|Home | About | Journals | Submit | Contact Us | Français|
The fusion glycoprotein (F) of respiratory syncytial virus (RSV), which mediates membrane fusion and virus entry, was shown to bind RhoA, a small GTPase, in yeast two-hybrid interaction studies. The interaction was confirmed in vivo by mammalian two-hybrid assay and in RSV-infected HEp-2 cells by coimmunoprecipitation. Furthermore, the interaction of F with RhoA was confirmed in vitro by enzyme-linked immunosorbent assay and biomolecular interaction analysis. Yeast two-hybrid interaction studies with various deletion mutants of F and with RhoA indicate that the key binding domains of these proteins are contained within, or overlap, amino acids 146 to 155 and 67 to 110, respectively. The biological significance of this interaction was studied in RSV-infected HEp-2 cells that were stably transfected to overexpress RhoA. There was a positive correlation between RhoA expression and RSV syncytium formation, indicating that RhoA can facilitate RSV-induced syncytium formation.
Human respiratory syncytial virus (RSV) belongs to the Pneumovirus genus of the Paramyxoviridae family. RSV is the major cause of acute lower respiratory tract illness in infants and young children (reviewed in reference 9). RSV isolates have been classified into two antigenic subgroups (A and B) on the basis of differences in reactivity with panels of monoclonal antibodies to attachment (G) protein (1, 34). The RSV envelope contains two major glycoproteins, the G and fusion (F) glycoproteins. The G glycoprotein is thought to mediate virus attachment (29), but the cell receptor has not been defined. The F glycoprotein promotes fusion of the viral and cellular membranes with subsequent transfer of viral genetic material into the cell. The F glycoprotein also promotes fusion of infected cell membrane with adjacent cell membrane, leading to the formation of syncytia. A third protein, the small hydrophobic (SH) protein, is also present in the envelope, but its function is unknown.
The F glycoprotein is synthesized as an inactive precursor, F0, which is cotranslationally modified by the addition of N-linked glycosylation in the endoplasmic reticulum. The F0 precursor is thought to assemble as a homooligomer into a tetramer (8). The F0 precursor is cleaved by cellular trypsin-like endoproteases into two disulfide-linked subunits, F1 and F2, before reaching the cell surface. The RSV F protein is structurally similar to the F proteins of other paramyxoviruses (6).
Three virus-encoded proteins, the nucleocapsid (N) protein, the phosphoprotein (P), and the RNA polymerase (L), are associated with the nucleocapsid to form a transcribing ribonucleoprotein (RNP) complex. RSV uses an additional protein expressed from the M2 gene open reading frame 1 as a transcription elongation factor (7). Previous studies indicate that the RNP complex requires cellular actin and possibly other proteins for RSV transcription (5, 19, 24). Similar involvement of cytoskeleton proteins in transcription has been observed in several other paramyxoviruses, namely, Sendai virus, measles virus, and parainfluenza virus type 3 (12, 32, 33). The interaction of RNP and the polymeric form of actin results in the alteration of structure of RNP from a loosely coiled to a moderately condensed form which appears to be favorable for transcription (12).
In addition, many enveloped viruses utilize cellular actin during the process of budding and maturation of virus particles released from the infected cells (4, 11, 44, 50). Furthermore, actin microfilaments have recently been shown to be involved in the spread of vaccinia virus between cells (10). Therefore, many enveloped viruses in general may use a common strategy for their transcription, morphogenesis, and cell-to-cell spread by utilizing cellular cytoskeletal components.
RhoA, a small GTPase of the Ras superfamily, has been shown to control a plethora of biological functions, including actin reorganization, gene expression, cell morphology, cell motility, and cell proliferation (35). RhoA is a common target for bacterial toxins and is of major importance for the entry of bacteria such as Shigella and Salmonella spp. into mammalian host cells (26, 51). It is also important in cell transformation by polyomaviruses (48). Further, it has been shown that adenovirus endocytosis requires actin cytoskeleton reorganization mediated by Rho family GTPases (30).
RhoA cycles between two states, i.e., an inactive, GDP-bound form and an active, GTP-bound form. RhoA in its active form is bound to GTP and undergoes a series of posttranslational modifications of its C-terminal end that include isoprenylation, C-terminal proteolytic cleavage, and carboxymethylation in the endoplasmic reticulum (43). The processed RhoA is then translocated to the plasma membrane, where it binds to phosphatidylserine moieties and acts upon various effector molecules (46).
The cytoskeleton requirements in paramyxovirus infection have long been recognized (14, 15). However, to this point there has been no evidence of any viral protein involvement in the direct or indirect stimulation of actin filament reorganization. The cellular proteins involved in the interaction with RSV proteins and the nature of their interactions with the cytoskeleton have not been defined. Since RSV F protein is important for RSV-induced syncytium formation, we attempted to identify F-interacting cellular proteins. A yeast two-hybrid screen was performed with RSV F as bait and a HeLa cDNA library as prey. We detected a small GTPase, RhoA, as an interacting partner of F. The interaction of F and RhoA was confirmed by various methods both in vivo and in vitro. The binding domains of F and RhoA were mapped. Further, we have shown that RhoA expression in the cell correlates with the number of RSV-induced plaques in cell culture.
The A2 strain of RSV was provided by R. Chanock, National Institutes of Health (NIH), Bethesda, Md. RSV stocks were prepared as previously described (21). HEp-2 cells were maintained in Eagle’s minimal essential medium supplemented with glutamine, gentamicin, penicillin G, and 10% fetal bovine serum.
The extracellular domain of F gene was amplified by PCR and cloned into the EcoRI and BamHI sites of the pAS2-BD (encodes Gal4 DNA-binding domain [BD]) vector (Clontech, Palo Alto, Calif.) such that a fusion between the Gal4 DNA-BD and the N terminus of the F gene is generated. Likewise, a HeLa cell cDNA library that had been constructed in the pGAD GH-AD (encodes Gal4 activation domain [AD]) vector to generate fusions between proteins encoded by the library cDNAs and the Gal4 AD was obtained from Clontech. The cotransformation and screening procedures were done as described in the manufacturer’s protocol. Briefly, the two types of hybrid plasmids were cotransformed into Saccharomyces cerevisiae Y190 reporter host strain and the cotransformants expressing interacting proteins were selected on synthetic dropout media deficient in His, Leu, and Trp. To confirm the protein interaction, primary His+ transformants were tested for expression of the second reporter gene lacZ by using a β-galactosidase assay. All positive transformants were then retested to eliminate false positives.
The extracellular domain of the F gene was cloned into the EcoRI and BamHI sites of the pM vector (Clontech) to generate fusions of F protein with the Gal4 DNA-BD (named pM-F). Similarly, the RhoA gene was cloned into the EcoRI and XbaI sites of pVP16 (Clontech) to generate fusions of the protein RhoA with the VP16 AD (VP16 transcriptional activation domain, derived from the VP16 protein of the herpes simplex virus) (named pVP16-RhoA). A third vector, pG5CAT, provides a chloramphenicol acetyltransferase (CAT) reporter gene under control of a Gal4-responsive element and the minimal promoter of adenovirus E1b (Clontech). The three vectors were cotransfected into the HEp-2 human epithelial cell line by using Lipofectamine (Gibco BRL, Grand Island, N.Y.) by standard methods. pM-F and pVP16-RhoA were also transfected alone or in combination with the vectors pM and pVP16, to determine autonomous activation of the Gal4 reporter pG5CAT by the expression plasmids and also to determine the basal transcription potential of each plasmid. The interaction between the proteins F and RhoA was assayed by measuring CAT gene expression by using a CAT enzyme-linked immunosorbent assay (ELISA) kit (Boehringer Mannheim, Indianapolis, Ind.). The level of CAT expression was determined by measuring the absorbance at 405 nm by using a microtiter plate reader (Dynatech, Chantilly, Va.).
Immunoaffinity-purified RSV F glycoprotein (a gift from Wyeth-Lederle-Praxis Biologicals, West Henrietta, N.Y.) was diluted to 200 ng/ml in carbonate buffer (pH 9.6). One hundred microliters of F suspension was applied to wells of Immulon II 96-well plates (NUNC, Roskilde, Denmark). Blocking was performed with 3% bovine serum albumin and 3% nonfat dry milk for 1 h. One hundred microliters of RhoA or Rac1, another Rho family GTPase (CalBiochem, La Jolla, Calif.), at 200 ng/ml, was added separately and incubated at room temperature for 2 h, followed by addition of a 1:4,000 dilution of anti-RhoA or anti-Rac1 monoclonal antibodies (Santa Cruz Biotech, Santa Cruz, Calif.) after washing with phosphate-buffered saline–0.1% Tween 20. After 1 h, plates were washed and a 1:7,000 dilution of goat anti-mouse immunoglobulin G (IgG) conjugated to horseradish peroxidase was added. After washing, the substrate 3,3′,5,5′-tetramethylbenzidine (Sigma, St. Louis, Mo.) was added and the color was read at 450 nm by using a microtiter plate reader. Immunoaffinity-purified RSV G glycoprotein (Wyeth-Lederle-Praxis Biologicals) was used instead of RSV F glycoprotein as a control.
Assays were designed according to the biomolecular interaction analysis (BIA) technology manual supplied by the manufacturer (Pharmacia Biosensor AB). Briefly, a capturing molecule, anti-F1 monoclonal antibody (kindly provided by Brian Murphy, NIH), was immobilized by amine coupling by using carbodiimide reaction, on the surface of a carboxymethylated dextran sensor chip. Immunoaffinity-purified F ligand was allowed to flow onto the surface of the immobilized monoclonal antibody so that the F protein was captured. Then, an analyte, RhoA protein, was allowed to flow onto the surface of immobilized ligand and the interaction was recorded on the sensorgram as resonance units (RU). Rac1 and the RSV surface glycoprotein G were used as negative analyte and ligand controls, respectively.
[35S]methionine-labeled RSV stock was prepared as previously described (21) with modification. After 24 h postinfection, the cell monolayer was washed with methionine-free medium, incubated in this medium for 45 min, and then labeled with 200 μCi of [35S]methionine (Amersham, Piscataway, N.J.) per ml of methionine-free medium for 24 h. For preparing RSV-infected and mock-infected cell lysates, HEp-2 cells were labeled with [35S]methionine 2 h before addition of [35S]methionine-labeled RSV and medium without RSV, respectively, and harvested 4 h after infection. Cells were lysed in RIPA buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.25% sodium deoxycholate, and protease inhibitors), and insoluble material was pelleted by centrifugation at 16,000 × g for 5 min. Supernatants were incubated with 2 μg of each antibody overnight at 4°C. Immune complexes were bound to protein G-Sepharose beads (Sigma) for 2 h at 4°C, washed three times with lysis buffer, and eluted with 2× sample buffer (10 mM Tris-HCl [pH 6], 10% glycerol, 2% sodium dodecyl sulfate [SDS], 10% 2-mercaptoethanol, and 0.006% phenol red). The proteins were resolved by SDS–12% polyacrylamide gel electrophoresis. For reimmunoprecipitation of the eluted fraction, the elution from the first immune complex was combined with 20 μl of 10% SDS and heated at 95°C for 5 min. The eluted proteins were then dissolved in 700 μl of lysis buffer and incubated with the appropriate antibody followed by protein G-Sepharose beads. The second elution was done as described above.
Clones encoding different RhoA deletion mutants were constructed by PCR amplification. All forward primers contained an EcoRI site, and all reverse primers carried an XhoI site. The respective forward and reverse primers were as follows: for RhoAN32 amplification, 5′-GTCCCGGAATTCGATGGAGGTGTATGTGCCCACAGTGTTTG-3′ and 5′-GCCGCTCGAGGCCAAGACAAGGCAACCAGA-3′; for RhoAN67 amplification, 5′-GTCCCGGAATTCGATGGATCGCCTGAGGCCCCTCTCCTAC-3′ and the same reverse primer as that for RhoAN32; for RhoAN110 amplification, 5′-GTCCCGGAATTCGATGGTGCCCATCATCCTGGTTGGGAATAAG-3′ and the same reverse primer as that for RhoAN32; for RhoAC155 amplification, 5′-GTCCCGGAATTCGATGGCTGCCATCCGGAAGAAACTGGTG-3′ and 5′-ATGCCGCTCGAGTCAGGTCTTTGCTGAACACTCCATGTAC-3′; and for RhoAN67-C110 amplification, the same forward primer as that for RhoAN67 and 5′-CCGCCGCTCGAGGATGGGCACGTTGGGACAGAAATGC-3′. PCR amplifications were carried out by using the pGAD GH-RhoA plasmid as a template and 30 cycles with steps of 1 min at 94°C, 1 min at 42°C, and 2 min at 72°C. PCR products were isolated and purified by agarose gel electrophoresis and were digested with EcoRI and XhoI. The resulting fragments were cloned into a pGAD GH vector that had been digested with EcoRI and XhoI. All constructs were sequenced by using a Sequenase sequencing kit (United States Biochemicals) to confirm that the correct bases were present. All primers were synthesized by IDT, Coralville, Iowa.
The F deletion mutants were constructed by PCR amplification. All forward primers contained an EcoRI site, and all reverse primers carried a BamHI site. The respective forward and reverse primers were as follows: for FN550 amplification, 5′-GCAGCATGCCATGGAGTTGCTAATCCTCAAAGC-3′ and 5′-GCGGCGCCTAGGATTTGTGGTGGATTTACCAGC-3′; for FN137 amplification, 5′-GTCCCGGAATTCATGGGATTTCTTGGTTTTTTGTTAGGTGTTGG-3′ and the same reverse primer as that as for FN550; for FN146 amplification, 5′-GTCCCGGAATTCATGTCTGCAATCGCCAGTGGCGTTGC-3′ and the same reverse primer as that for FN550; for FN155 amplification, 5′-GTCCCGGAATTCATGTCTAAGGTCCTGCACCTAGAAGGG-3′ and the same reverse primer as that for FN550; for FN224 amplification, 5′-GCATCGCGGATCCAATGTCAAATATAGAAACTGTGATAGAGTTCC-3′ and the same reverse primer as that for FN550; for FN283 amplification, 5′-GGTAGGACTAGTTAGTGGTAATTGTACTACATATGCTAAG-3′ and the same reverse primer as that for FN550; and for FN155-C467 amplification, 5′-GTCCCGGAATTCATGTCTAAGGTCCTGCACCTAGAAGGG-3′ and 5′-GCGGCGCCTAGGTCATATTATTGGTTCACCTTTTACATAGAG-3′. PGem7z-F plasmid containing the RSV F gene from strain A2 (gift from P. L. Collins, NIH) was used as a template and PCR amplified as described for RhoA constructs. The resulting fragments were cloned into the EcoRI and BamHI sites of the pAS2-BD vector. All constructs were sequenced to confirm that the correct bases were present. For FN146-C155, the following complementary oligonucleotides were synthesized with 5′ extension of EcoRI and 3′ extension of BamHI sites: forward, 5′-AATTCTCTGCAATCGCCAGTGGCGTTGCTGTATCTAAGGTG-3′; reverse, 5′-GATCCACCTTAGATACAGCAACGCCACTGGCGATTGCAGAG-3′. After annealing, the double-stranded FN146-C155 was ligated in EcoRI- and BamHI-digested pAS2-BD.
The Ecdysone-Inducible Expression system (Invitrogen, Carlsbad, Calif.) is based on the molting induction system found in Drosophila but is modified for inducible expression in mammalian cells. The system uses the steroid hormone ecdysone analog ponasterone A to activate expression of the gene of interest via a heterodimeric nuclear receptor. Ponasterone A has no detectable effect on mammalian cell physiology (36). Briefly, the human RhoA gene was PCR amplified by two external primers containing BamHI and EcoRI restriction sites by using a plasmid containing the RhoA gene as a template. The amplified product obtained after restriction digestion was cloned into the BamHI and EcoRI restriction sites of the pIND plasmid, which contains five modified ecdysone response elements upstream of a minimal heat shock promoter. The resulting construct, pIND-RhoA, was cotransfected with pVgRXR (which encodes the receptor subunits) into mammalian cells by using Lipofectamine (Gibco BRL) according to the manufacturer’s protocol. After 48 h posttransfection, cells were split into fresh media containing Zeocin (300 μg/ml) and G418 (600 μg/ml).
After stable cell lines were established, intracellular RhoA expression was induced by treating the cells with ponasterone A for 24 h. HEp-2 cells and stably RhoA-transfected HEp-2 cells that were untreated with ponasterone A were used as controls for endogenous RhoA expression. RhoA expression was confirmed by Western blot analysis by using anti-RhoA antibodies. Cells were lysed in RIPA buffer, and insoluble material was pelleted by centrifugation at 16,000 × g for 5 min. Supernatants were mixed in 2× sample buffer, and the proteins were resolved on SDS–12% polyacrylamide gel. Separated proteins were transferred to a polyvinylidene difluoride membrane by standard methods. After blocking with 3% bovine serum albumin, a 1:2,000 dilution of anti-RhoA monoclonal antibodies was added, followed by addition of a 1:4,000 dilution of alkaline phosphatase-conjugated anti-mouse antibodies. A substrate, Fast Red TR/Naphthol AS (Sigma), was added and washed after the color development.
Two-day-old HEp-2 cell monolayers, 80% confluent in 12-well plates (Costar, Cambridge, Mass.) were used for plaque assay (21). Intracellular RhoA expression was induced by treating the cells with ponasterone A for 24 h. HEp-2 cells and stably RhoA-transfected HEp-2 cells that were untreated with ponasterone A were used as controls. Twenty-four hours after treatment with ponasterone A, 100 μl of a solution containing RSV at 103 PFU/ml was added to HEp-2 cells in 12-well plates. After 3 days plates were fixed with 10% formalin and hematoxylin and eosin staining was performed. The number of RSV plaques in each well was determined.
The yeast two-hybrid system was used to screen a HeLa cell cDNA library for encoded proteins capable of binding to a fusion protein containing RSV F protein. Of 3 × 106 clones screened from the library, one positive clone that had strong β-galactosidase activity, when the plasmid encoding RSV F protein and the plasmid encoding a protein from the HeLa cell library were coexpressed, was identified (Fig. (Fig.1).1). Comparison with the Swiss-Prot and Protein Data Bank databases indicated that the amino acid sequence was identical to that of RhoA. The pGAD GH-RhoA containing RhoA gene from the HeLa cDNA library had sequences from the RhoA 5′ noncoding region, a coding region that was in frame with the Gal4 transcriptional activation domain, and the 3′ noncoding region including a poly(A) tail. To demonstrate the specificity of the interaction, the identities of the bait and prey proteins were reversed such that the extracellular domain of RSV F protein was now expressed as a Gal4 AD-F fusion while RhoA protein was expressed as Gal4 BD fusion proteins. The results confirmed that there was an interaction between F and RhoA. Interaction was not detected when the RSV G protein was coexpressed with RhoA.
We next asked whether the interaction of F with RhoA could be demonstrated in mammalian cells. To address this issue, we performed a mammalian two-hybrid analysis using transient transfection of the HEp-2 cells (Fig. (Fig.2).2). The extracellular domain of F fused to the Gal4 BD in the vector pM was cotransfected into HEp-2 cells with RhoA fused to the VP16 AD in the vector pVP16 and a reporter plasmid, pG5CAT. There was a significantly higher level (40-fold increase) of CAT expression in HEp-2 cell extracts expressing F and RhoA fusion proteins compared to the levels of CAT expression in HEp-2 cell extracts cotransfected with negative control plasmids. Thus, RSV F can interact with RhoA in vivo not only in yeast but also in mammalian cells.
To characterize further the in vivo association of RhoA and F, we examined whether the two proteins could be coimmunoprecipitated with anti-F1 monoclonal antibodies from RSV-infected or mock-infected HEp-2 cells (Fig. (Fig.3).3). HEp-2 cells were labeled with [35S]methionine 2 h before addition of [35S]methionine-labeled RSV and harvested 4 h after infection. RSV-infected cell lysates immunoprecipitated with anti-F1 monoclonal antibody (lane 1) showed F0, F1, F2, and RhoA proteins. Of the eluted precipitate from lane 1, 75% was reimmunoprecipitated with anti-RhoA antibodies (lane 2) and showed RhoA protein. Cell lysates from mock-infected HEp-2 cells immunoprecipitated with anti-F1 monoclonal antibody (lane 3) showed no F protein but when immunoprecipitated with anti-RhoA antibodies (lane 4) showed RhoA protein. Immunoprecipitation of mock-infected cell lysates with anti-F1 antibodies followed by reimmunoprecipitation of the eluted fraction with anti-RhoA antibodies did not produce F0, F1, F2, or RhoA protein (lane 5). In vitro-translated RhoA is shown in lane 6. Mouse isotype control was also used to coimmunoprecipitate RSV-infected cells. However, no protein with a molecular weight corresponding to that of F0, F1, F2, or RhoA protein was seen on the gel (data not shown). The data demonstrate that RhoA associates with the RSV F protein, thus confirming the interaction of RhoA and F in an RSV-infected mammalian cell. These data collectively demonstrate the interaction of RSV F with RhoA in vivo.
Since the in vivo experiments did not reveal whether F interacts with RhoA-GTP or RhoA-GDP, we examined whether RSV F interacts with a RhoA-GDP form in vitro by ELISA and by BIA. For in vitro assays, we used immunoaffinity-purified full-length F protein derived from RSV and purified recombinant RhoA protein or Rac1 protein expressed in Escherichia coli (Calbiochem).
Purified RSV F (20 ng/well) was applied to wells of Immulon II 96-well plates. After blocking, 20 ng of either RhoA or Rac1 was added separately to each well and the binding to F protein was detected by anti-RhoA or anti-Rac1 monoclonal antibodies. RSV F and RhoA interaction resulted in a 60-fold increase in absorbance compared to those of controls in which either recombinant Rac1 protein was added instead of RhoA or immunoaffinity-purified RSV surface glycoprotein G was added instead of RSV F protein (Fig. (Fig.4).4). This indicates that RSV F can bind RhoA-GDP and also confirms the interaction of both proteins in vitro.
To characterize further in vitro association of purified RSV F and RhoA-GDP, we examined the interaction of both proteins by real-time BIA (Fig. (Fig.5)5) using the BIAcore 2000 instrument. F protein was captured by anti-F1 monoclonal antibodies immobilized on the surface of the carboxymethylated dextran sensor chip. RhoA was allowed to flow onto the surface of immobilized F, and the interaction was recorded on the sensorgram as resonance units. As controls, Rac1 protein was used instead of RhoA and RSV surface glycoprotein G was used instead of RSV F protein. F-RhoA interaction gave a response of 976 RU (corresponding to binding of approximately 0.97 ng of RhoA to F per mm2 on the sensor chip surface) above the baseline RU. Experiments in which RSV G was used as a ligand control and Rac1 protein was used as an analyte control gave values similar to their respective baseline RU (phase a to b). These data support the ability of RSV F to interact with RhoA-GDP in vitro.
In order to map the binding domain of F protein, we constructed N-terminal and C-terminal deletion mutants of F protein by PCR amplification methods and studied the interactions in a yeast two-hybrid system (Fig. (Fig.6).6). Yeast transformants coexpressing FN550 or FN137 and RhoA proteins gave positive blue color as measured by β-galactosidase assay. A yeast transformant designated FN146 having an N-terminal nine-amino-acid deletion from F1 fusion peptide gave intense blue color, and colonies grew more rapidly and larger than FN137, indicating strong interaction with the RhoA protein. This also suggests that deleting the hydrophobic amino terminus may have resulted in a conformational change in the F1 protein which increased interaction with RhoA. Alternatively, the FN146 constructed without part of the fusion domain may be less toxic than FN137 to the yeast cells and allow more rapid growth. Yeast transformants with FN155, FN224, FN283, and FN155-C467 deletion mutant proteins did not interact with the RhoA protein as measured by β-galactosidase assay. These results suggest that the RhoA binding domain in RSV F is contained within or overlaps the region between amino acids 146 and 155 of the F protein. We then constructed FN146-C155, encoding the amino acids (146 to 155) which showed interaction with RhoA protein. The interaction suggests that the RhoA binding domain is contained within this nine-amino-acid region of RSV F.
Next, the RhoA binding domain was mapped (Fig. (Fig.7A).7A). We constructed N-terminal and C-terminal deletion mutants of RhoA protein by PCR amplification methods and studied the interactions in a yeast two-hybrid system. The yeast transformants with pAS2-FN550 and pGAD GH plasmids encoding various deletion mutants, designated RhoAN32, RhoAN67, and RhoAC155, gave blue color in a β-galactosidase assay, suggesting that binding to F was not affected. However, yeast transformant colonies with pAS2-FN550 and plasmid encoding RhoAN110 did not grow, indicating that the F protein did not bind to RhoA protein and that the binding site had been deleted. This result suggests that the F binding domain lies between amino acids 67 and 110 of RhoA. The interaction of this binding domain with F was confirmed by β-galactosidase assay by using the sequence encoding RhoAN67-C110.
We next correlated RhoA expression in HEp-2 cells with RSV-induced syncytium formation. This was carried out using a stably transfected HEp-2 cell line in which RhoA was expressed from an ecdysone-inducible promoter (Fig. (Fig.8).8). There was some leaky expression of RhoA in uninduced stably transfected HEp-2 cells (lane 2) compared to the levels of RhoA in normal HEp-2 cells (lane 1) (Fig. (Fig.8A).8A). To study the effect of RhoA overexpression on RSV infection, the induction of RhoA expression was initiated 24 h before RSV infection with the ecdysone analog, ponasterone A, and continued for 48 h after infection. RhoA expression correlated with the number (Fig. (Fig.8B)8B) and size (data not shown) of RSV-induced plaques in each cell line. There were statistically significant differences between plaque numbers for induced, RhoA-transfected cells and those for uninduced, RhoA-transfected cells and normal HEp-2 cells (P < 0.001 and P < 0.0001, respectively; two-tailed t test). Syncytium formation could be seen as early as 20 h after infection of induced, RhoA-transfected cells, whereas no syncytia were seen in ponasterone A-treated normal HEp-2 cells until after 48 h. These results demonstrate that upregulation of intracellular RhoA expression facilitates RSV-induced syncytium formation.
Our report is the first to describe a viral protein interacting with RhoA and also the first to describe a cellular ligand found to interact with an RSV protein. We have shown that RhoA interacts with RSV F both in vivo by yeast two-hybrid assay, mammalian two-hybrid assay, and coimmunoprecipitation and in vitro by ELISA and BIAcore. We have mapped the binding domains of RSV F and RhoA which may represent potential targets for the development of novel antiviral therapy. In addition, we have shown that RhoA expression correlates with the formation of RSV plaques.
Since viral surface glycoproteins with transmembrane regions are not transported to the nucleus, where the protein-protein interaction occurs in the yeast two-hybrid system, we used the extracellular domain of F to screen the HeLa cell cDNA library. This was confirmed by using a full-length F construct in a yeast two-hybrid screening in addition to the extracellular domain of F. The full-length F protein did not interact with any of the HeLa cDNA proteins as no yeast colonies grew on the selective media. The mapping studies of F and RhoA further confirmed that the extracellular domain of F is transported to the nucleus for interaction with RhoA in the yeast two-hybrid system.
We used a mammalian two-hybrid assay to confirm the interaction of F with RhoA. Because the assay is performed in mammalian cells, interactions between the proteins are more likely to be biologically significant. The results obtained in the mammalian two-hybrid assay showing a 40-fold increase in the CAT expression levels indicate that the RSV F can interact with RhoA in vivo in mammalian cells. They also support the probability that the interactions between the proteins are authentic and that the folding of binding domains of both proteins in yeast is similar to that in mammalian cells.
Unlike in yeast or mammalian two-hybrid systems, where interaction occurs in the nucleus, the site of interaction of F and RhoA in RSV-infected cells resembles natural infection, and the interaction of both the proteins is more authentic. Since RhoA is endogenously expressed in HEp-2 cells and to characterize further the in vivo association of F and RhoA in a native condition, we examined whether the two proteins could be coimmunoprecipitated with anti-F1 monoclonal antibodies from mock-infected and RSV-infected HEp-2 cells. Coimmunoprecipitation revealed that the interaction of F with RhoA occurs in RSV-infected HEp-2 cells; this is significant evidence of a direct association of the molecules during the process of infection. It is possible that RSV F and RhoA interaction may have occurred after detergent lysis of the cells. To address this concern, we mixed detergent-lysed RSV-infected cells with in vitro-translated RhoA, but no interaction was observed. These data suggest that after the F protein is treated with detergent and boiled, it is unable to interact with RhoA.
RhoA cycles between two states, i.e., an active, GTP-bound form and an inactive, GDP-bound form (3). In its inactive state, RhoA localizes to the cytoplasm in a complex with RhoA-GDP dissociation inhibitor (GDI) but translocates to the plasma membrane upon activation (46). As shown by crystallographic studies of RhoA, the structure of RhoA bound to GTP reveals a fold similar to that of RhoA-GDP but shows conformational differences localized in switch I (amino acids 28 to 38) and switch II (amino acids 61 to 78) (25, 52). The locus of binding of GTP or GDP to RhoA is in a phosphate-binding loop (amino acids 13 to 20) and the switch I region. The in vivo data from the yeast two-hybrid assay, the mammalian two-hybrid assay, and coimmunoprecipitation do not reveal whether RhoA is in the GDP- or GTP-bound form. Therefore, in order to determine whether F can bind RhoA in its GDP-bound form and whether RhoA-GDP can alter the structure of the F binding domain, we carried out in vitro binding experiments using recombinant RhoA. The results from ELISA indicate that F can bind RhoA-GDP and further suggest that the structure of the F binding domain in RhoA-GDP may not be affected. Although GTP or GDP binding to RhoA may not be a prerequisite for the association of F with RhoA in vitro, GTP binding to RhoA may be necessary for biological functions of the F and RhoA complex in vivo. It will therefore be important in future studies to define whether F binds equally well to Rho-GTP and whether this has relevance for downstream signaling events that can be mediated by RhoA. Although a number of mitogens, namely, lysophosphatidic acid, growth factors, and thrombin, are known to activate RhoA, an upstream ligand for RhoA has not been identified. The RSV F protein may therefore have value as a reagent in future studies of RhoA activation and signal transduction.
In ELISA, the F protein was immobilized by adsorption to plastic and this may at least partially alter the conformation of the protein. To address this concern and to further confirm the in vitro association of F and RhoA, we used the BIA, which is based on the surface plasmon resonance phenomenon. This method makes it possible to visualize the binding process as a function of time by monitoring the increase in refractive index that occurs when RhoA interacts with F that is captured by immobilized anti-F1 monoclonal antibody on the surface of a sensor chip. The native conformation of the immobilized F protein may be better preserved since it is bound to anti-F1 antibody rather than directly to the chip. The other advantage is that none of the proteins needs to be labeled or conjugated, which avoids the artifactual changes in binding properties that often result when proteins are labeled or conjugated (49). The results from BIA suggest that F binds RhoA with strong affinity and that the dissociation of bound RhoA is very slow in comparison to that of controls. The density of RhoA bound to F was approximately 0.97 ng/mm2 on the sensor chip surface, which is significant compared to controls. This high-affinity interaction between F and RhoA may be essential to the integrity and stability of the complex in biological systems to compete with host proteins for this domain of RhoA.
Mapping of the binding domains of F and RhoA has made it possible to determine critical domains of both proteins involved in the biological functions that may unravel new pathways involved in the replication of RSV. In mapping the domain for F in a yeast two-hybrid system, a weak interaction of RhoA with FN137 was seen, in contrast to a strong interaction of RhoA with FN146 deletion mutants. This suggests that there may be conformational determinants that affect access to the RhoA binding domain, thereby preventing RhoA interaction. These data also suggest that a conformational change or unfolding of fusion peptide may have to occur prior to or during RhoA binding. The binding domain (amino acids 146 to 154) is present within a part of the fusion peptide (amino acids 137 to 154), indicating that this region may be important for events involving binding with RhoA leading to virus entry. It is also possible that by stimulating RhoA, the various biological functions of RhoA, such as actin bundling, may be utilized for cell-to-cell spread by syncytium formation, virus assembly, and maturation. Since RSV infection in cell culture can occur despite the lack of the putative G and SH proteins (27), it is not surprising that F may have additional unknown functions. It is well known that F is involved in virus entry and syncytium formation, but it is possible that F may also be involved indirectly in virus maturation. For example, RhoA activation promotes reorganization of actin which could potentially serve as scaffolding in the formation of filamentous RSV particles.
It has been well established that the key molecular determinants for RhoA-effector protein binding are the switch I and switch II domains (17). Recently, a determinant for effector binding located between RhoA residues 75 and 92 was identified (53). The crystal structure of RhoA (25, 41, 42) indicates that the F protein binding domain in RhoA between amino acids 67 and 110 is a groove on the molecule bounded by a helical coil. This suggests that the binding domain in F could extend into the RhoA “pocket.” This region of RhoA is important for interaction of various downstream effectors (Fig. (Fig.7B)7B) that regulate multiple cellular processes. This region of RhoA was previously shown to bind a Rho GTPase-activating protein (GAP) (18, 41, 42), suggesting the potential for F binding to either increase or decrease GTPase activation during virus infection. Loop 6 (amino acids 87 to 90) of RhoA has been shown to bind two classes of effector kinases, represented by PKN or PRK2 (serine/threonine protein kinases) and Rho kinases (ROKα/ROCK-II/Rho-kinase and ROKβ/ROCK-I/p160ROCK), that mediate Rho-induced stress fiber formation and cellular transformation (53). In addition, loop 6 also binds two nonkinase molecules, rhophilin and rhotekin (17). It is possible that the F protein by interacting with RhoA may be simulating or blocking one of the effector functions mediated by RhoA. Alternatively, binding of this region of RhoA may confer conformational changes in F. There is considerable evidence that RhoA induces a rapid reorganization of actin into stress fibers in a variety of cell lines (39). In RSV infection, there is a stress fiber formation early in the infection (20). The stimulation of RhoA by F may lead to actin reorganization leading to stress fiber formation.
Previous work has shown that in RSV infection actin filaments are necessary for transcription (24), syncytium formation, and virus maturation (5, 19). In the last few years, actin has also been shown to play important roles in gene transcription, syncytium formation, and maturation of many other viruses (4, 11, 44, 50). Precise temporal and spatial control of actin filament organization is essential for these activities, but how this organization is achieved is not known. The interaction of viral protein with RhoA may have significance in virus infection to ensure a coordinated control of cellular activities required for virus replication, such as determination of the stage of the cell cycle and cytokinesis during syncytium formation and virus maturation. Although RhoA is ubiquitously expressed in all tissues, lung tissue expresses RhoA at a very high level (16). This may be one of the explanations for an efficient replication of RSV in lung tissue. This hypothesis is supported by the increased number (Fig. (Fig.8B)8B) and size of RSV plaques and the speed of their formation (data not shown) when RhoA is overexpressed in stably RhoA-transfected HEp-2 cells and clearly indicates physiological significance of RhoA in RSV infection. It is possible that increased RhoA expression enhances the number of RSV plaques by facilitating virus entry and cell-to-cell spread. Alternatively, increased RhoA activity may alter the cytoskeleton structure to indirectly improve the efficiency of virus infection. It is also possible that efficient virus replication and maturation may be affected by RhoA influences on gene transcription, rapid actin bundling, and regulation of cell morphology at the level of virus assembly.
Although the interaction of RSV F and RhoA is surprising, given its wide range of biological functions essential for survival of the cell and possibly for the virus to replicate within a short time, it is not surprising for a virus to target such a key molecule. In RSV infection of A549 cells (an airway epithelial cell line) and primary bronchial cells, there are increased levels of interleukin-8 (2, 37) and NF-κB (31), in addition to actin reorganization (5, 19) and stress fiber formation (20). Previous studies have shown that thrombin increases RSV-induced syncytium formation (13) and that inhibitors of thrombin inhibit RSV-induced syncytium formation in cell culture (47). RhoA expression is increased by treatment with thrombin (35), and activation of RhoA results in the increased levels of NF-κB (38, 40) and bradykinin (38) and actin reorganization and stress fiber formation (22, 35). RhoA is critical in actomyosin-based contractility as it increases calcium sensitivity in smooth muscles (23). Indeed RhoA activation has been shown to promote myosin kinase activity which induces bronchiolar smooth muscle contraction (28) and has been suggested to play a role in asthma (45). It is therefore intriguing to consider RhoA activation as a possible step in the process of RSV-induced wheezing caused by smooth muscle contraction. Thus, biological effects of RhoA activation add a new dimension to RSV pathogenesis.
Although our data support the involvement of RhoA during RSV infection, it is unclear where in the cell RSV F binds to RhoA. Therefore, additional work is needed to investigate not only the precise locus of interaction of F and RhoA but also the downstream signaling events potentially triggered by RhoA interaction with the RSV F protein during virus infection. Since many enveloped viruses have similar needs for utilizing actin for various steps in the virus life cycle, a common mechanism involving RhoA GTPase may be shared among some viruses of different families. The binding domain of F and RhoA may be an important target for developing novel therapy and designing a better vaccine for RSV. RhoA-derived peptides from the F binding domain or other molecules that interfere with the F and RhoA interaction may provide a novel therapeutic approach. One could also envision that an appropriate mutation in the RhoA binding region of F may attenuate live recombinant RSV to produce a candidate vaccine.
We thank Peter Collins, NIH, Bethesda, Md., for providing plasmid pGEM7z-F containing the F gene and Brian Murphy, NIH, for providing an RSV F hybridoma. We also thank Wyeth-Lederle-Praxis Biologicals, West Henrietta, N.Y., for providing immunoaffinity-purified F and G proteins for binding studies. Tara Gower, John Exton, Sandra Aung, and Teresa Johnson contributed through editorial comments and helpful discussions.
This work was supported in part by NIH grant RO1-AI-33933.