Premature Activation of the Paramyxovirus Fusion Protein before Target Cell Attachment with Corruption of the Viral Fusion Machinery*

Paramyxoviruses, including the childhood pathogen human parainfluenza virus type 3, enter host cells by fusion of the viral and target cell membranes. This fusion results from the concerted action of its two envelope glycoproteins, the hemagglutinin-neuraminidase (HN) and the fusion protein (F). The receptor-bound HN triggers F to undergo conformational changes that render it competent to mediate fusion of the viral and cellular membranes. We proposed that, if the fusion process could be activated prematurely before the virion reaches the target host cell, infection could be prevented. We identified a small molecule that inhibits paramyxovirus entry into target cells and prevents infection. We show here that this compound works by an interaction with HN that results in F-activation prior to receptor binding. The fusion process is thereby prematurely activated, preventing fusion of the viral membrane with target cells and precluding viral entry. This first evidence that activation of a paramyxovirus F can be specifically induced before the virus contacts its target cell suggests a new strategy with broad implications for the design of antiviral agents.

Parainfluenza viruses are important agents of lower respiratory tract disease in children (1,2), as well as in older adults and immunocompromised individuals (3). There are no vaccines available for the parainfluenza viruses (4 -12), and any vaccine is unlikely to protect the very youngest infants. Although other causes of viral respiratory disease in children have yielded in part to vaccination programs and antiviral use, children are still virtually unaided in their battle against the major causes of croup and bronchiolitis.
Paramyxoviruses, including parainfluenza viruses, enter their target cells by binding to a receptor molecule and fusing their viral envelope with the cell membrane to reach the cytoplasm. Both binding and fusion are critical for infection, and thwarting these essential events would suffice to prevent disease. Binding to the target cell is mediated via interaction of the viral receptor-binding molecule (hemagglutinin-neuraminidase (HN) 3 ) with sialic acid-containing receptor molecules on the cell surface. The viral envelope then fuses directly with the plasma membrane of the cell, a process mediated by the viral fusion protein (F), releasing the viral nucleocapsid into the cytoplasm (13)(14)(15). During fusion, the F protein must undergo a structural rearrangement (16 -18) requiring an activation step, which must occur at the right time and place in order for fusion and entry to occur.
The paramyxovirus F proteins are activated when the adjacent HN binds to a sialic acid-containing receptor, permitting fusion to occur (19). Once activation occurs, the fusion protein undergoes a coordinated series of conformational changes so that protein refolding generates the final, most stable form of the protein. HN, in addition to its role in activating fusion, possesses both hemagglutinating (sialic acid receptor-binding) and neuraminidase (sialic acid receptor-cleaving) activities. These three activities are tightly regulated within a type II membrane protein that has a cytoplasmic domain, a membrane-spanning region, a stalk region, and a globular head that contains the primary sialic acid-binding site and the neuraminidase active site. Structural analysis of the HNs of the avian paramyxovirus Newcastle Disease virus (NDV) (20,21), the HPIV3 HN (22), and the PIV5 HN (23) has identified the locations of the primary binding/neuraminidase active site residues on the globular head of the molecule, and several key domains required for the fusion-triggering function of HN have been identified (19, 24 -26). The viral fusion process is modulated by a balance between these three properties of HN; receptor binding, neuraminidase, and fusion triggering. Although F-activation is key for viral entry (19), triggering must occur only when F is close enough to then make contact with the target cell membrane. We suggest that this timing of activation represents a potential target for intervention with paramyxovirus infection (15,25,27).
For HPIV3, we proposed that, with the correct receptor mimic, it is possible to stimulate HN to activate F prematurely, and thus inactivate F after budding and release, incapacitating the virion before it can reach its target. To test this idea, we searched for small molecules that would interact with HN to activate its triggering activity and selected candidate molecules for further testing in vitro. In this report we show that a molecule predicted to bind to the HN globular domain inactivates parainfluenza virions before they contact target cell receptors. The mechanism of inactivation is via premature triggering of the fusion process.

EXPERIMENTAL PROCEDURES
Cells and Viruses-293T (human kidney epithelial), and CV-1 (African green monkey kidney) cells were grown in Dulbecco's modification of Eagle's medium (DMEM, Mediatech Cellgro) supplemented with 10% fetal bovine serum and antibiotics. Stocks of HPIV3 were made in CV-1 cells from virus that was plaque-purified four times. PIV5 was purchased from the American Type Culture Collection. Virus was collected 36 -48 h post-infection and stored at Ϫ80°C. Viral titers were determined by a previously described plaque assay with CV-1 cells (28).
HAE Culture-The EpiAirway AIR-100 tissue model (Mat-Tek Corp.) consists of normal human-derived tracheobronchial epithelial cells that have been cultured to form a pseudostratified, highly differentiated mucociliary epithelium closely resembling that of epithelial tissue in vivo. Upon receipt from the manufacturer, human airway epithelium (HAE) cultures were transferred into six-well plates (containing 0.9 ml of medium per well), with the apical surface remaining exposed to air, and incubated at 37°C in 5% CO 2 overnight, as previously described (28).
Infection and Treatment of HAE Cells and Measurement of Viral Titers from Infected HAE Cells-HAE cultures were infected by applying 200 l of EpiAirway medium containing 4000 plaque-forming units of wild-type (wt) HPIV3. Viral suspensions were applied to the apical surface of HAE tissues for a 90-min adsorption period at 37°C in the presence or absence of 3 mM CSC11 or CSC7 or zanamivir. At 90 min, the medium containing the inoculum was removed, and cultures were placed at 37°C and fed each day with 0.9 ml of medium via the basolateral surface. Viruses were harvested by adding 200 l of medium per well to the HAE cultures' apical surface and allowed to equilibrate for 30 min at 37°C. The suspension was then collected, and viral titers were determined as previously described (28).
Chemicals-Zanamivir was prepared from Relenza Rotadisks (5 mg of zanamivir with lactose). A 50 mM stock solution was prepared by dissolving each 5-mg blister capsule in 285 l of Opti-Mem (Invitrogen). CSC compounds were purchased from ChemDiv. 100 mM stock solutions were prepared in DMSO. Stock solutions were stored at Ϫ20°C.
Transient Expression of HPIV3 HN and F Constructs-HPIV3 HN and F cDNAs were digested with SacI or EcoRI and BamHI and ligated into digested pCAGGS and pEGFP mammalian expression vectors as previously described (19, 25, 29 -32). Constructs used in the assay for F protein triggering, HN-N-Venus and F-C-CFP, were prepared as previously described (33). NDV AV wt HN in pCAGGS was obtained from Dr. Ronald Iorio. Chimeric glycoprotein HPIV3 HN (from amino acids 1-144) and NDV HN (from amino acids 124 -571) were cloned in pCAGGS. Transfections were performed with Lipofectamine and Plus or Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions.
Plaque Reduction Assay-The effect of the compounds on plaque number was assessed as described (34,35). Briefly, CV-1 cell monolayers were infected with 100 plaque-forming units of WT HPIV3 in the presence of serial dilutions of the indicated compounds. After 90 min, the plates were overlaid with either agarose or methylcellulose; 24 h later the agarose was removed, cells were fixed and stained, and the plaques counted.
Hemadsorption Assays-Hemadsorption was performed and quantified as previously described (30). Briefly, growth medium from HN/F-cotransfected 293T cell monolayers in 48-well Biocoat plates (BD Biosciences Labware) was aspirated and replaced with 150 l of CO 2 -independent medium (pH 7.3, Invitrogen) containing different concentrations of compounds and of 1% sialic acid receptor-bearing erythrocytes (RBCs) in serum-free, CO 2 -independent medium, and placed at 4°C for 30 min. The wells were then washed three times with 150 l of cold CO 2 -independent medium. The bound RBCs were lysed with 200 l of RBC lysis solution (0.145 M NH 4 Cl, 17 mM Tris HCl), and absorbance was read at 405 nm using a Spectramax M5 (Molecular Devices) microplate reader.
Measurement of Neuraminidase Activity-Assays were performed in transiently transfected 293T cell monolayers, as described previously (26,30). Briefly, 293T cells expressing WT HN were added to 96-well plates in CO 2 -independent medium at pH 5.0. After adding reaction mixtures containing 20 mM (2Ј-(4-methylumbelliferyl)-␣-D-N-acetylneuraminic acid, Toronto Research Chemicals Inc.) substrate and different concentrations of CSC compounds, the plates were incubated at 37°C for 1 h. Throughout this period, fluorescence resulting from hydrolysis of the substrate was read at 365 nm excitation wavelength and 450 nm emission wavelength using a Spectramax M5 microplate reader.
␤-Galactosidase Complementation-based Fusion Assay-We previously adapted a fusion assay based on alpha complementation of ␤-galactosidase (29,36); receptor-bearing cells expressing the omega peptide of ␤-galactosidase are mixed with HPIV3 HN/F-coexpressing cells that also express the alpha peptide of ␤-galactosidase, in the presence of various compound concentrations. Cell fusion leads to complementation. Fusion is stopped by lysing the cells and, after addition of the substrate, fusion is quantitated on a Spectramax M5 microplate reader.

Measurement of Fusion of RBCs with HN/F-expressing Cells-
Monolayers of HN/F-expressing 293T cells were washed three times with CO 2 -independent medium. One set of cell monolayers was incubated with RBC suspensions for 30 min at 4°C; after a rinse to remove unbound RBC, the plates were transferred to 37°C with simultaneous addition of 3 mM of compound, or zanamivir, or fusion inhibitory peptide derived from the C-terminal region of HPIV3 F (HRC) (37) (pre-bound). A second set of cell monolayers was incubated with RBC in the presence of 3 mM compound or control substances directly at 37°C for 60 min (not pre-bound). After discarding the unbound RBC and lysis of the unfused RBCs with lysis solution (0.145 M NH 4 Cl, 17 mM Tris-HCl), the RBCs that had undergone fusion were lysed in 200 l of 0.2% Triton X-100-PBS and transferred to flat-bottom 96-well plates for quantification by absorption measurement at 405 nm.
Virucidal Assay and Quantitative PCR-Aliquots of HPIV3 viral preparations were incubated for 1 h at 37°C in Opti-Mem supplemented with the indicated compounds or DMSO. After incubation, the samples were diluted with Opti-Mem and cleared of compounds using Ultrafree MC filters (Millipore). The viral particles were then collected from the filters in Opti-Mem, and their infectivity was determined by plaque reduction assay. Using a One-Step Quantitative RealTime-PCR HPIV3 detection kit (Primer Design Ltc.) and Realplex2 MasterCycler (Eppendorf), the total amount of viral RNA in each sample was determined.
Detection of Fusion Protein Triggering-293T cell monolayers, transiently transfected with HN-N-Venus and F-C-CFP, or F-C-CFP alone, were incubated overnight in DMEM supplemented with 2 mM zanamivir to prevent fusion. The transfected cells were washed with Opti-Mem and incubated for 1 h at 37°C in Opti-Mem supplemented with 100 mg/ml cycloheximide (to prevent de novo protein synthesis) (38) only, or additionally supplemented with 2 mM zanamivir or 2 mM zanamivir and 3 mM of the desired compound. The cells were then lysed in DH buffer (50 mM HEPES, 100 mM NaCl, 0.005 g/ml dodecyl maltoside, and Complete protease inhibitor mixture (Roche Applied Science)). HN and F were immunoprecipitated from post-nuclear lysates with anti-GFP antibody-conjugated agarose beads (Santa Cruz Biotechnology) for 2 h at 4°C, washed, and resuspended in PBS, and then incubated in the absence or presence of Proteinase K (Sigma) (0.64 ϫ 10 Ϫ5 mg/ml (1ϫ) or 1.28 ϫ 10 Ϫ5 mg/ml (2ϫ)) for up to 90 min. Complete protease inhibitor mixture was added to stop Proteinase K digestion after 90 min. The protein-bead complexes were supplemented with Laemmli sample buffer, heated to 99°C for 5 min, then subjected to SDS-PAGE and Western blot analysis with polyclonal antibodies to the HRC region of F. Relative protein levels were quantified using a Kodak ImageStation and Kodak Molecular Imaging software.
Virtual Screening, Docking, and Scoring-Using the key interactions of zanamivir in its crystal structure (PDB 1v3e) with HN and of 2,3-dehydro-2-deoxy-N-acetylneuraminic acid in its crystal structure (PDB 1v3d) with HN (22) pharmacophoric models were obtained. These models were used to create queries for virtual screening. 3D-Flex searches (39) were conducted on the ZINC databases (40, 41) using the UNITY mod-ule of Sybyl 7.2 (Tripos, available online). The hits from the UNITY search were visually observed to identify molecules that were drug-like and GOLD (Genetic Optimization for Ligand Docking) 3.0 (42) was utilized to dock the hits into the binding pocket (site I) of HPIV3 HN, using 50 iterations per ligand and no early termination. All docking solutions were scored using the HINT (Hydropathic INTeractions) forcefield (43,44).
Docking and Scoring Zanamivir and CSC11 at Site II-The two ligands were docked into binding site II of HPIV3 HN (29) using GOLD 3.0, employing 100 iterations and no early termination. All the conformations of both ligands were scored using HINT, and the best positions were minimized using 1000 iterations of the Powell method with Gasteiger-Hückel charges in the Tripos Forcefield of Sybyl 7.2 and finally rescored using HINT.

Compounds That Interact with the Active Site of HN Are
Potent Inhibitors of HPIV3 Infectivity-To test the hypothesis that small molecules may interact with one or more active sites of HN in such a way as to inactivate the virus, we assessed small molecules predicted to bind to HPIV3 HN. A virtual screening experiment using the ZINC database (40,41) identified several hundred small molecule compounds predicted to interact with HN. Among these, 50 high scoring compounds were selected and purchased for testing. From this set, we found 17 (34%) that inhibited HPIV3 infectivity in a plaque reduction assay. Fig. 1 shows a plaque reduction assay in which the three most effective compounds (shown in Structure 1) were tested in parallel with zanamivir, a small molecule that reduces HPIV3 infectivity by inhibiting HN-receptor binding (35). (See supplemental  Table 1 for numerical values from a representative experiment.) The three new compounds tested were 4-[bis(4-hydroxy-2methyl-6-oxo-1,6-dihydropiridimin-5-yl)methyl]benzoic acid (CSC5),  NOVEMBER 4, 2011 • VOLUME 286 • NUMBER 44 sulfonamide (CSC11). CSC11 and CSC7 caused 50% inhibition of plaque formation (IC 50 , viral entry) at a concentration of 0.8 mM (compared with 1.7 mM for zanamivir), whereas CSC5 was less effective (1.5 mM). As a control for specificity, we assessed the effect of the compounds on plaque formation by a closely related paramyxovirus, simian virus 5 (PIV5); this virus was not sensitive to the inhibitory effect of the three molecules (supplemental Fig. S1), suggesting the inhibitory properties of the identified molecules were specific for HPIV3.

New Antiviral Strategy for Paramyxoviruses
CSC11 Inhibits Viral Infection in Human Airway Epithelium-For clinical utility, the ability of an antiviral agent to limit the viral lifecycle and prevent multiple rounds of infection is key. We recently established that a HAE model closely reflects the in vivo behavior of a panel of HPIV3 variants (28). The HAE model was previously used to characterize the cell specificity of respiratory syncytial virus (45) and HPIV3 (46), studies that confirmed the model replicates those paramyxovirus-HAE interactions occurring in the human lung. We also recently employed this system to test a sialidase-based inhibitor, and showed a direct correlation between the HAE and in vivo results (47). We therefore hypothesized that peptide efficacy in this ex vivo model would correlate with in vivo efficacy. We assessed the effect of the two better compounds (CSC11 and CSC7) on viral infection in HAE (Fig. 2). The pseudostratified epithelium was infected at the apical surface with wt HPIV3 in the presence of 3 mM of each compound. After a 90-min adsorption period, the liquid from the apical surface, containing virus and compound, was aspirated, and the titer of virus emerging from the apical surface was measured at 72 h post-infection (28). Fig. 2 shows that at 72 h post-infection, the viral titer in CSC11-treated tissues was reduced by almost 100%, whereas in those treated with either CSC7 or zanamivir, viral titer was reduced by ϳ75%. These results provided support for the efficacy of CSC11 in the natural host and will lead to future experiments with more potent analogs of this compound in an animal model.
New Effective Inhibitors of Viral Entry Also Inhibit Fusion-In beginning to explore the mode of action of the inhibitory compounds, we tested their effect on cell fusion mediated by HN/F using a ␤-galactosidase complementation assay. Fig. 3 shows that fusion was sensitive to CSC11 and CSC7 (and far less sensitive to CSC5). The concentrations of CSC11 and CSC7 required for 50% inhibition of fusion were similar to those causing similar inhibition of viral entry in Fig. 1. In contrast, as shown previously (35), zanamivir required lower concentrations to prevent fusion than to inhibit plaque formation (Figs. 1 and 3). We have previously discussed the notion that the glycoprotein receptor contact during viral glycoprotein-mediated cell-cell fusion differs from that during viral entry (48). It is logical therefore that a compound like zanamivir, whose effect depends on inhibiting HN-receptor binding, might require different concentrations to inhibit cell-cell fusion than to inhibit viral entry. The fact that the new compounds CSC11 and CSC7 do not show this difference in dose required to inhibit entry and cell fusion suggested that they do not act by inhibiting receptor binding but by a different mechanism. The next set of experiments addressed this question directly.
CSC11 Does Not Block HN-Receptor Interaction-We assessed the effect of the new compounds on HPIV3-HN receptor binding in an hemadsorption assay, which quantitates the binding of RBCs to HN/F-coexpressing cells. Fig. 4 shows that CSC11 and CSC7 (in contrast to zanamivir) did not have any   CSC11 and CSC7 Exert a Virucidal Effect-Because CSC11 and CSC7 do not interfere with HN receptor binding (Fig. 4), we hypothesized that their inhibition of viral entry might be attributable to a direct virucidal effect independent of the virus-target cell interaction. To test this hypothesis, virions were incubated with the compounds at 37°C or 4°C for 60 min, and, after removal of the compounds, the infectivity of the treated virions was assessed by plaque reduction assay. Fig. 5A shows that pretreatment of the virus with CSC11 and CSC7 (but not with zanamivir) had an irreversible effect on infectivity. Despite removal of these compounds prior to the assay, viral entry was reduced by 70 -80%. Incubation with the compounds at 4°C did not lead to a reduction in infectivity (not shown), consistent with the hypothesis that these compounds affect F-triggering, which cannot occur at this temperature. CSC5 had only a weak effect on infectivity.
To determine whether the loss in infectivity by effectively treated virions entailed destruction of the particles themselves, we assessed viral RNA in infectious and non-infectious preparations. The viral RNA levels in samples pretreated with CSC11 were similar to viral RNA levels in samples treated with either DMSO or zanamivir (Fig. 5B), indicating that the decreased infectivity caused by CSC11 was due to viral inactivation and not attributable to a loss of viral particles. The absolute values for the titers of infectious particles after each treatment, and the corresponding RNA copies, are provided in table form in supplemental Table S2. Although the number of infectious particles is ϳ6-fold less in samples treated with CSC11, the RNA copy number remains similar to control samples, and therefore the particle to plaque-forming units ratio is ϳ6-fold higher.
From this point forward we focused the studies on CSC11, which had emerged as a compound that effectively blocked both viral entry and viral glycoprotein-mediated fusion, and powerfully inactivated virus, clearly acting via a mechanism other than receptor blockade; we proceeded to investigate the mechanism of action.
CSC11 Inactivates the HN/F Fusion Machinery Prior to Receptor Engagement-To investigate the point in the fusion process at which CSC11 acts, we modified an assay that we previously designed to distinguish between different states of  A, aliquots of HPIV3 viral preparations were incubated with 3 mM solutions of the indicated compounds or DMSO/vehicle control for 60 min at 37°C. The compounds were removed using filtration columns, and the resulting viral infectivity was determined by plaque reduction assay. The effect of pretreatment with test compounds on infectivity is shown as plaque number compared with that obtained with virus that was not preincubated with test compounds. B, the effect of pretreatment with the test compounds on viral RNA copy number was determined using quantitative PCR. The results in A and B are means (ϮS.D.) of triplicate experiments. F-activation (19,26,29). This assay allowed us to address whether fusion and infectivity is modulated by CSC11 before or after HN engages the target cell receptor. To assess the stage of CSC11 action, cells coexpressing HPIV3 F and HN were allowed to bind to RBCs for 0 or 30 min at 4°C, at which point the cells were washed and medium containing 3 mM CSC11 or control compounds was added. The cells were transferred to 37°C for 60 min to allow F-activation, and the amount of target RBCs that had undergone fusion was quantified. Varying the length of RBC incubation allowed us to assess CSC11 activity after HN-receptor engagement (30-min RBC incubation) or at the same time as HN-receptor engagement (0-min RBC incubation). In each set of cells, the control compounds were zanamivir (which acts by inhibiting receptor binding) and a fusioninhibitory HRC peptide (derived from the C-terminal heptad of repeat region of F) that prevents the activated F from forming the six-helix bundle structure required for fusion, and thus acts after fusion activation (49 -51). Fig. 6A shows the striking finding that CSC11 inhibits only if exposed to the viral glycoproteins before HN has engaged its receptor. After HN-receptor interaction, as in the case of preincubation with target RBCs, the inhibitory effect of CSC11 was not seen. Blocking HN-receptor engagement with zanamivir, or blocking post-activation F-protein folding with inhibitory peptides, resulted in equal levels of fusion inhibition, because these mechanisms did not require the compounds to interact with HN or F prior to receptor engagement. (See supplemental Fig. S2 for absorbance values.) The results shown in Fig. 6A indicate that CSC11 acted on the viral glycoproteins before the moment when HN engaged the receptor. The inhibition was virus-specific; CSC11 did not inhibit the related paramyxovirus PIV5 (see supplemental Fig.  S1) even if preincubated with the virus (data not shown).
To further assess CSC11 activity prior to receptor engagement and to dissect its likely mechanism of action, we repeated this experiment with an additional step: we first incubated the HN/F-coexpressing cells with CSC11 or control compounds at 4°or 37°C for 60 min, after which we washed the cells and incubated them with RBCs at 4°C. After 30 min, we transferred the cells to 37°C to allow F-activation and quantitated the level of fusion. Fig. 6B reveals that CSC11 did not inhibit RBC fusion if exposed to HN/F-coexpressing cells at 4°C, a temperature that does not permit F-activation (29). If, however, CSC11 was exposed to the HN/F-expressing cells at 37°C (a temperature that permits F-activation (29)) and before receptor engagement, then cell fusion was ultimately inhibited. Binding to RBCs occurred after the CSC11 incubation, unperturbed (data not shown). As expected, the removal of zanamivir and fusion inhibitory peptides prior to the addition of RBCs resulted in a complete lack of inhibition at both 4°and 37°C, consistent with the proposed mechanism of action of these compounds. (See supplemental Fig. S2 for absorbance values.) Taken together, these results indicate that CSC11 inactivates the HN/F fusion machinery prior to receptor contact, while leaving the binding pocket of HN intact. We postulated that CSC11 interacts with HN to induce F-activation, leading to viral inactivation. The next set of experiments directly addressed this mechanism.
CSC11 Acts by Causing Untimely Structural Rearrangement of the F Protein-To confirm that the mechanism of action was in fact premature activation of the fusion machinery, we devised a biochemical assay to detect conformational change in F upon fusion activation, a strategy based upon similar assays used for detection of conformational change in influenza HA upon pH activation (52,53). In these experiments we first established that the native F protein is resistant to proteinase K treatment, while the post-triggered F is sensitive to degradation, and showed that receptor engagement by HN causes F to become sensitive to protease degradation. We then showed that CSC11 treatment causes the same change in the protease sensitivity of F. For the experiment shown in Fig. 7A, cells expressing HN and F were preincubated with or without receptor engagement, and with or without CSC11, then transferred to 37°C for 1 h to allow fusion to occur. The HN/F complexes were then immu- noprecipitated and subjected to proteinase K digestion. A Western blot of immunoprecipitated F shows that if receptor engagement is permitted in the absence of any compound (first lane of each triplet) the F protein becomes susceptible to protease degradation and is increasingly degraded by the higher concentrations of protease. In the absence of receptor engagement (blocked by zanamivir, second lane of each triplet) the levels of F protein remain nearly constant, indicating that, before receptor engagement, F is protected from proteinase K degradation. In the presence of CSC11 (third lane of each triplet) the F protein becomes susceptible to protease digestion and is degraded as fully as when activated by HN-receptor engagement. Note that, in the samples treated with CSC11, zanamivir is also present to block HN-receptor engagement (in the second lanes), indicating that it is solely the interaction with CSC11 that accounts for the conversion to protease susceptibility. Although it would seem an appropriate control to show samples treated with CSC11 alone, the presence of zanamivir in the assay was required to prevent receptor-triggered F-activation, which would itself induce protease susceptibility. The observed conformational change in F induced by CSC11 required the presence of the receptor-binding protein, HN; the experiment in Fig. 7B shows that, if only F was expressed, the F did not become more susceptible to protease digestion whether exposed to receptor or to CSC11. The set of experiments in Fig.  7B is identical to those in Fig. 7A, but now with only F expressed. Without HN, the conformational change in F did not occur. These results confirm our hypothesis, that CSC11 inactivates infectivity by inducing activation of the fusion mechanism.
CSC11 Inactivates Infectivity via Interaction with HN-We next determined whether, as we hypothesized, CSC11 interacts with HN to produce its effect. We considered the alternative interpretation of the experiment shown in Fig. 7; that CSC11 might bind a site on F, inducing conformational change that requires the presence of HN to be manifested. To address this question we used a chimeric receptor-binding protein composed of the stalk domain of HPIV3 HN and the globular domain of NDV. We and others have shown that the specificity for F-activation is conferred by the HPIV3 stalk (19) and that receptor engagement by the NDV globular head suffices for this chimeric HN to mediate activation of HPIV3 F (54,55). We could thus determine whether interaction of CSC11 with the HPIV3 globular head is specifically required for its effect on F-activation, in which case the same effect should not be seen with the NDV head, or whether CSC11 might bind F instead.
In the experiment shown in Fig. 8, cells were coexpressing HPIV3 F and either HPIV3 HN or the chimeric receptor-binding protein (HPIV3 HN stalk 1-144/NDV HN globular head 124 -571) (56), both constructs untagged for proper comparison. The cells were incubated with cycloheximide for 1 h and then exposed for 1 h to either CSC11 or control medium at either 4°C (a temperature at which F-activation does not occur) or 37°C. At 37°C, the HN/F pairs that mediate F-activation should do so. The cells were then washed and allowed to bind RBCs at 4°C for 30 min. The cells were then transferred to 37°C for 120 min during which activation of HPIV3 F and fusion could occur. After the 120-min incubation at 37°C, the supernatant fluids were collected, and the cells were lysed for quantitation of the fused RBC population. The bar graph in Fig. 8   shows the percent inhibition of fusion under each condition, obtained from the absorbance values of the RBCs that had undergone fusion under each condition. The data reveal that CSC11 treatment led to a drastic failure of fusion mediated by the HPIV3 HN/F pairs during subsequent incubation at 37°C; however, incubation of the chimeric HN/F pairs with CSC11 did not inhibit subsequent fusion. (See supplemental Fig. S3 for absorbance values.) These results indicate that in all other respects the chimeric HN is functioning like the HPIV3 HN in fusion promotion, but is not affected by CSC11, indicating that the globular head of HPIV3 HN is required for the inactivation of F that is mediated by CSC11.

DISCUSSION
The first step in paramyxovirus infection is binding of the receptor-binding protein (HN for HPIV3) to cell surface receptors (sialic acid-containing molecules). Receptor engagement irreversibly activates the viral fusion proteins (F) to their fusion-ready conformation, and F then inserts and fuses the viral envelope with the target cell's membrane, allowing viral entry and infection. Here, we have shown that it is possible for HN to induce F triggering prematurely, thus inactivating the viruses before they can enter the target cells. The finding, that paramyxovirus receptor mimics induce premature triggering of F distant from the target cell, provides the proof of concept for a new antiviral platform, with clinical significance for not only paramyxoviruses, but also other viruses utilizing a Class I fusion mechanism for viral entry.
The timing of F-activation is critical for paramyxovirus entry (15,25,28,57,58). In the case of PIV5, a mutated F protein with a hyperactive fusion phenotype releases the fusion peptide before reaching the target cell, thus preventing fusion with the target cell (58). Inappropriate triggering of F can also be induced by agents, such as tert-butyl hydroquinone (TBHQ), an effective inhibitor of influenza, which was shown to promote the conformational change of the hemagglutinin molecule (59). For HIV, cellulose acetate phthalate inhibits viral fusion by triggering the formation of so-called "dead-end bundles" of the helical coiled-coiled rods in gp41, suggesting that cellulose acetate phthalate inhibits fusion by inducing premature exposure of the fusion peptide (60). We recently showed that, for a related paramyxovirus, Nipah virus, artificial cell-like particles (protocells) that bear Nipah virus receptor molecules (EFNB2) on their surfaces can inactivate Nipah virus envelope glycoprotein pseudotyped virus particles, preventing infection, after only a transient interaction with the viral particles. Nipah virus, like parainfluenza virus, enters target cells via the concerted action of the two surface glycoproteins; in this case the binding protein (G) is responsible for activating the fusion protein to generate the final fusion-ready state. The antiviral protocells inactivate pseudotyped virus only under conditions that permit triggering of the viral fusion protein (27). These results supported our proposal (15) that receptor mimics can interact with the receptor-binding protein at the physiological binding site and cause F-triggering prior to target cell contact, thereby preventing paramyxoviral infection.
The experimental evidence presented here confirms that the timing of F-activation is key for successful viral entry and pro-vides proof that small receptor mimics can induce HN to prematurely trigger the fusion process, without contact with a target cell receptor. To elucidate the mechanism by which fusion is inactivated, we drew upon previous work that examined conformational changes of the fusion proteins of influenza (52), Sendai virus (61), and avian sarcoma/leukosis virus (ASLV) (62,63). We developed a biochemical assay to detect a conformational change in F upon activation, observing that the proteaseresistant, native conformation of HPIV3 F transitioned to a protease-sensitive state after activation. The compound CSC11 promotes protease sensitivity of F only in the presence of the HN molecule, indicating that CSC11 prematurely activates the HN/F fusion machinery specifically via an interaction with HN. Note that, without coexpression of HN, in Fig. 7B, increasing concentrations of protease treatment resulted in some degradation of F; however, the degradation was not greater in the CSC11-treated lanes. The degradation in the experiment shown in Fig. 7B was consistent with our finding that HN and F are complexed before receptor engagement (data not shown) and that F is protected from degradation within this complex. The biochemical assay for protease sensitivity provides direct evidence that a structural rearrangement of HPIV3 F can be induced before contact with any target cell, indicating that the transition to the extended, protease-sensitive transitional state may occur independently of a target cell membrane.
The proposed mechanism of action by CSC11 is contingent on its interaction with the active sites on the HN globular head. Site I is not only responsible for HN receptor-binding function, but also for neuraminidase activity (22). Zanamivir, which blocks receptor interaction, binds HN at this primary site. CSC11 affects neither receptor-binding nor neuraminidase activity. We have identified a second functional binding site (site II) that contains both binding and F-triggering activities (29,32). For NDV, our experimental data (32) and the revised crystal structure (21) revealed a second receptor binding site in the HN head, and we showed that this second binding site is activated by engagement of the primary binding site (64). For HPIV3, however, the dynamics are different. Although the crystal structure of HPIV3 HN revealed only one binding site (22), we experimentally identified a second receptor binding site that works totally differently from that of NDV; its binding is weak (whereas for NDV activation of site II leads to stronger receptor binding), its binding activity is pH-regulated (26), and it plays a direct role in triggering F (29). To evaluate whether the F-triggering activity of CSC11 might act via an interaction with site II, the structure of CSC11 (and in parallel zanamivir, for comparison) were docked into site II of HN. The best computationally docked poses of these compounds at site II yielded HINT scores (43,44) of 1100 for CSC11 and 600 for zanamivir, corresponding to an approximate binding energy difference of 1 kcal mol Ϫ1 (65), favoring CSC11. This computational analysis suggests that CSC11 might act via interaction with the binding site II of HN.
Compounds that inactivate a virus before host cell attachment by prematurely triggering fusion are of potential therapeutic significance as anti-paramyxovirus therapies. If efficient receptor mimics trigger and thus inactivate F after budding and release, then released virus could be rendered non-infectious and viral spread halted. The irreversible loss of infectivity of HPIV3 virions caused by exposure to premature F-triggering agents points to the possibility of developing a new class of antivirals to combat the severe disease and high lethality associated with paramyxoviruses. We contend that both inhibition testing and mechanistic studies are likely to reflect the authentic biology if carried out in tissue resembling the natural host, and for this reason we employ the human airway epithelium model that closely reflects the in vivo behavior of HPIV3 (28) and HPIV3 antiviral molecules (47). The efficacy of the premature triggering antiviral strategy in a physiologically relevant ex vivo system that mimics the human airway supports the notion that this strategy may be useful clinically. These data thus open the avenue to a new antiviral strategy for paramyxoviruses and other viruses that enter cells via related fusion mechanisms, for which receptor binding causes activation of the fusion/entry mechanism.