Binding of cellular p32 protein to the rubella virus P150 replicase protein via PxxPxR motifs

Abstract

A proline-rich region (PRR) within the rubella virus (RUBV) P150 replicase protein that contains three SH3 domain-binding motifs (PxxPxR) was investigated for its ability to bind cell proteins. Pull-down experiments using a glutathione S-transferase–PRR fusion revealed PxxPxR motif-specific binding with human p32 protein (gC1qR), which could be mediated by either of the first two motifs. This finding was of interest because p32 protein also binds to the RUBV capsid protein. Binding of p32 to P150 was confirmed and was abolished by mutation of the first two motifs. When mutations in the first two motifs were introduced into a RUBV cDNA infectious clone, virus replication was significantly impaired. However, virus RNA synthesis was found to be unaffected, and subsequent immunofluorescence analysis of RUBV-infected cells revealed co-localization of p32 and P150 but little overlap of p32 with RNA replication complexes, indicating that p32 does not participate directly in virus RNA synthesis. Thus, the role of p32 in RUBV replication remains unresolved.

Introduction

Rubella virus (RUBV) is the sole member of the genus Rubivirus in the family Togaviridae and is the aetiological agent of a series of birth defects termed congenital rubella syndrome (Frey, 1994). RUBV is a positive-sense, ssRNA virus of approximately 9.7 kb and encodes two ORFs (Frey, 1994). The first ORF, the non-structural protein ORF (NS-ORF), is translated to produce a non-structural precursor protein, P200. P200 is cleaved post-translationally by an embedded protease into the mature non-structural proteins P150 and P90. The second ORF, the structural protein ORF (SP-ORF), is translated from a subgenomic RNA and encodes the virion structural proteins, capsid protein (CP) and envelope glycoproteins E1 and E2. The non-structural proteins catalyse virus RNA synthesis, which occurs in membrane-bound replication complexes (RCs) in infected cells. P150 contains the protease domain that mediates this cleavage, as well as the predicted methyl/guanylyl transferase, the Q domain of undefined function and the X domain, which is an ADP–ribose-binding domain. P90 contains helicase and RNA-dependent RNA polymerase domains. The Q domain has been defined as a region of P150 that, when deleted, results in a non-viable mutant that, remarkably, can be rescued by the viral CP (Tzeng & Frey, 2003, 2009; Tzeng et al., 2006). Notably, the Q domain within P150 has a proline-rich region (PRR) stretching from residue 716 to 782 of the protein (which contains 1301 aa in total) (Koonin et al., 1992). Using bioinformatics tools, we found that the PRR contains three motifs that fulfil the criteria for the class II binding ligand involving SH3 domains, i.e. PxxPxR.

Proteins with SH3 domains often bind to ligands containing PxxP motifs in their amino acid sequence (Zarrinpar et al., 2003). SH3 domains are 50–70 aa and are usually present in proteins in eukaryotic cells involved in signal transduction (Kay et al., 2000). In addition to this, SH3 domains in eukaryotes are thought to play a role in directing cell compartmentalization (Bar-Sagi et al., 1993; Koch et al., 1991). Viral proteins that contain PxxP consensus sequences have been shown to bind host proteins with SH3 domains, examples of which include human immunodeficiency virus (HIV) Nef protein, bovine leukemia virus gp30 transmembrane protein, influenza A virus NS1 protein and hepatitis C virus NS5A protein (Aiken & Trono, 1995; Ehrhardt et al., 2007; Linnemann et al., 2002; Reichert et al., 2001; Saksela et al., 1995). These interactions have been shown to be important for virus replication, as well as for pathogenicity (Aiken & Trono, 1995; Ehrhardt et al., 2007; Linnemann et al., 2002; Reichert et al., 2001).

The goal of this study was to investigate whether the putative PxxPxR SH3-binding domains in the RUBV P150 PRR are of significance in the virus replication cycle, putatively via their ability to interact with cell proteins. To accomplish this, we performed mutagenesis and proteomics analyses to identify candidate host proteins that bound to the PRR through these SH3-binding domains. We identified cellular p32 protein (also known as gC1qR) as one of the cell proteins that binds to the PRR, specifically through either of the first two of the three PxxPxR motifs. Although this protein has no reported SH3 domains, this finding was intriguing, as p32 has already been implicated in the RUBV life cycle through its interaction with RUBV CP (Beatch & Hobman, 2000; Beatch et al., 2005; Claus et al., 2011; Fontana et al., 2007; Ilkow et al., 2010; Mohan et al., 2002). Mutagenesis of the first two PxxPxR motifs did not have an effect on virus RNA synthesis and, although p32 was found to co-localize with P150 in infected cells, it was mostly with the population of P150 not associated with RCs. Thus, the binding of p32 to P150 appears to play a role outside of viral RNA synthesis.

Results

Identification of class II SH3-binding domains in the Q domain of RUBV P150

Fig. 1(a) shows a diagram of the domains within RUBV P150, including the PRR within the Q domain. The PRR is located between residues 716 and 782 of P150. This region is also known as the hypervariable region, as it is the most variable region of the RUBV genome among natural RUBV isolates (Zhou et al., 2007). Although this region contains many core PxxP motifs, analysis of the PRR using ScanProsite software (on the Expasy Proteomics Server) led to the identification of three putative class II SH3-binding motifs with the consensus amino acid sequence PxxPxR, in which ‘x’ denotes any amino acid. These putative SH3-binding motifs are referred to as motifs 1, 2 and 3, from the N to the C terminus of the PRR, respectively. An alignment of sequences of the PRR from viruses of seven genotypes (Zhou et al., 2007) showed that all three motifs were conserved (data not shown). In a preliminary scan in which proline-to-alanine mutations (PxxPxR→AxxAxR) were introduced into each of these motifs independently in the Robo502 infectious cDNA clone (Tzeng & Frey, 2003) (these mutations were termed Mut1, Mut2 and Mut3, respectively; see Methods), it was found that in transfected Vero cells none of the mutations was lethal but that the Mut1 and Mut2 mutants reverted to wild type (wt) within one passage, whilst Mut3 did not. Thus, we concentrated on motifs 1 and 2 for the remainder of the study.

Fig. 1.

Identification of cell proteins interacting with the P150 PRR. (a) RUBV genome organization and location and sequence of the PRR. At the top is shown a diagram of the RUBV genome with the NS-ORF and SP-ORF shown as boxes containing the proteins that each encodes. An expanded version of P150 shows its four defined domains: MT, methyl/guanylyl transferase; Q, Q domain; X, X domain (ADP–ribose-binding domain); and P, protease. Below the P150 diagram is the sequence of the PRR, with its three predicted class II SH3-binding domains (underlined) and a poly-arginine cluster (shaded), mutation of which is lethal but which can be rescued by CP. Amino acid coordinates of P150, the domains within P150 and the PRR containing the PxxPxR motifs are shown. (b) GST–PRR pull-down of cell proteins. The P150 PRR and its mutant with motifs 1 and 2, both mutated from PxxPxR to AxxAxR, were cloned into the pEBG mammalian expression vector to produce GST–PRR/wt or GST–PRR/Mut1+2 fusion proteins, respectively, when expressed in transfected HEK293T cells. At 3 days post-transfection, the cells were lysed and the GST fusion proteins were captured on gluthathione–Sepharose 4B beads. Bound proteins were eluted and resolved by 6 (left gel) or 10 % (right gel) SDS-PAGE. Proteins that bound to the wt fusion protein but not to the mutant fusion protein (PRR1–PRR4) were excised and identified by trypsin digestion and mass spectroscopy. The determined identities and sizes of each of these proteins are indicated. (c) Confirmation of p32 binding to the PRR. HEK293T cells were transfected with the empty pEBG vector, GST–PRR/wt or GST–PRR/Mut1+2. At 3 days post-transfection, the cells were lysed and the GST or GST–PRR proteins and interacting cell proteins were isolated on glutathione–Sepharose 4B and resolved by 10 % SDS-PAGE followed by Western blotting with an anti-p32 antibody. The position of migration of p32 is indicated. (d) Determination of the motif to which p32 binds. Vero cells were mock transfected (lane 1) or transfected with a plasmid vector expressing an HA-tagged P150 Q domain (HA–P150Q) containing wt (lane 2), Mut1 (lane 3), Mut2 (lane 4) or Mut1+2 (lane 5) sequences. On day 1 post-transfection, the cells were lysed and the lysates immunoprecipitated (IP) with anti-p32 antibody. Following resolution by 10 % SDS-PAGE, Western blot (WB) analysis was performed with an anti-HA antibody. The upper panel (lysate) shows that all HA–P150Q constructs were expressed at similar levels, whilst the lower panel (immunoprecipitation with anti-p32 antibody) shows that wt, Mut1 and Mut2 HA–P150Q were co-immunoprecipitated with p32, whilst Mut1+2 HA–P150Q was not.

Identification of host cellular factors that bind to the P150 PRR

We next investigated host proteins that potentially bind to the PRR within P150. To this end, we made a glutathione S-transferase (GST) fusion expression construct in which GST was fused to the PRR (GST–PRR). Another construct containing mutations in motifs 1 and 2, GST–PRR/Mut1+2, was also constructed. Following transfection of human embryonic kidney (HEK) 293T cells, proteins binding to the wt GST–PRR/wt, but not the GST–PRR/Mut1+2, were considered as candidate binding partners. As shown in Fig. 1(b), a total of four putative host proteins were identified, termed PRR-1, -2, -3 and -4. These were identified as human Mdn1 protein, Gcn1L protein, nucleopore complex protein Nup205 and p32 protein, respectively. More than two peptides were identified at 95 % confidence for all of these proteins. Further characterization was confined to p32, as this protein was shown previously to bind to RUBV CP (Beatch & Hobman, 2000; Mohan et al., 2002). As shown in Fig. 1(c), a Western blot probed with anti-p32 antibody confirmed the binding of p32 to GST–PRR/wt but not GST–PRR/Mut1+2 or the GST control. To determine whether both PxxPxR motifs 1 and 2 were required for p32 binding, or if either one would suffice, plasmid constructs were generated that expressed a haemagglutinin (HA)-tagged Q domain with wt, Mut1, Mut2 or Mut1+2 sequences. As shown in Fig. 1(d), when transfected into Vero cells followed by lysis and immunoprecipitation with anti-p32 antibody, the wt, Mut1 and Mut2 constructs co-immunoprecipitated with p32 whilst the Mut1+2 construct did not, indicating that p32 could bind to either motif. Interestingly, these motifs within the Q domain of P150 overlap a poly-arginine motif, mutation of which is lethal but which can be rescued by CP (Tzeng & Frey, 2009). As CP also binds p32 through an arginine-rich domain (Beatch et al., 2005), to confirm that the binding of p32 by P150 was specific to the SH3-binding domains, plasmid constructs that expressed either an HA-tagged Q domain or an HA-tagged Q domain with the arginines in the poly-arginine cluster mutated to glutamine were generated. When transfected into Vero cells followed by lysis and immunoprecipitation with anti-p32 antibody, both constructs co-precipitated with p32 (data not shown), confirming that P150–p32 binding is mediated through the prolines within the PRR rather than through the poly-arginine cluster.

It has been reported that isolated PxxP domains can bind to SH3 domains with a higher affinity than to the full-length protein (Nguyen et al., 1998, 2000). To confirm that p32 bound to P150, a plasmid construct expressing an HA-tagged P150 was used to transfect Vero cells. As shown in Fig. 2(a), the expressed HA-tagged P150 was immunoprecipitated by anti-HA (positive control) and anti-p32 antibody, demonstrating that expressed P150 interacts with native p32. Thus, p32 is an authentic binding partner of P150. Cells were next co-transfected with plasmids independently expressing HA-tagged P150 and c-myc-tagged simian p32. As shown in Fig. 2(b and c), reciprocal co-immunoprecipitation was observed. Similar results were obtained using a Robo502 construct expressing a FLAG-tagged P150 in infected cells transfected with a plasmid construct expressing a tagged simian p32 (data not shown). When the co-transfection was carried out using an HA-tagged P150 with mutations in both PxxPxR motifs (Mut1+2), c-myc antibody did not co-precipitate the Mut1+2 HA-tagged P150 (Fig. 2d), showing that P150–p32 binding is mediated by these motifs within the PRR.

Fig. 2.

Co-immunoprecipitation of P150 and p32. (a) Vero cells were mock transfected or transfected with a plasmid vector expressing an HA-tagged P150 (CMV-P150–HA) as indicated. At 2 days post-transfection, the cells were lysed and the lysates immunoprecipitated (IP) with anti-HA or anti-p32 antibody followed by 6 % SDS-PAGE and Western blotting (WB) with an anti-HA antibody to detect HA-tagged P150. Expressed P150 was precipitated by both the control anti-HA antibody and the anti-p32 antibody and thus interacts with native p32. (b, c) Vero cells were either mock transfected or co-transfected with CMV-P150-HA and pcDNA-p32-c-myc as indicated. At 2 days post-transfection, the cells were lysed and the lysates were immunoprecipitated with anti-HA or anti-c-myc antibody. Following resolution by 6 and 10 % SDS-PAGE (b and c, respectively), Western blot analysis was performed using anti-HA (b) or anti-c-myc (c) antibodies. The expressed HA-tagged P150 and c-myc-tagged p32 were reciprocally co-immunoprecipitated. (d) Vero cells were co-transfected with CMV-P150-HA and p32-c-myc or co-transfected with CMV-P150-HA/Mut1+2 and pcDNA-p32-c-myc as indicated. At 2 days post-transfection, the cells were lysed and the lysates immunoprecipitated with anti-HA or anti-c-myc antibody. Following resolution by 6 % SDS-PAGE, Western blot analysis was performed with an anti-HA antibody. Expressed wt P150 (*) was co-immunoprecipitated with p32, but expressed Mut1+2 P150 was not.

Effect of PxxPxR mutations on virus phenotype

The motif 1 and 2 mutations were introduced into Robo502, a construct termed Robo502/Mut1+2. Titres obtained following transfection of in vitro Robo502/Mut1+2 transcripts into Vero cells were 3×10⁴ p.f.u. ml⁻¹ compared with 2×10⁶ p.f.u. ml⁻¹ produced by wt Robo502 transcripts. After one passage of the transfected culture fluids at an m.o.i. of 1, the Mut1+2 titre produced was 5×10² p.f.u. ml⁻¹ compared with 6×10⁶ p.f.u. ml⁻¹ for wt. When the passage 1 virus was sequenced, the mutations in both motifs were found to have been retained.

We next investigated the RNA phenotype of viruses bearing the PxxPxR mutations. Vero cells were transfected with transcripts from Robo502/Mut1+2 as well as wt Robo502. However, as can be seen in Fig. 3(a) the mutations had no effect on RNA accumulation. As CP binds both p32 and P150, the experiment was repeated by introducing Mut1+2 mutations into RUBrep/GFP, a replicon in which the structural protein ORF, including the CP gene, was replaced by GFP. In this experiment, replicon RNA replication was measured by GFP expression, which is dependent on synthesis of the subgenomic RNA (Tzeng et al., 2001). However, no differences were detected between the wt and mutant replicon in transfected Vero cells (Fig. 3b), and thus the presence of CP did not lead to rescue of the PxxPxR mutations.

Fig. 3.

Phenotype of wt and mutant RNA synthesis and replicon replication. (a) Vero cells were transfected with wt Robo502 and Robo502/Mut1+2 transcripts. At 3 days post-transfection, the cells were lysed and the viral RNAs analysed by agarose gel electrophoresis and Northern blotting. (b) Transcripts from two replicon constructs (infectious cDNA constructs with the SP-ORF replaced by the reporter gene GFP), wt RUBrep/GFP and RUBrep/GFP-Mut1+2, were used to transfect Vero cells, and GFP expression was recorded on day 3 post-transfection by fluorescence microscopy. As GFP is expressed from the subgenomic RNA, its intensity is a measure of replicon replication, i.e. RNA synthesis (Tzeng et al., 2001).

Considering the lack of effect of the PxxPxR mutations on RNA accumulation by the replicon, we next analysed the distribution of P150, p32, CP and RCs in RUBV-infected Vero cells by immunofluorescence. As shown in Fig. 4(a), in mock-infected cells the p32 signal was strictly cytoplasmic and indicative of mitochondrial localization (i.e. rod-like), indicating that most of the p32 was localized to mitochondria. In cells infected with Robo502/P150–HA and co-stained for CP and p32 (Fig. 4b, upper panels) or RCs (visualized by staining for dsRNA) and p32 (Fig. 4b, lower panels), an extensive overlap was apparent between CP in perinuclear clusters, but little overlap between RCs and p32 could be detected. The p32 signal appeared to surround the RCs more than overlap them (Fig. 4b, lower panels). Cells infected with Robo502/P150–GFP, a virus expressing a GFP-tagged P150, and stained for RCs and p32 revealed that P150 overlapped the RCs (Fig. 4c, yellow arrows), although some P150 was localized separately from RCs (Fig. 4c, white arrow), as reported previously (Matthews et al., 2009). p32 overlapped to some extent with RCs and P150 (Fig. 4c, purple arrows), but overlapped to a greater extent with the P150 not in the RCs (Fig. 4c, red arrows). Whilst the overlap of P150 and p32 is consistent with the immunoprecipitation results, the lack of localization of p32 with P150 in RCs is consistent with the mutagenesis results, indicating that the interaction of p32 and P150 plays no role in RNA synthesis.

Fig. 4.

Subcellular localization of p32, CP, P150 and RCs. (a) Merged image of mock-infected Vero cells stained with rabbit anti-p32 and Alexa Fluor 595-conjugated donkey anti-rabbit (red). Nuclei were stained with Hoechst 33342 (blue). (b) Vero cells infected with Robo502/P150–HA at an m.o.i. of 1 and stained for p32 (red) as in (a), along with CP (green, top row; mouse anti-CP and Alexa Fluor 488-conjugated chicken anti-mouse antibodies) or dsRNA (green, bottom row; mouse anti-dsRNA and Alexa Fluor 488-conjugated chicken anti-mouse antibodies) to visualize RCs at 48 h p.i. Merged images are shown on the right. Yellow arrows indicate co-localizing CP and p32 clusters. Green arrows indicate an RC surrounded by p32 clusters. Nuclei were stained with Hoechst 33342 (blue). (c) Multicolour analysis of Robo502/P150–GFP-infected Vero cells at 48 h p.i. (m.o.i. of 1). Fluorescence microscopy was used to visualize p32 (blue; rabbit anti-p32 and Cascade Blue-conjugated goat anti-rabbit), dsRNA (red; mouse anti-dsRNA and Alexa Fluor 595-conjugated donkey anti-mouse) and P150 (green; labelled with GFP). Red-outlined inset, p32 and P150 merge; blue-outlined inset, P150 and dsRNA merge; green-outlined inset, p32 and dsRNA merge. Arrows indicate P150 and p32 overlap (red arrows), P150 and dsRNA overlap (yellow arrows), P150 not with dsRNA (white arrow) and P150, p32 and dsRNA overlap (purple arrows). Bars, 10 µm.

Discussion

In this study, we investigated the three potential PxxPxR class II SH3-binding motifs located within the PRR region residing within the Q domain of the RUBV P150 replicase protein. The Q domain is the only vacant space within P150 that has not been accounted for by computer homology searches (Koonin et al., 1992) and was only defined as a domain by its ability to be deleted from replicons, resulting in a non-viable construct that is fully rescued by CP (Tzeng & Frey, 2003). In fact, CP can be used to replace the Q domain in replicons with retention of viability (Tzeng & Frey, 2009). Rescue involves, at least in part, a poly-arginine cluster within the Q domain (located within the PRR), which is compensated for by a poly-arginine cluster in the CP. However, the function of the Q domain is unknown. The PRR was originally defined by computer scanning, and a corresponding region was found in hepatitis E virus ORF1, which corresponds to the RUBV NS-ORF, but interestingly was not found in the NS-ORF of alphaviruses, the other genus in the family Togaviridae (Koonin et al., 1992). The PRR of hepatitis E virus ORF1 contains neither PxxPxR motifs nor a poly-arginine cluster. Whilst the PRR is within the so-called hypervariable region of the RUBV genome, the PxxPxR motifs are well conserved in that the only variation detected was within the genome of one genotype 2B virus, which had amino acid substitutions in motifs 1 and 3. As PxxPxR motifs are involved in binding proteins with class II SH3 homology domains, we hypothesized that these motifs within P150 bound to cell proteins important in the virus replication cycle, either because they actively participated in virus replication or because they constituted a signalling pathway necessary for virus replication.

As preliminary experiments indicated that mutations in the third PxxPxR motif were stable but mutations in the first two motifs rapidly reverted, we concentrated on the first two motifs for proteomic analysis. Binding experiments using wt and mutant versions of the PRR fused to GST as bait led to identification of four candidate host cellular proteins, namely the p32 protein, human Mdn1 protein, Gcn1L protein and nucleopore complex protein Nup205. The human Mdn1 protein is a homologue of the yeast Rea1 protein involved in ribosome biogenesis (Galani et al., 2004), Gcn1L protein has protein-binding and ribosome-binding domains that are involved in regulation of translation (Sattlegger & Hinnebusch, 2000) and the Nup205 protein is part of the nuclear pore complex and, like other members of this complex, is involved in regulating the permeability of the nuclear membrane (Grandi et al., 1997). Interestingly, studies have shown that some RNA viruses target members of the nuclear pore complex to disrupt normal trafficking of host RNA and proteins (Gustin, 2003), a mechanism that probably targets host-cell protein synthesis. In this regard, RUBV does not exhibit a marked shutdown of either cellular RNA or protein synthesis (Hemphill et al., 1988).

We decided to concentrate on investigation of the interaction of p32 with P150, as there was precedence for the binding of this protein to RUBV CP (Beatch & Hobman, 2000; Beatch et al., 2005), as well as to proteins of other viruses (Luo et al., 1994; Matthews & Russell, 1998b; Wang et al., 1997). In particular, a tantalizing prospect was that binding to p32 was the redundant function shared by the Q domain and CP that would provide the mechanism behind the CP-mediated rescue phenomenon. However, the PxxPxR mutations that failed to bind p32 were not lethal when introduced into the RUBV replicon, as are mutations in the Q domain rescued by CP, and the critical poly-arginine cluster was not involved in binding of p32 to the Q domain. p32 binding could be mediated by PxxPxR motif 1 or 2. Whilst preliminary experiments demonstrated rapid reversion of mutations of either motif when introduced into a RUBV infectious cDNA clone, for unknown reasons we found in subsequent experiments that these mutations were stable, a finding consistent with the ability of either motif to bind p32 (data not shown). We demonstrated through various methods that p32 is an authentic binding partner of RUBV P150 mediated through the first two PxxPxR motifs. Thus, despite the inconsistency in the reversion results, the correct PxxPxR motifs responsible for the P150/p32 interaction were identified.

The p32 protein was first isolated as part of the ASF/SF2 human splicing factor complex in the nucleus of HeLa cells (Krainer et al., 1991). In addition to its function in the nucleus, p32 was later shown to be a resident of the mitochondria (Muta et al., 1997). Some of the functions include maintaining oxidative phosphorylation, and earlier work on p32 showed that it functioned as a phosphate translocator in mitochondria (Guérin et al., 1990; Phelps et al., 1991). Additionally, it has been shown that proteins destined for the mitochondria are bound to p32 through a mitochondrial signal sequence, which allows p32 to act as an import receptor. It was later shown that p32 contains a 73 aa mitochondrial targeting sequence at the N terminus that is efficiently removed by proteolytic processing (Muta et al., 1997). p32 has also been identified as a potential binding partner for a variety of proteins that would have to bind p32 at the cell surface, and p32 serves as the receptor for hantavirus (Choi et al., 2008; Ghebrehiwet et al., 1994; Herwald et al., 1996; Joseph et al., 1996). The interaction of p32 with other viral proteins has been documented extensively. The interaction of p32 with the HIV Tat and Epstein–Barr virus EBNA-1 proteins was shown to be important for enhancement of transcriptional activity (Wang et al., 1997; Yu et al., 1995). Additionally, p32 was shown to bind to the HIV Rev protein (Luo et al., 1994), which facilitates nucleus-to-cytoplasm transport of unspliced HIV RNA, and to the adenovirus core protein to potentially mediate the shuttling of adenoviral DNA into the nucleus (Matthews & Russell, 1998a, b).

The interaction of interest to us was that between p32 and RUBV CP (Beatch & Hobman, 2000; Beatch et al., 2005). Recently, Claus et al. (2011) reported that small interfering RNA (siRNA)-mediated knockdown of p32 in Vero cells prior to RUBV infection reduced the titre of virus produced tenfold, and we have obtained similar results (H. A. Mousa and T. K. Frey, unpublished observations). Thus, p32 appears to be important for RUBV replication, although, given the impact that p32 knockout has been shown to have on metabolism, namely a shift from oxidative phosphorylation to glycolysis (Fogal et al., 2010), it is not clear that this effect on RUBV replication is direct. Nevertheless, CP and p32 are both important for the redistribution of mitochondria to the perinuclear region in RUBV-infected cells and the propensity of mitochondria to cluster around RCs, which is a unique hallmark of RUBV infection (Beatch et al., 2005; Lee et al., 1996). This reorganization of mitochondria was thought perhaps to serve as a source of energy for the RNA synthesis process, a notion supported by a recent finding that RUBV infection increases levels of cellular ATP (Claus et al., 2011). It was also hypothesized that the reorganization of mitochondria may facilitate the incorporation of cardiolipin into the mature virion (Bardeletti & Gautheron, 1976). In contrast, RUBV infection blocks mitochondrial transport, lowering the amount of p32 within the mitochondria (Ilkow et al., 2010) and CP inhibits the induction of apoptosis through its interaction with mitochondria (Ilkow et al., 2011). Finally, mutagenesis of the poly-arginine cluster in CP critical for p32 binding also specifically decreased subgenomic RNA synthesis (Beatch et al., 2005), although it is hard to sort out the precise mechanism for this observation, as these poly-arginine clusters are also involved in binding RNA (Liu et al., 1996). Thus, CP–p32 binding may play a number of roles during RUBV infection.

Mutagenesis of the two PxxPxR motifs involved in p32 binding to P150 led to a crippled virus that replicated four logs less efficiently than wt virus. Of course, this decrease in replication efficiency is not necessarily specifically the result of p32 binding and may be, fully or in part, due to reduced binding to another protein that binds to these motifs (such as one of the other three that we identified). The Mut1+2 virus exhibited no defect in RNA synthesis, and we showed that this was not due to rescue by CP because replication of CP-lacking replicons was not affected by this mutation. This finding led us to investigate the intracellular distribution of p32 in RUBV-infected cells. Whilst we confirmed overlap of P150 and p32, only occasional overlap of p32 with RCs was detected, indicating that p32 does not play a role in virus RNA synthesis. Fontana et al. (2010) detected p32 associated with cytopathic vacuoles in RUBV replicon-transfected cells by immunoelectron microscopy. However, as Mut1+2 replicons display no defect in replication, the presence of p32 at this location would seem to be adventitious. It thus appears that the PxxPxR motifs in P150 are important in a step in the virus life cycle subsequent to RNA synthesis. We have recently identified perinuclear clusters that appear in RUBV-infected cells late in infection in which P150, CP, ssRNA (but not dsRNA) and cellular G3BP protein accumulate that we speculate are sites where encapsidation occurs (Matthews & Frey, 2012). Potentially, it is within this site that binding to P150 by p32 or other cell proteins is critical for the virus replication cycle.

Methods

Cell lines and virus.

The 293T line of transformed human embryonic kidney cells (HEK293T; ATCC) was grown and maintained at 37 °C with 5 % CO₂, in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 % FBS and penicillin/streptomycin (100 µg ml⁻¹). African green monkey kidney (Vero) cells were grown and maintained at 35 °C with 5 % CO₂ in DMEM containing 5 % FBS and gentamicin (10 µg ml⁻¹). Two RUBV strains were used: F-Therein, a wild-type strain, referred to here as RUBV, and Robo502 virus and its mutants, derived from an infectious cDNA clone (Tzeng & Frey, 2003).

Plasmid construction.

Detailed cloning strategies and primer sequences are available on request from the authors. Briefly, site-directed mutagenesis was carried out to change both prolines in a PxxPxR motif to alanines in the Robo502 infectious clone (Tzeng & Frey, 2003). There are three of these motifs in the proline hinge region, termed motifs 1, 2 and 3, and proline-to-alanine mutations in each motif (i.e. PxxPxR was mutated to AxxAxR) were introduced into Robo502 digested with the same restriction enzymes to replace the corresponding wt fragment. The mutant constructs were named Robo502/Mut1, -Mut2 and -Mut3 corresponding to mutations made in motifs 1, 2 and 3, respectively. To prepare constructs with mutations in motifs 1 and 2 in combination, Mut1 DNA was used as DNA template and Mut2 was added. This construct was named Robo502/Mut1+2. RUBrep/GFP-Mut1+2 was created by replacing the restriction fragment containing the coding sequences for the PRR in RUBrep/GFP with the corresponding restriction fragment from Robo502/Mut1+2.

GST–PRR fusion constructs were created as follows. The PRR in P150, between nt 2189 and 2461 of the RUBV genome (between aa 716 and 782 of P150), was amplified from Robo502 by PCR using primers flanking the PRR-coding sequences. The amplified PCR product was ligated into a pEBG mammalian expression vector (kindly provided by Dr Tom Hobman, University of Alberta, Canada), creating the construct GST–PRR. A pEBG-Mut1+2 construct was created using the strategy described above with the GST–PRR construct as the template, the forward and reverse primers used to amplify the sequence introduced into PRR, and the Mut1 and Mut2 mutagenic primers. This construct was named GST–PRR/Mut1+2.

HA-tagged RUBV P150 Q domain constructs were made as follows. The Q domain of RUB P150 (between aa 497 and 803 of P150) and its R_Q mutant (Tzeng & Frey, 2009) were amplified from the HA-tagged P150 RUB infectious clone (Robo502–HA) (Matthews et al., 2009) and RUBrep-HA/GFP R_Q, respectively. VR1062 and purified PCR products were ligated together, resulting in constructs termed HA–P150Q wild-type and HA–P150Q R_Q. Constructs expressing P150Q with the Mut1, Mut2 and Mut1+2 sequences were constructed similarly.

To construct a plasmid expressing p32 tagged at the C terminus with a c-myc epitope, the simian p32 gene in a pCB6 vector (Beatch & Hobman, 2000; kindly provided by Dr Tom Hobman) was PCR amplified using primers that added the c-myc epitope and introduced into the pcDNA vector. The resulting construct was named pcDNA-p32-c-myc. Mutations in motifs 1 and 2 were introduced into a plasmid expressing HA-tagged P150, cytomegalovirus (CMV)-P150–HA (Matthews et al., 2009), resulting in CMV-P150–HA/Mut1+2.

In vitro transcription, transfection and analysis of RNA synthesis and replicon replication.

Wt Robo502, Robo502/Mut1+2, wt RUBrep/GFP and RUBrep/GFP/Mut1+2 were transcribed in vitro followed by transfection of Vero cells, as described previously (Tzeng & Frey, 2002). The medium was collected 1 week after transfection; this was designated passage 0 (P0). Titres of virus in the P0 medium were determined by plaque assay (Pugachev et al., 1997). P0 medium was subsequently passaged by infecting Vero cells at an m.o.i. of 1. At 7 days p.i., the P1 medium was collected and titrated by plaque assay. Sequencing of the PRR region was carried out using RNA extracted from P1 cells after the medium had been harvested. Virus-specific RNA was analysed in cells transfected with wt Robo502 or Robo502/Mut1+2 on day 3 post-transfection by Northern blotting using VR-C-E2-E1 as a probe, as described previously (Tzeng et al., 2001). To analyse replicon replication, live cells transfected with wt RUBrep/GFP or RUBrep/GFP-Mut1+2 were imaged using a Zeiss Axioplan microscope with epifluorescence capacity under a ×10 objective on day 3 post-transfection.

Transfection with pEBG constructs and identification of interacting cell proteins.

Dishes (60 mm diameter) of confluent HEK293T cells were transfected with 5–10 µg pEBG construct DNA using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. At 3 days post-transfection, the cells were washed twice with lysis wash buffer [50 mM Tris/HCl (pH 7.4), 1 % NP-40, 150 mM NaCl, 2 mM EDTA and 150 mM sodium salicylate] and lysed with 1 % NP-40 buffer [50 mM Tris/HCl (pH 7.4), 1 % NP-40, 150 mM NaCl and 2 mM EDTA], and GST fusion proteins were isolated from the lysate using gluthathione–Sepharose 4B beads (Amersham Pharmacia Biotech). Before incubation of the transfected cell lysates with the GST beads, the lysates were pre-cleared with an equal volume of unconjugated Sepharose beads (Sigma). Following the pre-clearing step, 0.5 ml gluthathione beads was incubated with 5 ml lysate for 3 h at 4 °C, after which the beads were washed seven times with lysis buffer and adherent proteins were eluted with 2× SDS gel loading buffer [50 mM Tris/HCl (pH 6.8), 100 mM DTT, 2 % SDS, 0.1 % bromophenol blue and 10 % glycerol]. The proteins were resolved by SDS-PAGE (6 or 10 % acrylamide) and proteins were visualized by using GelCode blue staining reagent (Pierce, Thermo Scientific). Proteins of interest were excised and analysed by mass spectrometry at the Morehouse School of Medicine (GA, USA) or the Scripps Institute (TSRI Center for Mass Spectrometry, CA, USA).

Immunoprecipitation and Western blot analysis.

Dishes (60 mm diameter) of Vero cells transfected with CMV-P150 or co-transfected with pcDNA-p32-c-myc and either CMV-P150 or CMV-P150-Mut1+2 using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol were lysed at 2 days post-transfection with 0.5 ml modified RIPA buffer [10 mM Tris/HCl (pH 7.4), 150 mM NaCl, 3 mM EDTA, 1 % Triton X-100, 0.1 % SDS and 0.5 % sodium deoxycholate] in the presence of 1× Mini Complete Protease Inhibitor (Roche) and 250 µl lysate was used for immunoprecipitation studies. Briefly, 1 µl mouse anti-HA or anti-c-myc mAb (Roche) or rabbit anti-p32 polyclonal antibody (CeMines) was added and immunoprecipitation was carried out as described previously (Pugachev et al., 1997). Immunoprecipitates were resolved by 6 or 10 % SDS-PAGE and Western blotting was carried out as described previously (Wu & Tai, 2004). The blot was probed with mouse anti-HA mAb (Roche) or rabbit anti-p32 polyclonal antibody. Anti-mouse or anti-rabbit antibody (Promega) conjugated with alkaline phosphatase was used as a secondary antibody, and bound antibodies were detected by using a BCIP/nitroblue tetrazolium colorimetric assay (Roche).

Microscopy.

Vero cells grown on glass coverslips and infected with Robo502/P150–HA (Matthews et al., 2009) or Robo502/P150–GFP (Matthews et al., 2010) (these viruses express a P150 tagged with HA or GFP, respectively) were analysed at 48 h p.i. The cells were fixed in 4 % formaldehyde for 10 min at room temperature and then, after several washes with PBS, were permeabilized with 0.5 % Triton X-100 (10 min at room temperature) and exposed to antibodies. Primary antibody was diluted 1 : 500 and secondary antibody was diluted 1 : 2000. Coverslips were incubated for 1 h with each antibody solution, with several PBS washes in between each step. Coverslips were mounted and analysed by confocal microscopy using a Zeiss LSM 700 microscope and imaging software. Antibodies used in these experiments not listed above were: mouse anti-dsRNA (Scientific Consulting), Cascade Blue-conjugated goat anti-rabbit (Molecular Probes), Alexa Fluor 595-conjugated donkey anti-mouse or anti-rabbit and Alexa Fluor 488-conjugated chicken anti-mouse (Molecular Probes). Hoechst stain (Molecular Probes) was used to stain nuclear DNA.

p32 knockdown and infection.

Knockdown of p32 was accomplished by mock transfecting or transfecting Vero cells with siRNA specific for p32 (sense sequence 5′-GCAUCCCACCAACAUUUGATT-3′, 100 nM final concentration; Ambion) using Lipofectamine 2000 according to the manufacturer’s instructions. At 24, 36, 72 and 96 h post-transfection, the cells were lysed and the lysate was processed for Western blotting (10 % SDS-PAGE) and probed with rabbit anti-p32 and anti-calnexin antibodies (Sigma). To test the effect of knockdown on RUBV replication, at 48 h post-transfection, cells were infected with RUBV at an m.o.i. of 1. Medium was harvested at 24 and 48 h p.i. and viral titres were determined by plaque assay. This experiment was repeated four times.