Interactions between Rubella Virus Capsid and Host
http://www.100md.com
病菌学杂志 2005年第16期
Departments of Cell Biology
Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta T6G 2H7, Canada
ABSTRACT
The distribution and morphology of mitochondria are dramatically affected during infection with rubella virus (RV). Expression of the capsid, in the absence of other viral proteins, was found to induce both perinuclear clustering of mitochondria and the formation of electron-dense intermitochondrial plaques, both hallmarks of RV-infected cells. We previously identified p32, a host cell mitochondrial matrix protein, as a capsid-binding protein. Here, we show that two clusters of arginine residues within capsid are required for stable binding to p32. Mutagenic ablation of the p32-binding site in capsid resulted in decreased mitochondrial clustering, indicating that interactions with this cellular protein are required for capsid-dependent reorganization of mitochondria. Recombinant viruses encoding arginine-to-alanine mutations in the p32-binding region of capsid exhibited altered plaque morphology and replicated to lower titers. Further analysis indicated that disruption of stable interactions between capsid and p32 was associated with decreased production of subgenomic RNA and, consequently, infected cells produced significantly lower amounts of viral structural proteins under these conditions. Together, these results suggest that capsid-p32 interactions are important for nonstructural functions of capsid that include regulation of virus RNA replication and reorganization of mitochondria during infection.
INTRODUCTION
Rubella virus (RV) is the causative agent of the disease known as rubella or German measles in humans. Symptoms are typically mild when contracted postnatally; however, prenatal exposure in the first trimester of pregnancy results in a series of congenital defects known as congenital rubella syndrome (CRS). Symptoms of CRS include low birth weight, deafness, cataracts, mental retardation, and heart disease (10). Despite the widespread use of an effective vaccine, RV remains the leading cause of congenital defects by an infectious agent worldwide. It is estimated that over 100,000 cases of CRS occur every year (46).
RV is a positive-strand RNA virus of the family Togaviridae. Its genome is approximately 9,755 nucleotides in length and encodes two nonstructural proteins and three structural proteins (12). Like other togaviruses, the RV structural proteins are translated as a polyprotein precursor from a subgenomic RNA (44). The structural proteins consist of a capsid phosphoprotein and two envelope glycoproteins that form the spike complexes on the surface of the virion (32). The major function of the capsid involves homo-oligomerization and binding of genomic RNA to form the nucleocapsid (36). Interestingly, capsid has been shown to have numerous functions in addition to formation of the nucleocapsid. For instance, capsid expression has been reported to cause apoptosis in certain cell types, suggesting that this protein may play a role in the cytopathic effect of the virus (13). Evidence from two recent studies suggests that capsid may also play a role in regulating replication of genomic RNA. In one study, capsid expression was shown to complement the replication defects of replicons with deletions in the nonstructural genes (52). In another study, it was reported that capsid expression was able to modulate genomic RNA replication (9).
Another interesting feature of RV capsid is that it displays multiple intracellular localizations. In cells transiently expressing all three structural proteins, a pool of capsid is localized to the Golgi region, where budding of virions typically occurs (4, 22, 39). In the absence of E2 and E1, capsid has been reported to associate largely with the endoplasmic reticulum (4). Subsequently, immunogold electron microscopy and indirect immunofluorescence studies revealed that pools of capsid are also associated with mitochondria and viral replication complexes (6, 35). Colocalization of capsid with the RV nonstructural protein p150 has also been observed (28). The presence of capsid at multiple intracellular sites suggests that it may have functions in addition to forming nucleocapsids.
The association of capsid with mitochondria is intriguing, given that a significant role for this organelle in RV replication was first reported more than 25 years ago (2). Specifically, cardiolipin, a phospholipid associated exclusively with mitochondria, was found to be a component of the RV envelope. More recent studies have shown that mitochondrial distribution and morphology are affected in RV-infected cells. For example, mitochondria cluster in association with viral replication complexes in the perinuclear region (32). Late in infection, electron-dense plaques, or confronting membranes, are observed between the outer membranes of opposing mitochondria (33). In contrast, confronting membranes have not been observed in cells infected with other togaviruses. In some cases, loss of mitochondrial cristae has also been observed in RV-infected cells (32).
We previously demonstrated that perinuclear clustering of mitochondria does not require virus replication but, rather, can be induced by expression of the RV structural proteins (6). Moreover, a major host-encoded capsid-binding protein is p32, a mitochondrial matrix protein (6, 41). The close association of capsid with mitochondria led us to investigate the nature of this relationship further. Here we show that expression of capsid, in the absence of the other RV proteins, is sufficient to induce clustering of mitochondria to the juxtanuclear region and formation of confronting membranes between mitochondria. Furthermore, we demonstrate that the stable interaction of capsid with p32 is important for redistribution of mitochondria and for virus replication.
MATERIALS AND METHODS
Reagents, antibodies, and cDNA clones. Reagents and supplies were from the following sources. Protein A- and G-Sepharose and glutathione-Sepharose were purchased from Pharmacia (Alameda, CA). Phenylmethylsulfonyl fluoride, fibronectin, sodium dodecyl sulfate (SDS), bovine serum albumin, and general lab chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). 14C-labeled protein standards were purchased from Amersham Corp. (Arlington Heights, IL). Radiolabeled inorganic phosphate H333PO4 (4,000 Ci/mmol) was purchased from ICN (Costa Mesa, CA). Media and serum for cell culture were purchased from Life Technologies-Invitrogen Inc. or Sigma. Pwo polymerase and PerFectin transfection reagent were purchased from Roche Molecular Biochemicals (Laval, Quebec, Canada) and Gene Therapy Systems Inc. (San Diego, CA), respectively. Vero, COS, RK13, and BHK cells were obtained from the American Type Culture Collection (Rockville, MD). The M33 strain of RV and an infectious cDNA clone (55) were kindly provided by Shirley Gillam (University of British Columbia). The C1 anticapsid monoclonal antibody was a gift from Jerry Wolinsky (University of Texas, Houston). Rabbit polyclonal antibodies to the nonstructural protein p150 were kindly provided by Tero Ahola (University of Helsinki).
Mammalian cell culture and transfections. BHK, COS, and Vero cells were cultured in Dulbecco's modified Eagle's medium (high glucose) containing 10% fetal bovine serum, 2 mM glutamine, 1 mM HEPES, and antibiotics. RK13 cells were cultured in minimal essential medium containing 10% fetal bovine serum, 2 mM glutamine, 1 mM HEPES, 0.1 mM nonessential amino acids, and antibiotics. Cells were transiently transfected using PerFectin transfection reagent as described by the manufacturer.
Plasmid construction. Recombinant plasmids were constructed as described below using standard subcloning techniques. Where indicated, constructs were generated by PCR using the primers listed in Table 1. Generally, reverse primers contained in-frame stop codons. Typical reaction mixtures (100 μl) included 2 to 4 U DNA polymerase, 100 to 500 ng template DNA, forward and reverse oligonucleotide primers (15 pmol each), and deoxynucleoside triphosphates (10 μM). Reaction mixtures were subjected to 30 cycles in a Robocycler Gradient 40 (Stratagene) or in a DeltaCycler II system (ERICOMP). All constructs were sequenced to verify their authenticity and to ensure the absence of second site mutations. DNA sequences were amplified using Pwo (Roche) or Pfx (Stratagene) DNA polymerases according to the manufacturers' instructions.
Constructs used in yeast two-hybrid analysis: capsid constructs. The cDNA encoding amino acid residues 1 to 277 of capsid was amplified by PCR using primers Capsid-F and C-E2SP-R. The resulting product was digested with EcoRI and BglII and then subcloned into pGBT9 (Clontech). Capsid deletion constructs generated by PCR were digested with EcoRI and BglII and subcloned into the EcoRI and BamHI sites of pGBT9 or pGBKT7 (Clontech). The constructs and the primers used for their creation were as follows: pGBT9-C1-88, Capsid-F and CM11; pGBT9-C87-171, CM12 and CM13; pGBT9-C167-277, CM14 and C-E2SP-R; pGBKT7-C1-45, Capsid-F and CR45; pGBKT7-C46-89, Capsid46 and CR89; pGBKT7-C30-65, RVC30 and RVC65R; pGBKT7-C30-69, RVC30 and RVC69R; pGBKT7-C30-79, RVC30 and RVC79R; pGBKT7-C30-89, RVC30 and CR89.
Constructs for expression in mammalian cells. Plasmids for constitutive expression of RV structural proteins (pCMV5-CapE2SP, pCMV5-E2E1, and pCMV5-24S) have been described previously (20, 30). For regulated expression of capsid, pcDNA 5/T0-Capsid was created. Briefly, the capsid cDNA, including the region coding for the E2 signal peptide, was amplified by PCR using primers AV11H and AV10 with pCMV5-CapE2SP as template. The PCR product was digested with HindIII and subcloned into pcDNA 5/T0 (Invitrogen).
The arginine residues in the p32-binding region of capsid were changed to alanine residues by PCR-mediated mutagenesis of a capsid cDNA cassette. The mutated cDNAs were subsequently used to replace analogous regions in both pCMV5-CapE2SP and pCMV5-24S. For the 5RA mutant, the plasmid pCMV5-24S (20) was used as a template in a mutagenic PCR with the forward primer 5RAP and the reverse primer Capsid-R. The resulting PCR fragment was digested with NarI and used to replace the NarI capsid fragment in pCMV5-CapE2SP. The resulting plasmid was named pCMV5-C5RA. For the 6RA mutant, using pCMV5-24S as template, the forward primer AV11 and the reverse primer 6RAP2 were used in a mutagenic PCR. The resulting PCR fragment was digested using EcoRI and ApaI and used to replace the analogous capsid fragment in pCMV5-CapE2SP. The resulting plasmid was named pCMV5-C6RA. For the 11RA mutant, using pCMV5-24S6RA as template, the forward primer 5RAP and the reverse primer Capsid-R were used in a mutagenic PCR. The resulting PCR fragment was digested with NarI and ligated into NarI-cut pCMV5-CapE2SP to create pCMV5-C11RA. To introduce the mutagenized regions of capsid into the 24S cDNA, the EcoRI/SphI fragment of the mutant capsids was excised by endonuclease restriction digestion and used to replace the analogous region (EcoRI/SphI) in pCMV5-24S.
The arginine-to-alanine mutations in capsid were introduced into the RV M33 infectious clone (pBRM33) (55) by replacing the NcoI and SphI fragments from the capsid-encoding region of pBRM33 with the analogous fragments from pCMV5-C5RA, pCMV5-C6RA, or pCMV5-C11RA.
The plasmid used for expression of full-length p32 in cultured cells (pCB6+p32) was described previously (6). The pEGFP-p32m plasmid was created by fusing the region encoding the mature form of p32, in frame, to the carboxyl terminus of green fluorescent protein (GFP). The cDNA sequence encoding mature p32 was amplified by PCR from pCB6+p32 using primers p32mG and p32R3. The PCR product was digested with EcoRI and BamHI and subcloned into pEGFP-C1 (Clontech) cut with the same restriction enzymes.
Radioimmunoprecipitation and RNA-binding assay. Transfected COS cells were metabolically labeled with [35S]methionine-cysteine and radioimmunoprecipitated as described previously (23). Immune complexes were washed with phosphate-buffered saline containing 0.1% Triton X-100 to preserve protein-protein interactions. Radioimmunoprecipitation of 33P-labeled capsid and RNA-binding assays were performed as described elsewhere (31).
Northern blot analyses. Vero cells were infected with wild-type and mutant viruses (multiplicity of infection [MOI] = 1), and total RNA was harvested using TRIzol reagent (Invitrogen) at various times postinfection. The RNA (10 μg/lane) was separated on 1% agarose gels containing 3.7% formaldehyde and then transferred to Hybond N membranes using a TurboBlotter apparatus (Schleicher & Schuell). Membranes were probed with a capsid cDNA that had been radiolabeled with [32P]dCTP using a Random Primers DNA labeling system (Invitrogen). Hybridization and washing protocols were performed with reagents corresponding to entire capsid coding region and NorthernMax (Ambion) reagents. After washing, membranes were exposed to a phosphorimager screen overnight.
Generation of stable cell lines inducibly expressing capsid. To create stable cell lines expressing capsid, subconfluent T-REx 293 (Invitrogen) cells (1 x 106 cells/60-mm dish) stably expressing the tetracycline repressor protein were transfected with 8 μg pcDNA5T/0-capsid. At 48 h posttransfection, cells were split 1:10 and cultured in medium containing hygromycin B (5 μg/ml) and blasticidin (100 μg/ml). Surviving colonies were isolated using cloning cylinders and tested for inducible expression of capsid by indirect immunofluorescence and by immunoblot analyses. Multiple clones were analyzed; however, clone 293TRC 2B was used for all experiments described in this study.
Immunofluorescence microscopy. Vero cells grown on coverslips were processed for indirect immunofluorescence microscopy 24 h after transfection. To visualize mitochondria, Mitotracker Red CMXRos was added to the cell culture medium at a final concentration of 30 ng/ml for 20 min at 37°C prior to fixation. Samples were processed as described previously (6) and then examined using a Zeiss 510 confocal microscope. Images from optical sections (0.8 μm) were processed using Adobe Photoshop 7.0.
Electron microscopy. HEK293TRC 2B cells were induced to express capsid by the addition of doxycycline (1 μg/ml) to the culture medium. Uninduced HEK293TRC 2B cultures served as the negative controls. Cells were processed for electron microscopy 45 h postinduction as described previously (16). Images were captured using a Megaview 3 charge-coupled device camera (Soft Imaging System).
Yeast two-hybrid analysis. Interactions between capsid deletions (subcloned into pGBT9 or pGBTK7) and full-length p32 (in pGAD10 [6]) were assayed by cotransforming AH109 cells and testing for growth on medium lacking adenine, histidine, leucine, and tryptophan (4DO). The positive control was a Clontech system control that utilizes two strongly interacting proteins, p53 (in pGBKT7) and simian virus 40 large T antigen (in pGADT7). As a negative control, AH109 cells were transformed with two plasmids encoding proteins that do not interact (pGBKT7-p53 and pGAD-ABP280). Growth on the 4DO plates within 4 days (at 30°C) was interpreted as evidence of a positive interaction. All media and transformations were performed according to the protocols described in the Clontech MATCHMAKER system.
Synthesis of infectious viral RNA and production of mutant viruses. Plasmids containing full-length RV cDNA clones were linearized with HindIII and used as templates for transcription of capped RNAs using the mMessage mMachine kit (Ambion). RNAs were quantitated by electrophoresis and by spectroscopy at 260 nm. RNA was introduced into BHK cells by electroporation as described elsewhere (31). Virus-containing culture media were collected at various time points and then clarified at 7,000 x g for 10 min before storage at –80°C. Virus titers were determined by plaque assay as described previously (31).
One-step growth curves were determined as follows: Vero cells (2 x 105/35-mm dish) were infected with recombinant viruses (MOI = 5) for 2 h at 35°C. Cells were washed once with phosphate-buffered saline and then incubated for 1 h in growth medium. Fresh medium was then added, and incubation continued at 35°C. Medium was removed at 12-h intervals and stored at –80°C, and titers were determined by plaque assay.
To assess the synthesis of viral proteins, Vero cells were infected with M33 or mutant strains (5RA and 6RA) at an MOI of 1.0. Cells were harvested at various times postinfection, and lysates were prepared. Samples were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting with antibodies to capsid and the nonstructural protein p150.
Virus particle assembly assay. COS cells (1.5 x 105/35-mm dish) were transiently transfected with 2 μg of the pCMV5-24S constructs encoding wild-type and mutant capsids. Assembly and secretion of RV-like particles was assayed as described elsewhere (30).
RESULTS
Capsid expression induces formation of intermitochondrial plaques. Mitochondria have long been thought to have an important role in the replication of RV (1, 2, 33). Moreover, we recently demonstrated that expression of the RV structural proteins induces relocalization of mitochondria to the perinuclear region (6). Given that the capsid protein but neither of the RV glycoproteins is intimately associated with mitochondria (6, 35), we rationalized that capsid is most likely to be the virus protein responsible for this process. To test this hypothesis, Vero cells were transiently transfected with expression vectors encoding either capsid alone or the RV glycoproteins (E2 and E1). Indeed, perinuclear clustering of mitochondria was evident in transfected cells expressing capsid alone but not in untransfected cells (Fig. 1A to D). In contrast, expression of E2 and E1 did not significantly affect the distribution of mitochondria (Fig. 1E to F). Together, these observations support our hypothesis that capsid expression induces reorganization of mitochondria.
To further examine the effects of capsid expression on mitochondrial morphology, a stable cell line (HEK 293TRC 2B) was created in which capsid expression was under control of the tetracycline repressor. Addition of doxycycline to the culture medium resulted in robust expression of capsid protein (Fig. 2A). HEK 293TRC 2B cells were cultured for 48 h under inducing or noninducing conditions and then processed for routine electron microscopy. Electron-dense structures were observed between apposing mitochondria in cells cultured in the presence of doxycycline (Fig. 2C to E) but were not present in the noninduced cells (Fig. 2B). These electron-dense structures are reminiscent of the confronting membranes found in RV-infected cells (34). The width of confronting membranes in RV-infected cells was reported to be 22 to 25 nm. In accordance with this observation, the average width of the electron-dense zones in capsid-expressing cells was 24 nm. These results indicate that expression of the RV capsid is sufficient to induce the formation of confronting membranes, a hallmark of RV infection.
Capsid-p32 interactions. Analysis of the RV capsid sequence using the Web-based algorithm PSORT II Prediction (http://psort.nibb.ac.jp/form2.html) revealed that classical mitochondrial targeting sequences are absent from this protein. As such, the mechanism of capsid targeting to mitochondria is not clear. However, capsid is known to bind to the carboxyl terminus of mitochondrial matrix protein p32 (reference 41 and unpublished data), a situation that would leave the amino-terminal mitochondrial targeting sequence of p32 unobstructed and presumably functional. Accordingly, it is tempting to speculate that targeting of capsid to mitochondria involves piggybacking onto nascent p32. If this is the mechanism by which capsid is targeted to the mitochondria, it should be possible to block association of capsid with mitochondria by expressing a form of p32 that is not targeted to the mitochondria. To create such a construct, the mitochondrial targeting sequence of p32 was replaced with GFP to create the fusion protein GFPp32m. It was expected that GFPp32m would act in a dominant negative fashion and abrogate targeting of capsid to mitochondria.
Vero cells were transiently transfected with plasmids encoding GFP or GFPp32m either alone or together with capsid. When coexpressed with GFP, capsid displayed a characteristic juxtanuclear localization, while GFP was diffuse throughout the cytoplasm and nucleus (Fig. 3D to F). In addition, the localization of GFP was not altered by the expression of capsid, indicating that these two proteins do not interact (Fig. 3A to F). When expressed alone, GFPp32m was not associated with mitochondria-like structures but, rather, was localized throughout the cytoplasm in diffuse as well as punctate structures (Fig. 3G to I). As p32 is normally present in homo-oligomeric complexes (24), it is likely that the punctate structures are aggregates of p32. As expected, coexpression of capsid with GFPp32m resulted in the colocalization of GFPp32m with capsid (Fig. 3J to L). However, unexpectedly, coexpression of these proteins resulted in a dramatic relocalization of GFPp32m to compact juxtanuclear structures that resembled those of mitochondrially targeted capsid (Fig. 3F).
To ascertain whether the capsid-GFPp32m interactions were occurring at mitochondria, Vero cells were transfected with plasmids encoding either GFPp32m alone or GFPp32m and capsid. Mitotracker Red CMXRos was used to visualize mitochondria for these experiments. When expressed alone, GFPp32m did not associate with mitochondria but, instead, exhibited a largely diffuse localization with some puncta throughout the cytoplasm. In cells expressing this protein, mitochondria appeared identical to those in untransfected control cells (Fig. 4A to C). In contrast, when GFPp32m was coexpressed with capsid it was confined to structures that were stained with the Mitotracker dye (Fig. 4D to F). These results suggest that capsid is directed to mitochondria by a mechanism that is independent of the p32 mitochondrial targeting sequence.
An arginine-rich region of capsid is required for binding to p32. Our strategy to block binding of capsid to mitochondria using a "dominant negative" form of p32 proved unsuccessful, since targeting of capsid to this organelle appears to be independent of the p32 mitochondrial targeting sequence. In lieu of this approach, we elected to create replication-competent RV strains encoding capsid proteins that do not bind to p32. Accordingly, the first step in this process required mapping of the p32-binding site within capsid. Previously, using the yeast two-hybrid assay, we localized a p32-binding site within the amino-terminal region of capsid (6). To more precisely map the p32-binding site, we assayed the interactions between full-length p32 and a series of overlapping capsid constructs. Results from this analysis are shown in Fig. 5A. These data indicate that the minimum region of capsid that strongly interacts with p32 is comprised of amino acid residues 30 to 69.
Inspection of the p32-binding region in capsid revealed that it contains 11 arginine residues that are distributed into two clusters (Fig. 5B). Interestingly, it has been reported that the p32-binding sites of several other proteins are rich in arginine residues (8, 18, 37, 43, 54). For this reason, we focused our subsequent analysis on the arginine residues in this region of capsid. Site-directed mutagenesis was used to change the arginine residues in the two clusters to alanine residues. The resulting capsid constructs were named 5RA, 6RA, and 11RA (Fig. 5B).
Coimmunoprecipitation assays were used to determine if the arginine residue clusters in the p32-binding domain of capsid are required for stable binding to p32. COS cells were transiently transfected with expression vectors encoding wild-type or mutant capsids and the RV glycoproteins E2 and E1. At 40 h posttransfection, cells were biosynthetically labeled with [35S]methionine-cysteine and cell lysates were prepared under nondenaturing conditions. Goat anti-p32 serum was used to isolate immune complexes from lysates, which were then analyzed by SDS-PAGE and fluorography. As expected, wild-type capsid coimmunoprecipitated with p32 (Fig. 6A, upper panel). In contrast, capsids containing the 5RA, 6RA, or 11RA mutations did not form stable complexes with p32. To ensure that the mutant capsid constructs were stably expressed in transfected cells, capsid was also immunoprecipitated from cell lysates using rabbit anticapsid serum (Fig. 6A, lower panel). Together, these results indicate that the mutant capsids are expressed at levels comparable to the wild-type M33 capsid and that both of the arginine clusters within the p32-binding domain are necessary for stable binding to p32. This was not unexpected, since our results obtained using the yeast two-hybrid system indicated that portions of capsid containing only the amino-terminal arginine residue cluster (amino acid residues 1 to 45) or the carboxyl-terminal arginine residue cluster (amino acid residues 46 to 89) were unable to interact with p32. In contrast, a construct encoding amino acid residues 30 to 69, which contains both arginine clusters, interacted with p32 as well as full-length capsid (Fig. 5A).
Functional analysis of capsid RA mutants. Before proceeding to construction of recombinant viruses, it was important to ensure that the capsid RA mutants were functional, since the distribution and presumably intracellular binding interactions of capsid are influenced by the presence of glycoproteins E2 and E1 (16). Therefore, most of the following experiments were conducted under conditions that mimic virus assembly; specifically, capsid was coexpressed together with E2 and E1. The targeted RA mutations in the p32-binding site are in close proximity to serine 46, which is the primary phosphorylation site of capsid. Regulated phosphorylation of serine 46 has been shown to be important for virus replication, an effect that is likely mediated through its role in modulating the RNA-binding activity of capsid (31). The levels of phosphorylation of the various capsid mutants were compared to wild-type capsid using radioimmunoprecipitation and fluorography assays. Transfected cells were biosynthetically labeled with H333PO4 for 12 h before capsid proteins were immunoprecipitated, separated by SDS-PAGE, and then subjected to fluorography. Wild-type capsid and capsids containing either the 5RA or the 6RA mutations were phosphorylated to the same extent (Fig. 6B). In contrast, the 11RA mutant was not stably phosphorylated. Immunoblot analysis showed that all of the mutant capsids were expressed at comparable levels in the transfected cells. These results indicate that alteration of one, but not both, of the arginine clusters does not appreciably alter the phosphorylation state of capsid.
Next, we investigated whether the arginine-to-alanine mutations affected the ability of capsid to assemble into virions. As a convenient method to study parameters that govern virus assembly, we have routinely used an RV-like particle (RLP) assay. Coordinated expression of the RV structural proteins results in the formation of RLPs, which resemble native virions in terms of morphology, antigenicity, and immunogenicity (15, 16, 21, 30). Wild-type and mutant capsids were coexpressed with or without RV glycoproteins in COS cells, and RLPs were isolated from the precleared media of transfected cells by ultracentrifugation at 100,000 x g. The presence of pelletable capsid in the medium indicates that RLPs have been assembled and secreted. When coexpressed with the RV glycoproteins, wild-type capsid and capsids containing the 5RA or the 6RA cluster mutations were able to support assembly and secretion of RLPs (Fig. 6C, lower panel). In contrast, RLPs were not formed when wild-type capsid was expressed in the absence of E2 and E1. Comparable amounts of wild-type, 5RA, and 6RA capsids were detected in the medium, suggesting that RLPs are produced and secreted with similar efficiencies in each case. However, transfected cells expressing the 11RA capsid mutant did not secrete detectable amounts of RLPs. The presence of intracellular capsid in the transfected cells was confirmed by immunoblotting of cell lysates (Fig. 6C, upper panel). Together, these results indicate that mutation of either the proximal or the distal arginine clusters does not affect capsid phosphorylation, or its ability to support virus particle assembly, and by extrapolation, the formation of RV virions. These results were not surprising given that, by analogy to alphavirus capsid proteins (14) and analyses using protein structure algorithms, the amino-terminal one-third of capsid is thought to be relatively unstructured.
We next examined the RNA-binding activities of the arginine-alanine cluster mutants in an in vitro RNA-binding assay (31, 36). Since the p32-binding site overlaps with the RNA-binding site, we were concerned that ablating the arginine residues in this region would negatively affect RNA binding. To increase the affinity of capsids for RNA, samples were first treated with phosphatase prior to binding analyses (31). Interestingly, no loss of RNA-binding activity was associated with the capsid 5RA mutant (Fig. 6D). Rather, this mutant appeared to bind genomic RNA with higher affinity than wild-type capsid but similarly to the hypophosphorylated mutant S46A (31). However, it should be noted that these in vitro binding assays are somewhat variable, and in previous experiments we have documented that dephosphorylated wild-type capsid binds RNA as well as the S46A mutant (31). Accordingly, the important result here is that the five proximal arginine residues in the p32-binding site of capsid are not required for RNA binding. In contrast, the 6RA capsid bound RNA only very weakly (data not shown), indicating that the distal arginine cluster is important for binding to genomic RNA.
The p32-binding site of capsid is important for virus replication. The results described above suggest that the capsid 5RA and 6RA mutants are structurally intact and can function normally in assembly of virus particles. We next sought to determine the importance of the p32-binding site in capsid function during virus replication. The mutant capsid genes were subcloned into an RV infectious clone (55), and capped genomic RNAs were synthesized in vitro. BHK-21 cells were electroporated with equal amounts of RNA (10 μg), and media were collected at scheduled intervals starting at 12 h postelectroporation. Infectious virus released from the cells was quantitated by plaque assay in RK13 cells. Wild-type M33 RV produced large, clear plaques that were easily visible by crystal violet staining (Fig. 7A). The 5RA strain produced plaques that were smaller and more opaque than M33 plaques. The 6RA strain produced plaques that were even smaller than 5RA plaques. Not surprisingly, the 11RA mutant did not produce detectable amounts of infectious virus (data not shown). These results indicate that viruses encoding capsid proteins that do not interact with p32 exhibit decreased abilities to produce and/or secrete infectious virions.
To confirm that viral proteins were adequately expressed in the electroporated cells and that decreased virus titers were not the result of inefficient electroporation, BHK cells electroporated with equivalent amounts of infectious RNA (10 μg) were processed for double labeling immunofluorescence 3 days postelectroporation. Cells were stained with antibodies against the RV envelope protein E1 and p32 to label cells expressing viral proteins and as a counterstain, respectively. The percentages of cells expressing E1 were determined in six randomly selected fields from two independent experiments (Table 2). Similar levels of E1-positive cells were observed in samples electroporated with wild-type, 5RA, and 6RA genomic RNAs (88 to 97%). However, a much smaller fraction (26%) of cells electroporated with 11RA genomic RNA was positive for E1 expression. The lower percentage of cells expressing E1 in the 11RA mutant probably results from the inability of the mutant capsids to support assembly of infectious virus particles and, thus, the percentage of E1-expressing cells in this case likely reflects the electroporation efficiency. These results also confirm that the 5RA and the 6RA mutant viruses are able to replicate and secrete infectious virus particles that infect neighboring cells. Therefore, these results also provide indirect evidence that both the 5RA and 6RA capsids are able to bind to and incorporate genomic RNA into virions. However, the virus titers obtained from electroporated BHK cells were consistently lower (100- to 1,000-fold) for the 5RA and 6RA RNAs compared to the M33 RNA (data not shown).
To rule out the possibility that the lower titers associated with the 5RA and 6RA viruses were not due to replication or secretion defects specific to BHK cells, conditions that approach one-step growth conditions in Vero cells were employed. Although infected BHK cells produce more intracellular infectious virus than Vero cells, the latter typically release more infectious virus than BHK cells (3). Consequently Vero cells are most often used for propagation of RV. Titers for virus stocks obtained from electroporated BHK cells were determined and used to infect Vero cells (MOI = 5 PFU/cell). Media were collected every 12 h postinfection, and virus titers were determined by plaque assay. In all cases, viral titers peaked between 60 and 72 h; however, the 5RA and 6RA titers were almost 3 orders of magnitude lower than M33 (Fig. 7B). These results suggest that the arginine residues in the p32-binding site are important for efficient virus replication in two different cell types.
M33 and 5RA strains exhibit different kinetics in viral protein and RNA synthesis. To better understand how the lack of a stable capsid-p32 interaction leads to replication defects, we chose to examine early and late steps in the virus life cycle. The nonstructural protein p150 is translated directly from the genomic RNA and is therefore an early event (14a). In contrast, capsid is synthesized from a subgenomic mRNA and is a later event in the replication scheme (44). Typically, nonstructural and structural proteins are first detected in Vero cells at 12 and 16 h postinfection, respectively (19). Since the 6RA capsid does not bind well to genomic RNA, we used the 5RA strain for these experiments. Vero cells were infected with either M33 or 5RA RV strains, and cell lysates were prepared at various times postinfection. Samples were run on SDS-PAGE, and the levels of p150 and capsid were determined by immunoblotting (Fig. 8A). The profiles of p150 expression were similar in M33- and 5RA-infected cells, although levels of p150 were slightly lower in the latter case. At 24 h postinfection, cells infected with M33 and 5RA viruses contained similar levels of p150 protein. Between 24 and 48 h, levels of p150 increased in both groups of infected cells (Fig. 8A). Consistent with its structural role in virus assembly at a late step, the levels of capsid in M33-infected cells increased significantly in a time-dependent manner. In contrast, the intracellular levels of the 5RA capsid did not increase over time. This was not due to inherent instability in these mutant capsids, as transfection experiments revealed that both 5RA and 6RA capsids were stable when expressed from plasmids (Fig. 6). Moreover, levels of the two other RV structural proteins E2 and E1 were dramatically lower in 5RA-infected cells (data not shown). Together, these results suggest that ablation of the p32-binding site within capsid results in replication defects at a relatively late step in the virus life cycle.
To investigate this defect further, the levels of virus-specific RNA were monitored at various times postinfection. Whereas the genomic RNA was seen to accumulate with similar kinetics in the M33- and 5RA-infected cells, the accumulation of subgenomic RNA was impaired in the latter group of infected cells (Fig. 8B). These results suggest that the relatively low level of structural protein synthesis observed in 5RA-infected cells is due mainly to a defect in subgenomic RNA production.
Capsid-p32 interactions mediate perinuclear clustering of mitochondria. To gain other insights into how abolishing capsid-p32 interactions inhibits virus replication, we next examined how mitochondrial clustering was affected by expression of the mutant capsid proteins. Vero cells were transiently transfected with vectors encoding wild-type capsid or the capsid RA mutants and processed as described above. As described previously, capsid expression induced a dramatic rearrangement of mitochondria (Fig. 9). Mitochondria were distributed in a lacey pattern throughout the cytoplasm of untransfected cells but were noticeably compacted in the juxtanuclear regions of the cells expressing wild-type capsid. In cells expressing the capsid 5RA or 6RA mutants, mitochondrial staining patterns were more punctate and condensed into juxtanuclear regions than mock-transfected cells, but not as pronounced as in cells expressing wild-type capsid. These results suggest that the interaction of capsid with p32 facilitates the recruitment of mitochondria to the juxtanuclear region.
DISCUSSION
Nonstructural roles for the capsid protein. The RV capsid protein assembles with the viral genome to form the structural core of the virion interior. In addition to this structural function, there is increasing evidence that the capsid plays integral roles in several other important processes, including induction of apoptosis (13) and replication of virus-specific RNA (9, 52). In keeping with its structural function, a large pool of capsid is expected to associate with organelles of the secretory pathway where nucleocapsid assembly and virus budding occur. However, recent studies by a number of laboratories including our own have revealed the existence of capsid pools that associate with nonsecretory organelles. These sites include the cytoplasmic surface of mitochondria (6, 35) and endosome-associated viral replication complexes (28, 35). Conceivably, the cohort of capsid associated with replication complexes functions in some aspect of RNA metabolism, such as transcription, stabilization, and/or packaging of nascent virus-encoded RNAs. The association of capsid with mitochondria appears to be unique to RV among togaviruses (32) and, until now, has not been formally addressed.
In the present study, we found that capsid expression in the absence of other viral proteins has dramatic effects on the morphology and distribution of mitochondria. Specifically, perinuclear clustering of mitochondria and formation of electron-dense zones between adjacent cisternae were observed, both of which are hallmarks of RV-infected cells (33). The underlying mechanism that leads to these defects is not clear, but a potential link to RV-induced mitochondrial abnormalities came from the finding that capsid interacts with the host-encoded protein p32 (6). The bulk of this host protein resides in the matrix of mitochondria (11, 42, 48). Despite the intense efforts of numerous laboratories, the normal physiological function of p32 has remained elusive. However, a recent study suggests that this protein is required for the proapoptotic function of the BH3-only protein Hrk (50). Interestingly, like the RV capsid, Hrk binds to the highly conserved carboxyl-terminal region of p32. It is not known if the binding sites for these two proteins overlap.
A large number of studies have documented the interactions between p32 and cellular proteins as well as virus proteins (7, 8, 18, 25, 29, 37, 38, 54, 56). In most cases, interactions between virus-encoded proteins and p32 occur at nonmitochondrial sites. Given that the proapoptotic function of p32 is dependent upon its localization to mitochondria (50), it is conceivable that some p32-dependent virus-host interactions serve to prevent or delay virus-induced cell death.
Of particular relevance to the present study is the documented role of p32 in RNA metabolism (27, 45). Indeed, interactions between p32 and virus proteins are known to be important for virus replication. For example, expression of human p32 in murine cells overcomes a posttranscriptional block in replication of human immunodeficiency virus (HIV) (57). In addition, p32 is required for latent-cycle DNA replication of Epstein-Barr virus (53). By analogy, interactions between p32 and the RV capsid appear to be important for transcription of the RV subgenomic RNA. Consistent with our present findings, overexpression of p32 reportedly enhances replication of RV by an as-yet-undefined mechanism (41). Interestingly, the level of capsid expression and presumably other structural proteins was dramatically increased when cells that overexpress p32 were infected with RV.
The capsid p32-binding site is critical for replication. We were somewhat puzzled that our mapping studies of the p32-binding region in capsid conflicted with those of a previous study (41). Mohan et al. reported that the p32-binding site resides with amino acid residues 1 to 28 of capsid. In contrast, our analysis indicated that the p32-binding site is located between amino acid residues 30 to 69. It seems unlikely that there are two separate p32-binding sites since, using the same type of assay (yeast two hybrid), we did not detect interactions between p32 and amino acid residues 1 to 45 of capsid (6). While we are currently unable to reconcile the differences between these two studies, our findings are consistent with those of several groups who have reported that the p32-binding regions of other proteins contain multiple clusters of arginine residues (8, 18, 54). For example, two arginine-rich clusters within the p32-binding domain of Epstein-Barr virus EBNA-1 are necessary for stable interaction with p32 in cultured cells (53, 54). Similarly, we observed nearly identical results for the RV capsid. Specifically, two separate arginine clusters (5RA and 6RA) between amino acid residues 30 and 69 are required for stable capsid-p32 interactions.
The p32-binding site is located in a region of capsid that is thought to be relatively unstructured. Accordingly, ablation of this site was not expected to compromise the structural integrity of capsid. Indeed, both 5RA and 6RA mutants were phosphorylated normally and functioned at wild-type capacity in the virus assembly assay. However, viruses containing the 5RA and 6RA mutations replicated to titers that were between 100- and 1,000-fold lower than the wild-type M33 strain. The defect in replication for the 5RA mutant virus, at least, probably occurs at a relatively late step but before nucleocapsid assembly. Whereas accumulation of the nonstructural protein p150 was not greatly affected in cells infected by the 5RA virus strain, accumulation of structural proteins was greatly diminished. Further analysis indicated that the levels of subgenomic RNA were significantly lower in cells infected with the 5RA virus strain. Accordingly, we favor the hypothesis that the RNA replication-specific roles of capsid (9, 52) are dependent upon interactions with p32. However at this point, we cannot rule out the possibility that the decreased levels of structural proteins in 5RA-infected cells are partly the result of translational defects. Specifically, we observed that the 5RA capsid binds RV RNA better than the wild-type capsid. Thus, it is also possible that tight binding of the 5RA capsid to the 24S RNA inhibits translation.
Effects of capsid on mitochondria may be mediated by interactions with p32. In addition to being less effective in supporting virus replication, expression of the 5RA and 6RA capsids resulted in reduced perinuclear clustering of mitochondria relative to M33 capsid. Presumably, capsid is targeted to the cytoplasmic surface of mitochondria independently of p32 and, once associated with the cytoplasmic surface of this organelle, is then able to complex with p32. Under certain circumstances, p32 is able to leave the mitochondrial matrix (49). It is therefore possible that capsid binds to a pool of p32 as it leaves the mitochondria, although it is not at all clear how the association of capsid with this organelle would induce release of p32. Accordingly, we favor the scenario that capsid binds to the carboxyl-terminal region of newly synthesized p32 shortly before it is translocated into mitochondria. This would be expected to leave the amino-terminal mitochondrial targeting sequence of p32 exposed, thereby allowing at least partial translocation of p32 into the mitochondrial matrix. Alternatively, its association with capsid may inhibit import of p32 into mitochondria. A similar mechanism prevents import of the cellular enzyme fumarase into the mitochondria of Saccharomyces cerevisiae (26). Thus, retention of p32 in the cytosol would result in the formation of capsid-p32 complexes at the cytoplasmic surface of mitochondria. Since both capsid and p32 are known to form homo-oligomers (5, 24), it is tempting to speculate that capsid-p32 complex functions as a scaffold or a molecular adhesive to drive aggregation of mitochondria.
The function of mitochondrial aggregation in RV-infected cells is intriguing but not yet understood. Cells infected with other viruses, including Semliki Forest virus (33), Flock House virus (40), and African swine fever (47), exhibit similar mitochondrial defects. As with the RV capsid, expression of the hepatitis B virus X protein is sufficient to induce clustering of mitochondria to a juxtanuclear region in the absence of other viral proteins (51). While X protein expression has been linked to apoptotic cell death, clustering of mitochondria in itself is not likely to be the cause of death, since Vero cells expressing the RV capsid display aggregated mitochondria but do not show visible signs of apoptosis. Rather, it is thought that aggregation of mitochondria to sites in close proximity to virus replication or assembly factories serves to provide a source of cellular energy to assist virus replication (17, 33, 40, 47). A second but not mutually exclusive possibility is that the capsid-induced association of mitochondria with virus replication sites facilitates the incorporation of cardiolipin into virions (2).
In conclusion, our data are consistent with a scenario in which capsid interaction with p32 and consequent aggregation of mitochondria are beneficial for virus replication. Importantly, these data are in agreement with a corollary study from another laboratory, which showed that replication of RV is enhanced in cells overexpressing p32 (41). The challenge now is to understand the mechanism by which capsid-p32 interactions influence nonstructural functions of capsid, in particular the synthesis of subgenomic RNA.
ACKNOWLEDGMENTS
We thank Shirley Gillam, Jerry Wolinsky, and Terry Ahola for their gifts of reagents. Margaret Hughes is acknowledged for excellent technical assistance.
The work was funded by a grant to T.C.H. from the Canadian Institutes of Health Research (CIHR). T.C.H. is the recipient of a Senior Medical Scholarship from the Alberta Heritage Foundation for Medical Research (AHFMR). M.D.B. was supported by graduate studentship awards from AHFMR and CIHR. J.C.E. is supported by a Canada graduate scholarship. L.M.J.L. is supported by a graduate studentship award from AHFMR.
REFERENCES
Bardeletti, G. 1977. Respiration and ATP level in BHK21/13S cells during the earliest stages of rubella virus replication. Intervirology 8:100-109.
Bardeletti, G., and D. C. Gautheron. 1976. Phospholipid and cholesterol composition of rubella virus and its host cell BHK 21 grown in suspension cultures. Arch. Virol. 52:19-27.
Bardeletti, G., J. Tektoff, and D. Gautheron. 1979. Rubella virus maturation and production in two host cell systems. Intervirology 11:97-103.
Baron, M. D., T. Ebel, and M. Suomalainen. 1992. Intracellular transport of rubella virus structural proteins expressed from cloned cDNA. J Gen. Virol. 73:1073-1086.
Baron, M. D., and K. Forsell. 1991. Oligomerization of the structural proteins of rubella virus. Virology 185:811-819.
Beatch, M. D., and T. C. Hobman. 2000. Rubella virus capsid associates with host cell protein p32 and localizes to mitochondria. J. Virol. 74:5569-5576.
Bruni, R., and B. Roizman. 1996. Open reading frame P—a herpes simplex virus gene repressed during productive infection encodes a protein that binds a splicing factor and reduces synthesis of viral proteins made from spliced mRNA. Proc. Natl. Acad. Sci. USA 93:10423-10427.
Bryant, H. E., D. A. Matthews, S. Wadd, J. E. Scott, J. Kean, S. Graham, W. C. Russell, and J. B. Clements. 2000. Interaction between herpes simplex virus type 1 IE63 protein and cellular protein p32. J. Virol. 74:11322-11328.
Chen, M. H., and J. P. Icenogle. 2004. Rubella virus capsid protein modulates viral genome replication and virus infectivity. J. Virol. 78:4314-4322.
Cooper, L. Z. 1969. Rubella in pregnancy. Postgrad. Med. 46:106-107.
Dedio, J., W. Jahnen-Dechent, M. Bachmann, and W. Muller-Esterl. 1998. The multiligand-binding protein gC1qR, putative C1q receptor, is a mitochondrial protein. J. Immunol. 160:3534-3542.
Dominguez, G., C. Y. Wang, and T. K. Frey. 1990. Sequence of the genome RNA of Rubella virus: evidence for genetic rearrangement during Togavirus evolution. Virology 177:225-238.
Duncan, R., A. Esmaili, L. M. Law, S. Bertholet, C. Hough, T. C. Hobman, and H. L. Nakhasi. 2000. Rubella virus capsid protein induces apoptosis in transfected RK13 cells. Virology 275:20-29.
Forsell, K., M. Suomalainen, and H. Garoff. 1995. Structure-function relation of the NH2-terminal domain of the Semliki Forest virus capsid protein. J. Virol. 69:1556-1563.
Frey, T. K. 1994. Molecular biology of rubella virus. Adv. Virus Res. 44:69-160.
Garbutt, M., H. Chan, and T. C. Hobman. 1999. Secretion of rubella virions and virus-like particles in cultured epithelial cells. Virology 261:340-346.
Garbutt, M., L. M. Law, H. Chan, and T. C. Hobman. 1999. Role of rubella virus glycoprotein domains in assembly of virus-like particles. J. Virol. 73:3524-3533.
Garzon, S., H. Strykowski, and G. Charpentier. 1990. Implication of mitochondria in the replication of Nodamura virus in larvae of the Lepidoptera, Galleria mellonella (L.) and in suckling mice. Arch. Virol. 113:165-176.
Hall, K. T., M. S. Giles, M. A. Calderwood, D. J. Goodwin, D. A. Matthews, and A. Whitehouse. 2002. The herpesvirus Saimiri open reading frame 73 gene product interacts with the cellular protein p32. J. Virol. 76:11612-11622.
Hemphill, M. L., R. Y. Forng, E. S. Abernathy, and T. K. Frey. 1988. Time course of virus-specific macromolecular synthesis during rubella virus infection in Vero cells. Virology 162:65-75.
Hobman, T. C., M. L. Lundstrom, and S. Gillam. 1990. Processing and intracellular transport of rubella virus structural proteins in COS cells. Virology 178:122-133.
Hobman, T. C., M. L. Lundstrom, C. A. Mauracher, L. Woodward, S. Gillam, and M. G. Farquhar. 1994. Assembly of rubella virus structural proteins into virus-like particles in transfected cells. Virology 202:574-585.
Hobman, T. C., N. O. Seto, and S. Gillam. 1994. Expression of soluble forms of rubella virus glycoproteins in mammalian cells. Virus Res. 31:277-289.
Hobman, T. C., L. Woodward, and M. G. Farquhar. 1993. The rubella virus E2 and E1 spike glycoproteins are targeted to the Golgi complex. J. Cell Biol. 121:269-281.
Jiang, J., Y. Zhang, A. R. Krainer, and R. M. Xu. 1999. Crystal structure of human p32, a doughnut-shaped acidic mitochondrial matrix protein. Proc. Natl. Acad. Sci. USA 96:3572-3577.
Kittlesen, D. J., K. A. Chianese-Bullock, Z. Q. Yao, T. J. Braciale, and Y. S. Hahn. 2000. Interaction between complement receptor gC1qR and hepatitis C virus core protein inhibits T-lymphocyte proliferation. J. Clin. Investig. 106:1239-1249.
Knox, C., E. Sass, W. Neupert, and O. Pines. 1998. Import into mitochondria, folding and retrograde movement of fumarase in yeast. J. Biol. Chem. 273:25587-25593.
Krainer, A. R., A. Mayeda, D. Kozak, and G. Binns. 1991. Functional expression of cloned human splicing factor SF2: homology to RNA-binding proteins, U1 70K, and Drosophila splicing regulators. Cell 66:383-394.
Kujala, P., T. Ahola, N. Ehsani, P. Auvinen, H. Vihinen, and L. Kaariainen. 1999. Intracellular distribution of rubella virus nonstructural protein P150. J. Virol. 73:7805-7811.
Laine, S., A. Thouard, J. Derancourt, M. Kress, D. Sitterlin, and J. M. Rossignol. 2003. In vitro and in vivo interactions between the hepatitis B virus protein P22 and the cellular protein gC1qR. J. Virol. 77:12875-12880.
Law, L. M., R. Duncan, A. Esmaili, H. L. Nakhasi, and T. C. Hobman. 2001. Rubella virus E2 signal peptide is required for perinuclear localization of capsid protein and virus assembly. J. Virol. 75:1978-1983.
Law, L. M., J. C. Everitt, M. D. Beatch, C. F. Holmes, and T. C. Hobman. 2003. Phosphorylation of rubella virus capsid regulates its RNA binding activity and virus replication. J. Virol. 77:1764-1771.
Lee, J. Y., and D. S. Bowden. 2000. Rubella virus replication and links to teratogenicity. Clin. Microbiol. Rev. 13:571-587.
Lee, J. Y., D. S. Bowden, and J. A. Marshall. 1996. Membrane junctions associated with rubella virus infected cells. J. Submicrosc. Cytol. Pathol. 28:101-108.
Lee, J. Y., D. Hwang, and S. Gillam. 1996. Dimerization of rubella virus capsid protein is not required for virus particle formation. Virology 216:223-227.
Lee, J. Y., J. A. Marshall, and D. S. Bowden. 1999. Localization of rubella virus core particles in Vero cells. Virology 265:110-119.
Liu, Z., D. Yang, Z. Qiu, K. T. Lim, P. Chong, and S. Gillam. 1996. Identification of domains in rubella virus genomic RNA and capsid protein necessary for specific interaction. J. Virol. 70:2184-2190.
Luo, Y., H. Yu, and B. M. Peterlin. 1994. Cellular protein modulates effects of human immunodeficiency virus type 1 Rev. J. Virol. 68:3850-3856.
Matthews, D. A., and W. C. Russell. 1998. Adenovirus core protein V interacts with p32—a protein which is associated with both the mitochondria and the nucleus. J. Gen. Virol. 79:1677-1685.
McDonald, H., T. C. Hobman, and S. Gillam. 1991. The influence of capsid protein cleavage on the processing of E2 and E1 glycoproteins of rubella virus. Virology 183:52-60.
Miller, D. J., M. D. Schwartz, and P. Ahlquist. 2001. Flock house virus RNA replicates on outer mitochondrial membranes in Drosophila cells. J. Virol. 75:11664-11676.
Mohan, K. V., B. Ghebrehiwet, and C. D. Atreya. 2002. The N-terminal conserved domain of rubella virus capsid interacts with the C-terminal region of cellular p32 and overexpression of p32 enhances the viral infectivity. Virus Res. 85:151-161.
Muta, T., D. Kang, S. Kitajima, T. Fujiwara, and N. Hamasaki. 1997. p32 protein, a splicing factor 2-associated protein, is localized in mitochondrial matrix and is functionally important in maintaining oxidative phosphorylation. J. Biol. Chem. 272:24363-24370.
Nikolakaki, E., G. Simos, S. D. Georgatos, and T. Giannakouros. 1996. A nuclear envelope-associated kinase phosphorylates arginine-serine motifs and modulates interactions between the lamin B receptor and other nuclear proteins. J. Biol. Chem. 271:8365-8372.
Oker-Blom, C., I. Ulmanen, L. Kaariainen, and R. F. Pettersson. 1984. Rubella virus 40S genome RNA specifies a 24S subgenomic mRNA that codes for a precursor to structural proteins. J. Virol. 49:403-408.
Petersen-Mahrt, S. K., C. Estmer, C. Ohrmalm, D. A. Matthews, W. C. Russell, and G. Akusjarvi. 1999. The splicing factor-associated protein, p32, regulates RNA splicing by inhibiting ASF/SF2 RNA binding and phosphorylation. EMBO J. 18:1014-1024.
Robertson, S. E., D. A. Featherstone, M. Gacic-Dobo, and B. S. Hersh. 2003. Rubella and congenital rubella syndrome: global update. Rev. Panam. Salud Publica 14:306-315.
Rojo, G., M. Chamorro, M. L. Salas, E. Vinuela, J. M. Cuezva, and J. Salas. 1998. Migration of mitochondria to viral assembly sites in African swine fever virus-infected cells. J. Virol. 72:7583-7588.
Seytter, T., F. Lottspeich, W. Neupert, and E. Schwarz. 1998. Mam33p, an oligomeric, acidic protein in the mitochondrial matrix of Saccharomyces cerevisiae is related to the human complement receptor gC1q-R. Yeast 14:303-310.
Soltys, B. J., D. Kang, and R. S. Gupta. 2000. Localization of P32 protein (gC1q-R) in mitochondria and at specific extramitochondrial locations in normal tissues. Histochem. Cell Biol. 114:245-255.
Sunayama, J., Y. Ando, N. Itoh, A. Tomiyama, K. Sakurada, A. Sugiyama, D. Kang, F. Tashiro, Y. Gotoh, Y. Kuchino, and C. Kitanaka. 2004. Physical and functional interaction between BH3-only protein Hrk and mitochondrial pore-forming protein p32. Cell Death Differ. 11:771-781.
Takada, S., Y. Shirakata, N. Kaneniwa, and K. Koike. 1999. Association of hepatitis B virus X protein with mitochondria causes mitochondrial aggregation at the nuclear periphery, leading to cell death. Oncogene 18:6965-6973.
Tzeng, W. P., and T. K. Frey. 2003. Complementation of a deletion in the rubella virus p150 nonstructural protein by the viral capsid protein. J. Virol. 77:9502-9510.
Van Scoy, S., I. Watakabe, A. R. Krainer, and J. Hearing. 2000. Human p32: a coactivator for Epstein-Barr virus nuclear antigen-1-mediated transcriptional activation and possible role in viral latent cycle DNA replication. Virology 275:145-157.
Wang, Y., J. E. Finan, J. M. Middeldorp, and S. D. Hayward. 1997. p32/TAP, a cellular protein that interacts with EBNA-1 of Epstein-Barr virus. Virology 236:18-29.
Yao, J., and S. Gillam. 1999. Mutational analysis, using a full-length rubella virus cDNA clone, of rubella virus E1 transmembrane and cytoplasmic domains required for virus release. J. Virol. 73:4622-4630.
Yu, L., Z. Zhang, P. M. Loewenstein, K. Desai, Q. Tang, D. Mao, J. S. Symington, and M. Green. 1995. Molecular cloning and characterization of a cellular protein that interacts with the human immunodeficiency virus type 1 Tat transactivator and encodes a strong transcriptional activation domain. J. Virol. 69:3007-3016.
Zheng, Y. H., H. F. Yu, and B. M. Peterlin. 2003. Human p32 protein relieves a posttranscriptional block to HIV replication in murine cells. Nat. Cell Biol. 5:611-618.(Martin D. Beatch, Jason C)
Medical Microbiology and Immunology, University of Alberta, Edmonton, Alberta T6G 2H7, Canada
ABSTRACT
The distribution and morphology of mitochondria are dramatically affected during infection with rubella virus (RV). Expression of the capsid, in the absence of other viral proteins, was found to induce both perinuclear clustering of mitochondria and the formation of electron-dense intermitochondrial plaques, both hallmarks of RV-infected cells. We previously identified p32, a host cell mitochondrial matrix protein, as a capsid-binding protein. Here, we show that two clusters of arginine residues within capsid are required for stable binding to p32. Mutagenic ablation of the p32-binding site in capsid resulted in decreased mitochondrial clustering, indicating that interactions with this cellular protein are required for capsid-dependent reorganization of mitochondria. Recombinant viruses encoding arginine-to-alanine mutations in the p32-binding region of capsid exhibited altered plaque morphology and replicated to lower titers. Further analysis indicated that disruption of stable interactions between capsid and p32 was associated with decreased production of subgenomic RNA and, consequently, infected cells produced significantly lower amounts of viral structural proteins under these conditions. Together, these results suggest that capsid-p32 interactions are important for nonstructural functions of capsid that include regulation of virus RNA replication and reorganization of mitochondria during infection.
INTRODUCTION
Rubella virus (RV) is the causative agent of the disease known as rubella or German measles in humans. Symptoms are typically mild when contracted postnatally; however, prenatal exposure in the first trimester of pregnancy results in a series of congenital defects known as congenital rubella syndrome (CRS). Symptoms of CRS include low birth weight, deafness, cataracts, mental retardation, and heart disease (10). Despite the widespread use of an effective vaccine, RV remains the leading cause of congenital defects by an infectious agent worldwide. It is estimated that over 100,000 cases of CRS occur every year (46).
RV is a positive-strand RNA virus of the family Togaviridae. Its genome is approximately 9,755 nucleotides in length and encodes two nonstructural proteins and three structural proteins (12). Like other togaviruses, the RV structural proteins are translated as a polyprotein precursor from a subgenomic RNA (44). The structural proteins consist of a capsid phosphoprotein and two envelope glycoproteins that form the spike complexes on the surface of the virion (32). The major function of the capsid involves homo-oligomerization and binding of genomic RNA to form the nucleocapsid (36). Interestingly, capsid has been shown to have numerous functions in addition to formation of the nucleocapsid. For instance, capsid expression has been reported to cause apoptosis in certain cell types, suggesting that this protein may play a role in the cytopathic effect of the virus (13). Evidence from two recent studies suggests that capsid may also play a role in regulating replication of genomic RNA. In one study, capsid expression was shown to complement the replication defects of replicons with deletions in the nonstructural genes (52). In another study, it was reported that capsid expression was able to modulate genomic RNA replication (9).
Another interesting feature of RV capsid is that it displays multiple intracellular localizations. In cells transiently expressing all three structural proteins, a pool of capsid is localized to the Golgi region, where budding of virions typically occurs (4, 22, 39). In the absence of E2 and E1, capsid has been reported to associate largely with the endoplasmic reticulum (4). Subsequently, immunogold electron microscopy and indirect immunofluorescence studies revealed that pools of capsid are also associated with mitochondria and viral replication complexes (6, 35). Colocalization of capsid with the RV nonstructural protein p150 has also been observed (28). The presence of capsid at multiple intracellular sites suggests that it may have functions in addition to forming nucleocapsids.
The association of capsid with mitochondria is intriguing, given that a significant role for this organelle in RV replication was first reported more than 25 years ago (2). Specifically, cardiolipin, a phospholipid associated exclusively with mitochondria, was found to be a component of the RV envelope. More recent studies have shown that mitochondrial distribution and morphology are affected in RV-infected cells. For example, mitochondria cluster in association with viral replication complexes in the perinuclear region (32). Late in infection, electron-dense plaques, or confronting membranes, are observed between the outer membranes of opposing mitochondria (33). In contrast, confronting membranes have not been observed in cells infected with other togaviruses. In some cases, loss of mitochondrial cristae has also been observed in RV-infected cells (32).
We previously demonstrated that perinuclear clustering of mitochondria does not require virus replication but, rather, can be induced by expression of the RV structural proteins (6). Moreover, a major host-encoded capsid-binding protein is p32, a mitochondrial matrix protein (6, 41). The close association of capsid with mitochondria led us to investigate the nature of this relationship further. Here we show that expression of capsid, in the absence of the other RV proteins, is sufficient to induce clustering of mitochondria to the juxtanuclear region and formation of confronting membranes between mitochondria. Furthermore, we demonstrate that the stable interaction of capsid with p32 is important for redistribution of mitochondria and for virus replication.
MATERIALS AND METHODS
Reagents, antibodies, and cDNA clones. Reagents and supplies were from the following sources. Protein A- and G-Sepharose and glutathione-Sepharose were purchased from Pharmacia (Alameda, CA). Phenylmethylsulfonyl fluoride, fibronectin, sodium dodecyl sulfate (SDS), bovine serum albumin, and general lab chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). 14C-labeled protein standards were purchased from Amersham Corp. (Arlington Heights, IL). Radiolabeled inorganic phosphate H333PO4 (4,000 Ci/mmol) was purchased from ICN (Costa Mesa, CA). Media and serum for cell culture were purchased from Life Technologies-Invitrogen Inc. or Sigma. Pwo polymerase and PerFectin transfection reagent were purchased from Roche Molecular Biochemicals (Laval, Quebec, Canada) and Gene Therapy Systems Inc. (San Diego, CA), respectively. Vero, COS, RK13, and BHK cells were obtained from the American Type Culture Collection (Rockville, MD). The M33 strain of RV and an infectious cDNA clone (55) were kindly provided by Shirley Gillam (University of British Columbia). The C1 anticapsid monoclonal antibody was a gift from Jerry Wolinsky (University of Texas, Houston). Rabbit polyclonal antibodies to the nonstructural protein p150 were kindly provided by Tero Ahola (University of Helsinki).
Mammalian cell culture and transfections. BHK, COS, and Vero cells were cultured in Dulbecco's modified Eagle's medium (high glucose) containing 10% fetal bovine serum, 2 mM glutamine, 1 mM HEPES, and antibiotics. RK13 cells were cultured in minimal essential medium containing 10% fetal bovine serum, 2 mM glutamine, 1 mM HEPES, 0.1 mM nonessential amino acids, and antibiotics. Cells were transiently transfected using PerFectin transfection reagent as described by the manufacturer.
Plasmid construction. Recombinant plasmids were constructed as described below using standard subcloning techniques. Where indicated, constructs were generated by PCR using the primers listed in Table 1. Generally, reverse primers contained in-frame stop codons. Typical reaction mixtures (100 μl) included 2 to 4 U DNA polymerase, 100 to 500 ng template DNA, forward and reverse oligonucleotide primers (15 pmol each), and deoxynucleoside triphosphates (10 μM). Reaction mixtures were subjected to 30 cycles in a Robocycler Gradient 40 (Stratagene) or in a DeltaCycler II system (ERICOMP). All constructs were sequenced to verify their authenticity and to ensure the absence of second site mutations. DNA sequences were amplified using Pwo (Roche) or Pfx (Stratagene) DNA polymerases according to the manufacturers' instructions.
Constructs used in yeast two-hybrid analysis: capsid constructs. The cDNA encoding amino acid residues 1 to 277 of capsid was amplified by PCR using primers Capsid-F and C-E2SP-R. The resulting product was digested with EcoRI and BglII and then subcloned into pGBT9 (Clontech). Capsid deletion constructs generated by PCR were digested with EcoRI and BglII and subcloned into the EcoRI and BamHI sites of pGBT9 or pGBKT7 (Clontech). The constructs and the primers used for their creation were as follows: pGBT9-C1-88, Capsid-F and CM11; pGBT9-C87-171, CM12 and CM13; pGBT9-C167-277, CM14 and C-E2SP-R; pGBKT7-C1-45, Capsid-F and CR45; pGBKT7-C46-89, Capsid46 and CR89; pGBKT7-C30-65, RVC30 and RVC65R; pGBKT7-C30-69, RVC30 and RVC69R; pGBKT7-C30-79, RVC30 and RVC79R; pGBKT7-C30-89, RVC30 and CR89.
Constructs for expression in mammalian cells. Plasmids for constitutive expression of RV structural proteins (pCMV5-CapE2SP, pCMV5-E2E1, and pCMV5-24S) have been described previously (20, 30). For regulated expression of capsid, pcDNA 5/T0-Capsid was created. Briefly, the capsid cDNA, including the region coding for the E2 signal peptide, was amplified by PCR using primers AV11H and AV10 with pCMV5-CapE2SP as template. The PCR product was digested with HindIII and subcloned into pcDNA 5/T0 (Invitrogen).
The arginine residues in the p32-binding region of capsid were changed to alanine residues by PCR-mediated mutagenesis of a capsid cDNA cassette. The mutated cDNAs were subsequently used to replace analogous regions in both pCMV5-CapE2SP and pCMV5-24S. For the 5RA mutant, the plasmid pCMV5-24S (20) was used as a template in a mutagenic PCR with the forward primer 5RAP and the reverse primer Capsid-R. The resulting PCR fragment was digested with NarI and used to replace the NarI capsid fragment in pCMV5-CapE2SP. The resulting plasmid was named pCMV5-C5RA. For the 6RA mutant, using pCMV5-24S as template, the forward primer AV11 and the reverse primer 6RAP2 were used in a mutagenic PCR. The resulting PCR fragment was digested using EcoRI and ApaI and used to replace the analogous capsid fragment in pCMV5-CapE2SP. The resulting plasmid was named pCMV5-C6RA. For the 11RA mutant, using pCMV5-24S6RA as template, the forward primer 5RAP and the reverse primer Capsid-R were used in a mutagenic PCR. The resulting PCR fragment was digested with NarI and ligated into NarI-cut pCMV5-CapE2SP to create pCMV5-C11RA. To introduce the mutagenized regions of capsid into the 24S cDNA, the EcoRI/SphI fragment of the mutant capsids was excised by endonuclease restriction digestion and used to replace the analogous region (EcoRI/SphI) in pCMV5-24S.
The arginine-to-alanine mutations in capsid were introduced into the RV M33 infectious clone (pBRM33) (55) by replacing the NcoI and SphI fragments from the capsid-encoding region of pBRM33 with the analogous fragments from pCMV5-C5RA, pCMV5-C6RA, or pCMV5-C11RA.
The plasmid used for expression of full-length p32 in cultured cells (pCB6+p32) was described previously (6). The pEGFP-p32m plasmid was created by fusing the region encoding the mature form of p32, in frame, to the carboxyl terminus of green fluorescent protein (GFP). The cDNA sequence encoding mature p32 was amplified by PCR from pCB6+p32 using primers p32mG and p32R3. The PCR product was digested with EcoRI and BamHI and subcloned into pEGFP-C1 (Clontech) cut with the same restriction enzymes.
Radioimmunoprecipitation and RNA-binding assay. Transfected COS cells were metabolically labeled with [35S]methionine-cysteine and radioimmunoprecipitated as described previously (23). Immune complexes were washed with phosphate-buffered saline containing 0.1% Triton X-100 to preserve protein-protein interactions. Radioimmunoprecipitation of 33P-labeled capsid and RNA-binding assays were performed as described elsewhere (31).
Northern blot analyses. Vero cells were infected with wild-type and mutant viruses (multiplicity of infection [MOI] = 1), and total RNA was harvested using TRIzol reagent (Invitrogen) at various times postinfection. The RNA (10 μg/lane) was separated on 1% agarose gels containing 3.7% formaldehyde and then transferred to Hybond N membranes using a TurboBlotter apparatus (Schleicher & Schuell). Membranes were probed with a capsid cDNA that had been radiolabeled with [32P]dCTP using a Random Primers DNA labeling system (Invitrogen). Hybridization and washing protocols were performed with reagents corresponding to entire capsid coding region and NorthernMax (Ambion) reagents. After washing, membranes were exposed to a phosphorimager screen overnight.
Generation of stable cell lines inducibly expressing capsid. To create stable cell lines expressing capsid, subconfluent T-REx 293 (Invitrogen) cells (1 x 106 cells/60-mm dish) stably expressing the tetracycline repressor protein were transfected with 8 μg pcDNA5T/0-capsid. At 48 h posttransfection, cells were split 1:10 and cultured in medium containing hygromycin B (5 μg/ml) and blasticidin (100 μg/ml). Surviving colonies were isolated using cloning cylinders and tested for inducible expression of capsid by indirect immunofluorescence and by immunoblot analyses. Multiple clones were analyzed; however, clone 293TRC 2B was used for all experiments described in this study.
Immunofluorescence microscopy. Vero cells grown on coverslips were processed for indirect immunofluorescence microscopy 24 h after transfection. To visualize mitochondria, Mitotracker Red CMXRos was added to the cell culture medium at a final concentration of 30 ng/ml for 20 min at 37°C prior to fixation. Samples were processed as described previously (6) and then examined using a Zeiss 510 confocal microscope. Images from optical sections (0.8 μm) were processed using Adobe Photoshop 7.0.
Electron microscopy. HEK293TRC 2B cells were induced to express capsid by the addition of doxycycline (1 μg/ml) to the culture medium. Uninduced HEK293TRC 2B cultures served as the negative controls. Cells were processed for electron microscopy 45 h postinduction as described previously (16). Images were captured using a Megaview 3 charge-coupled device camera (Soft Imaging System).
Yeast two-hybrid analysis. Interactions between capsid deletions (subcloned into pGBT9 or pGBTK7) and full-length p32 (in pGAD10 [6]) were assayed by cotransforming AH109 cells and testing for growth on medium lacking adenine, histidine, leucine, and tryptophan (4DO). The positive control was a Clontech system control that utilizes two strongly interacting proteins, p53 (in pGBKT7) and simian virus 40 large T antigen (in pGADT7). As a negative control, AH109 cells were transformed with two plasmids encoding proteins that do not interact (pGBKT7-p53 and pGAD-ABP280). Growth on the 4DO plates within 4 days (at 30°C) was interpreted as evidence of a positive interaction. All media and transformations were performed according to the protocols described in the Clontech MATCHMAKER system.
Synthesis of infectious viral RNA and production of mutant viruses. Plasmids containing full-length RV cDNA clones were linearized with HindIII and used as templates for transcription of capped RNAs using the mMessage mMachine kit (Ambion). RNAs were quantitated by electrophoresis and by spectroscopy at 260 nm. RNA was introduced into BHK cells by electroporation as described elsewhere (31). Virus-containing culture media were collected at various time points and then clarified at 7,000 x g for 10 min before storage at –80°C. Virus titers were determined by plaque assay as described previously (31).
One-step growth curves were determined as follows: Vero cells (2 x 105/35-mm dish) were infected with recombinant viruses (MOI = 5) for 2 h at 35°C. Cells were washed once with phosphate-buffered saline and then incubated for 1 h in growth medium. Fresh medium was then added, and incubation continued at 35°C. Medium was removed at 12-h intervals and stored at –80°C, and titers were determined by plaque assay.
To assess the synthesis of viral proteins, Vero cells were infected with M33 or mutant strains (5RA and 6RA) at an MOI of 1.0. Cells were harvested at various times postinfection, and lysates were prepared. Samples were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting with antibodies to capsid and the nonstructural protein p150.
Virus particle assembly assay. COS cells (1.5 x 105/35-mm dish) were transiently transfected with 2 μg of the pCMV5-24S constructs encoding wild-type and mutant capsids. Assembly and secretion of RV-like particles was assayed as described elsewhere (30).
RESULTS
Capsid expression induces formation of intermitochondrial plaques. Mitochondria have long been thought to have an important role in the replication of RV (1, 2, 33). Moreover, we recently demonstrated that expression of the RV structural proteins induces relocalization of mitochondria to the perinuclear region (6). Given that the capsid protein but neither of the RV glycoproteins is intimately associated with mitochondria (6, 35), we rationalized that capsid is most likely to be the virus protein responsible for this process. To test this hypothesis, Vero cells were transiently transfected with expression vectors encoding either capsid alone or the RV glycoproteins (E2 and E1). Indeed, perinuclear clustering of mitochondria was evident in transfected cells expressing capsid alone but not in untransfected cells (Fig. 1A to D). In contrast, expression of E2 and E1 did not significantly affect the distribution of mitochondria (Fig. 1E to F). Together, these observations support our hypothesis that capsid expression induces reorganization of mitochondria.
To further examine the effects of capsid expression on mitochondrial morphology, a stable cell line (HEK 293TRC 2B) was created in which capsid expression was under control of the tetracycline repressor. Addition of doxycycline to the culture medium resulted in robust expression of capsid protein (Fig. 2A). HEK 293TRC 2B cells were cultured for 48 h under inducing or noninducing conditions and then processed for routine electron microscopy. Electron-dense structures were observed between apposing mitochondria in cells cultured in the presence of doxycycline (Fig. 2C to E) but were not present in the noninduced cells (Fig. 2B). These electron-dense structures are reminiscent of the confronting membranes found in RV-infected cells (34). The width of confronting membranes in RV-infected cells was reported to be 22 to 25 nm. In accordance with this observation, the average width of the electron-dense zones in capsid-expressing cells was 24 nm. These results indicate that expression of the RV capsid is sufficient to induce the formation of confronting membranes, a hallmark of RV infection.
Capsid-p32 interactions. Analysis of the RV capsid sequence using the Web-based algorithm PSORT II Prediction (http://psort.nibb.ac.jp/form2.html) revealed that classical mitochondrial targeting sequences are absent from this protein. As such, the mechanism of capsid targeting to mitochondria is not clear. However, capsid is known to bind to the carboxyl terminus of mitochondrial matrix protein p32 (reference 41 and unpublished data), a situation that would leave the amino-terminal mitochondrial targeting sequence of p32 unobstructed and presumably functional. Accordingly, it is tempting to speculate that targeting of capsid to mitochondria involves piggybacking onto nascent p32. If this is the mechanism by which capsid is targeted to the mitochondria, it should be possible to block association of capsid with mitochondria by expressing a form of p32 that is not targeted to the mitochondria. To create such a construct, the mitochondrial targeting sequence of p32 was replaced with GFP to create the fusion protein GFPp32m. It was expected that GFPp32m would act in a dominant negative fashion and abrogate targeting of capsid to mitochondria.
Vero cells were transiently transfected with plasmids encoding GFP or GFPp32m either alone or together with capsid. When coexpressed with GFP, capsid displayed a characteristic juxtanuclear localization, while GFP was diffuse throughout the cytoplasm and nucleus (Fig. 3D to F). In addition, the localization of GFP was not altered by the expression of capsid, indicating that these two proteins do not interact (Fig. 3A to F). When expressed alone, GFPp32m was not associated with mitochondria-like structures but, rather, was localized throughout the cytoplasm in diffuse as well as punctate structures (Fig. 3G to I). As p32 is normally present in homo-oligomeric complexes (24), it is likely that the punctate structures are aggregates of p32. As expected, coexpression of capsid with GFPp32m resulted in the colocalization of GFPp32m with capsid (Fig. 3J to L). However, unexpectedly, coexpression of these proteins resulted in a dramatic relocalization of GFPp32m to compact juxtanuclear structures that resembled those of mitochondrially targeted capsid (Fig. 3F).
To ascertain whether the capsid-GFPp32m interactions were occurring at mitochondria, Vero cells were transfected with plasmids encoding either GFPp32m alone or GFPp32m and capsid. Mitotracker Red CMXRos was used to visualize mitochondria for these experiments. When expressed alone, GFPp32m did not associate with mitochondria but, instead, exhibited a largely diffuse localization with some puncta throughout the cytoplasm. In cells expressing this protein, mitochondria appeared identical to those in untransfected control cells (Fig. 4A to C). In contrast, when GFPp32m was coexpressed with capsid it was confined to structures that were stained with the Mitotracker dye (Fig. 4D to F). These results suggest that capsid is directed to mitochondria by a mechanism that is independent of the p32 mitochondrial targeting sequence.
An arginine-rich region of capsid is required for binding to p32. Our strategy to block binding of capsid to mitochondria using a "dominant negative" form of p32 proved unsuccessful, since targeting of capsid to this organelle appears to be independent of the p32 mitochondrial targeting sequence. In lieu of this approach, we elected to create replication-competent RV strains encoding capsid proteins that do not bind to p32. Accordingly, the first step in this process required mapping of the p32-binding site within capsid. Previously, using the yeast two-hybrid assay, we localized a p32-binding site within the amino-terminal region of capsid (6). To more precisely map the p32-binding site, we assayed the interactions between full-length p32 and a series of overlapping capsid constructs. Results from this analysis are shown in Fig. 5A. These data indicate that the minimum region of capsid that strongly interacts with p32 is comprised of amino acid residues 30 to 69.
Inspection of the p32-binding region in capsid revealed that it contains 11 arginine residues that are distributed into two clusters (Fig. 5B). Interestingly, it has been reported that the p32-binding sites of several other proteins are rich in arginine residues (8, 18, 37, 43, 54). For this reason, we focused our subsequent analysis on the arginine residues in this region of capsid. Site-directed mutagenesis was used to change the arginine residues in the two clusters to alanine residues. The resulting capsid constructs were named 5RA, 6RA, and 11RA (Fig. 5B).
Coimmunoprecipitation assays were used to determine if the arginine residue clusters in the p32-binding domain of capsid are required for stable binding to p32. COS cells were transiently transfected with expression vectors encoding wild-type or mutant capsids and the RV glycoproteins E2 and E1. At 40 h posttransfection, cells were biosynthetically labeled with [35S]methionine-cysteine and cell lysates were prepared under nondenaturing conditions. Goat anti-p32 serum was used to isolate immune complexes from lysates, which were then analyzed by SDS-PAGE and fluorography. As expected, wild-type capsid coimmunoprecipitated with p32 (Fig. 6A, upper panel). In contrast, capsids containing the 5RA, 6RA, or 11RA mutations did not form stable complexes with p32. To ensure that the mutant capsid constructs were stably expressed in transfected cells, capsid was also immunoprecipitated from cell lysates using rabbit anticapsid serum (Fig. 6A, lower panel). Together, these results indicate that the mutant capsids are expressed at levels comparable to the wild-type M33 capsid and that both of the arginine clusters within the p32-binding domain are necessary for stable binding to p32. This was not unexpected, since our results obtained using the yeast two-hybrid system indicated that portions of capsid containing only the amino-terminal arginine residue cluster (amino acid residues 1 to 45) or the carboxyl-terminal arginine residue cluster (amino acid residues 46 to 89) were unable to interact with p32. In contrast, a construct encoding amino acid residues 30 to 69, which contains both arginine clusters, interacted with p32 as well as full-length capsid (Fig. 5A).
Functional analysis of capsid RA mutants. Before proceeding to construction of recombinant viruses, it was important to ensure that the capsid RA mutants were functional, since the distribution and presumably intracellular binding interactions of capsid are influenced by the presence of glycoproteins E2 and E1 (16). Therefore, most of the following experiments were conducted under conditions that mimic virus assembly; specifically, capsid was coexpressed together with E2 and E1. The targeted RA mutations in the p32-binding site are in close proximity to serine 46, which is the primary phosphorylation site of capsid. Regulated phosphorylation of serine 46 has been shown to be important for virus replication, an effect that is likely mediated through its role in modulating the RNA-binding activity of capsid (31). The levels of phosphorylation of the various capsid mutants were compared to wild-type capsid using radioimmunoprecipitation and fluorography assays. Transfected cells were biosynthetically labeled with H333PO4 for 12 h before capsid proteins were immunoprecipitated, separated by SDS-PAGE, and then subjected to fluorography. Wild-type capsid and capsids containing either the 5RA or the 6RA mutations were phosphorylated to the same extent (Fig. 6B). In contrast, the 11RA mutant was not stably phosphorylated. Immunoblot analysis showed that all of the mutant capsids were expressed at comparable levels in the transfected cells. These results indicate that alteration of one, but not both, of the arginine clusters does not appreciably alter the phosphorylation state of capsid.
Next, we investigated whether the arginine-to-alanine mutations affected the ability of capsid to assemble into virions. As a convenient method to study parameters that govern virus assembly, we have routinely used an RV-like particle (RLP) assay. Coordinated expression of the RV structural proteins results in the formation of RLPs, which resemble native virions in terms of morphology, antigenicity, and immunogenicity (15, 16, 21, 30). Wild-type and mutant capsids were coexpressed with or without RV glycoproteins in COS cells, and RLPs were isolated from the precleared media of transfected cells by ultracentrifugation at 100,000 x g. The presence of pelletable capsid in the medium indicates that RLPs have been assembled and secreted. When coexpressed with the RV glycoproteins, wild-type capsid and capsids containing the 5RA or the 6RA cluster mutations were able to support assembly and secretion of RLPs (Fig. 6C, lower panel). In contrast, RLPs were not formed when wild-type capsid was expressed in the absence of E2 and E1. Comparable amounts of wild-type, 5RA, and 6RA capsids were detected in the medium, suggesting that RLPs are produced and secreted with similar efficiencies in each case. However, transfected cells expressing the 11RA capsid mutant did not secrete detectable amounts of RLPs. The presence of intracellular capsid in the transfected cells was confirmed by immunoblotting of cell lysates (Fig. 6C, upper panel). Together, these results indicate that mutation of either the proximal or the distal arginine clusters does not affect capsid phosphorylation, or its ability to support virus particle assembly, and by extrapolation, the formation of RV virions. These results were not surprising given that, by analogy to alphavirus capsid proteins (14) and analyses using protein structure algorithms, the amino-terminal one-third of capsid is thought to be relatively unstructured.
We next examined the RNA-binding activities of the arginine-alanine cluster mutants in an in vitro RNA-binding assay (31, 36). Since the p32-binding site overlaps with the RNA-binding site, we were concerned that ablating the arginine residues in this region would negatively affect RNA binding. To increase the affinity of capsids for RNA, samples were first treated with phosphatase prior to binding analyses (31). Interestingly, no loss of RNA-binding activity was associated with the capsid 5RA mutant (Fig. 6D). Rather, this mutant appeared to bind genomic RNA with higher affinity than wild-type capsid but similarly to the hypophosphorylated mutant S46A (31). However, it should be noted that these in vitro binding assays are somewhat variable, and in previous experiments we have documented that dephosphorylated wild-type capsid binds RNA as well as the S46A mutant (31). Accordingly, the important result here is that the five proximal arginine residues in the p32-binding site of capsid are not required for RNA binding. In contrast, the 6RA capsid bound RNA only very weakly (data not shown), indicating that the distal arginine cluster is important for binding to genomic RNA.
The p32-binding site of capsid is important for virus replication. The results described above suggest that the capsid 5RA and 6RA mutants are structurally intact and can function normally in assembly of virus particles. We next sought to determine the importance of the p32-binding site in capsid function during virus replication. The mutant capsid genes were subcloned into an RV infectious clone (55), and capped genomic RNAs were synthesized in vitro. BHK-21 cells were electroporated with equal amounts of RNA (10 μg), and media were collected at scheduled intervals starting at 12 h postelectroporation. Infectious virus released from the cells was quantitated by plaque assay in RK13 cells. Wild-type M33 RV produced large, clear plaques that were easily visible by crystal violet staining (Fig. 7A). The 5RA strain produced plaques that were smaller and more opaque than M33 plaques. The 6RA strain produced plaques that were even smaller than 5RA plaques. Not surprisingly, the 11RA mutant did not produce detectable amounts of infectious virus (data not shown). These results indicate that viruses encoding capsid proteins that do not interact with p32 exhibit decreased abilities to produce and/or secrete infectious virions.
To confirm that viral proteins were adequately expressed in the electroporated cells and that decreased virus titers were not the result of inefficient electroporation, BHK cells electroporated with equivalent amounts of infectious RNA (10 μg) were processed for double labeling immunofluorescence 3 days postelectroporation. Cells were stained with antibodies against the RV envelope protein E1 and p32 to label cells expressing viral proteins and as a counterstain, respectively. The percentages of cells expressing E1 were determined in six randomly selected fields from two independent experiments (Table 2). Similar levels of E1-positive cells were observed in samples electroporated with wild-type, 5RA, and 6RA genomic RNAs (88 to 97%). However, a much smaller fraction (26%) of cells electroporated with 11RA genomic RNA was positive for E1 expression. The lower percentage of cells expressing E1 in the 11RA mutant probably results from the inability of the mutant capsids to support assembly of infectious virus particles and, thus, the percentage of E1-expressing cells in this case likely reflects the electroporation efficiency. These results also confirm that the 5RA and the 6RA mutant viruses are able to replicate and secrete infectious virus particles that infect neighboring cells. Therefore, these results also provide indirect evidence that both the 5RA and 6RA capsids are able to bind to and incorporate genomic RNA into virions. However, the virus titers obtained from electroporated BHK cells were consistently lower (100- to 1,000-fold) for the 5RA and 6RA RNAs compared to the M33 RNA (data not shown).
To rule out the possibility that the lower titers associated with the 5RA and 6RA viruses were not due to replication or secretion defects specific to BHK cells, conditions that approach one-step growth conditions in Vero cells were employed. Although infected BHK cells produce more intracellular infectious virus than Vero cells, the latter typically release more infectious virus than BHK cells (3). Consequently Vero cells are most often used for propagation of RV. Titers for virus stocks obtained from electroporated BHK cells were determined and used to infect Vero cells (MOI = 5 PFU/cell). Media were collected every 12 h postinfection, and virus titers were determined by plaque assay. In all cases, viral titers peaked between 60 and 72 h; however, the 5RA and 6RA titers were almost 3 orders of magnitude lower than M33 (Fig. 7B). These results suggest that the arginine residues in the p32-binding site are important for efficient virus replication in two different cell types.
M33 and 5RA strains exhibit different kinetics in viral protein and RNA synthesis. To better understand how the lack of a stable capsid-p32 interaction leads to replication defects, we chose to examine early and late steps in the virus life cycle. The nonstructural protein p150 is translated directly from the genomic RNA and is therefore an early event (14a). In contrast, capsid is synthesized from a subgenomic mRNA and is a later event in the replication scheme (44). Typically, nonstructural and structural proteins are first detected in Vero cells at 12 and 16 h postinfection, respectively (19). Since the 6RA capsid does not bind well to genomic RNA, we used the 5RA strain for these experiments. Vero cells were infected with either M33 or 5RA RV strains, and cell lysates were prepared at various times postinfection. Samples were run on SDS-PAGE, and the levels of p150 and capsid were determined by immunoblotting (Fig. 8A). The profiles of p150 expression were similar in M33- and 5RA-infected cells, although levels of p150 were slightly lower in the latter case. At 24 h postinfection, cells infected with M33 and 5RA viruses contained similar levels of p150 protein. Between 24 and 48 h, levels of p150 increased in both groups of infected cells (Fig. 8A). Consistent with its structural role in virus assembly at a late step, the levels of capsid in M33-infected cells increased significantly in a time-dependent manner. In contrast, the intracellular levels of the 5RA capsid did not increase over time. This was not due to inherent instability in these mutant capsids, as transfection experiments revealed that both 5RA and 6RA capsids were stable when expressed from plasmids (Fig. 6). Moreover, levels of the two other RV structural proteins E2 and E1 were dramatically lower in 5RA-infected cells (data not shown). Together, these results suggest that ablation of the p32-binding site within capsid results in replication defects at a relatively late step in the virus life cycle.
To investigate this defect further, the levels of virus-specific RNA were monitored at various times postinfection. Whereas the genomic RNA was seen to accumulate with similar kinetics in the M33- and 5RA-infected cells, the accumulation of subgenomic RNA was impaired in the latter group of infected cells (Fig. 8B). These results suggest that the relatively low level of structural protein synthesis observed in 5RA-infected cells is due mainly to a defect in subgenomic RNA production.
Capsid-p32 interactions mediate perinuclear clustering of mitochondria. To gain other insights into how abolishing capsid-p32 interactions inhibits virus replication, we next examined how mitochondrial clustering was affected by expression of the mutant capsid proteins. Vero cells were transiently transfected with vectors encoding wild-type capsid or the capsid RA mutants and processed as described above. As described previously, capsid expression induced a dramatic rearrangement of mitochondria (Fig. 9). Mitochondria were distributed in a lacey pattern throughout the cytoplasm of untransfected cells but were noticeably compacted in the juxtanuclear regions of the cells expressing wild-type capsid. In cells expressing the capsid 5RA or 6RA mutants, mitochondrial staining patterns were more punctate and condensed into juxtanuclear regions than mock-transfected cells, but not as pronounced as in cells expressing wild-type capsid. These results suggest that the interaction of capsid with p32 facilitates the recruitment of mitochondria to the juxtanuclear region.
DISCUSSION
Nonstructural roles for the capsid protein. The RV capsid protein assembles with the viral genome to form the structural core of the virion interior. In addition to this structural function, there is increasing evidence that the capsid plays integral roles in several other important processes, including induction of apoptosis (13) and replication of virus-specific RNA (9, 52). In keeping with its structural function, a large pool of capsid is expected to associate with organelles of the secretory pathway where nucleocapsid assembly and virus budding occur. However, recent studies by a number of laboratories including our own have revealed the existence of capsid pools that associate with nonsecretory organelles. These sites include the cytoplasmic surface of mitochondria (6, 35) and endosome-associated viral replication complexes (28, 35). Conceivably, the cohort of capsid associated with replication complexes functions in some aspect of RNA metabolism, such as transcription, stabilization, and/or packaging of nascent virus-encoded RNAs. The association of capsid with mitochondria appears to be unique to RV among togaviruses (32) and, until now, has not been formally addressed.
In the present study, we found that capsid expression in the absence of other viral proteins has dramatic effects on the morphology and distribution of mitochondria. Specifically, perinuclear clustering of mitochondria and formation of electron-dense zones between adjacent cisternae were observed, both of which are hallmarks of RV-infected cells (33). The underlying mechanism that leads to these defects is not clear, but a potential link to RV-induced mitochondrial abnormalities came from the finding that capsid interacts with the host-encoded protein p32 (6). The bulk of this host protein resides in the matrix of mitochondria (11, 42, 48). Despite the intense efforts of numerous laboratories, the normal physiological function of p32 has remained elusive. However, a recent study suggests that this protein is required for the proapoptotic function of the BH3-only protein Hrk (50). Interestingly, like the RV capsid, Hrk binds to the highly conserved carboxyl-terminal region of p32. It is not known if the binding sites for these two proteins overlap.
A large number of studies have documented the interactions between p32 and cellular proteins as well as virus proteins (7, 8, 18, 25, 29, 37, 38, 54, 56). In most cases, interactions between virus-encoded proteins and p32 occur at nonmitochondrial sites. Given that the proapoptotic function of p32 is dependent upon its localization to mitochondria (50), it is conceivable that some p32-dependent virus-host interactions serve to prevent or delay virus-induced cell death.
Of particular relevance to the present study is the documented role of p32 in RNA metabolism (27, 45). Indeed, interactions between p32 and virus proteins are known to be important for virus replication. For example, expression of human p32 in murine cells overcomes a posttranscriptional block in replication of human immunodeficiency virus (HIV) (57). In addition, p32 is required for latent-cycle DNA replication of Epstein-Barr virus (53). By analogy, interactions between p32 and the RV capsid appear to be important for transcription of the RV subgenomic RNA. Consistent with our present findings, overexpression of p32 reportedly enhances replication of RV by an as-yet-undefined mechanism (41). Interestingly, the level of capsid expression and presumably other structural proteins was dramatically increased when cells that overexpress p32 were infected with RV.
The capsid p32-binding site is critical for replication. We were somewhat puzzled that our mapping studies of the p32-binding region in capsid conflicted with those of a previous study (41). Mohan et al. reported that the p32-binding site resides with amino acid residues 1 to 28 of capsid. In contrast, our analysis indicated that the p32-binding site is located between amino acid residues 30 to 69. It seems unlikely that there are two separate p32-binding sites since, using the same type of assay (yeast two hybrid), we did not detect interactions between p32 and amino acid residues 1 to 45 of capsid (6). While we are currently unable to reconcile the differences between these two studies, our findings are consistent with those of several groups who have reported that the p32-binding regions of other proteins contain multiple clusters of arginine residues (8, 18, 54). For example, two arginine-rich clusters within the p32-binding domain of Epstein-Barr virus EBNA-1 are necessary for stable interaction with p32 in cultured cells (53, 54). Similarly, we observed nearly identical results for the RV capsid. Specifically, two separate arginine clusters (5RA and 6RA) between amino acid residues 30 and 69 are required for stable capsid-p32 interactions.
The p32-binding site is located in a region of capsid that is thought to be relatively unstructured. Accordingly, ablation of this site was not expected to compromise the structural integrity of capsid. Indeed, both 5RA and 6RA mutants were phosphorylated normally and functioned at wild-type capacity in the virus assembly assay. However, viruses containing the 5RA and 6RA mutations replicated to titers that were between 100- and 1,000-fold lower than the wild-type M33 strain. The defect in replication for the 5RA mutant virus, at least, probably occurs at a relatively late step but before nucleocapsid assembly. Whereas accumulation of the nonstructural protein p150 was not greatly affected in cells infected by the 5RA virus strain, accumulation of structural proteins was greatly diminished. Further analysis indicated that the levels of subgenomic RNA were significantly lower in cells infected with the 5RA virus strain. Accordingly, we favor the hypothesis that the RNA replication-specific roles of capsid (9, 52) are dependent upon interactions with p32. However at this point, we cannot rule out the possibility that the decreased levels of structural proteins in 5RA-infected cells are partly the result of translational defects. Specifically, we observed that the 5RA capsid binds RV RNA better than the wild-type capsid. Thus, it is also possible that tight binding of the 5RA capsid to the 24S RNA inhibits translation.
Effects of capsid on mitochondria may be mediated by interactions with p32. In addition to being less effective in supporting virus replication, expression of the 5RA and 6RA capsids resulted in reduced perinuclear clustering of mitochondria relative to M33 capsid. Presumably, capsid is targeted to the cytoplasmic surface of mitochondria independently of p32 and, once associated with the cytoplasmic surface of this organelle, is then able to complex with p32. Under certain circumstances, p32 is able to leave the mitochondrial matrix (49). It is therefore possible that capsid binds to a pool of p32 as it leaves the mitochondria, although it is not at all clear how the association of capsid with this organelle would induce release of p32. Accordingly, we favor the scenario that capsid binds to the carboxyl-terminal region of newly synthesized p32 shortly before it is translocated into mitochondria. This would be expected to leave the amino-terminal mitochondrial targeting sequence of p32 exposed, thereby allowing at least partial translocation of p32 into the mitochondrial matrix. Alternatively, its association with capsid may inhibit import of p32 into mitochondria. A similar mechanism prevents import of the cellular enzyme fumarase into the mitochondria of Saccharomyces cerevisiae (26). Thus, retention of p32 in the cytosol would result in the formation of capsid-p32 complexes at the cytoplasmic surface of mitochondria. Since both capsid and p32 are known to form homo-oligomers (5, 24), it is tempting to speculate that capsid-p32 complex functions as a scaffold or a molecular adhesive to drive aggregation of mitochondria.
The function of mitochondrial aggregation in RV-infected cells is intriguing but not yet understood. Cells infected with other viruses, including Semliki Forest virus (33), Flock House virus (40), and African swine fever (47), exhibit similar mitochondrial defects. As with the RV capsid, expression of the hepatitis B virus X protein is sufficient to induce clustering of mitochondria to a juxtanuclear region in the absence of other viral proteins (51). While X protein expression has been linked to apoptotic cell death, clustering of mitochondria in itself is not likely to be the cause of death, since Vero cells expressing the RV capsid display aggregated mitochondria but do not show visible signs of apoptosis. Rather, it is thought that aggregation of mitochondria to sites in close proximity to virus replication or assembly factories serves to provide a source of cellular energy to assist virus replication (17, 33, 40, 47). A second but not mutually exclusive possibility is that the capsid-induced association of mitochondria with virus replication sites facilitates the incorporation of cardiolipin into virions (2).
In conclusion, our data are consistent with a scenario in which capsid interaction with p32 and consequent aggregation of mitochondria are beneficial for virus replication. Importantly, these data are in agreement with a corollary study from another laboratory, which showed that replication of RV is enhanced in cells overexpressing p32 (41). The challenge now is to understand the mechanism by which capsid-p32 interactions influence nonstructural functions of capsid, in particular the synthesis of subgenomic RNA.
ACKNOWLEDGMENTS
We thank Shirley Gillam, Jerry Wolinsky, and Terry Ahola for their gifts of reagents. Margaret Hughes is acknowledged for excellent technical assistance.
The work was funded by a grant to T.C.H. from the Canadian Institutes of Health Research (CIHR). T.C.H. is the recipient of a Senior Medical Scholarship from the Alberta Heritage Foundation for Medical Research (AHFMR). M.D.B. was supported by graduate studentship awards from AHFMR and CIHR. J.C.E. is supported by a Canada graduate scholarship. L.M.J.L. is supported by a graduate studentship award from AHFMR.
REFERENCES
Bardeletti, G. 1977. Respiration and ATP level in BHK21/13S cells during the earliest stages of rubella virus replication. Intervirology 8:100-109.
Bardeletti, G., and D. C. Gautheron. 1976. Phospholipid and cholesterol composition of rubella virus and its host cell BHK 21 grown in suspension cultures. Arch. Virol. 52:19-27.
Bardeletti, G., J. Tektoff, and D. Gautheron. 1979. Rubella virus maturation and production in two host cell systems. Intervirology 11:97-103.
Baron, M. D., T. Ebel, and M. Suomalainen. 1992. Intracellular transport of rubella virus structural proteins expressed from cloned cDNA. J Gen. Virol. 73:1073-1086.
Baron, M. D., and K. Forsell. 1991. Oligomerization of the structural proteins of rubella virus. Virology 185:811-819.
Beatch, M. D., and T. C. Hobman. 2000. Rubella virus capsid associates with host cell protein p32 and localizes to mitochondria. J. Virol. 74:5569-5576.
Bruni, R., and B. Roizman. 1996. Open reading frame P—a herpes simplex virus gene repressed during productive infection encodes a protein that binds a splicing factor and reduces synthesis of viral proteins made from spliced mRNA. Proc. Natl. Acad. Sci. USA 93:10423-10427.
Bryant, H. E., D. A. Matthews, S. Wadd, J. E. Scott, J. Kean, S. Graham, W. C. Russell, and J. B. Clements. 2000. Interaction between herpes simplex virus type 1 IE63 protein and cellular protein p32. J. Virol. 74:11322-11328.
Chen, M. H., and J. P. Icenogle. 2004. Rubella virus capsid protein modulates viral genome replication and virus infectivity. J. Virol. 78:4314-4322.
Cooper, L. Z. 1969. Rubella in pregnancy. Postgrad. Med. 46:106-107.
Dedio, J., W. Jahnen-Dechent, M. Bachmann, and W. Muller-Esterl. 1998. The multiligand-binding protein gC1qR, putative C1q receptor, is a mitochondrial protein. J. Immunol. 160:3534-3542.
Dominguez, G., C. Y. Wang, and T. K. Frey. 1990. Sequence of the genome RNA of Rubella virus: evidence for genetic rearrangement during Togavirus evolution. Virology 177:225-238.
Duncan, R., A. Esmaili, L. M. Law, S. Bertholet, C. Hough, T. C. Hobman, and H. L. Nakhasi. 2000. Rubella virus capsid protein induces apoptosis in transfected RK13 cells. Virology 275:20-29.
Forsell, K., M. Suomalainen, and H. Garoff. 1995. Structure-function relation of the NH2-terminal domain of the Semliki Forest virus capsid protein. J. Virol. 69:1556-1563.
Frey, T. K. 1994. Molecular biology of rubella virus. Adv. Virus Res. 44:69-160.
Garbutt, M., H. Chan, and T. C. Hobman. 1999. Secretion of rubella virions and virus-like particles in cultured epithelial cells. Virology 261:340-346.
Garbutt, M., L. M. Law, H. Chan, and T. C. Hobman. 1999. Role of rubella virus glycoprotein domains in assembly of virus-like particles. J. Virol. 73:3524-3533.
Garzon, S., H. Strykowski, and G. Charpentier. 1990. Implication of mitochondria in the replication of Nodamura virus in larvae of the Lepidoptera, Galleria mellonella (L.) and in suckling mice. Arch. Virol. 113:165-176.
Hall, K. T., M. S. Giles, M. A. Calderwood, D. J. Goodwin, D. A. Matthews, and A. Whitehouse. 2002. The herpesvirus Saimiri open reading frame 73 gene product interacts with the cellular protein p32. J. Virol. 76:11612-11622.
Hemphill, M. L., R. Y. Forng, E. S. Abernathy, and T. K. Frey. 1988. Time course of virus-specific macromolecular synthesis during rubella virus infection in Vero cells. Virology 162:65-75.
Hobman, T. C., M. L. Lundstrom, and S. Gillam. 1990. Processing and intracellular transport of rubella virus structural proteins in COS cells. Virology 178:122-133.
Hobman, T. C., M. L. Lundstrom, C. A. Mauracher, L. Woodward, S. Gillam, and M. G. Farquhar. 1994. Assembly of rubella virus structural proteins into virus-like particles in transfected cells. Virology 202:574-585.
Hobman, T. C., N. O. Seto, and S. Gillam. 1994. Expression of soluble forms of rubella virus glycoproteins in mammalian cells. Virus Res. 31:277-289.
Hobman, T. C., L. Woodward, and M. G. Farquhar. 1993. The rubella virus E2 and E1 spike glycoproteins are targeted to the Golgi complex. J. Cell Biol. 121:269-281.
Jiang, J., Y. Zhang, A. R. Krainer, and R. M. Xu. 1999. Crystal structure of human p32, a doughnut-shaped acidic mitochondrial matrix protein. Proc. Natl. Acad. Sci. USA 96:3572-3577.
Kittlesen, D. J., K. A. Chianese-Bullock, Z. Q. Yao, T. J. Braciale, and Y. S. Hahn. 2000. Interaction between complement receptor gC1qR and hepatitis C virus core protein inhibits T-lymphocyte proliferation. J. Clin. Investig. 106:1239-1249.
Knox, C., E. Sass, W. Neupert, and O. Pines. 1998. Import into mitochondria, folding and retrograde movement of fumarase in yeast. J. Biol. Chem. 273:25587-25593.
Krainer, A. R., A. Mayeda, D. Kozak, and G. Binns. 1991. Functional expression of cloned human splicing factor SF2: homology to RNA-binding proteins, U1 70K, and Drosophila splicing regulators. Cell 66:383-394.
Kujala, P., T. Ahola, N. Ehsani, P. Auvinen, H. Vihinen, and L. Kaariainen. 1999. Intracellular distribution of rubella virus nonstructural protein P150. J. Virol. 73:7805-7811.
Laine, S., A. Thouard, J. Derancourt, M. Kress, D. Sitterlin, and J. M. Rossignol. 2003. In vitro and in vivo interactions between the hepatitis B virus protein P22 and the cellular protein gC1qR. J. Virol. 77:12875-12880.
Law, L. M., R. Duncan, A. Esmaili, H. L. Nakhasi, and T. C. Hobman. 2001. Rubella virus E2 signal peptide is required for perinuclear localization of capsid protein and virus assembly. J. Virol. 75:1978-1983.
Law, L. M., J. C. Everitt, M. D. Beatch, C. F. Holmes, and T. C. Hobman. 2003. Phosphorylation of rubella virus capsid regulates its RNA binding activity and virus replication. J. Virol. 77:1764-1771.
Lee, J. Y., and D. S. Bowden. 2000. Rubella virus replication and links to teratogenicity. Clin. Microbiol. Rev. 13:571-587.
Lee, J. Y., D. S. Bowden, and J. A. Marshall. 1996. Membrane junctions associated with rubella virus infected cells. J. Submicrosc. Cytol. Pathol. 28:101-108.
Lee, J. Y., D. Hwang, and S. Gillam. 1996. Dimerization of rubella virus capsid protein is not required for virus particle formation. Virology 216:223-227.
Lee, J. Y., J. A. Marshall, and D. S. Bowden. 1999. Localization of rubella virus core particles in Vero cells. Virology 265:110-119.
Liu, Z., D. Yang, Z. Qiu, K. T. Lim, P. Chong, and S. Gillam. 1996. Identification of domains in rubella virus genomic RNA and capsid protein necessary for specific interaction. J. Virol. 70:2184-2190.
Luo, Y., H. Yu, and B. M. Peterlin. 1994. Cellular protein modulates effects of human immunodeficiency virus type 1 Rev. J. Virol. 68:3850-3856.
Matthews, D. A., and W. C. Russell. 1998. Adenovirus core protein V interacts with p32—a protein which is associated with both the mitochondria and the nucleus. J. Gen. Virol. 79:1677-1685.
McDonald, H., T. C. Hobman, and S. Gillam. 1991. The influence of capsid protein cleavage on the processing of E2 and E1 glycoproteins of rubella virus. Virology 183:52-60.
Miller, D. J., M. D. Schwartz, and P. Ahlquist. 2001. Flock house virus RNA replicates on outer mitochondrial membranes in Drosophila cells. J. Virol. 75:11664-11676.
Mohan, K. V., B. Ghebrehiwet, and C. D. Atreya. 2002. The N-terminal conserved domain of rubella virus capsid interacts with the C-terminal region of cellular p32 and overexpression of p32 enhances the viral infectivity. Virus Res. 85:151-161.
Muta, T., D. Kang, S. Kitajima, T. Fujiwara, and N. Hamasaki. 1997. p32 protein, a splicing factor 2-associated protein, is localized in mitochondrial matrix and is functionally important in maintaining oxidative phosphorylation. J. Biol. Chem. 272:24363-24370.
Nikolakaki, E., G. Simos, S. D. Georgatos, and T. Giannakouros. 1996. A nuclear envelope-associated kinase phosphorylates arginine-serine motifs and modulates interactions between the lamin B receptor and other nuclear proteins. J. Biol. Chem. 271:8365-8372.
Oker-Blom, C., I. Ulmanen, L. Kaariainen, and R. F. Pettersson. 1984. Rubella virus 40S genome RNA specifies a 24S subgenomic mRNA that codes for a precursor to structural proteins. J. Virol. 49:403-408.
Petersen-Mahrt, S. K., C. Estmer, C. Ohrmalm, D. A. Matthews, W. C. Russell, and G. Akusjarvi. 1999. The splicing factor-associated protein, p32, regulates RNA splicing by inhibiting ASF/SF2 RNA binding and phosphorylation. EMBO J. 18:1014-1024.
Robertson, S. E., D. A. Featherstone, M. Gacic-Dobo, and B. S. Hersh. 2003. Rubella and congenital rubella syndrome: global update. Rev. Panam. Salud Publica 14:306-315.
Rojo, G., M. Chamorro, M. L. Salas, E. Vinuela, J. M. Cuezva, and J. Salas. 1998. Migration of mitochondria to viral assembly sites in African swine fever virus-infected cells. J. Virol. 72:7583-7588.
Seytter, T., F. Lottspeich, W. Neupert, and E. Schwarz. 1998. Mam33p, an oligomeric, acidic protein in the mitochondrial matrix of Saccharomyces cerevisiae is related to the human complement receptor gC1q-R. Yeast 14:303-310.
Soltys, B. J., D. Kang, and R. S. Gupta. 2000. Localization of P32 protein (gC1q-R) in mitochondria and at specific extramitochondrial locations in normal tissues. Histochem. Cell Biol. 114:245-255.
Sunayama, J., Y. Ando, N. Itoh, A. Tomiyama, K. Sakurada, A. Sugiyama, D. Kang, F. Tashiro, Y. Gotoh, Y. Kuchino, and C. Kitanaka. 2004. Physical and functional interaction between BH3-only protein Hrk and mitochondrial pore-forming protein p32. Cell Death Differ. 11:771-781.
Takada, S., Y. Shirakata, N. Kaneniwa, and K. Koike. 1999. Association of hepatitis B virus X protein with mitochondria causes mitochondrial aggregation at the nuclear periphery, leading to cell death. Oncogene 18:6965-6973.
Tzeng, W. P., and T. K. Frey. 2003. Complementation of a deletion in the rubella virus p150 nonstructural protein by the viral capsid protein. J. Virol. 77:9502-9510.
Van Scoy, S., I. Watakabe, A. R. Krainer, and J. Hearing. 2000. Human p32: a coactivator for Epstein-Barr virus nuclear antigen-1-mediated transcriptional activation and possible role in viral latent cycle DNA replication. Virology 275:145-157.
Wang, Y., J. E. Finan, J. M. Middeldorp, and S. D. Hayward. 1997. p32/TAP, a cellular protein that interacts with EBNA-1 of Epstein-Barr virus. Virology 236:18-29.
Yao, J., and S. Gillam. 1999. Mutational analysis, using a full-length rubella virus cDNA clone, of rubella virus E1 transmembrane and cytoplasmic domains required for virus release. J. Virol. 73:4622-4630.
Yu, L., Z. Zhang, P. M. Loewenstein, K. Desai, Q. Tang, D. Mao, J. S. Symington, and M. Green. 1995. Molecular cloning and characterization of a cellular protein that interacts with the human immunodeficiency virus type 1 Tat transactivator and encodes a strong transcriptional activation domain. J. Virol. 69:3007-3016.
Zheng, Y. H., H. F. Yu, and B. M. Peterlin. 2003. Human p32 protein relieves a posttranscriptional block to HIV replication in murine cells. Nat. Cell Biol. 5:611-618.(Martin D. Beatch, Jason C)