|Home | About | Journals | Submit | Contact Us | Français|
Parainfluenza virus 5 (PIV5) activates and is neutralized by the alternative pathway (AP) in normal human serum (NHS) but not by heat-inactivated (HI) serum. We have tested the relationship between the fusion activity within the PIV5 F protein, the activation of complement pathways, and subsequent complement-mediated virus neutralization. Recombinant PIV5 viruses with enhanced fusion activity were generated by introducing point mutations in the F fusogenic peptide (G3A) or at a distal site near the F transmembrane domain (S443P). In contrast to wild-type (WT) PIV5, the mutant G3A and S443P viruses were neutralized by both NHS and HI serum. Unlike WT PIV5, hyperfusogenic G3A and S443P viruses were potent C4 activators, C4 was deposited on NHS-treated mutant virions, and the mutants were neutralized by factor B-depleted serum but not by C4-depleted serum. Antibodies purified from HI human serum were sufficient to neutralize both G3A and S443P viruses in vitro but were ineffective against WT PIV5. Electron microscopy data showed greater deposition of purified human antibodies on G3A and S443P virions than on WT PIV5 particles. These data indicate that single amino acid changes that enhance the fusion activity of the PIV5 F protein shift the mechanism of complement activation in the context of viral particles or on the surface of virus-infected cells, due to enhanced binding of antibodies. We present general models for the relationship between enhanced fusion activity in the paramyxovirus F protein and increased susceptibility to antibody-mediated neutralization.
The lipid bilayer of a paramyxovirus particle contains the viral glycoproteins which are activators of complement (1, 2, 3), a powerful innate immune system that can play important roles in viral pathogenesis and immunity (4, 5). Importantly, the paramyxovirus envelope can also include strong inhibitors of complement such as CD55 and CD46 (6). We have shown that changes in the balance of these activating and inhibiting factors can be critical determinants of resistance of paramyxoviruses to neutralization (6, 7). An important goal is the identification of signatures within the viral glycoproteins that activate innate immune responses and lead to neutralization. Here, we have tested the relationship between the fusion activity within the paramyxovirus F protein, the activation of complement pathways, and subsequent complement-mediated virus neutralization.
Once activated, complement components are capable of direct neutralization of viruses, through mechanisms that include aggregation, opsonization, or virion lysis (4, 5). The complement proteolytic cascade can be initiated through three main pathways: the classical pathway, the lectin pathway, and the alternative pathway (8, 9). Activation of the classical pathway typically involves binding of the C1q component to virus-antibody (Ab) complexes but can also involve association of C1q by itself with virus particles. Human immunodeficiency virus (HIV) (10) and vesicular stomatitis virus (VSV) (11) are known to activate the classical pathway. The lectin pathway is activated through recognition of carbohydrate signatures on viral glycoproteins by the cellular mannan-binding lectin (MBL), and this is an important pathway in the pathogenesis of Ross River virus (12) and opsonization of influenza virus (13). Both the classical and the lectin pathways activate C4 cleavage to C4a and C4b. Compared to the classical and lectin pathways, the signals that activate the alternative pathway are less well understood, but they are thought to involve recognition of foreign surfaces by an antibody-independent mechanism (4, 14). Epstein-Barr virus (15) and Sindbis virus (16) are known examples of viruses that activate the alternative pathway.
We and others have previously shown that the paramyxoviruses mumps virus (MuV) and parainfluenza virus 5 (PIV5) (formerly known as simian virus 5 [SV5]) activate the alternative pathway (2, 6). However, this property is not common to all paramyxoviruses. For example, both measles virus and human parainfluenza virus 3 (HPIV3) activate either the classical or alternative pathway, depending on whether virions or cells expressing viral glycoproteins are assayed, and the absolute dependence on antibody for activation can differ (3, 17). These differences in pathways activated are even more evident for Newcastle disease virus (NDV), which can activate all three complement pathways, depending on the cell type used for virus growth (18).
Complement factors recognize and respond to foreign components such as microbial patterns and antigens (9). The complement-activating foreign components in PIV5 and MuV particles consist of two predominant viral glycoproteins: an attachment protein hemagglutinin-neuraminidase (HN) which binds to and cleaves sialic acid and a fusion protein (F) which fuses the viral envelope with the host cell or causes cell-cell fusion (19). Prior work has shown that the extent of alternative pathway activation by these viruses was inversely related to sialic acid concentrations on virion particles, due to the presence of viral neuraminidase activity (2, 20). In contrast, however, the signals in the viral F protein which could contribute to complement activation have not been extensively defined.
The PIV5 F protein is a prototype type I fusion protein, and much of our understanding of structure-function relationships and mechanisms of membrane fusion have come from the biochemical and crystallographic study of this trimeric membrane protein (21, 22, 23). The PIV5 F protein is synthesized as an F0 precursor, which is cleaved by a cellular protease at a stretch of arginine residues into F1 and F2 subunits (24), resulting in release of a hydrophobic fusion peptide (FP) (Fig. 1) involved in membrane fusion. When triggered for fusion, the F protein undergoes extensive conformational changes, and two heptad repeat regions (HRA and HRB) (Fig. 1) are aligned to form a very stable six-helix bundle (19). Prior studies by Horvath and Lamb (25) using a heterologous expression system showed that a glycine-to-alanine substitution at position 3 of the FP (G3A) resulted in a hyperfusogenic F protein with enhanced ability to promote cell-cell fusion (Fig. 1). Similarly, a serine-to-proline substitution at position 443 (distal to the FP) also created a hyperfusogenic F protein, which in transfection studies showed an enhanced ability to cause syncytium formation (26). Thus, these two distinct single point mutations in F result in mutants with enhanced capacity in cell-cell fusion when expressed out of the context of a bone fide PIV5 infection (26, 27).
Given the importance of complement in antiviral immunity and the limited knowledge of the features of paramyxoviral glycoproteins that drive complement activation, we have tested the hypothesis that the PIV5 F protein contributes to pathways of complement-mediated virus neutralization. Our results show that recombinant viruses harboring the hyperfusogenic F protein G3A or S443P are altered in their complement activation compared to wild-type (WT) PIV5, and this correlates with the increased sensitivity of these mutants to binding by antibodies contained in normal human serum (NHS). Our results have implications for the design of more efficacious therapeutic vectors and our understanding of selective immune pressures which could drive the evolution of paramyxovirus fusion proteins.
Normal human serum (NHS) was collected from healthy donors, processed and divided into small aliquots before freezing at −80°C as published previously (6). Where described, sera were heat inactivated (HI) by heating to 56°C for 30 min. On rare occasions, we found HI human serum which was still capable of neutralizing WT PIV5. Soluble Galα1-3Gal carbohydrate was from Toronto Research Chemicals (Toronto, Canada). Purified complement components were procured from Complement Technologies (Tyler, TX). Pooled NHS was purchased from Innovative Research (Novi, MI). Antibodies specific for C3 and C4 were from Complement Technologies (Tyler, TX), those for human IgG from Jackson ImmunoResearch, and those for actin from Sigma. Anti-C4 polyclonal antibody for use in electron microscopy (EM) and fluorescence-activated cell sorter (FACS) analysis was from MyBiosource. Mouse polyclonal anti-PIV5 serum was generated by three peritoneal injections of purified PIV5 particles and collection of serum over a 3-month period.
Monolayer MDBK or A549 cell cultures were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% HI fetal bovine serum (FBS). CHO cells were grown as described previously (6). WT PIV5 (W3A strain) carrying the green fluorescent protein (GFP) gene between the HN and L genes has been described previously (28). PIV5 encoding the hyperfusogenic G3A mutation was constructed by insertion of the mutant F gene from CPI-F-G3A (29) into the cDNA encoding WT PIV5 with the GFP gene at the HN-L gene junction. The S44P F protein mutant has been described previously (30), and PIV5 harboring this F protein mutant in the absence of a GFP gene was the kind gift of Robert A. Lamb and David Waning (Northwestern University). For all data except those in Fig. 6, WT and mutant viruses were grown in MDBK cells. The ratios of PFU to hemagglutinin (HA) units for the WT, G3A, and S443P viruses were 1.26 × 106, 1.6 × 106, and 4.13 × 105, respectively.
For purification, viruses were concentrated by centrifugation through a glycerol cushion (5 h; 25,000 rpm; SW28 rotor), and resuspended virus was further centrifugation on a 30 to 60% sucrose gradient (2 h; 23,000 rpm; SW28 rotor). The virus band was collected, pelleted, resuspended in buffer, and stored at −80°C. Time- and concentration-dependent neutralization by NHS was carried out as described previously (6), with remaining infectivity determined by plaque assay on CV-1 cells. Results are the average from six reactions, with the significance of data points calculated using one-way analysis of variance (ANOVA) followed by the Tukey test.
Preexisting antibodies in NHS (hAb) were purified using a protein G Mag-Sepharose column (GE Health Care, Piscataway, NJ) as described by the manufacturer. Briefly, HI serum was added to the protein G beads, incubated for 15 min, washed with buffer, and eluted with 2.5% acetic acid. Eluted proteins were dialyzed against multiple changes of phosphate-buffered saline (PBS) (pH 7.4) and quantitated using a NanoDrop instrument.
Electron microscopy (EM) analysis of sucrose gradient-purified WT or mutant PIV5 was carried as previously described (31). Samples were treated with NHS (1:10 dilution), blocked with 1% bovine serum albumin (BSA) in PBS, and then probed with anti-C4 polyclonal antibody at a 1:25 dilution in PBS with 1% BSA. Deposition was detected with 12-nm-gold-labeled donkey anti-goat antibody (Jackson ImmunoResearch Laboratories, PA). Binding of human antibody was analyzed using a 1:10 dilution of purified human antibody and 6-nm-gold-labeled goat anti-human secondary antibody. After labeling, particles were subjected to negative staining with 2% phosphotungstic acid (pH 6.6) and analyzed with a Technai transmission electron microscope as previously described (6).
Proteins were analyzed by SDS-PAGE followed by Western blotting as described previously (6). Blots were probed with polyclonal rabbit serum specific for the PIV5 NP, P, M, F, or HN protein. Details on the antibodies are available on request. Blots were visualized by horseradish peroxidase (HRP)-conjugated antibodies and enhanced chemiluminescence (Pierce Chemicals). Binding of C4 and human antibody was determined using anti-C4 (1:2,000 dilution) or polyclonal HRP-conjugated anti-human IgG (1:2,000 dilution). Equal loading was assessed by probing with mouse anti-human actin (1:4,000) followed by HRP–anti-mouse antibody (1:4,000).
For enzyme-linked immunosorbent assay (ELISA) measurement of complement activation, dilutions of purified virus in phosphate-buffered saline (PBS) (pH 7.4) were mixed with a 1:10 dilution of NHS (assay volume, 20 μl) and incubated for 1 h at 37°C. The samples were further diluted 1:500 and used in an ELISA specific for C3a or C4a as described by the manufacturer (BD Biosciences, San Jose, CA). Statistical significance was determined using Student's t test.
CHO cells were mock infected or infected at a multiplicity of infection (MOI) of 10 with WT, G3A, or S443P PIV. At 20 h postinfection (hpi), cells were removed from the dish and treated with a 1:20 dilution of pooled human sera or with purified hAb. After washing in PBS, cells were either analyzed by flow cytometry or lysed and analyzed by Western blotting as described above.
For flow cytometry, infected cells were analyzed as described previously (6) with mouse polyclonal anti-PIV5 sera, with anti-C4 polyclonal antibody, or with anti-human IgG antibody. Following washes, the cell pellets were suspended in PBS and incubated with anti-mouse Alexa Fluor 488 and/or anti-rabbit Alexa Fluor 633. Analysis was carried out on a FACSCalibur machine (BD Biosciences, San Diego, CA).
Recombinant PIV5 viruses were generated to encode point mutations in the F protein which have been shown in transfection experiments to result in enhanced fusion activity (26, 27). The PIV5 G3A and S443P mutants contained an F protein with a G-to-A substitution in the fusion peptide or an S-to-P substitution in a region near the C-terminal HRB (Fig. 1), respectively. As shown in Fig. 1B, CV-1 cells infected with WT PIV5 appeared by light microscopy to be similar to mock-infected control cells, and syncytia were not detected until very late in infection (not shown). In contrast, cells infected with the G3A mutant showed numerous large pockets of cell-cell fusion by 24 hpi, and by 30 hpi the majority of the cell monolayer was consumed by syncytia. Monolayers of cells infected with the S443P mutant also showed numerous syncytia, but the extent of cell-cell fusion was lower and was kinetically delayed relative to that of the G3A mutant (Fig. 1B).
To determine the ability of NHS to neutralize the G3A mutant, 100 PFU of WT or mutant PIV5 derived from MDBK cells was incubated with dilutions of NHS, and remaining infectivity was determined by plaque assay. As shown in Fig. 2A, WT PIV5 was effectively neutralized by NHS up to a 1:40 dilution, with an 80% neutralization at a 1:80 dilution (black bars). Results with the G3A mutant were similar to those with WT PIV5, in that NHS was effective at neutralization up to 1:80. However, a striking difference between viruses was seen with HI serum (Fig. 2A, hatched bars). Here, HI serum was not able to neutralize PIV5 even at a 1:2 dilution, a result contrasting with that for the G3A mutant, which was effectively neutralized by HI serum. Similar results were seen using sera from four other donors (not show). Thus, the G3A mutant differs from WT PIV5 by being sensitive to in vitro neutralization by HI serum.
We and others have previously shown that WT PIV5 is neutralized by the alternative complement pathway (2, 31), which is dependent on the activity of factor B (fB). As shown in Fig. 2B, PIV5 was neutralized in vitro by NHS (white bars) but not by HI (hatched bars) or by serum depleted of fB (gray bars). Neutralization of WT PIV5 was reconstituted by addition of physiological levels of fB to fB-depleted serum (crosshatched bars). These results contrasted sharply with those seen for G3A, where virus was neutralized by HI and the depletion or reconstitution of fB did not substantially alter the extent of neutralization (Fig. 2B).
The above findings raised the hypothesis that the G3A mutant was neutralized by the classical or lectin pathways, both of which are dependent on C4 for activity. As shown in Fig. 2C, the G3A mutant was neutralized by NHS and HI serum as shown above but was not neutralized by C4-depleted serum (hatched bar). Neutralization was reconstituted by addition of physiological levels of C4 to C4-depleted serum. At the serum dilution tested for Fig. 2C, HI serum did not completely neutralize G3A compared to NHS but was more neutralizing than C4-depleted serum. Thus, NHS has two distinct mechanisms for neutralization of G3A: C4-dependent and C4-independent pathways that can be distinguished by dilution.
The hyperfusogenic S443P mutant has a single point mutation near the C-terminal HRB (26), a region outside the FP where the G3A mutation is located (see schematic diagram in Fig. 1A). To determine if the above findings for G3A extended to other fusogenic F mutants, mutant S443P was tested for sensitivity to neutralization by NHS and HI serum. As shown in Fig. 2D, the S443P mutant was effectively neutralized by NHS up to a 160-fold dilution but was also neutralized by HI serum. Together, these data indicate that a single point mutation in the PIV5 F protein which enhances fusion activity alters the pathway of in vitro complement-mediated neutralization from the alternative pathway to classical/lectin pathway.
We tested the hypothesis that differences in complement-meditated neutralization of WT and hyperfusogenic PIV5 correlated with differential activation of C4. PIV5 viruses were grown in MDBK cells and purified by sucrose gradient centrifugation, and levels of viral components were compared by Western blotting. As shown in Fig. 3A, G3A particles contained levels of NP, M, and HN closely matching those seen with WT PIV5 but had ~3-fold-lower levels of F protein. Comparison of purified WT PIV5 and S443P proteins showed no substantial difference in the relative abundance of NP, M, HN, or F (Fig. 3A).
Complement activation was tested in vitro by incubation of various levels of purified G3A virus with NHS and analysis of the C3a cleavage product by ELISA. C3 was activated to similar levels by WT virus and the G3A mutant as evidenced by a dose-dependent appearance of C3a (Fig. 3B). In contrast, both the G3A and S443P mutants were more potent activators of C4 than WT PIV5. This is shown in Fig. 3C, where as little as 0.1 μg of the mutant virions led to the appearance of high levels of C4a cleavage product compared to that seen with WT PIV5. In time course experiments, the G3A and S443P mutants showed very rapid C4 activation in as little as 1 min of in vitro incubation with NHS, whereas WT PIV5 activated C4 with the same low level as seen with NHS alone (Fig. 4A).
Association of C4 with WT or fusogenic mutants was assayed by treatment of purified particles with NHS followed by immunogold EM using anti-C4 antibodies. As shown in Fig. 4B, WT PIV5 virions showed only rare examples of immunostaining, whereas nearly every G3A and S443P particle showed C4 staining. This was consistent with the dependence of the G3A mutant on C4 for complement-mediated in vitro neutralization and with the rapid activation of C4 cleavage by mutant but not WT PIV5 (Fig. 2C). In most cases, immunogold staining tended to cluster as described previously for the analysis of cellular proteins incorporated into PIV5, mumps virus, and VSV particles (6, 7). This may represent membrane microdomains which are incorporated into budding particles. Taken together, these data indicate that C4 is activated by and deposited on two mutant viruses harboring hyperfusogenic F proteins to a greater extent than that seen for WT PIV5.
The two findings of neutralization by HI serum and C4 activation by the two F mutants (G3A and S443P) raised the hypothesis that antibodies contained in human sera were responsible for neutralizing the fusogenic mutants but not WT PIV5. As shown in Fig. 5A, pooled commercially available serum behaved similarly to individual sera as shown above, since HI serum was ineffective at neutralizing WT PIV5 but retained some neutralizing capacity against both the G3A and S443P mutants. As such, antibodies were purified from HI pooled human serum by protein G-magnetic bead chromatography. Most importantly, this purified human antibody (hAb) showed a dose-dependent neutralization of both the G3A and S443P mutants in vitro (Fig. 5B). In contrast, infectivity of WT PIV5 was not affected by incubation with purified hAb.
To determine if hAb isolated from pooled HI serum was capable of binding to virus particles, purified WT PIV5 or the G3A or S443P mutant viruses were incubated with hAb followed by gold-labeled anti-human Ig and examined by EM. As shown in Fig. 5C, both mutant viruses showed extensive deposition of anti-Ig gold particles on their surfaces, whereas WT PIV5 showed relatively little deposition. Immunoprecipitation studies were not able to distinguish enhanced binding of pooled human antibodies to F or HN (data not shown). Taken together, these findings that purified hAb neutralized and was deposited on G3A and S443P viruses to a greater extent than WT PIV5 virus are consistent with the result that HI serum neutralizes the fusogenic viruses and not WT PIV5.
Previous work has shown that F protein can be converted from a prefusion to a postfusion form by elevated heat (32). To determine if heat treatment alone could enhance antibody binding, purified WT virions were incubated at 37°C or 50°C before reaction with purified hAb. As shown in the examples in Fig. 6, there was extensive deposition of anti-Ig gold particles on the surfaces of WT virions pretreated at 50°C compared to on those pretreated at 37°C. Thus, the increased reactivity of fusogenic mutants with purified hAb can be reproduced with WT virions under conditions that promote transition of F to the postfusion form.
Proteins from animals other than Old World monkeys and humans can be modified to contain a Galα1-3Galβ1-4GlcNAc-R (Galα1-3Gal) epitope, which is recognized by natural antibodies in human sera (33, 34). To determine if this modification contributed to neutralization of G3A mutant by HI sera, viruses were grown in human A549 lung epithelial cells, which lack the Gal transferase activity, and tested in a plaque reduction assay. As shown in Fig. 7A, HI sera neutralized G3A but not WT PIV5, a result very similar to the above data obtained using MDBK-derived viruses. Likewise, pretreatment of HI sera with 10 mg/ml of soluble Galα1-3Gal carbohydrate did not compete away the neutralizing capacity against the G3A mutant (Fig. 7B). Together, these data indicate that addition of Galα1-3Gal epitopes does not contribute to neutralization of the G3A mutant.
The above findings with PIV5 virions raised the hypothesis that complement would be differentially activated on the surfaces of cells infected with the hyperfusogenic mutants compared to those with WT PIV5 infection. To test this, CHO cells, which lack the expression of CD55 and CD46 regulators (6), were mock infected or infected at a high MOI with WT PIV5 or the hyperfusogenic mutants. At 20 hpi, cells were treated with NHS, washed, and analyzed for C4 deposition by flow cytometry. As shown in the two representative results in Fig. 8A, 60 to 80% of cells infected with the G3A or S443P mutant viruses had C4 deposition, compared to ~5% of cells infected with WT PIV5. These mutant-infected cells also had a higher mean fluorescence intensity of C4 staining (Fig. 8B). Similar results were seen in flow cytometric analysis of infected cells that had been treated with purified hAb, where >90% of cells infected with the fusogenic mutant viruses were positive for Ig staining (Fig. 8C). Western blot analysis confirmed the presence of higher levels of cell-associated C4 and Ig for the F mutant infection than for WT controls (Fig. 8D). Thus, the increased C4 activation and association with Ig seen with hyperfusogenic virus particles also extends to the surface of cells infected with G3A and S443P mutants.
In view of the potent role that complement can play in virus neutralization and immunity, we have tested the hypothesis that functional changes in the paramyxovirus F protein can impact interactions with the human complement system. Our work demonstrates that single amino acid substitutions shown previously to enhance the fusion activity of the PIV5 F protein can dramatically alter complement activation in the context of viral particles or on the surface of virus-infected cells, shifting the activation from the alternative pathway normally seen with WT PIV5 to a C4-dependent pathway. The most striking result was our finding of sensitivity of PIV5 harboring the fusogenic F protein mutants to neutralization by purified hAb compared to that of WT PIV5 and the extensive deposition of hAb on these mutant viral particles. Together, these data support a model whereby virions or cell surfaces containing a hyperfusogenic F protein are bound by antibodies in NHS which would otherwise bind poorly to WT proteins and this subsequently leads to activation of the classical complement pathway. These results have implications for the design of viral vectors and possible selective evolutionary pressures against a hyperfusogenic F protein.
It is well established that the extent of sialic acid modification on microbial surfaces may contribute to induction of the alternative pathway, and the neuraminidase activity in the PIV5 and MuV HN proteins can be a determinant of this activation (1, 19). In contrast, the contribution of the PIV5 F protein to complement activation and neutralization is not known. For measles virus, expression of the F protein on cell surfaces by transfection approaches leads to activation of the alternative pathway, and C3 was found conjugated to F protein (3). These findings are not inconsistent with our results, since we show that the PIV5 mutants harboring a hyperfusogenic F protein alter complement activation. Thus, in our results WT PIV5 particles are inducers of the alternative pathway (31), and this is shifted to the classical pathway by F mutations.
While there are examples of viruses which can be neutralized through the classical complement pathway in the absence of antibody binding (10), antibodies typically initiate this cascade. Here we show that in contrast to WT PIV5, the fusogenic mutants activate C4 cleavage, and they are neutralized by HI and by purified hAb. Thus, the available evidence is consistent with differential complement activation driven by differential Ab binding to particles containing fusogenic F versus WT F. The nature and glycoprotein target of Abs that differentially bind to mutant and WT virions are unknown, but it is clear that PIV5 can be neutralized by monoclonal antibodies against either HN or F (35). These NHS-derived Abs could consist of natural Abs in human serum, such as those described for the neutralization of influenza virus in the absence of prior virus exposure (36) or for nonneutralizing natural IgM, which cooperates with complement in neutralization of HPIV3 (37). Human sera contain very high levels of antibodies to the Galα1-3Gal epitope present on proteins from animals other than Old World monkeys and humans (33, 34). However, this difference in carbohydrate addition cannot account for our results, since the differential ability of HI serum to neutralize G3A versus WT PIV5 did not change with virus from A549 human cells or when HI serum was pretreated with high concentrations of Galα1-3Gal carbohydrate.
Alternatively, the antibodies which bind to and neutralize the fusogenic mutants could consist of cross-reactive antibodies generated during exposure to other closely related pathogens. For PIV5, these closely related pathogens would include human parainfluenza virus 2 (HPIV2) and/or MuV. In more than 20 human donor sera that we have tested, WT PIV5 was neutralized by NHS but not HI serum. On occasion, we have found donors whose HI serum did neutralize WT PIV5, and we have found that these donors also have high anti-MuV titers (not shown).
We propose two nonexclusive models to account for differential antibody binding and sensitivity of mutant and WT PIV5 particles to neutralization by NHS. It is known from transfection experiments that mutations in F that increase fusion activity can alter the binding of anti-F monoclonal antibodies (30, 38). Similar results were seen with pseudotypes containing the Nipah virus glycoproteins, in which changes in F protein fusion activity also changed the sensitivity to neutralization (39), and in the case of destabilizing mutations in the measles virus F protein (40). These data have been proposed to reflect a change in F protein conformation or availability of new antigenic epitopes. Similarly, the interactions of HN and F are proposed in many cases to dictate fusion taking place at the right place and the right time (19, 22, 41). As such, the F mutations G3A and S443P could reduce stable interactions with HN, resulting in exposure of new antigenic sites on either HN or F for antibody binding and neutralization.
Changes in F protein which increase fusion activity are thought to reduce the threshold of activation from the prefusion conformation (22). Thus, as an alternative model, the sensitivity of the G3A and S433P mutants to neutralization could reflect the binding of low-affinity antibodies to the sensitive G3A F protein, resulting in premature triggering of virion-associated F from the prefusion to the postfusion form and loss of infectivity. Consistent with this, we found enhanced binding of hAb to WT virions after pretreatment at 50°C, conditions that convert F from pre- to postfusion conformations. This model is similar to that proposed for inhibition of HPIV3 infection by inducing premature activation of the F protein prior to interactions with a host cell (42).
The finding that single point mutations can dramatically increase F protein fusion activity argues that the WT F protein is not in its most active fusogenic form (22, 25). This raises the question of why naturally occurring hyperfusogenic paramyxoviruses such as that seen with the G3A mutant are not a common finding. For PIV5, natural variants that differ in fusion activity have been described (e.g., SER ), but these variants are often less fusogenic than the prototype W3A strain of PIV5. Young et al. have shown that fusogenic variants arise during the propagation of cells persistently infected with PIV5 (44). Interestingly, these fusogenic variants were more rapidly cleared from in vitro cultures by neutralizing HN and F antibodies than WT PIV5, consistent with an inverse relationship between increased fusion activity and higher sensitivity to neutralization. Taking these results together, we speculate that there may be selective immune pressure to limit the growth of fusogenic viruses which may be more easily neutralized by antibodies or by complement-dependent mechanisms.
There is increasing interest in harnessing the paramyxovirus fusion activity for therapeutic vectors (see, e.g., references 29 and 45). We have previously shown that PIV5 expressing the G3A mutant F protein is highly effective at reducing tumor burden in a nude mouse model system (29). However, the relative ability of the oncolytic G3A mutant versus WT PIV5 to be neutralized in animal studies has not been examined. Our results have implications for the design of vectors that would be safer (e.g., more easily neutralized by preexisting antibody and complement), but they also raise the question of balancing this property with being cleared too quickly to provide therapeutic or immune-stimulatory effects (46).
We are grateful to Ellen Young for excellent technical help. The PIV5-S443P virus was constructed and made by David Waning and Robert A. Lamb at Northwestern University and kindly provided for use in these studies. We thank Elankumaran Subbiah for helpful insights and David Ornelles for help with statistical analyses.
This work was supported by NIH grant AI083253 and by the flow cytometry core and imaging core of the WFUCCC (NCI CCSG P30CA012197).
Published ahead of print 19 June 2013