BTV1-VP2₉₄ and BTV10-VP2₉₄ fusion proteins, respectively, were lysed, fractionated into soluble and insoluble fractions, and run on SDS-PAGE gels. Western blot analysis of the His-tagged products revealed the presence of the BTV1-VP2₉₄ and BTV10-VP2₉₄ fusion proteins in both soluble and insoluble fractions in both expression systems (Fig. 1B). Thus, these data suggest the presence of an exposed loop after the first 94 residues of VP2 that separates autonomous folded domains as soluble fractions. Hence, the insertion of a TC in this region of VP2 might allow the construction of a VP2 variant in which overall folding and biological activity are preserved.

In a second, alternate approach to identifying the presence of exposed loop regions in VP2, the similarity of BTV VP2 to the VP2 protein of African horse sickness virus (AHSV), a closely related orbivirus (38), was assessed. A database search of the available AHSV (4 of the 9 serotypes) and BTV (24 of the 26 serotypes) VP2 peptide sequences (data not shown) revealed that VP2 proteins from AHSV serotypes are generally longer (ranging from 1,051 to 1,060 residues) than those from BTV serotypes (ranging from 950 to 962 residues). Further, a careful sequence alignment of BTV and AHSV VP2 proteins revealed the presence in all BTV serotypes of two in-frame deletions, between residues 352 and 371 and between residues 420 and 452, that were absent in AHSV (Fig. 3A). This led us to hypothesize that the two deletions might be located at exposed loop-linker regions of the protein that could also be explored for the insertion of TC tags in VP2. Hence, based on these results, it was predicted that three regions in VP2 could potentially be exploited for the introduction of TC tags (Fig. 3A).

FIG 3.

Schematic demonstration of insertion of tags into VP2. (A) Schematic representation of deletions in the amino acid sequence of BTV1 VP2 (bottom) that are absent in AHSV4 VP2 (top). The numbers designate amino acid positions in the VP2 sequence. The arrow indicates the position of amino acid number 94. (B) Insertion of TC tags (boxed) into BTV1 VP2 sequences.

Generation and characterization of recovered mutant viruses containing TC-tagged VP2.

A TC tag was inserted into the coding region of BTV1 VP2 between the amino acids at positions 94 and 95, 352 and 353, or 420 and 421 (Fig. 3B). As described previously (33), uncapped S2 T7 transcripts (BTV1-S2) were generated for all constructs (BTV1-S2₉₄, BTV1-S2₃₅₂, and BTV1-S2₄₂₀) to recover TC-tagged mutant viruses. The reverse genetics system used transfection of BSR cells with 9 T7-derived RNA transcripts (S1 and S3 to S10) together with either wild-type (WT) BTV1-S2 or each of the tagged S2 transcripts. Like the wild-type virus, all mutant viruses were recovered successfully.

Plaque assays were undertaken to investigate the plaque morphologies of the newly generated tagged viruses BTV1-VP2TC1 (BTV1-S2₉₄), BTV1-VP2TC2 (BTV1-S2₃₅₂), and BTV1-VP2TC3 (BTV1-S2₄₂₀). Although clear plaques were visible 3 days posttransfection for all three BTV1 VP2-tagged viruses and also for the control WT BTV1 recovered at the same time, the plaques formed by BTV1-VP2TC3 were smaller than those of both wild-type BTV1 and mutants BTV1-VP2TC1 and BTV1-VP2TC2 (Fig. 4A). This suggested that, compared to BTV1-VP2TC1 and BTV1-VP2TC2, the TC tagging of VP2 in BTV1-VP2TC3 might have generated a mildly attenuated virus. To confirm the replication of the recovered viruses, genomic dsRNAs from cells infected with independent plaques were extracted, purified, and analyzed on a nondenaturing polyacrylamide gel (Fig. 4B). The results revealed that the 10 dsRNA segments synthesized by the three tagged viruses had dsRNA profiles indistinguishable from that of WT BTV1. Subsequently, using forward and reverse primers flanking full-length S2, cDNAs for both WT and tagged viruses were generated from viral dsRNA by reverse transcription-PCR (RT-PCR) and were sequenced. The data confirmed the presence of the TC tag sequence at the relevant position in the S2 segment of each tagged virus (Fig. 4C).

FIG 4.

Recovery of tagged BTV particles. (A) Plaque morphologies of WT BTV1 and TC-tagged viruses. (B) Genomic dsRNA from BSR cells infected with WT BTV1 (lane 1) or tagged BTV1-VP2TC1 (lane 2), BTV1-VP2TC2 (lane 3), or BTV1-VP2TC3 (lane 4) was purified and analyzed on a nondenaturing polyacrylamide gel. (C) Sequence electropherograms of segment 2 RT-PCR products from TC-tagged viruses. The positions of TC tags are indicated above each panel.

The difference in plaque morphology between BTV1-VP2TC3, on the one hand, and BTV1-VP2TC1, BTV1-VP2TC2, and WT BTV1, on the other, prompted us to investigate the infectivities and growth characteristics of the three newly generated viruses. For this purpose, BSR cells were infected at an MOI of 1 with BTV1-VP2TC1, BTV1-VP2TC2, or BTV1-VP2TC3 for different times. The growth and viral protein expression kinetics of each recovered tagged virus were then monitored by total plaque assay titers (Fig. 5A) and WB (Fig. 5B), respectively. Control cells were infected with WT BTV1 and were treated similarly to the tagged viruses. Monitoring of the total plaque assay titers at 0, 24, and 48 h p.i. demonstrated that although all three tagged viruses showed overall growth profiles similar to that of the WT (Fig. 5A), the total titer of BTV1-VP2TC3 was significantly lower (P < 0.05) than that of the WT (Fig. 5A) at each time point p.i. In contrast, the difference between the total titer of BTV1-VP2TC1 or BTV1-VP2TC2 and that of WT BTV1 was statistically insignificant (P > 0.05) at all times analyzed (Fig. 5A). This indicates that the insertion of a TC tag in VP2 after amino acid position 94 or 352 did not significantly impede the function of VP2, whereas TC tagging after amino acid position 420 caused some loss of VP2 function such that the growth of the tagged virus was impaired. WB was undertaken to analyze the production of two viral structural proteins (VP2 and VP5) and one nonstructural protein (NS2) encoded by BTV1 at 0, 24, and 48 h p.i. in cells infected with the tagged and WT viruses (Fig. 5B). BSR cells infected with BTV1-VP2TC1, BTV1-VP2TC2, or WT BTV1 showed similar expression profiles for VP2, VP5, and NS2 at all times p.i. (Fig. 5B). In all blots, the levels of tubulin, used as a loading control, were equivalent. Further, when the production of VP5, VP2, and NS2 was quantified and normalized to tubulin production, the levels of virus proteins produced by BTV1-VP2TC3 at 24 h (Fig. 5C) and 48 h (Fig. 5D) p.i. were statistically significantly different (P < 0.05) from those produced by the WT virus. In contrast, there was no statistically significant difference (P > 0.05) in the expression of VP5, VP2, and NS2 between cells infected with either BTV1-VP2TC1 or BTV1-VP12TC2 and cells infected with WT BTV1. Since BTV1-VP2TC1 and BTV1-VP2TC2 have growth curves and protein production levels similar to those of WT BTV1, these two tagged viruses were utilized for virus entry studies.

FIG 5.

Characterization of recovered mutant viruses containing TC-tagged VP2. (A) The total titers of mutant viruses or WT virus in BSR cells, expressed as PFU/ml, were determined at different times p.i. and were plotted on a logarithmic scale. Asterisks indicate that the titers of BTV1-VP2TC3 at 24 and 48 h p.i. are statistically significantly lower than those of WT BTV1 (P < 0.05). (B) Western blot analysis of VP2, VP5, NS2, and tubulin in infected-cell lysates at 0, 24, and 48 h p.i. The cells were infected either with WT BTV1 or with the tagged virus BTV1-VP2TC1, BTV1-VP2TC2, or BTV1-VP2TC3. Cell lysates from mock-infected cells (−) were included in each panel. Western blotting was performed using antibodies against tubulin and against BTV VP2, VP5, and NS2. (C and D) Quantitation of VP5, VP2, and NS2 produced by cells infected with WT BTV1 (filled bars), BTV1-VP2TC1 (dark shaded bars), BTV1-VP2TC2 (light shaded bars), and BTV1-VP2TC3 (open bars) (bars 1, 2, 3, and 4, respectively) at 24 h (C) and 48 h (D) p.i. The quantities of viral proteins produced by each virus were normalized to the quantity of cellular tubulin produced. Error bars indicate standard errors for three data sets. Asterisks indicate that the amounts of VP5, VP2, and NS2 produced by cells infected with BTV1-VP2TC3 at 24 and 48 h p.i. are statistically significantly different from those with WT BTV1 (P < 0.05).

Internalization of tagged viruses in cells.

Live-cell imaging was undertaken to assess whether the fluorescent labeling of the tagged viruses with the biarsenical dye FlAsH generated sufficient signal for investigation of the trafficking of tagged viruses during entry into host cells (Fig. 6). For this purpose, HeLa cells infected with BTV1-VP2TC2 were stained with the fluorescent biarsenical dye FlAsH, which binds specifically to tetracysteine tags, and live cells were imaged by confocal microscopy immediately after infection (Fig. 6A). Control uninfected cells stained with FlAsH showed no fluorescent signal for VP2 (Fig. 6B). When the movement of the fluorescently labeled TC-tagged virus BTV1-VP2TC2 was monitored over time (Fig. 6C), the movement of the labeled virus particle (indicated by white arrows) from its initial position at time zero p.i. (indicated by yellow arrows) confirmed that the movement of tagged BTV1-VP2TC2 can be tracked over time in an infected cell (Fig. 6C; see also Movie S1 in the supplemental material).

FIG 6.

Live-cell imaging of tagged BTV. (A) HeLa cells infected with BTV1-VP2TC2 were labeled with FlAsH immediately p.i. and were analyzed on the microscope stage in a closed chamber (37°C) for 20 min. (B) Control uninfected cells were labeled as for panel A and were analyzed on the microscope stage in a closed chamber (37°C). (C) Live tracking of tagged BTV. A single cell from the upper panel is magnified in the lower panels. Yellow and white arrows indicate the initial and final positions of the tagged-BTV particle over time postinfection. The initial position of the virus particle represents its position at time zero. Bars, 10 μm. The bottom right panel is a cartoon of the single cell showing the position of the tagged-virus particle (red) in the cellular cytoplasm.

It has been shown previously that BTV enters cells through the clathrin-mediated endocytic pathway (11). Cellular dynamin is known to mediate the pinching of the clathrin-coated pits to form the coated vesicles (39, 40), and it has been reported that inhibition of dynamin with Dynasore (31) impedes this process. Hence, further experiments were undertaken to determine whether the entry of the tagged viruses is also influenced by dynamin (Fig. 7A and B). For this purpose, BTV1-VP2TC1 or BTV1-VP2TC2 was adsorbed onto both mock-treated and Dynasore-treated HeLa cells for 30 min at 4°C. The cells were washed and were either processed at time zero p.i. or incubated at 37°C for 30 min. At both times, the cells were fixed with 4% paraformaldehyde, stained with FlAsH, and visualized by confocal imaging. While analysis of control untreated cells at time zero p.i. showed the presence of a majority of the tagged BTV1-VP2TC1 (94% ± 3.1%) or BTV1-VP2TC2 (95% ± 2.9%) on the plasma membranes (Fig. 7A, top) of infected cells, at the later time point (30 min) p.i., both BTV1-VP2TC1 (84% ± 1.9%) and BTV1-VP2TC2 (80% ± 3.2%) were observed within the cellular cytoplasm (Fig. 7A, center). In contrast, Dynasore-treated cells analyzed at 0 min (93% ± 3.3% for BTV1-VP2TC1 and 94% ± 3.1% for BTV1-VP2TC2) and 30 min (84% ± 2.3% for BTV1-VP2TC1 and 84% ± 0.6% for BTV1-VP2TC2) p.i. demonstrated the presence of a majority of tagged-virus particles on the plasma membrane (Fig. 7A, bottom).

FIG 7.

Internalization of tagged viruses in cells. (A) Effect of dynamin inhibition on entry. HeLa cells either mock treated (−) or treated with 80 μM Dynasore (+) were infected with BTV1-VP2TC1 (left) or BTV1-VP2TC2 (right) and were monitored at 0 min (top) and 30 min (center and bottom) p.i. VP2 in the tagged viruses was labeled with FlAsH dye and was visualized by confocal microscopy. Arrows indicate fluorescently tagged VP2. Bars, 10 μm. The white lines depict the margins of the cells and were drawn by superimposing the image of the cell taken under transmission light. (B) Effect of dynamin inhibition on NS2 expression. HeLa cells treated with Dynasore (+) or mock treated (−) were infected with BTV1-VP2TC1 (left) or BTV1-VP2TC2 (right) and were monitored for NS2 (green) expression 16 h p.i. Nuclei (blue) were stained with Hoechst 33258. Bars, 5 μm. (C) Localization of VP2 and VP5 in HeLa cells infected with BTV1-VP2TC1 (left) or BTV1-VP2TC2 (right). Tagged VP2 was visualized by FlAsH staining, and a polyclonal antibody was used to label VP5. The times p.i. at which the cells were processed for confocal microcopy are shown. Colocalization of VP2 (green) and VP5 (red) was visualized as yellow. Bars, 10 μm. (D) Effect of ammonium chloride on BTV replication. HeLa cells pretreated with 30 mM ammonium chloride (+) were infected with either BTV1-VP2TC1, BTV1-VP2TC2, or WT BTV1 and were assessed by plaque assays for total virus titers at 24 h p.i. Cells treated with the diluent (−) were used as controls. Significant differences in the virus titers are indicated by asterisks.

To further investigate whether dynamin is required for BTV entry, HeLa cells infected with BTV1-VP2TC1 or BTV1-VP2TC2 were monitored for the expression of one of the BTV nonstructural proteins, NS2, at 16 h p.i. in the presence (Fig. 7B, top) or absence (Fig. 7B, bottom) of Dynasore. Although, as in Fig. 7A, the presence of Dynasore did not completely abolish the expression of NS2 (Fig. 7B, top), NS2 levels in Dynasore-treated cells were clearly lower than those in untreated control cells (Fig. 7B, bottom). While the Dynasore-dependent decrease in NS2 expression was greater for cells infected with BTV1-VP2TC1 (94% ± 1%) than for cells infected with BTV1-VP2TC2 (90.5% ± 2.8%), a similar decrease in NS2 expression in Dynasore-treated cells infected with WT BTV1 (95.7% ± 0.3%) showed that tagged viruses and WT BTV1 behaved similarly. Further, statistical analysis also confirmed that the difference in NS2 expression between Dynasore-treated cells infected with either BTV1-VP2TC1 or BTV1-VP2TC2 and Dynasore-treated cells infected with WT BTV1 was statistically insignificant (P > 0.05). These data are consistent with an earlier study that has also noted a role for dynamin in BTV entry (41) and confirm that dynamin might play a functionally important role during tagged-BTV entry.

The canonical view of BTV entry is that the outer capsid (VP2 and VP5) is shed in an early endosome to release a fusion-competent core, which then fuses with the endocytic membrane and formally enters the cytoplasm (5, 8, 11, 26, 27). However, whether VP2 and VP5, which are juxtaposed on the surface of the virus particle, remain together during this period is unknown. To investigate this using tagged virions, cells were infected with BTV1-VP2TC1 or BTV1-VP2TC2, fixed with 4% paraformaldehyde, and immunostained, and the localization of VP2 and VP5 (Fig. 7C) was assessed at 2, 5, and 15 min p.i. FlAsH was used to label VP2 in cells infected with the tagged virus BTV1-VP2TC1- or BTV1-VP2TC2. The second outer capsid protein, VP5, was also immunolabeled with an anti-VP5 polyclonal antibody. Although colocalization (yellow) of VP2 and VP5 by confocal microscopy was observed at 2 min (95.8% ± 4.2% for BTV1-VP2TC1 and 96.7% ± 3.3% for BTV1-VP2TC2) and at 5 min (90.5% ± 4.8% for BTV1-VP2TC1 and 87.9% ± 2.4% for BTV1-VP2TC2) p.i. (Fig. 7C, top and center), from 15 min p.i. onward VP2 and VP5 were seen as separate entities (Fig. 7C, bottom) in the majority of infected cells (72.9% ± 1.5% for BTV1-VP2TC1 and 72.7% ± 2.03% for BTV1-VP2TC2). The separation of VP2 and VP5 after virus internalization suggests that the two proteins disengage early in virus entry.

Previous studies have shown that BTV entry is pH dependent (25) and that VP5 has pH-dependent fusogenic activity (27). Further, the VP5 structure has certain features analogous to those of fusion proteins of some enveloped virus proteins, in particular influenza virus hemagglutinin (HA) (5). To confirm biochemically that the TC-tagged virus particles behave similarly to WT virus particles, the effect of an acidic pH on BTV entry was explored by treating the cells with ammonium chloride (Fig. 7D), a lysosomotropic weak base that immediately raises the pH of intracellular acidic vesicles. Cell viability assays, used to assess possible cytotoxicity induced by ammonium chloride, showed no toxicity (data not shown). HeLa cells were exposed to 30 mM ammonium chloride prior to BTV infection, and BTV replication was examined at 24 h p.i. by determining the virus titers. A decrease of almost two and a half log units in virus titers was observed by plaque assays in cells pretreated with ammonium chloride and infected with BTV1-VP2TC1, BTV1-VP2TC2, or WT BTV1 (Fig. 7D), confirming that the entry of tagged viruses was similar to WT virus entry. Western blotting of infected cells also indicated that ammonium chloride had a strong inhibitory effect on BTV replication (data not shown). The accumulating results suggest that TC-tagged viruses behave in a similar manner to that of WT BTV1.

Segregation of VP2 and VP5 in endocytic pathways.

Since VP2 appeared to segregate from VP5 quite early during virus infection, further investigations were undertaken to identify the cellular compartments that might be involved in BTV entry. Since BTV particles enter cells by clathrin-mediated endocytosis (11), both early (EEA1)- and late (CD63)-endosome markers were used to ascertain the distribution of tagged VP2 during entry into mammalian cells (Fig. 8). HeLa cells infected at an MOI of 10 with either BTV1-VP2TC1 or BTV1-VP2TC2 were incubated at 4°C for 1 h to synchronize virus infection, followed by incubation at 37°C for 5, 15, or 30 min and processing for FlAsH labeling of the tagged VP2. Confocal analysis of tagged VP2 in cells infected with BTV1-VP2TC1 (Fig. 8A and B, left) or BTV1-VP2TC2 (Fig. 8A and B, right) demonstrated that for both the tagged viruses, VP2 colocalized with EEA1 (Fig. 8A) but not with CD63 (Fig. 8B). Further, quantification of colocalization for VP2 and EEA1 showed that both BTV1-VP2TC1 and BTV1-VP2TC2 were localized with EEA1 by 5 min p.i. (89.2% ± 0.8% for BTV1-VP2TC1 and 88.8% ± 2.3% for BTV1-VP2TC2). On further incubation, colocalization of VP2 and EEA1 was maintained at both 15 min (92.8% ± 3.73% for BTV1-VP2TC1 and 89.2% ± 0.4% for BTV1-VP2TC2) and 30 min (93.9% ± 3.1% for BTV1-VP2TC1 and 94.4% ± 2.8% for BTV1-VP2TC2) p.i. Since VP2 and VP5 colocalization studies revealed that VP2 and VP5 segregate from each other by 15 min p.i., the retention of VP2 in the EEA1-labeled early-endosomal compartments indicated that VP2 had been shed and that virus particles containing an outer layer of VP5 only might have trafficked to the CD63-labeled late-endosomal compartments.

FIG 8.

Immunofluorescence analysis of cells infected with the tagged viruses reveals the segregation of VP2 and VP5 in endocytic pathways. Shown is the localization of BTV VP2 with EEA1 (A) and CD63 (B) in HeLa cells infected with BTV1-VP2TC1 (left) or BTV1-VP2TC2 (right). HeLa cells were infected as described in Materials and Methods, and at 5, 15, and 30 min p.i., they were fixed, permeabilized, and processed for confocal microscopy. VP2 in tagged viruses was visualized by FlASH staining. Colocalization of VP2 (green) and EEA1/CD63 (red) was visualized as yellow and is highlighted by arrows. Bars, 10 μm. The white lines in panel A depict the margins of the cells and were drawn by superimposing the image of the cell taken under transmission light.

Trafficking of VP5 during virus entry.

The role of VP5 in BTV entry has been elucidated by two independent studies that disagree on the precise site of fusion. While one study, conducted with BTV10-infected HeLa cells, established that the virus entered through receptor-mediated endocytosis and early endosomes (11), a second study demonstrated that BTV1 particles entered BHK cells through clathrin-independent macropinocytosis and that late endosomes played a crucial role in this process (41). Since our data showed that VP2 is retained on early endosomes, the role of VP5 in BTV trafficking was investigated further. Since the early-endosome-based trafficking of the two tagged viruses BTV1-VP2TC1 and BTV1-VP2TC2 was similar to that of WT BTV1, only VP2TC1 was used to analyze the relationship of VP5 with early- and late-endosomal compartments in BTV entry (Fig. 9). HeLa cells infected with BTV1-VP2TC1 at an MOI of 10 were first incubated at 4°C for 1 h to synchronize virus infection, then incubated at 37°C for 5, 15, or 30 min, fixed, and further processed for immunolabeling of both VP5 and EEA1 or CD63 compartments by their respective antibodies. Confocal analysis of VP5 in cells infected with BTV1-VP2TC1 (Fig. 9A) demonstrated that colocalization of VP5 and EEA1 labeled compartments was observed only up to 15 min p.i. (Fig. 9A, center). In contrast, when cells fluorescently labeled for VP5 and CD63 were analyzed (Fig. 9B), no colocalization was observed for the tagged virus at 5 min p.i. (Fig. 9B, left). At 15 min p.i., some colocalization between VP5 and CD63 was apparent (Fig. 9B, center); by 30 min p.i., VP5 was entirely colocalized with CD63 (Fig. 9B, right). Further, quantitation of the colocalization of VP5 and EEA1 established that with an increase in the time p.i., there was a decrease in the colocalization of VP5 and EEA1 (Fig. 9C). In contrast, when cells fluorescently labeled for VP5 and CD63 were analyzed (Fig. 9B), almost negligible colocalization (about 2%) was observed at 5 min p.i. for the tagged virus (Fig. 9B, left). Subsequently, sequential increases in the colocalization of VP5 and CD63 were observed at 15 min (67%) and 30 min (87%) p.i. (Fig. 9C). These results therefore demonstrate that following infection of cells by BTV, VP2 is lost in an early-endosomal compartment while BTV particles containing an outer layer of VP5 traffic from the early to the late endosomes.

FIG 9.

Trafficking of VP5 during virus entry. (A and B) Colocalization of VP5 (red) with EEA1 (A) and CD63 (B). HeLa cells infected with BTV1-VP2TC1 were fixed at the times p.i. shown, permeabilized, and labeled with a polyclonal antibody for VP5. Colocalization of VP5 and CD63 was visualized as yellow. Bars, 10 μm. (C) Average colocalization of VP5 with EEA1 and CD63, plotted as percentages against times p.i. Error bars indicate standard errors for three data sets.

DISCUSSION

Viruses have developed different strategies to hijack intrinsic host cellular pathways for entry and to deliver their genomes to specific cellular locations for replication. With enveloped viruses, the fusion of viral and cellular membranes prompts the release of the capsid or genome into the cytoplasm. For nonenveloped viruses, the ability of the outer capsid proteins to disrupt cellular membranes or to form pores in them results in the delivery of the inner capsid or viral genome to the cytosol.

Fluorescent labeling of virus particles and cellular structures has made it possible to monitor live-virus trafficking in infected cells (15, 16, 42). Whereas enveloped viruses have been successfully labeled with genetically encoded fluorescent proteins (21–24, 43), the structural constraints of nonenveloped virus capsid structures have made the insertion of tags difficult. An alternate strategy involving nonspecific labeling of nonenveloped virus capsid structures with fluorescent dyes has helped to elucidate the entry pathway of viruses with naked capsids. Thus, live-cell imaging of cells infected with fluorescently labeled poliovirus has revealed that after internalization through a clathrin-, caveolin-, and flotillin-independent, but actin- and tyrosine kinase-dependent, pathway, the virus releases its RNA rapidly from vesicles located very close to the plasma membrane and does not require endocytic acidification or microtubule-dependent transport (14). In this study, however, for the first time, we have successfully identified exposed loop regions in the BTV outer capsid protein VP2 and have generated replication-competent tagged viruses via reverse genetics (28, 33). Since, to date, investigation of the functional role of VP2 in BTV entry has been limited to studies based mainly on recombinant protein expression (9), the tagging of VP2 and its successful incorporation into BTV particles have provided valuable insights into its role during virus entry and trafficking.

Among the different pathways reported for the internalization of virus particles, receptor-mediated endocytic pathways regulated by both clathrin-coated pits and cellular proteins, such as the AP-2 complex and the GTPase dynamin, have been held accountable for the majority of viral entrance mechanisms in cells. We have reported previously that BTV10 enters and infects HeLa cells via clathrin-mediated endocytosis and that the early but not the late endosomes play an essential role in the early stages of BTV entry (11). However, Gold et al. (41) found that clathrin-mediated entry is not the major entry route used by BTV1 to enter BHK-21 cells and that the virus particles are delivered directly to late endosomes through a pathway that shares certain common factors with macropinocytosis. Since it is well established that Dynasore, an inhibitor of dynamin 2, modulates clathrin-mediated entry, we used this drug to study BTV entry. Our data showed that although Dynasore did not completely block virus entry, it was able to block the entry of majority of the virus particles. Dynamin 2 is also involved in several exocytic traffic steps (44), including exit from the Golgi complex (45). Since BTV uses a nonlytic exocytic pathway for virus release (46), prolonged incubation with Dynasore was not carried out, and the study was limited to the first 30 min of virus infection. Further investigation of BTV trafficking revealed that VP2 and VP5, the two outer surface proteins, separate from each other by 15 min p.i. The involvement of endosomal vesicles was also confirmed by infecting cells pretreated with ammonium chloride, a chemical that raises the pH of the endosomes. This further substantiated the importance of an acidic pH in BTV replication (25, 27). When the BTV entry pathway was investigated for the involvement of early and late endosomes, fluorescently labeled VP2 localized to the early and not the late endosomes. In comparison, although the detection of VP5 in infected cells by fluorescently labeled antibody demonstrated the distribution of VP5 to both early and late endosomal compartments, the colocalization of VP5 with EEA1 was observed to occur at earlier times p.i. than its colocalization with CD63. Together, these results suggest that after the early removal of VP2, the virus particles retain an outer layer of VP5 and traffic from the early to the late endosomes for the putative pH-induced structural modification of VP5. This facilitates pore formation in the endosomal membranes, leading to the release of transcriptionally active cores into the cytoplasm. In addition, while it was difficult to infer from the live-cell data whether the moving BTV particles entered the cells, analysis of BTV entry with the same tagged viruses in fixed cells clearly showed that BTV particles enter cells within the first 5 min of infection. Although our current results have demonstrated a mechanism slightly different from that in the two published studies on BTV entry, an increasing number of viruses have also been revealed to use more than one pathway to enter cells (47–52). Since BTV is able to infect a wide variety of tissue culture cells, such as the BHK-21, Vero, HeLa, and C6/36 cell lines, as well as other cell types (11, 36, 53), it is possible that BTV might utilize different entry mechanisms to initiate infection in different cells. In addition, the serotype of BTV might also play a role in determining the pathway for viral entry.

Interestingly, studies with untagged rotaviruses have revealed that the virus strain determines the choice of endocytic entry pathway into MA104 gut cells and that VP4, the outer spike protein of rotavirus, is responsible for this phenomenon (52). While the bovine rotavirus UK strain enters cells through a clathrin-mediated endocytic process, the rhesus rotavirus strain uses a poorly defined endocytic pathway that is clathrin and caveolin independent (47, 50). Recently, an in-depth study on rotavirus entry has further shown that although both bovine and rhesus strains reach maturing endosomes to establish virus infection, bovine rotavirus, unlike the rhesus strain, has to traffic to late endosomes (54). This requirement for the late endosomes was also shared by other rotavirus strains of human and porcine origin. In another study, using mammalian reoviruses that were bound to fluorescent dyes in vitro, it was shown that both virus particles and infectious subvirion particles (ISVPs) were internalized by clathrin-mediated endocytosis in Madin-Darby canine kidney cells (55). However, virions were trafficked to both early and late endosomes, while ISVPs escaped the endocytic pathway from a location before early endosomes (55).

In this study, it was possible for us to identify accessible loop-linker regions in VP2 that could accommodate the insertion of a tag into a replication-competent viral genome, allowing visualization of virus trafficking. The creation of tagged viruses also provides a valuable tool for studying BTV pathogenesis, including virus entry pathways, uncoating, and capsid synthesis. By use of biarsenical labeling, we also report for the first time that VP2 and VP5 segregate from each other very early during BTV entry. Using the tagged virus, this study has paved the way for detailed analysis of interactions with intracellular markers in real time by live-cell imaging.