Determinants in the Maturation of Rubella Virus P200 Replicase Polyprotein Precursor

Abstract

Rubella virus (RUBV), a positive-strand RNA virus, replicates its RNA within membrane-associated replication complexes (RCs) in the cytoplasm of infected cells. RNA synthesis is mediated by the nonstructural proteins (NSPs) P200 and its cleavage products, P150 and P90 (N and C terminal within P200, respectively), which are processed by a protease residing at the C terminus of P150. In this study of NSP maturation, we found that early NSP localization into foci appeared to target the membranes of the endoplasmic reticulum. During maturation, P150 and P90 likely interact within the context of P200 and remain in a complex after cleavage. We found that P150-P90 interactions were blocked by mutational disruption of an alpha helix at the N terminus (amino acids [aa] 36 to 49) of P200 and that these mutations also had an effect on NSP targeting, processing, and membrane association. While the P150-P90 interaction also required residues 1700 to 1900 within P90, focus formation required the entire RNA-dependent RNA polymerase (aa 1700 to 2116). Surprisingly, the RUBV capsid protein (CP) rescued RNA synthesis by several alanine-scanning mutations in the N-terminal alpha helix, and packaged replicon assays showed that rescue could be mediated by CP in the virus particle. We hypothesize that CP rescues these mutations as well as internal deletions of the Q domain within P150 and mutations in the 5′ and 3′ cis-acting elements in the genomic RNA by chaperoning the maturation of P200. CP's ability to properly target the otherwise aggregated plasmid-expressed P200 provides support for this hypothesis.

INTRODUCTION

Rubella virus (RUBV) is a positive-strand RNA virus with an ∼10-kb genome that belongs to the Togaviridae family and is the sole member of the Rubivirus genus. RUBV virions are approximately 70 nm in diameter and composed of a single copy of genomic RNA that is surrounded by a nucleocapsid shell (nucleocapsids are made from multiple copies of the capsid protein [CP]). The nucleocapsid is wrapped by an envelope derived from host cell membranes containing the virus-encoded glycoproteins E1 and E2 that reside within the envelope as dimer spikes (E2-E1) (10). The RUBV genome contains two open reading frames (ORFs), and the 5′ ORF is directly translated from the genome and encodes the nonstructural proteins (NSPs) involved in viral RNA synthesis. The NSPs are initially translated as a polyprotein called P200 (5, 9). P200 is thought to function in properly targeting the genomic RNA to initial sites of replication complex (RC) assembly where viral RNA synthesis occurs (30); however, little is known about the mechanisms by which this occurs. Subsequently, P200 functions in the synthesis of negative-strand RNA using the incoming genome as a template (23). It is known that P200 possesses protease activity that cleaves at residue 1301 (out of 2,116 residues) to produce the two mature replicase proteins P150 and P90 (9, 24). P150 and P90 form a complex that synthesizes two positive-strand RNAs, genomic and subgenomic RNA (9, 22, 24), but it is not clear if this interaction takes place in the context of P200.

A subgenomic RNA that is identical to the 3′ terminal third of the genomic RNA is produced during RUBV RNA synthesis (3, 34, 44, 46), and this second ORF encodes the structural proteins N-CP-E2-E1-C. While serving as an mRNA for the structural proteins appears to be the only role of the subgenomic RNA, newly synthesized genomic RNAs subsequently either undergo translation, producing P200 to recapitulate RC assembly and RNA synthesis, or are packaged into virus particles. Besides its role in forming virus particles, CP performs several nonstructural functions during virus infection (15), the most intriguing of which is its ability to rescue lethal mutations in the Q domain (a proline and arginine-rich domain) of P150 as well as within the 5′ and 3′ cis-acting elements (CAEs) of the genomic RNA (42, 43, 45).

The precise molecular events that take place during RUBV infection from the time of genomic RNA translation to RC assembly remain poorly understood. However, replicon RNA constructs in which the structural protein ORF is replaced with a reporter gene, i.e., green fluorescent protein (GFP), are able to replicate, although they do not routinely spread from cell to cell (41), demonstrating that only the NSPs are necessary for RC formation and RNA replication. Early reports indicated that RUBV RCs form within invaginations on the outer membranes of lysosomes and late endocytic vesicles (6, 20, 27). However, recent examination of RUBV-infected cells revealed that P150 and double-stranded RNA (dsRNA; a standard marker for RCs) showed intracellular colocalization, but only marginally with late endosomes and lysosomes, as well as colocalization at the plasma membrane (30). Sindbis virus (SINV) and Semliki Forest virus (SFV), two other togaviruses belonging to the Alphavirus genus, though originally reported to be replicating their RNA in association with the endo-/lysosomal compartment, have now been found to replicate in membranous spherules which originate at the plasma membrane and migrate to the perinuclear region via endocytosis (11–13). It is likely that the biogenesis of RUBV RCs follows a similar pathway.

In a previous study, we found that mutagenesis of an alpha helix at the N terminus of P200 (amino acids [aa] 36 to 49) unexpectedly exerted long-range effects on P200 function, including decreasing the efficiency of its cleavage and altering its subcellular localization (29). In the current study, we extended this observation by finding that mutagenesis of the N-terminal alpha helix also disrupts the establishment of P150-P90 interactions, their targeting and membrane association, and ultimately, virus production, suggesting that the interaction(s) between the P150 and P90 domains is important for several NSP functions. Intriguingly, the virus CP could rescue one of the N-terminal alpha-helix mutants, the mutant with the E36A mutation, that interfered with P150-P90 interactions and NSP targeting, leading us to speculate that CP could serve as a chaperone in the process of NSP maturation and RC generation.

MATERIALS AND METHODS

Reagents.

Lipofectamine 2000, rabbit polyclonal antibodies to GFP used for immunoprecipitation, donkey anti-rabbit IgG Alexa Fluor 594, and Hoechst 33342 were obtained from Invitrogen. The protein A-agarose, nitroblue tetrazolium–5-bromo-4-chloro-3-indolylphosphate (NBT/BCIP), mouse anti-HA antibodies, and protease inhibitor cocktail were obtained from Roche. The rabbit antibodies to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and calnexin were obtained from Abcam and Sigma, respectively. The rabbit polyclonal GFP peptide antibodies used for Western blotting were obtained from Clontech. Goat anti-rabbit or -mouse alkaline phosphatase conjugate was obtained from Promega. The rabbit 7W7 anticapsid antibody was a kind gift from Tom Hobman. The mouse anti-dsRNA was obtained from Scientific Consultants.

Cells and viruses.

A line of African green monkey kidney (Vero) cells was used in these experiments and maintained in Dulbecco modified Eagle medium (DMEM; Cellgro) supplemented with 5% with fetal bovine serum (FBS; Atlanta Biologicals) and 10 μg/ml gentamicin (Gibco). An additional cell line, called C-Vero, that stably expresses the capsid protein (aa 1 to 277) was used and was described previously (45). C-Vero cells were maintained as Vero cells, except that the medium was supplemented with G418. Baby hamster kidney (BHK) cells were obtained from ATCC and maintained in DMEM with 10% FBS. BHK-SP cells, which are BHK cells stably expressing the RUBV structural proteins C-E2-E1 (BHK/CE2E1 cells), are described in reference 4 and were maintained accordingly with hygromycin. All cells were held within a humidified 5% CO₂ incubator at 35°C. Robo502/P150-GFP stocks were prepared as described in reference 29.

Plasmids.

The cDNA of the infectious clone Robo502/P150-GFP was created by inserting the GFP gene in frame at the downstream NotI restriction site within the P150 gene (between aa 717 and 718) (29). The parent plasmid, CMV/P200(GFP-HA), expressed GFP-tagged P150 and P90 tagged with a hemagglutinin (HA) epitope between aa 1661 and 1662. Proline-insertion mutagenesis was carried out using a three-round asymmetric PCR strategy described in reference 2. To this end, a single proline residue was inserted into P200-GFP between aa 41 and 42, creating the construct CMV/P200-GFP-P1a. Using the same three-round PCR strategy, proline insertions and alanine substitutions were introduced into the infectious clone; the forward primer in the final round contained a 5′ HindIII site followed by an SP6 promoter sequence and the first 25 nucleotides of the RUBV genome. The HindIII-AgeI-restricted fragments were ligated into a similarly cut Robo502/P150-GFP plasmid or substituted into the Robo502/P150-HA or RUBRep/P150-HA version. Creation of the HA epitope insert was described in reference 45. GCC was the codon used for alanine substitution, and eight mutations were made: E36A, etc. The P200(GFP-HA) deletion mutants D1 and D2 were created by PCR using a forward primer complementary to sequences upstream of the EcoRV site in the RUBV genome-encoding cDNA and a reverse primer complementary to the 3′ ORF deletion end site, which also contained a downstream BamHI restriction site. The EcoRV-BamHI-restricted PCR fragments were ligated into a similarly cut P200(GFP-HA) backbone to create D1 and D2. P200*, D1*, and D2* were created by introducing an RsrII-EcoRV-restricted fragment from RUBRep-GFP-NSP* that was described in reference 42 into a similarly cut P200(GFP-HA), D1, or D2 backbone.

Immunoprecipitation and Western blotting.

At 24 h posttransfection (hpt), Vero cells grown in 60-mm plates were lysed in 0.6 ml of lysis buffer (phosphate-buffered saline [PBS] with 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 1 mM CaCl₂, 0.5 mM MgCl₂, and EDTA-free protease inhibitor cocktail [Roche]). Insoluble debris was removed by centrifugation at 16,000 × g. For immunoprecipitation, 0.5 ml of supernatant was mixed with 2 μl of GU10 antibody (9), anti-GFP, anti-HA, or, as a control, no antibody. The lysates were then incubated for 1 h at room temperature before being mixed with 10 μl of protein A-agarose beads (Roche) for an additional hour, with mixing. The beads were washed twice with lysis buffer and twice with wash buffer (PBS with 0.5% Triton X-100, 0.5% deoxycholate, 1 mM CaCl₂, and 0.5 mM MgCl₂) before eluting with 75 μl of 2× sample buffer (20 mM Tris, pH 6.8, 10% glycerol, 2% SDS, and 0.1% bromophenol blue) and boiling for 2 min. For Western blotting, 10 to 15 μl of cell lysate or the same volume of immunoprecipitated solution was resolved by SDS-PAGE, followed by transfer to nitrocellulose membranes, the membranes were probed with anti-GFP or GU10 antibodies (both raised in rabbits), and proteins were detected with alkaline phosphatase-conjugated, goat anti-rabbit secondary antibodies (Promega) using NBT/BCIP (Roche) as a substrate.

Fluorescence microscopy.

Immunofluorescence was conducted as previously described (30), with some modifications. Briefly, Vero cells were grown on glass coverslips until 20 to 30% confluent and either infected with Robo502/P150-GFP or transfected with a plasmid(s) or in vitro transcripts. At the appropriate time, the cells were fixed in 4% formaldehyde and permeabilized with 1% Triton X-100. Antibodies were diluted 1:1,000 in PBS containing 2% bovine serum albumin, and the incubation period for each antibody was 1 h at 25°C. Hoechst 33342 was diluted 1/15,000 into the primary antibody solution. Cells that were not stained with antibodies did not get permeabilized and underwent only 4% formaldehyde fixation and Hoechst 33342 staining of nuclei. After washing away the antibodies and stain with PBS, the coverslips were mounted on microscope slides and analyzed using either a Zeiss epifluorescence microscope equipped with an AxioCam imaging system and integrated AxioVision software (see images in Fig. 2 and 7) or a Zeiss LSM 700 confocal microscope interfaced with a computer running LSM 700 software (see Fig. 1 and 5).

Fig 2.

Effect of P1 mutation on P200 cleavage, targeting, P150-P90 interaction, and infectivity. (A) Western blotting of lysates from mock- or Robo502/P150-GFP (wt or P1 mutant) transcript-transfected Vero cells at 3 and 72 h posttransfection probed with rabbit anti-GFP, -CP, or -calnexin antibodies. (B) Vero cells were analyzed by fluorescence microscopy at 3 h posttransfection with in vitro Robo502/P150-GFP transcripts from wt or mutant P200-P1 Robo502/P150-GFP constructs (top and bottom, respectively). Hoechst 33342 was used to stain nuclei. Bars, 10 μm. (C) Vero cells transfected with wild-type P200 or P200-P1 mutant plasmid constructs were lysed at 24 h posttransfection, and aliquots were subjected to immunoprecipitation with protein A-agarose without antibodies (X) or rabbit anti-P90 antibodies (GU10) and protein A-agarose. Lysates (L) and immunoprecipitates were analyzed side by side by Western blotting with rabbit GFP or GU10 antibody probes. The positions of migration of P200, P150, and P90 are shown. (D) Undiluted culture medium from 72-h-posttransfection cells (from wt or P1 viruses) was passaged on Vero cells and analyzed for GFP expression and RC (red) production at 48 h postinfection using antibodies for dsRNA. Bars, 100 μm.

Fig 7.

Replicon packaging assay. wt or E36A RUBrep/P150-HA/GFP replicon RNAs (construct diagrams are under the respective micrographs) were transfected into Vero, C-Vero, BHK, or BHK-SP (stably expressing the RUBV structural proteins C-E2-E1) cells and examined microscopically for GFP expression at 2 days posttransfection (d.p.t.). Additionally, culture fluid from BHK transfected cells was undiluted, and that from BHK-SP transfected cells (at 2 days posttransfection) was diluted 10-fold and used independently to infect a fresh monolayer of Vero or C-Vero cells. At 2 days postinfection (d.p.i.), GFP expression was examined microscopically for each infected Vero and C-Vero cell culture. Images were acquired with a ×20 objective and with the same exposure time (1 s).

Fig 1.

NSP localization in Robo502/P150-GFP-infected cells. (A) Diagram of infectious clone Robo502/P150-GFP expressing an in-frame GFP insertion within the P150 gene. The genome is approximately 10 kb and contains a 5′ cap and 3′ poly(A) tail, along with cis-acting elements (CAEs) important for RNA replication and the Q (arginine-rich) domain. The two ORFs within the genome are designated by boxes, while the untranslated regions are designated by lines. (B) Mock-infected Vero cells stained with rabbit anti-calnexin and donkey anti-rabbit Alexa Fluor 594 (red). (C) Vero cells infected with Robo502/P150-GFP (multiplicity of infection, 1) stained at 40 h postinfection (hpi) with rabbit anti-calnexin. (Left) NSPs (green from the GFP tag); (middle) calnexin (red); (right) merge. The yellow arrows point to regions of NSP colocalization with the ER, the light blue arrows point to a region where short NSP fibers overlap with the ER, and the green arrows point to where NSPs have accumulated at the cell edges, while the dark blue arrows point to spherical perinuclear aggregates. Bar, 10 μm. The localization of the NSPs was quantified and is displayed below the infected cell images. (D) z-stack analysis of Robo502/P150-GFP-infected cells. Panels a to g show z slices (right side of each panel) and the corresponding intensity profiles for calnexin (red line) and NSPs (green line) on the left side of the panel, which corresponds to the region denoted by the red arrow (9.5 μm long) in the cytoplasm. Intensity is measured in arbitrary units and is noted to the left in panel a. Black arrows point to profiles of overlap between calnexin and NSPs. The yellow arrow points to extensive calnexin and NSP overlap, and this image in panel e is expanded and split to the right, showing a side-view three-dimensional reconstruction of the infected cells.

Fig 5.

Mapping the targeting domains within P200. P200 with a protease site mutation, called P200* (A), and C-terminal deletion constructs D1* (B) and D2* (C) were expressed via plasmids. The protease (P), helicase (Hel), and RNA-dependent RNA polymerase (RdRp) domains are denoted in each construct diagram. Each construct contains a GFP insert in the P150 gene and an HA epitope tag within P90 between aa 1661 and 1662. Below each construct diagram is shown its expression in representative micrographs of transfected Vero cells (green from GFP). Vero cells were transfected with the P200*, D1*, or D2* plasmids, and after 24 h the cultures were fixed, permeabilized, and then stained with rabbit anti-calnexin and goat anti-rabbit TRITC (tetramethyl rhodamine isocyanate)-conjugate antibodies. Calnexin staining (red) is shown in the left panels; P200*, D1*, or D2* distribution (green from the GFP tag) is shown in the center panels; and merged images are shown in the right panels. Hoechst 33342 was used to stain nuclei. Bars, 10 μm. Blue arrows point to spherical perinuclear aggregates, red arrows point to foci, purple arrows point to a more diffuse phenotype, orange arrows point to small scattered spheres, and aqua arrows point to short fibers. These subcellular distributions are quantified in panel D.

Northern blotting.

Cells grown in 35-mm plates were harvested at 48 h posttransfection and lysed with TRI reagent (Molecular Research Center, Inc.) according to the manufacturer's instructions. Approximately 1/5 of the resuspended RNA pellet was heat denatured (55°C) in ∼50% formamide and then resolved on a 1% agarose gel in TAE (Tris-acetate-EDTA) buffer. The fractioned RNA was then transferred through downward capillary action onto a BrightStar-Plus nylon membrane (Ambion) using NorthernMax transfer buffer (Ambion). After transfer, RNA was cross-linked to the membrane with the optimal setting on a Fisher Scientific FB-UVXL-100 UV cross-linker and then rinsed briefly in water. Membranes were prehybridized with Ultrahyb (Ambion) hybridization buffer at 65°C for 1 h, after which 1 μg of nick-translated Robo502 template labeled for 1.5 h at 15°C in the presence of 70 μCi of [³²P]dCTP (Perkin Elmer) and heat denatured by boiling for 8 min was added. Hybridizations were carried out overnight at 65°C, and membranes were washed twice with Ambion low-stringency buffer (5 to 10 min each wash) and then twice with Ambion high-stringency buffer (15 to 20 min each wash), with each wash performed at 65°C, and were then exposed to Kodak X-ray film. Generally, blots were exposed for 2 to 4 h for C-Vero samples and Vero samples were exposed overnight before film development.

Replicon packaging assay.

Packaging of replicons was performed as described in reference 4, with minor differences. Briefly, Vero, C-Vero, BHK, or BHK-SP cells were transfected with wild-type (wt) or E36A replicons and at 48 h posttransfection were analyzed for GFP production. Cell medium was also collected and used to infect Vero and C-Vero cells at various dilutions. Infected cells were analyzed microscopically at 48 h postinfection for GFP production.

Membrane pelleting assay.

Separation of proteins from transfected cultured cells (at 24 h posttransfection) into soluble and insoluble fractions by high-speed centrifugation (16,000 × g) was carried out as described in reference 30.

RESULTS

Intracellular targeting of RUBV NSPs in infected cells.

The RUBV NSPs are initially distributed in a punctate pattern in the cytoplasm of infected cells (17, 30). A unique NSP fiber network forms late during infection and was the subject of a previous communication (29). The punctate pattern by cells at 40 h postinfection with a recombinant RUBV expressing a GFP-tagged P150 (Fig. 1A; described in reference 29) is illustrated in Fig. 1. A previous study (21) found that RUBV RCs formed in close proximity to the endoplasmic reticulum (ER), and we investigated this potential interaction by staining the ER with calnexin antibodies (Fig. 1). Compared to mock-infected cells, RUBV infection did not change the structure of the ER (Fig. 1B and C), but some of the ER stain was not as bright in the RUBV-infected cells as it was in the mock-infected cells (∼15% of infected cells). The punctate foci formed by the NSPs were present throughout the cytoplasm in approximately 60% of the infected cells (Fig. 1C, bottom) but accumulated in the perinuclear region and along the edges of the cell. As expected, overlap of the NSP foci with the ER was apparent (marked by yellow arrows in Fig. 1C) in many of the infected cells containing NSP foci. Consistent with previous reports (29), ∼10% of the infected cells had large spherical perinuclear aggregates (sPNAs) and ∼30% of the infected cells had fibers. Infected cells with different NSP localization patterns, for the most part, had relatively similar levels of RCs (data not shown).

The infected cells with punctate NSP foci were further analyzed for NSP/ER colocalization by examining z stacks of RUBV-infected cells and quantifying the relative intensities of the ER and NSP signals (Fig. 1D). As shown in Fig. 1D, moving from the top of the infected cell (panel a) to the bottom (panel g), the greatest degree of overlap in the intensity profile for NSPs and the ER appeared to be at the interior region of the cytoplasm (panel e), adjacent to the nucleus. The perinuclear overlap of NSPs with the ER can also be seen from the side-view reconstruction of the infected cell (Fig. 1D, far right).

Requirement of the N-terminal alpha helix in P200 for efficient proteolysis, targeting, and P150-P90 interaction.

We previously reported that proline-insertion mutagenesis of a putative alpha helix between residues 36 and 49 of P200 unexpectedly exhibited long-range effects on P200 by reducing its proteolytic cleavage efficiency and disrupting its subcellular targeting upon plasmid expression (29). To extend these findings, the effects of this mutation (termed P1) on P200 cleavage and subcellular localization upon expression from an in vitro RNA transcript were studied (Fig. 2A and B). Consistent with the plasmid expression findings (29), the P1 mutation in the virus context reduced the proteolytic cleavage efficiency of P200 (Fig. 2A) and disrupted its subcellular localization from the normally cytoplasmic punctate focus pattern seen in wt-transfected cells (Fig. 2B, top) into a nonspecific cytoplasmic localization.

The effect of the P1 mutation on the P150-P90 interaction was assessed by coimmunoprecipitation. As shown in Fig. 2C, when expressed from a plasmid and processed from the wt P200, P150 was coimmunoprecipitated by anti-P90 antibody, while P150 processed from the P200-P1 mutant was not. Thus, disruption of the N-terminal alpha helix in P200 interfered with the interaction between P150 and P90.

Additionally, Robo502/P150-GFP virus containing the P1 mutation was determined to be nonviable due to the lack of CP synthesis by 72 hpt (Fig. 2A). The lack of GFP expression or RC formation following passage of the 72-h-posttransfection fluid onto uninfected cells (Fig. 2D), both of which were evident when a control wt construct was employed (these assays were used to detect infectivity since Robo502/P150-GFP does not reproducibly form distinct plaques), further demonstrated the lethality of the P1 mutation.

Next, a series of alanine-substitution mutants described previously (29) and shown in Fig. 3A was used to test whether any of these residues are important for the interaction between P150 and P90. P150, either wt or one of its mutants, and P90 were coexpressed from different plasmids, followed by immunoprecipitation with anti-P90 antibodies. Except for the E36A and I49A mutants, wt P150 or P150 containing the other alanine-substitution mutations were coimmunoprecipitated with P90 (Fig. 3B). The T42A and Q45A mutants had slightly reduced coimmunoprecipitation efficiency compared to wt. Thus, E36 and I49 play a role in the interaction between P150 and P90. Consistent with our original findings for the mutated Robo502/P150-GFP viruses, only the T42A and Q45A mutants were viable, as determined by the production of RCs (visualized by dsRNA staining) or CP (detected by Western blotting) by 48 or 72 h posttransfection, respectively, along with the undiluted passage of 72-hpt culture fluid on Vero cells that were subsequently monitored for GFP expression and dsRNA production, which occurred by 2 or 3 days postinfection (data not shown).

Fig 3.

Effect of alanine scanning of the N-terminal alpha helix on P150-P90 interaction. (A) Diagram of linear amino acid sequence of the N-terminal alpha helix (between residues 36 and 49 of P150) and alanine-substitution mutations. (B) Vero cells were cotransfected with independent plasmid constructs expressing P150-GFP (wt or one of the alanine-scanning mutants) and P90. At 24 h posttransfection, the cells were lysed and aliquots were immunoprecipitated (IP) with GU10 (anti-P90) antibody, followed by Western blotting (WB) with blots probed with rabbit anti-GFP antibodies to detect coimmunoprecipitated P150 (top). (Middle and bottom) Western blots of aliquots of the total lysate probed with anti-GFP and GU10 antibodies, respectively (the arrow points to the P90 band).

Determinants in P90 required for P150-P90 interaction.

Having established that the N-terminal alpha helix in P150 was a determinant in establishment of the P150-P90 interaction, we used C-terminal deletion mapping to localize determinants of this interaction in P90. To this end, two plasmid-based constructs, P200-D1 and P200-D2, were generated which removed ∼200 and 400 aa from the C terminus of P200, corresponding to the C-terminal half of the RNA-dependent RNA polymerase (RdRp) domain and the complete RdRp domain, respectively (Fig. 4A). In addition to a GFP tag in P150, these constructs also contained an HA tag between the helicase and RdRp domains (between aa 1661 and 1662). The expression of the wt, D1, and D2 constructs by Western blotting is shown in Fig. 4B. In Fig. 4C, coimmunoprecipitation with P150-GFP using GFP antibody was demonstrated for both wt P90 and P90-D1. However, the P150-P90 interaction was lost by the D2 construct. Thus, a determinant in the P150-P90 interaction appears to be present between approximately aa 1700 and 1900, and without the N-terminal domain of the RdRp, P150 and P90 fail to interact.

Fig 4.

Coimmunoprecipitation of P150 and P90 expressed from P200, D1, or D2. (A) Diagrams of P200 deletion constructs. The protease (P), helicase (Hel), and RNA-dependent RNA polymerase (RdRp) domains are denoted. Each construct contains a GFP insert in the P150 gene and an HA epitope tag within P90 between aa 1661 and 1662. (B and C) Vero cells transfected with P200, D1, or D2 constructs were lysed at 24 h posttransfection, and the total lysate was subjected to Western blotting with rabbit GFP (top blot), mouse HA (middle blot), or rabbit GAPDH (bottom blot) antibody probes (B) or subjected to immunoprecipitation (IP) with no antibody (X; protein A-agarose only), HA, or GFP antibodies, followed by Western blotting with mouse anti-HA (C). The arrowheads mark the P90 product of each construct. HC, heavy chain.

Since mutations within the N-terminal alpha helix that abrogate P150-P90 interactions also disrupt NSP targeting, we hypothesized that loss of the putative P150 binding site in P90 by C-terminal deletion of P200 should also disrupt targeting. Therefore, the effects of the C-terminal deletions on the localization of P200 were investigated using plasmid-based constructs that have a mutated protease cleavage site (designated by an asterisk). This mutation of 1299-GGG-1301 to 1299-VVV-1301 makes analysis of the uncleaved precursor possible, since without the mutation it would rapidly be processed (22, 24, 30). Upon expression (Fig. 5A), P200* formed extended perinuclear aggregates in approximately half of the transfected cells, which we attributed to being an artifact of overexpression. Approximately 25% of the transfected cells displayed P200* localizing into foci similar in shape and size to the NSP structures observed during the early stages of RUBV infection. Also similar to that observed for NSPs during infection, approximately half the cells with foci showed overlap of the foci with the ER. Consistent with the infected cell NSP phenotype (Fig. 1C, dark blue arrow) was the appearance of spherical perinuclear aggregates (sPNAs) of P200* in approximately 10% of the transfected cells. Apparent in about 10% of the transfected cells were small scattered cytoplasmic spheres of P200*. The percentage of cells with each of these patterns of localization is shown in Fig. 5D.

Upon expression of P200-D1*, only about 10% of the transfected cells produced punctate foci, and about 60% of these transfected cells exhibited the D1* product as a diffuse pattern but still retained overlap with the ER (Fig. 5B). Furthermore, D1* aggregated less than P200*. Figure 5C shows that expression of D2* resulted in the formation of a similar number of foci and aggregates as expression of D1*; however, like P200*, there were no obvious diffuse patterns. In contrast to both P200* and D1*, about 25% of the D2*-transfected cells appeared to have short fiber-like structures, distinct from the foci and spheres, in their cytoplasm. Also, more spheres were observed in the cytoplasm of D2*-transfected cells than in the cytoplasm of those expressing P200*. Although the localization of P200 changed when the RdRp domain was partially or completely deleted (D1 and D2), overlap with the ER was never completely absent for the deletion mutants. In fact, expression of aa 1 to 461 of P200, the membrane binding domain (30), also resulted in accumulation with the ER signal in most transfected cells (data not shown). Thus, localization and possibly three-dimensional accumulation of the precursor, but not apparent membrane association with the ER, were affected by deletion of the RdRp domain.

A role of CP in NSP maturation.

The RUBV CP rescues lethal internal deletions of the Q domain in P150 (42) at an early stage of RUBV replication (i.e., before the generation of RCs or accumulation of viral RNAs), suggesting that it may influence a step in P200 maturation. In fact, the Q domain can be functionally replaced by CP insertion (43). The series of Robo502 constructs harboring the alanine-scanning substitutions between aa 36 and 49 made ideal constructs to test the effect of CP on P200 maturation, since most of these mutations (with the exceptions of T42A and Q45A) were lethal by the criterion that the mutants failed to synthesize detectable CP by 72 h posttransfection (data not shown). To this end, we tested the mutant series in C-Vero cells, a line that stably expresses CP and that we used in previous studies (4, 28, 45). The HA-tagged P150 version of Robo502 (Robo502/P150-HA) was used in these studies because it replicates more robustly than the GFP version and forms distinct plaques, likely due to the smaller degree of disruption caused by insertion of the HA epitope rather than GFP into the Q domain.

First, lysates from transfected Vero and C-Vero cells at 48 h posttransfection were examined by Northern blotting to assay virus-specific RNA accumulation. In Vero cells, only the wt, T42A, Q45A, and R47A transcripts produced detectable levels of genomic RNA (Fig. 6A); however, in C-Vero cells, wt and all of the mutants except the K46A and I49A mutants produced detectable amounts of genomic RNA, indicating rescue of E36A, R38A, and D39A by CP. When quantified by densitometry (shown in arbitrary units above each Northern blot), the amount of genomic RNA accumulated in Vero cells by the T42A and Q45A constructs was approximately twice as high as that for wt. In C-Vero cells, wt, E36A, R38A, D39A, T42A, and R47A genomic RNA levels were roughly equivalent, while the amount produced by the Q45A construct was ∼2.5 times higher than that of wt. The subgenomic RNA pattern (data not shown) for the series was consistent with the corresponding genomic RNA pattern, and thus, no effect on subgenomic RNA synthesis by the alanine substitutions could be observed for the replicating viruses. It is worth noting that the Northern blot from Vero cells was exposed approximately five times longer than the C-Vero blot, and thus, there was a large increase in RNA synthesis by all of the constructs in the presence of CP, including the wt, as we have previously reported (45).

Fig 6.

Capsid-mediated RC formation and RNA synthesis enhancement. Vero or C-Vero cells transfected with in vitro transcripts of wild-type Robo502/P150-HA (expressing an HA-tagged P150) with sequences from either wild type or a mutant with one of the alanine-scanning mutations. (A) Northern blots probed for RUBV genomic RNA (gRNA) with a ³²P-labeled nick-translated Robo502 plasmid. The Vero cell blot was exposed to film overnight, while the C-Vero cell blot was exposed 2 to 3 h before film development. Arbitrary densitometry units measured by an ImageQuant imager are shown above each blot, while the 28S rRNA band (stained by ethidium bromide and visualized with UV light) from each sample is shown as a loading control. (B) At 48 h posttransfection, the number of dsRNA-positive cells per field of microscopic view (magnification, ×20) was quantified and averaged for Vero and C-Vero transfected cells (numbers are averaged from two experiments, with counts taken from approximately 5 fields of view per experiment). dsRNA foci were detected by immunofluorescence microscopy using anti-dsRNA. (C) The number of log₁₀ PFU/ml was determined for each construct in both cell lines by plaque assay using culture fluid harvested from cells in experiments from panel A at 48 h postinfection.

Second, the number of Vero and C-Vero cells with RCs after 48 h posttransfection with one of the wt or mutant viral RNAs was determined by staining for dsRNA. As shown in Fig. 6B, there was an increase in the number of dsRNA-positive C-Vero cells over Vero cells for each of the constructs, including the wt, in whose case the increase was roughly 5-fold. Thus, the increase in the amount of virus RNA exhibited on Northern blots in both this and previous studies is due to the number of cells successfully producing RNA. By cursory inspection (counting the number of RCs in a single cell from Vero or C-Vero cells, n = 2 from each, ∼900 RCs/cell), there did not appear to be an increase in the number of RCs in C-Vero versus Vero cells.

Finally, the yield of PFU from transfected Vero and C-Vero cells was determined by plaque assay (Fig. 6C). Interestingly, several of the alanine-scanning mutations (E36A, R38A, D39A, and R47A) that were judged to be lethal in the P150-GFP virus by production of CP at 72 hpt yielded PFU following transfection of Vero cells with transcripts from the Robo502/P150-HA versions. However, the level of PFU production by these mutants was 2 to 3 log units lower than that by wt, likely explaining why CP production by them was not detectable. Surprisingly, equal or fewer numbers of PFU were produced by transfected C-Vero than Vero cells. In the case of the wt, the difference was 10-fold less in C-Vero versus Vero cells. We speculate that the CP expressed in C-Vero cells, which lacks the C-terminal 23 aa that form the signal sequence for E2 and are necessary for proper association of CP with the glycoproteins during virion morphogenesis (19), interferes with production of virus particles.

We recently showed that CP in the infecting virus particle can rescue a mutation in the Q domain (4). To see if this was also the case with the N-terminal alpha-helix mutants, the E36A mutant was introduced into the RUBrep/P150-HA/GFP replicon. In this construct, GFP is expressed from the subgenomic RNA, and thus, GFP intensity is a measure of RNA synthesis. The wt and E36A replicon transcripts were transfected into BHK or BHK/CE2E1 cells, the latter of which is a line expressing the RUBV SP ORF that allows packaging of the replicon. As shown in Fig. 7, the expression of GFP by the wt replicon is enhanced, and the E36A replicon is rescued in the BHK/CE2E1 cells in comparison to BHK cells, due to the expression of the CP in the former cell line. The same differences in GFP intensity were observed for Vero versus C-Vero cells independently transfected with the two different replicons. However, when 2-day-posttransfection packaged replicons were used to infect Vero or C-Vero cells, no difference in GFP expression was observed in those cells at 2 days postinfection, indicating that CP in the incoming packaged replicon particle can enhance wt replicon replication and rescue E36A replicons, similar to what was previously observed with the Q-domain mutant (4).

Mechanism of CP in P200 maturation.

To test the hypothesis that CP plays a role in P200 maturation and to gain insight into the possible mechanism, we tested parameters of P200 maturation affected by the N-terminal alpha-helix mutations, namely, P200 processing, P150-P90 interaction, and NSP targeting. First, these parameters were tested comparatively in Vero and C-Vero cells using P200 constructs with the wt, E36A, and I49A sequences, with the last two representing mutations that are rescued and not rescued by CP, respectively. As shown in Fig. 8A, processing of plasmid-expressed P200 was less efficient for both the E36A and I49A mutants than the wt but showed no differences in Vero and C-Vero cells. Second, the interaction of coexpressed P150 and P90 was detected by coimmunoprecipitation for the wt P150 construct and even appeared to be more efficient in C-Vero than in Vero cells (Fig. 8B). However, the P150-P90 interaction between the E36A or I49A mutant P150 and P90 was not detectable in either Vero or C-Vero cells.

Fig 8.

Biochemical analysis of CP effects on P150-P90 interaction and P200 processing, membrane attachment, and localization. (A) Processing of P200 (wt or E36A or I49A mutant) during expression in Vero and C-Vero cells. At 24 h posttransfection, lysates were analyzed by Western blotting against GFP (P200 and P150) and calnexin. The ratio of plasmid-expressed P150-P200 was determined by densitometry, and the values are shown below each respective lane. The blot and corresponding values represent three independent experiments. (B) Immunoprecipitation of P150 (wt or E36A or I49A mutant) by P90-HA using HA antibodies after expression in Vero and C-Vero cells. At 24 h posttransfection (P150 and P90 were coexpressed from different plasmids), lysates were analyzed by Western blotting using antibodies against calnexin, GFP (P150), HA (P90), and FLAG (CP) or immunoprecipitated by anti-HA, followed by Western blotting against GFP and HA. (C) Membrane fractionation of P200 (wt or E36A or I49A mutant) after plasmid expression in Vero and C-Vero cells. At 24 h posttransfection, cells were broken by 20 passages through a 26.5-gauge needle, and the postnuclear supernatant was separated into insoluble (pellet) and soluble (S) fractions by centrifugation at 16,000 × g. Western blotting was used to assess fractionation; calnexin (insoluble control), GAPDH (soluble control), and GFP (P200 and P150) antibody probes were used. (D) Analysis of P200* in the presence or absence of CP. The phenotypes of P200* in transfected cells (Vero or C-Vero) were counted by microscopic observation at 24 h posttransfection from 3 different experiments, with approximately 10 fields of view examined from each experiment. Bars represent standard deviation from the mean.

We also examined the membrane association of P200 and P150 using a membrane pelleting assay (30). As shown in Fig. 8C, P200 and P150 fractioned primarily with membranes, with no differences in fractionation of P200 or P150 observed between Vero and C-Vero cells. For both the E36A and I49A P200 constructs, membrane accumulation of both P200 and P150 was reduced, as shown by the roughly equivalent levels of the E36A or I49A mutant proteins in the membrane and soluble fractions, regardless of whether they were expressed in Vero or C-Vero cells (Fig. 8C). Thus, CP did not seem to affect the membrane association of P200 and P150 or the E36A or I49A mutants.

Finally, compared to Vero cells, expression of P200* in C-Vero cells caused a reduction in P200* aggregation while concomitantly increasing the amount of foci and diffuse patterns (Fig. 8D). The increase in the amount of properly targeted P200* in the presence of CP provides evidence that CP plays a role in P200 maturation. On the other hand, the localization of P200-E36A and -I49A, which do not produce wt localization patterns (29), was unaffected when expressed in C-Vero cells (data not shown).

DISCUSSION

It has been well documented (18, 26, 31–33, 35, 37, 40) that the RCs of positive-strand RNA viruses form associations with cellular membranes through protein-membrane contacts. While the localization of RUBV RCs was initially reported to be on the membranes of endocytic vacuoles (20, 27), we have detected RCs at other intracellular locations, including the perinuclear matrix and the plasma membrane (30). Recent studies with the alphaviruses have indicated that RCs initially form on the plasma membrane and are subsequently endocytosed, leading to the formation of the characteristic intracellular structures (termed cytopathic vacuoles) that consist of apparent endocytotic vacuoles lined internally by RCs (11, 13). Some RNA viruses, including poliovirus (39), a picornavirus, and hepatitis C virus (HCV) (8), a flavivirus, establish RCs on the surface of ER membranes. The pathway by which the RUBV NSPs become associated with membranes is poorly understood, but it has been established that the NSPs are associated with membranes by a domain within the N-terminal region of P150 (30). We hypothesize that after genome translation to P200, the NSPs become membrane associated in foci at the ER prior to migration to the plasma membrane. Further, sometime after the membrane-bound complexes migrate to the plasma membrane, RNA synthesis proceeds, producing dsRNA detectable upon accumulation at the plasma membrane. It is likely that the events at the plasma membrane for RUBV RC internalization are similar to those for SINV, where these structures are endocytosed and form the traditional modified endo-/lysosome structures originally observed in RUBV-infected cells (27). As RUBV infection proceeds past 36 h, P150 accumulates in membrane compartments that are resistant to nonionic detergent extraction (30). Additionally, P150 also assembles into long, cytoplasmic fibers with membranous characteristics (17, 30). Short P150 fibers in association with the ER were observed in this study, suggesting that ER membranes are incorporated into P150 fibers, creating the membranous fibers observed at the electron microscopic level during RUBV infection (17).

A long alpha helix at the N terminus of P200 is critically important in NSP maturation, as its interruption by proline-insertion mutagenesis compromises proteolytic cleavage of P200, the association of P150 and P90, and the intracellular localization of the NSPs. Analogously, brome mosaic virus uses an alpha helix in its 1a protein to control multiple roles in targeting and RC assembly, including ER association (25). Further, the hepatitis C virus nonstructural protein NS4B contains an alpha helix that controls targeting and membrane binding to the ER (7, 14), whereupon it oligomerizes and modifies the cellular membranes prior to RC formation. As mentioned previously, the RUBV NSP membrane binding domain resides within P150 between aa 101 and 461, but this domain does not bind membranes without the region from aa 1 to 100 that contains the alpha helix (30). The alpha helix likely imparts a close-range structural effect on membrane binding, but in contrast, as in the case of both P200 cleavage and the P150-P90 interaction, the N-terminal alpha helix is involved in a long-range interaction. It is possible that the N-terminal alpha helix is the binding domain on P150 that directly interacts with a putative binding domain within P90, an interaction that would structurally juxtapose the RdRp in P90 with the methyl-/guanylyltransferase activity predicted to reside at the N terminus of P150. However, this alpha helix may be analogous to the alpha helix in protein 1a of brome mosaic virus that modulates binding of proteins 1a (the methyl-/guanylyltransferase and helicase) and 2a (the RdRp) without itself serving as a binding domain (1, 25).

Interestingly, alanine-scanning mutagenesis revealed that only the residues on the ends of the alpha helix (E36 and I49) abrogated P150-P90 binding, compromised the efficiency of P200 cleavage, and altered subcellular localization of the NSPs, although several, but not all, of the mutations within the alpha helix were lethal. It is possible that mutagenesis of these residues individually does not disrupt the P150-P90 interaction, as detected in a coimmunoprecipitation assay, but does interrupt the interaction in a more subtle, functional fashion (helical wheel projections were not illustrative in this regard) (29). Certainly, an intricate balance exists for P150-P90 interactions in regard to RNA synthesis, considering that T42A and Q45A mutants had weakened P150-P90 interactions but produced more genomic RNA than wt, whereas E36A and I49A mutants had no detectable P150-P90 interactions and suffered detrimental defects in RNA synthesis. Regardless, all of the mutants tested produced virus titers lower than those of wt, ultimately demonstrating the importance of this N-terminal alpha helix in multiple steps during RUBV production.

The correlation of mutations in the N-terminal alpha helix that abrogate P150-P90 binding and interfere with P200 processing suggests that P150-P90 binding is necessary for efficient proteolysis. However, the correlation was not absolute since the P1 insertion, E36A, and I49A mutations in the N-terminal alpha helix diminished the effectiveness of protease cleavage, but the D2 C-terminal deletion mutation of P90 had no effect on protease efficiency. It has been shown with the alphaviruses that acquisition of three-dimensional conformations of the NSPs within the context of the precursor is necessary for proper localization (36). Given many of the similarities between RUBV and alphaviruses, interference with conformational acquisition, e.g., lack of P150-P90 binding, could explain the effect that the N-terminal alpha-helix mutations have on RUBV NSP localization. However, it cannot be ruled out that some other function necessary for proper localization may have also been affected by the mutagenesis.

Through C-terminal deletion mutagenesis of a cleavable P200 construct, we were able to localize a domain between aa 1700 and 1900, a region comprising the N-terminal half of the RdRp necessary for P150-P90 binding. While it is recognized that large-scale deletions can have large effects on secondary structure, this type of mapping was successfully used to identify three alpha helices (one of them is the focus of the current study) in P150 that contribute to fiber formation (29, 30) and the ΔNotI domain (42, 43). In both cases, finer-tuned mapping has been conducted to reveal valuable insight into those domain functions, and the current study aids in extending those findings. We are currently mapping aa 1700 to 1900 more precisely.

The profound effects on the subcellular localization of an uncleavable P200 when portions of RdRp were deleted (leaving the putative P150 binding domain [D1] or eliminating it [D2]) suggest that P200 in its entirety, likely through multiple conformations, controls localization. The ability of these interactions between regions of P200 to drive localization was exemplified by expression of aa 1 to 461, which stimulated extensions of the plasma membrane, a process that was accentuated by coexpression with P90 but suppressed by coexpression with P150 (data not shown). The complexity of these different localization phenotypes exhibited by deletion, mutation, and coexpression testifies to the multifunctional properties of the different domains in these large viral proteins that must have multirole functions during infection, given the limited coding capacity of RNA viruses.

Finally, we obtained evidence that the RUBV CP assists in maturation of the NSPs. The CP not only rescues a diversity of lethal mutations, including a deletion of the Q domain in P150 and mutations in the 5′ and 3′ cis-acting elements of the genomic RNA, but it also potentiates wt RNA synthesis. In this study, we extended the panel of rescued mutations to some of the alanine-scanning mutations within the N-terminal alpha helix. While in previous studies the effect of CP on wt potentiation/mutant rescue had been assayed by Northern blotting and GFP expression of replicon constructs (45), in this study, we also analyzed the number of cells with RCs in the presence or absence of CP. With this assay, it was observable that the potentiation of wt constructs was due in part to an increase in the number of cells with RCs (Fig. 6A), which correlates with the results of the GFP expression assay (see the difference in GFP expression between both the Vero and BHK cells with and without CP in Fig. 7). The latter assay indicates that transfected cells synthesize more RNA in the presence of CP, as reflected in the increase in GFP intensity. However, it is possible that CP may facilitate the translation of subgenomic RNA, considering its multiple roles during infection (15).

Recently, we showed that CP in the virion could rescue a replicon with a lethal mutation in the Q domain (4). In this study, we found that CP rescued a replicon bearing the E36A mutation. The implication of this finding is that a limited number of CP molecules can elicit the rescue effect. Given the diversity of mutations rescued by CP, we hypothesize that CP plays a chaperone-like function that assists in proper maturation of wt P200 and restores proper maturation of mutant P200. In this model, the N-terminal alpha helix, the Q domain, and the viral RNA all play a role in P200 maturation and establishment of RCs. There is precedence for the involvement of the viral RNA in the RC maturation process. For example, it has been shown that the viral RNA along with the NSPs is necessary for formation of RCs in Semliki Forest virus-infected cells (38), and this is also the case for nodaviruses (16). In testing our hypothesis that CP chaperones P200, we examined the steps in P200 maturation that we had been studying in this and previous studies (29, 30), finding no difference in processing efficiency, P150-P90 interaction, or membrane association by P200 or P150 in the presence or absence of CP. However, CP exerted a marked effect on intracellular localization of expressed P200, reducing the number of cells with aggregates and increasing the number of cells with a diffuse or focal distribution, the latter of which at least resembles the distribution of NSPs in infected cells. This finding explains the increase in the number of cells forming RCs in the presence of CP, since CP properly positions P200 for RC assembly, and suggests that our CP chaperone hypothesis is correct. CP from the incoming virion may also play a role in binding genomic RNA on the ribosome and regulating translation of the genomic RNA by diverting the genomic RNA-P200-CP complex toward RC maturation.