|Home | About | Journals | Submit | Contact Us | Français|
Flock house virus (FHV) is a bipartite, positive-strand RNA insect virus that encapsidates its two genomic RNAs in a single virion. It provides a convenient model system for studying the principles underlying the copackaging of multipartite viral RNA genomes. In this study, we used a baculovirus expression system to determine if the uncoupling of viral protein synthesis from RNA replication affected the packaging of FHV RNAs. We found that neither RNA1 (which encodes the viral replicase) nor RNA2 (which encodes the capsid protein) were packaged efficiently when capsid protein was supplied in trans from nonreplicating RNA. However, capsid protein synthesized in cis from replicating RNA2 packaged RNA2 efficiently in the presence and absence of RNA1. These results demonstrated that capsid protein translation from replicating RNA2 is required for specific packaging of the FHV genome. This type of coupling between genome replication and translation and RNA packaging has not been observed previously. We hypothesize that RNA2 replication and translation must be spatially coordinated in FHV-infected cells to facilitate retrieval of the viral RNAs for encapsidation by newly synthesized capsid protein. Spatial coordination of RNA and capsid protein synthesis may be key to specific genome packaging and assembly in other RNA viruses.
Flock house virus (FHV), a member of the Nodaviridae family, is a nonenveloped, icosahedral insect virus whose genome consists of two positive-strand RNA molecules (3, 28). The bipartite nature of the FHV genome organizes its nonstructural and structural genes onto RNA1 (3.1 kb) and RNA2 (1.4 kb), respectively. An open reading frame (ORF) that nearly spans the length of RNA1 encodes protein A (112 kDa), the RNA-dependent RNA polymerase which directs the replication of RNA1, RNA2, and a 387-nucleotide (nt) subgenomic RNA3 (7). RNA3 corresponds to the 3′ end of RNA1 and encodes two small proteins, B1 and B2 (10). No function has been ascribed to B1, while B2 functions as an inhibitor of RNA silencing (15). Interestingly, RNA3 is also required as a transactivator of RNA2 replication, but the mechanism by which this is achieved is not completely understood (6). RNA2 encodes the 43-kDa capsid precursor protein α, which is the only structural protein required for the assembly of FHV provirions (7, 9). Each provirion consists of 180 subunits of protein α, arranged with T=3 quasiequivalent symmetry, and the two genomic RNAs. Provirions are not infectious and undergo an autocatalytic maturation process in which protein α cleaves into proteins β (38 kDa) and γ (5 kDa). Both cleavage products remain associated with the matured, infectious virion (9, 26).
FHV RNA1 and RNA2 are packaged into a single virion but the mechanism by which FHV copackages its multipartite genome is still unknown (14). One hypothesis is that an interaction between RNA1 and RNA2 occurs prior to packaging, allowing the genome to be encapsidated as a complex. The absence of significant complementarity between RNA1 and RNA2, however, does not support this model. Moreover, it was recently shown that RNA1 can be packaged in the absence of RNA2 by a capsid protein variant that lacks N-terminal residues 2 through 31, further suggesting that a noncovalent complex between RNA1 and RNA2 is not formed prior to packaging (17). An alternate hypothesis is that the packaging of genomic RNAs occurs in a sequential manner, similar to what has been described for bacteriophage 6 (19). In this model, one genomic RNA strand would interact with a specific binding site on the capsid protein and the formation of this intermediate would allow for the binding of the second strand. No experimental evidence has been obtained to date to support this model.
We recently showed that FHV replication can be initiated in Sf21 cells from recombinant baculovirus vectors (13). Specifically, we demonstrated that the coinfection of Sf21 cells with baculoviruses containing the cDNA for RNA1 and RNA2 launches self-directed RNA replication that leads to the formation of progeny FHV particles. Many of these particles contain an accurate complement of the FHV genome. We have now used this system to uncouple the synthesis of protein A and capsid protein from the replication of RNA1 and RNA2 to investigate whether RNA1 can be packaged in the absence of RNA2 and vice versa. Our results show that capsid protein translated from a replication-competent RNA2 can package RNA2 in the absence of RNA1. However, capsid protein synthesized in trans from a nonreplicating template packages neither RNA1 nor RNA2, even when both RNAs are subject to replication by protein A. This indicates a previously unrecognized type of coupling of RNA replication and translation on one hand and RNA packaging on the other. We suspect that accurate assembly of infectious FHV particles normally occurs in cellular microenvironments where viral RNA and capsid protein synthesis are spatially coordinated. This constraint represents a new aspect of viral assembly that could be important for many other RNA viruses.
Spodoptera frugiperda cells (line IPLB-Sf21) (30) were propagated in TC100 medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin at 27°C as described previously (24).
The construction of DNA transfer vectors for the generation of recombinant baculoviruses AcR1δ and AcR2δ with the BacPAK expression system kit (BD Biosciences) was previously described (13). These vectors contain the cDNA of FHV RNA1 and RNA2, respectively, with the hepatitis delta virus (HDV) ribozyme sequence (22) located immediately adjacent to the 3′ end of each cDNA insert.
To generate transfer vectors for AcR1δ[−5′UTR], AcR1[−3′UTR], and AcR1[−5′3′UTR], the plasmid FHV[1,0] (2), in which the cDNA of RNA1 is fused to the HDV ribozyme sequence at its 3′ end, was utilized. For AcR1δ[−5′UTR], a DNA fragment containing RNA1 and the HDV sequence was amplified by PCR with Pfu-Turbo polymerase (Stratagene) using primers that were designed to specifically delete the 5′ untranslated region (UTR) of RNA1. For AcR1[−3′UTR] and AcR1[−5′3′UTR], DNA fragments containing RNA1 were PCR amplified using primers that were designed to specifically delete the 3′ UTR or both the 5′ and 3′ UTRs of FHV RNA1, respectively. The 3′ UTR of RNA1 was defined as the 3′-terminal 71 nucleotides not encoding protein A. The forward primers used for the amplification of the previously described DNA fragments with 5′ and/or 3′ UTR deletions contained a 5′-terminal EcoRI restriction site, while the reverse primers contained an XbaI restriction site. Following PCR, these DNA fragments were electrophoresed through an agarose gel in Tris-acetate-EDTA (TAE) buffer, purified using the Qiaex II purification kit (QIAGEN), digested with EcoRI and XbaI, and ligated into the EcoRI-XbaI-digested gel-purified transfer vector pBacPAK9 (BD Biosciences).
To generate transfer vectors for AcR2[−5′UTR], AcR2[−3′UTR], and AcR2[−5′3′UTR], the plasmid p2BS(+)-wt (26), containing the cDNA of RNA2, was utilized. For AcR2[−5′UTR], a DNA fragment containing RNA2 was PCR amplified using primers that were designed to specifically delete the 5′ UTR of RNA2. For AcR2[−3′UTR] and AcR2[−5′3′UTR], DNA fragments containing RNA2 were PCR amplified using primers that were designed to specifically delete the 3′ UTR or both the 5′ and 3′ UTRs of FHV RNA2, respectively. The forward and reverse primers used for the amplification of these DNA fragments contained 5′-terminal BamHI and 3′-terminal XbaI restriction sites, respectively. Following PCR, the DNA fragments were purified, digested with BamHI and XbaI, and ligated into the BamHI-XbaI-digested gel-purified pBacPAK9.
To generate transfer vectors for AcΔ31δ and AcΔ31[−3′UTR], plasmid p2BS(+)Δ31 (4), which lacks the coding sequence for amino acids 2 through 31 of the FHV capsid protein, was utilized as a template for PCR. The forward primer corresponded to nucleotides 1 through 18 and had a BamHI site, while the reverse primer corresponded to nucleotides 1088 through 1068 of FHV RNA2. The resultant PCR product was purified and digested with BamHI and HpaI. The transfer vector for AcΔ31δ was constructed by a three-way ligation reaction that involved (i) the BamH1-HpaI-digested PCR product, (ii) a HpaI-BamH1 DNA fragment containing the remaining 3′-end-proximal 453 nt of FHV RNA2 flanked by the HDV sequence (obtained by digesting the AcR2δ transfer vector with BamH1 and HpaI and gel purifying a 542-bp DNA fragment), and (iii) BamHI-digested pBacPAK9. The transfer vector for AcΔ31[−3′UTR] was constructed by a ligation reaction that involved (i) the BamH1-HpaI-digested PCR product and (ii) a HpaI-BamH1 DNA fragment containing the remaining 300 nt of the FHV capsid protein reading frame followed by the pBacPAK9 vector sequence (obtained by digesting the AcR2[−3′UTR] transfer vector with BamH1 and HpaI and gel purifying a 5,838-bp DNA fragment).
The construction of a transfer vector for AcR2δ[AUG→Stop] involved two steps. First, the plasmid p2BS(+)-wt was used as a template for an inverse PCR (11) with primers that mutated the DNA sequence encoding the start (AUG) codon of the capsid protein ORF to a stop (UAG) codon. Second, the resultant plasmid, p2BSR2δ[AUG→Stop], was used as a template for a PCR with a forward primer that contained a BamHI site and a reverse primer corresponding to nucleotides 1088 through 1068 of FHV RNA2. The PCR product was digested with BamHI and HpaI and the transfer vector for AcR2δ[AUG→Stop] was constructed utilizing a similar three-way ligation strategy to the one described for AcΔ31δ.
Following the transformation of competent DH5α cells, plasmid DNA was isolated from several clones and the presence and orientation of the inserted DNA were determined by diagnostic restriction endonuclease mapping. Upon restriction endonuclease screening, positive clones were sequenced to confirm that no errors were introduced by PCR.
Recombinant baculoviruses were generated following the protocols of the manufacturer (BD Biosciences). In brief, each transfer vector was transfected into Sf21 cells together with Bsu36I-linearized BacPAK6 baculoviral DNA (BD Biosciences), and cell supernatants were harvested 3 days posttransfection. Putative recombinant baculoviruses were purified by plaquing the cell supernatants once on Sf21 cell monolayers and amplified following confirmation of the expression of the inserted genes. The titers of the recombinant baculovirus stocks were determined by plaque assay and ranged from 0.1 × 108 to 2 × 108 PFU/ml.
Monolayers consisting of 1.5 × 106 Sf21 cells in six-well plates were infected at a multiplicity of infection (MOI) of either 5 or 10 PFU per cell by the addition a 0.5-ml sample containing recombinant baculovirus stocks. This was followed by a 1-h incubation at room temperature (with rocking), the removal of the unattached baculovirus, and the addition of 2 ml complete growth medium to each well. Incubation was continued at 27°C for 3 to 5 days without agitation.
Capped RNA2 transcripts were synthesized by in vitro transcription from XbaI-linearized p2BS(+)-wt as described previously (25). RNA3 was purified from Sf21 cells infected with AcR1δ as follows: total cellular RNA was purified from these cells with the RNeasy mini kit 3 days postinfection (QIAGEN). The RNA was electrophoresed through a nondenaturing 1% agarose gel in TAE and visualized with ethidium bromide present in the gel. A band of the expected size for RNA3 (387 nt) that was not present in uninfected Sf21 cells was excised from the gel and purified using the RNaid kit (Qbiogene).
An infection/transfection protocol for the production of infectious FHV particles was described previously (13). In brief, monolayers consisting of 1.5 × 106 Sf21 cells in six-well plates were infected with recombinant baculoviruses at an MOI of 5 for the expression of RNA1 or RNA1 UTR deletion mutants. This was followed by liposome-mediated transfection of 100 ng in vitro-synthesized capped RNA2 at 24 h postinfection. In some instances, approximately 100 ng gel-purified RNA3 was cotransfected with 100 ng RNA2 (the molar ratio of RNA3 to RNA2 was approximately 3.6:1). Incubation was continued at 27°C for 2 to 4 days posttransfection.
At 3 days postinfection, baculovirus-infected or baculovirus-infected/RNA-transfected Sf21 cells in each well of a six-well plate were lysed by the addition of Nonidet P-40 to a final concentration of 0.5% (vol/vol). After an incubation of 10 min on ice, cell debris was pelleted at 16,000 × g for 10 min in a microfuge, and particles in the resultant supernatant were in turn pelleted through a 30% (wt/wt) sucrose cushion in 50 mM HEPES (pH 7)-0.1% bovine serum albumin-5 mM CaCl2 at 243,000 × g for 45 min in a Beckman SW50.1 rotor. The pellet was resuspended in 50 μl of 50 mM HEPES (pH 7)-5 mM CaCl2, centrifuged at 16,000 × g for 10 min in a microfuge for the exclusion of insoluble matter, and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis or immunoblotting.
FHV particles were purified from Sf21 cells 5 days postinfection by pelleting through a 30% (wt/wt) sucrose cushion followed by sedimentation though a 10 to 40% (wt/wt) sucrose gradient as described previously (13). The particles were harvested from the sucrose gradient by needle puncture or on an ISCO gradient fractionator at 0.75 ml/min and 0.5 min/fraction.
Total cellular RNA was extracted from Sf21 cells 3 days postinfection using the RNeasy kit (QIAGEN), while RNA was extracted from purified FHV particles by means of phenol-chloroform extraction as described previously (25).
Total cellular RNA was analyzed by denaturing agarose gel electrophoresis in the presence of formaldehyde as described previously (27), with one modification: a 2% Seakem LE agarose gel (FMC) was used instead of a 1% agarose gel. RNA extracted from purified FHV particles was analyzed by electrophoresis through a nondenaturing 2% Seakem LE agarose gel in TAE buffer. In both cases, RNA was visualized with ethidium bromide. For Northern blot analysis, RNA was electrophoresed through a denaturing 2% Seakem LE agarose-formaldehyde gel and transferred to a nylon membrane as described previously (27). The probe used for hybridization was digoxigenin-UTP labeled and complementary to nucleotides 124 through 1400 of RNA2. The synthesis of this probe was previously described (13) and prehybridization, hybridization, and chemiluminescent detection were carried out according to the manufacturer's protocols (Roche Molecular Biochemicals).
RT-PCR was performed on RNA extracted from FHV particles using the OneStep RT-PCR kit (QIAGEN). The forward and reverse primers that were used corresponded to nucleotides 1 through 18 and 488 through 468 of RNA2, respectively. The resultant cDNA fragments were cloned into the pCR4-TOPO vector (Invitrogen) and sequenced. As a control, cDNA fragments were synthesized by RT-PCR from a 1:1 mixture of capped wild-type (wt) RNA2 and capped AUG→Stop RNA2 in vitro transcripts, cloned into pCR4-TOPO, and sequenced. Capped AUG→Stop RNA2 was synthesized as described for capped wild-type RNA2 (25), with the exception that XbaI-linearized p2BSR2δ[AUG→Stop] was used as a template for in vitro transcription.
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblot analysis with rabbit anti-FHV serum were carried out as described previously (4).
We recently demonstrated that FHV RNA1 replicates to high levels in Sf21 cells when it is launched from the recombinant baculovirus AcR1δ (13). AcR1δ contains the cDNA of RNA1 directly followed by the hepatitis delta virus ribozyme sequence (Fig. (Fig.1A)1A) (13). After the transcription of the RNA1 gene from the baculovirus DNA and self-cleavage of the HDV sequence, an RNA1 product that contains an authentic 3′ end and only a few heterologous nucleotides at the 5′ end is obtained. Translation of this RNA1 yields protein A, which in turn amplifies the RNA1 message and also produces the subgenomic RNA3 (Fig. (Fig.1B,1B, lane 2). To produce protein A in the absence of RNA1 replication, we generated three new recombinant baculoviruses in which the 5′ UTR, 3′ UTR, together with the HDV site, or all were deleted. The rationale was based on previous studies that showed that the deletion of the 5′ UTR severely reduces the efficiency of RNA1 replication, whereas deletion of the 3′ UTR completely abolishes it (1, 2, 16). All UTR deletion mutants carried an intact ORF for protein A and were expected to produce functional RNA polymerase. Figure Figure1A1A shows a schematic representation of the baculovirus deletion mutants AcR1δ[−5′UTR], AcR1[−3′UTR], and AcR1[−5′3′UTR].
To determine if the elimination of the UTRs inhibited RNA1 replication, Sf21 cells were infected with each of the AcR1 deletion mutants described above, and total cellular RNA was extracted 3 days after infection. Electrophoretic analysis of the RNA samples demonstrated that, relative to a control of AcR1δ-infected cells, AcR1δ[−5′UTR]-infected cells contained only faintly detectable levels of RNA1 and reduced levels of the subgenomic RNA3 (Fig. (Fig.1B,1B, lane 3). This result was in line with previous observations that in the absence of the 5′ UTR, protein A does not efficiently replicate RNA1 (2). While the absence of the 5′ UTR was not expected to affect the synthesis of RNA3 (5), the overall reduced level of RNA1 resulted in a concomitant decrease of the subgenomic RNA. As expected, neither RNA1 nor RNA3 was detected in AcR1[−3′UTR]- and AcR1[−5′3′UTR]-infected cells (Fig. (Fig.1B,1B, lanes 4 and 5). Analysis of cell lysates by protein gel electrophoresis showed that the levels of protein A were similar regardless of the absence or presence of the UTRs (data not shown), indicating that amplification of RNA1 by replication did not lead to a corresponding increase in protein production. This was reminiscent of the situation for FHV-infected Drosophila cells, which contain very high levels of RNA1 but only low levels of protein A (7).
To confirm that all three UTR deletion mutant types, in particular the 3′ UTR deletion mutants, directed the synthesis of functional protein A, Sf21 cells were infected with one of the three recombinant baculoviruses and transfected 24 h later with a mixture of RNA2 and RNA3. The rationale was that if functional RNA polymerase were present, then it would replicate and thereby amplify RNA2 and RNA3 in trans as we had shown previously for AcR1δ (13). This, in turn, would result in the synthesis of capsid protein and assembly of virus particles, which could be detected by immunoblot analysis. In a separate set of experiments, baculovirus-infected Sf21 cells were transfected with RNA2 but not RNA3. Recall that the replication of RNA2 requires RNA3 as a transactivator (6). Since the detection of capsid protein by immunoblot analysis is possible only after the amplification of RNA2 by replication (unpublished results), this set of experiments served as a convenient negative control.
As shown in Fig. Fig.22 (lanes 1 and 2), cells infected with AcR1δ[−5′UTR] and transfected with RNA2 in either the presence or absence of RNA3 contained abundant amounts of capsid protein α and the cleavage product β. The presence of protein β indicated that capsid protein had formed particles, since the maturation cleavage is assembly dependent (9). The other cleavage product, protein γ, is too small (5 kDa) to be visualized on standard Laemmli polyacrylamide gels. The fact that high levels of capsid protein were synthesized in the absence of transfected RNA3 was expected, since AcR1δ[−5′UTR] gave rise to sufficient amounts of RNA3 itself (Fig. (Fig.1B,1B, lane 3). In contrast, Sf21 cells that were infected with one of the 3′ UTR deletion constructs produced capsid proteins α and β only when cotransfected with RNA2 and RNA3 (Fig. (Fig.2,2, lanes 3 to 6). The levels of coat protein in these cells differed from experiment to experiment and are lower in the particular experiment shown here, but this was not consistently observed. Taken together, the results provided indirect confirmation that the cells contained functional protein A.
The infection/transfection system described above provided an opportunity to determine whether capsid protein could package RNA2 in the absence of replicating RNA1. This allowed us to address the question of whether the viral RNAs were packaged independently of each other. To test this, FHV particles were sucrose gradient purified from cells supporting the replication of RNA2 but not RNA1 (infected with AcR1[−3′UTR] or AcR1[−5′3′UTR] and transfected with RNA2 and RNA3) and, as a control, from cells supporting replication of both RNA1 and RNA2 (infected with AcR1δ or AcR1δ[−5′UTR] and transfected with RNA2). RNA was then extracted from the particles and analyzed on a nondenaturing agarose gel. RNA1 and RNA2 were packaged by FHV capsid protein in cells that supported the replication of both RNAs (Fig. (Fig.3,3, lanes 1 and 2). The population of purified particles also contained random cellular RNA, as we observed previously (13). The slight decrease in the ratio of RNA1 to RNA2 for particles purified from AcR1δ[−5′UTR]-infected cells (Fig. (Fig.3,3, lane 2) compared to AcR1δ-infected cells (Fig. (Fig.3,3, lane 1) was probably the result of the proportionately lower levels of RNA1 available for packaging in AcR1δ[−5′UTR]-infected cells (Fig. (Fig.1B,1B, lane 3). In contrast, considerable quantities of RNA2 (and cellular RNA), but no RNA1, were detected in particles isolated from cells supporting the replication of RNA2 but not RNA1 (Fig. (Fig.3,3, lanes 3 and 4). This result demonstrated that RNA2 could be packaged in the absence of RNA1 replication.
We next wanted to test the reverse scenario: could RNA1 be packaged in the absence of RNA2 replication? This required an expression system in which replication-competent RNA1 was coexpressed with replication-incompetent RNA2 (for the synthesis of FHV capsid protein). Accordingly, the 5′ UTR and/or 3′ UTR sequences present in AcR2δ were deleted with a strategy analogous to that described above for RNA1 except that the HDV site was not retained at the 3′ end of the 5′ UTR deletion construct.
Figure Figure4A4A shows a schematic representation of recombinant baculoviruses for the expression of RNA2 with a 5′ UTR deletion (AcR2[−5′UTR]), a 3′ UTR deletion (AcR2[−3′UTR]), and the deletion of both UTRs (AcR2[−5′3′UTR]). Immunoblot analysis of lysates from Sf21 cells individually infected with these constructs showed no difference in the expression levels of the coat protein (data not shown). To determine if the UTR deletions inhibited RNA2 replication, Sf21 cells were coinfected with AcR1δ and AcR2δ, AcR2[−5′UTR], AcR2[−3′UTR], or AcR2[−5′3′UTR]. Electrophoretic analysis of total cellular RNA 3 days after infection showed that RNA1 was replicated with equal efficiency in all cases (Fig. (Fig.4B,4B, top panel). A band comigrating with wild-type RNA2 was detected in cells infected with AcR2δ and AcR2[−5′UTR] but not in those infected with the 3′ UTR deletion constructs (Fig. (Fig.4B,4B, top panel). Northern blot analysis with a negative-strand RNA2 probe confirmed that this band represented RNA2 (Fig. (Fig.4B,4B, bottom panel). It was surprising that the AcR2[−5′UTR] mutant gave rise to replicating RNA2. Since this clone did not carry the HDV site, its 3′ end was not identical to that of native RNA2. Therefore, it was expected that the FHV polymerase would not recognize it as a template for RNA replication (1). The fact that it did suggested that protein A is capable of recognizing replication elements within the 3′ UTR of RNA2 irrespective of the presence of additional heterologous nucleotides at the 3′ end.
Several additional bands of higher molecular mass were also detected and most probably represented primary RNA2 transcripts that had terminated at one of several polyhedrin gene termination signals and that had not cleaved at the HDV ribozyme site as observed previously (13). Such primary RNA transcripts could also be detected for the 3′ UTR deletion mutants upon longer exposure of the blot (not shown).
It has been demonstrated previously that not only does RNA3 transactivate RNA2 replication but its synthesis is repressed at the onset of RNA2 replication (8, 32). Therefore, additional proof that RNA2 was not replicated in cells infected with AcR2[−3′UTR] or AcR2[−5′3′UTR] was manifested in the detection of higher levels of RNA3 in these cells than in AcR2δ-infected cells (Fig. (Fig.4B,4B, top panel, compare lanes 4 and 5 to lane 2). Taken together, these results demonstrated that RNA2 replication, like that of RNA1, was slightly inhibited by the 5′ UTR deletion but completely inhibited by the 3′ UTR deletion. As shown below, FHV capsid protein was, however, expressed from these nonreplicating RNA2 templates.
To determine if RNA1 could be packaged in the absence of replicating RNA2, FHV particles were purified from Sf21 cells supporting the replication of RNA1 in the absence of RNA2 replication (coinfected with AcR1δ and either AcR2[−3′UTR] or AcR2[−5′3′UTR]) and, as a control, from cells supporting the replication of both RNA1 and RNA2 (coinfected with AcR1δ and either AcR2δ or AcR2[−5′UTR]). As expected, both RNA1 and RNA2 were packaged by FHV capsid protein, together with random cellular RNA, in cells that supported the replication of RNA1 and RNA2 (Fig. (Fig.5,5, lanes 2 and 3). Surprisingly, the examination of the RNA packaged into particles from cells that supported only the replication of RNA1 revealed the presence of random cellular RNA only (Fig. (Fig.5,5, lanes 4 and 5). No RNA1 was detectable, despite the fact that high levels of this RNA were available for packaging in the cells in which these particles were assembled (Fig. (Fig.4B,4B, top panel, lanes 4 and 5).
There were two possible explanations for the inability of capsid protein to package RNA1 in the absence of RNA2 replication. First, it was possible that the packaging of RNA1 was dependent on the packaging of RNA2, i.e., that the packaging of FHV genomic RNAs occurs in a sequential manner, starting with RNA2. Second, it was possible that RNA1 was not packaged efficiently because it was somehow dependent on the replication of RNA2. To distinguish between these two possibilities, we decided to take advantage of a capsid protein mutant (Δ31) that lacks N-terminal residues 2 through 31. This mutant is able to package RNA1 independently of RNA2 in FHV-infected Drosophila cells (17). Thus, we reasoned that this mutant would package RNA1 in the absence of replicating RNA2 unless RNA2 replication was key to packaging FHV RNAs.
Two baculovirus expression vectors were constructed for the synthesis of replicating and nonreplicating Δ31 RNA2. The baculoviruses for the synthesis of replicating Δ31 RNA2 (AcΔ31δ) and nonreplicating Δ31 RNA2 (AcΔ31[−3′UTR]) were identical to AcR2δ and AcR2[−3′UTR], respectively, with the exception that sequences encoding residues 2 to 31 of protein α had been removed (compare Fig. Fig.4A4A and Fig. Fig.6A).6A). Analogous to our previous observations with constructs expressing wt capsid protein, Sf21 cells coinfected with AcR1δ and either AcΔ31δ or AcΔ31[−3′UTR] contained equivalent amounts of RNA1 (Fig. (Fig.6B,6B, lanes 1 and 2). A band representing Δ31 RNA2 transcripts, however, could not be clearly distinguished from other bands in the total RNA sample. We therefore used Northern blot analysis to confirm that Δ31 RNA2 replication products were present at high levels in AcR1δ/AcΔ31δ-infected cells (Fig. (Fig.6C,6C, lane 1) and were absent in AcR1δ/AcΔ31[−3′UTR]-infected cells (Fig. (Fig.6C,6C, lane 2). Higher-molecular-weight primary Δ31 RNA2 transcripts did become detectable upon longer exposure of this blot (not shown). As also observed previously, the levels of RNA3 present in AcR1δ/AcΔ31[−3′UTR]-infected cells (Fig. (Fig.6B,6B, lane 2) were higher than those in AcR1δ/AcΔ31δ-infected cells (Fig. (Fig.6B,6B, lane 1), indicating that Δ31 RNA2 inhibited RNA3 accumulation in AcR1δ/AcΔ31δ-infected cells but not in AcR1δ/AcΔ31[−3′UTR]-infected cells. Taken together, these results demonstrated that the deletion of the 3′ UTR of Δ31 RNA2 abolished Δ31 RNA2 replication in a fashion similar to what has been described for the deletion of the 3′ UTR from wt RNA2.
FHV particles were purified from AcR1δ/AcΔ31δ-infected cells and AcR1δ/AcΔ31[−3′UTR]-infected cells to determine if Δ31 capsid protein could package replicating RNA1 in the absence of replicating Δ31 RNA2. We focused on the packaging phenotype of particles displaying T=3 symmetry and ignored the collection of smaller particles that this mutant capsid protein also forms because the internal volume of the smaller particles is insufficient to accommodate a genome segment the size of RNA1 (4).
As shown in Fig. Fig.77 (lane 1), RNA1 and trace amounts of Δ31 RNA2 (together with random cellular RNA) were detected in particles derived from cells supporting RNA1 and Δ31 RNA2 replication. However, neither RNA1 nor Δ31 RNA2 was detected in particles derived from cells supporting the replication of RNA1 but not Δ31 RNA2 (Fig. (Fig.7,7, lane 4). The failure of Δ31 capsid protein to package RNA1, therefore, had to be a reflection of the fact that capsid protein was produced from a nonreplicating template.
The results obtained so far pointed to a link between RNA2 replication and FHV RNA packaging. However, it was unclear whether replication of RNA2 per se was sufficient for packaging of the FHV genome or whether there was an additional link between replication of RNA2 and subsequent translation into capsid protein. To address this issue, we generated a recombinant baculovirus that encoded an RNA2 replicon with a closed capsid protein reading frame. This baculovirus, AcR2δ[AUG→Stop], was identical to AcR2δ (Fig. (Fig.4A)4A) with the exception that the start (AUG) codon of the capsid protein ORF was mutated to a stop (UAG) codon. To confirm that capsid protein synthesis but not RNA2 replication was disabled, Sf21 cells were infected with AcR2δ[AUG→Stop] in the presence or absence of AcR1δ. The detection of an RNA species with a similar molecular weight to RNA2 for AcR1δ/AcR2δ[AUG→Stop]-infected cells (Fig. (Fig.8A,8A, lane 2) but not for AcR2δ[AUG→Stop]-infected cells (Fig. (Fig.8A,8A, lane 1) demonstrated that AUG→Stop RNA2 was replicated to high levels in Sf21 cells supporting FHV RNA replication. As expected, FHV capsid protein was undetectable in these cells as demonstrated by immunoblot analysis (data not shown).
In the following analysis, Sf21 cells were coinfected with baculoviruses AcR2δ[AUG→Stop], AcR1δ, and AcR2δ to confirm that AUG→Stop RNA2 could be packaged by capsid protein that was synthesized in trans from replicating wt RNA2. In the more critical experiment, Sf21 cells were infected with AcR2δ[AUG→Stop], AcR1δ, and AcR2[−5′3′UTR]. The latter experiment would reveal whether there was any difference in FHV RNA packaging when capsid protein was synthesized in trans from a nonreplicating template. FHV particles were purified from the infected cells, and RNA was extracted and subjected to agarose gel electrophoresis. Northern blot analysis with probes for RNA2 and RT-PCR were used to examine the nature of the packaged RNA.
As shown in Fig. Fig.8B,8B, particles purified from cells that were coinfected with viruses that included AcR2δ contained both RNA1 and RNA2 (lane 1). To determine whether the packaged RNA2 included both wt and AUG→Stop products, it was amplified by RT-PCR with primers specific for RNA2 and the resulting products ligated into DNA vectors. Sequence analysis of 10 independent clones showed that the ratio of plasmids with an inserted AUG→Stop sequence to plasmids with an inserted wild-type RNA2 sequence was 1.5:1. This result proved that (i) AUG→Stop RNA2 could be packaged by capsid protein synthesized in trans from a replicating wt RNA2 and (ii) that the packaging efficiencies of the two RNAs were similar if not identical. We also performed a control experiment in which an equimolar mixture of in vitro-synthesized wt RNA2 and AUG→Stop RNA2 transcripts was subjected to RT-PCR, cloning, and sequencing. The same ratio (1.5:1) for wt RNA2 to AUG→Stop RNA2 was obtained.
Interestingly, particles purified from cells infected with the baculovirus that produced capsid protein from a nonreplicating template contained predominantly random cellular RNA (Fig. (Fig.8B,8B, top panel, lane 2). This was despite the fact that high levels of RNA1 and AUG→Stop RNA2 were available for packaging in these cells (Fig. (Fig.8A,8A, lane 3). A faint band that appeared to comigrate with RNA2 was visible in the collection of packaged RNAs, but it did not hybridize with an RNA2-specific probe in the Northern blot analysis (Fig. (Fig.8B,8B, bottom panel, lane 2). Taken together, these results showed conclusively that packaging of the FHV genomic RNAs requires capsid protein synthesized from a replicating RNA2.
This study describes the application of a baculovirus expression system for the synthesis of functional FHV proteins from either replicating or nonreplicating genomic RNA templates. This system was reliant on the removal of the 3′ UTRs from RNA1 and RNA2, which was shown to result in the inhibition of RNA replication with retention of protein A and capsid protein synthesis, respectively. The ability to effectively uncouple viral protein synthesis from RNA replication in this baculovirus expression system was used to determine if RNA2 could be packaged in the absence of RNA1 and vice versa. Examination of the RNA content of particles assembled in the presence of nonreplicating RNA1 and replicating RNA2 demonstrated that RNA2 was packaged efficiently in the absence of RNA1 replication. The reverse was, however, not true, as replicating RNA1 was not readily detected in particles that were assembled in the absence of RNA2 replication. We initially interpreted the results to indicate that the packaging of the two RNAs is sequential and that it must begin with RNA2. However, subsequent experiments revealed that the inability of the capsid protein to package RNA1 was due to the fact that the capsid protein had been produced from a nonreplicating template. This became obvious when neither replicating RNA1 nor replicating RNA2 were packaged by capsid protein translated from a primary baculovirus transcript. The RNAs were packaged, however, when capsid protein was synthesized from replicating wild-type RNA2. This result suggested some sort of coupling of replication translation of RNA2 on one hand and FHV RNA packaging on the other.
Coupling of RNA replication and packaging has previously been observed for poliovirus (20). In this case, nonreplicating poliovirus subgenomic RNAs were not trans-encapsidated by capsid protein generated from a replicating poliovirus genome. This prompted the investigators to propose that the encapsidation of poliovirus RNAs requires a direct physical interaction between the replication complexes and the assembling capsid protein. Such a direct link between replication complexes and assembling capsid protein is unlikely for FHV because pulse-chase experiments have shown that the genomic RNAs are packaged long after they have been synthesized by the FHV polymerase (9). Also, in FHV, the replication of the viral RNAs is not sufficient for packaging, based on our finding that replicating RNA1 and AUG→Stop RNA2 were not packaged by capsid protein synthesized in trans from a nonreplicating template. At the same time, the possibility of a linkage in cis between RNA packaging and translation could also be excluded because nontranslating AUG→Stop RNA2 was efficiently packaged by capsid protein made in trans from wild-type, replicating RNA2.
Our results can be rationalized by hypothesizing that RNA2 replication is required for the packaging of FHV RNAs because it guarantees that capsid protein is synthesized in a cellular location that permits immediate access to progeny RNA1 and RNA2. More specifically, we suspect that the translation of replicated RNA2 molecules occurs near or at the site of replication itself, thereby ensuring that capsid protein does not have to traffic through the cytosol in order to retrieve the viral RNAs for assembly. This model, illustrated schematically in Fig. Fig.9,9, explains why capsid protein translated from a nonreplicating mRNA, such as a baculovirus transcript, fails to encapsidate FHV RNAs. Presumably, such transcripts, after exiting the nucleus, are translated in a location distal from the RNA replication site, thereby leading to packaging of cellular RNAs. We find this model intriguing because it sheds light on cellular requirements for assembly. We suspect that the events associated with FHV infection normally lead to establishment of cellular microdomains or compartments in which replication, translation, and assembly are spatially coordinated and that this coordination is crucial for the accurate formation of progeny virions. Although there is no experimental evidence for the existence of such microdomains during nodaviral replication, microdomains are known to exist for other viruses, for example, reoviruses. In reoviruses, the assembly of progeny virions occurs in so-called viroplasms, which are special cellular microenvironments containing structural as well as nonstructural reovirus proteins (29). Interestingly, reoviruses, like FHV, package a multipartite genome, and although the genomic segments represent double-stranded RNAs in the mature virion, there is evidence that the segments are packaged as single-stranded mRNAs (21). It is conceivable that the coordinated packaging of multiple genomic segments into a single virion requires unique cellular environments that are formed during the course of the replication cycle.
Our results with FHV are also reminiscent of a situation that was recently described for the positive-strand virus Kunjin virus, a member of the family Flaviviridae (12). Specifically, full-length nonreplicating Kunjin virus RNA was not packaged by capsid protein derived in cis from a nonreplicating template. However, when replication of the full-length RNA was restored by providing the replication proteins in trans, packaging of the genome occurred as revealed by the formation of infectious virions. The authors of that study concluded that replication of viral RNA on cellular membranes somehow allowed relocation of the replication complexes to the site of assembly. Alternatively, it is possible that capsid protein, now translated from the replicated progeny RNA, was in the appropriate location for selecting the viral genome.
The scenario that we imagine for specific packaging of FHV RNAs also explains why the population of FHV particles produced in Sf21 cells from baculovirus vectors yielding replicating RNA1 and RNA2 always included particles that contain cellular RNA. This is because only a fraction of the primary FHV transcripts ultimately become subject to replication by the FHV polymerase. A large portion of the transcripts are immediately translated, and the resulting capsid protein invariably packages cellular RNA as we showed many years ago (24). What this implies is that capsid proteins translated from replicating as opposed to nonreplicating RNAs assemble into particles that segregate into distinct populations and that only the population derived from replicating RNA contains the FHV genome. We are currently in the process of confirming this by using capsid proteins that display differential epitopes depending on what kind of template they are derived from.
The hypothesis that capsid protein must be synthesized near the site of RNA replication to ensure specific packaging of the FHV genome calls into question the existence of bona fide encapsidation signals on the viral RNAs. It could be argued that in the type of situation that we propose, RNA packaging is simply a result of mass action, i.e., nonspecific interaction of capsid protein with progeny RNA located in the immediate vicinity. Indeed, packaging of nonviral RNA is very efficient, as indicated by the high yields of FHV virus-like particles that are obtained from Sf21 cells expressing coat protein from a nonreplicating template (24). On the other hand, in vitro assembly experiments have clearly shown that the coat protein preferentially selects viral RNA for packaging from a mixture containing nonviral nucleic acids (23). This suggests the presence of specific signals on the viral RNAs that promote packaging of the FHV genome. The notion that FHV RNAs contain specific packaging signals is also supported by the fact that an encapsidation signal has already been identified for a defective interfering RNA derived from RNA2 (31). When this signal is deleted, the defective interfering RNA2 is no longer packaged. Secondly, the principle of mass action does not satisfyingly explain the high fidelity with which RNA1 and RNA2 are copackaged, and thirdly, it does not explain why the subgenomic RNA3 is not packaged. Thus, further experiments are required to elucidate the molecular mechanism that ensures the copackaging of FHV RNA1 and RNA2 into progeny virions.
This work was supported by grants from the National Institutes of Health (GM53491 and C027181) and a National Institutes of Health fellowship award (AI10262) to N.K.K.
We thank J. K. Lanman for helpful suggestions during the preparation of the manuscript.
†Manuscript 16850-MB from the Scripps Research Institute.