Structure of the cleavage-activated prefusion form of the parainfluenza virus 5 fusion protein

Abstract

The paramyxovirus parainfluenza virus 5 (PIV5) enters cells by fusion of the viral envelope with the plasma membrane through the concerted action of the fusion (F) protein and the receptor binding protein hemagglutinin-neuraminidase. The F protein folds initially to form a trimeric metastable prefusion form that is triggered to undergo large-scale irreversible conformational changes to form the trimeric postfusion conformation. It is thought that F refolding couples the energy released with membrane fusion. The F protein is synthesized as a precursor (F0) that must be cleaved by a host protease to form a biologically active molecule, F1,F2. Cleavage of F protein is a prerequisite for fusion and virus infectivity. Cleavage creates a new N terminus on F1 that contains a hydrophobic region, known as the FP, which intercalates target membranes during F protein refolding. The crystal structure of the soluble ectodomain of the uncleaved form of PIV5 F is known; here we report the crystal structure of the cleavage-activated prefusion form of PIV5 F. The structure shows minimal movement of the residues adjacent to the protease cleavage site. Most of the hydrophobic FP residues are buried in the uncleaved F protein, and only F103 at the newly created N terminus becomes more solvent-accessible after cleavage. The conformational freedom of the charged arginine residues that compose the protease recognition site increases on cleavage of F protein.

The Paramyxoviridae are enveloped, negative-strand RNA viruses that are significant pathogens in humans and animals (1). The family includes parainfluenza viruses 1–5 (PIV1–5), mumps virus, measles virus, Newcastle disease virus, Sendai virus, Hendra virus, Nipah virus, respiratory syncytial virus, and metapneumovirus. To enter cells, paramyxoviruses, like all enveloped viruses, must fuse the viral envelope with a membrane of a host cell. For paramyxoviruses, this process involves two viral spike glycoproteins: a receptor binding protein, variously called HN, H, or G, and the fusion protein, F (2, 3).

The paramyxovirus F protein is a class I viral fusion protein that initially folds in the endoplasmic reticulum into a trimeric metastable prefusion form and on triggering undergoes major irreversible conformational changes (refolding) to form the trimeric postfusion conformation. F protein refolding couples the energy released with membrane fusion (4). The F protein is synthesized as a precursor (F0) that must be cleaved either by a host protease (furin or furin-like protease) in the trans Golgi apparatus or by an extracellular trypsin-like enzyme to form the biologically active molecule F1,F2. Cleavage creates a new N terminus on F1 that contains a highly conserved hydrophobic region known as the fusion peptide (FP) (5). Cleavage of F is a prerequisite for fusion and virus infectivity (6, 7), and intracellular cleavage of F correlates with virus pathogenicity (8).

For PIV5, the receptor-binding protein hemagglutinin-neuraminidase (HN) binds to sialic acid moieties on the cell surface and is required for the activation of F occurring at the plasma membrane and at neutral pH. It is thought that F interacts with the stalk domain of HN (9–17). On fusion activation, F undergoes refolding, resulting in formation of a trimeric coiled coil composed of a heptad repeat A region that extends away from the viral membrane (18–20).

Peptide inhibitor studies and available atomic structures indicate that many of the key elements of this entry mechanism are common to other class I viral fusion proteins, such as the hemagglutinin (HA) of influenza virus, gp120/41 of HIV, S protein of severe acute respiratory syndrome coronavirus, and glycoprotein (GP) of Ebola virus (reviewed in ref. 4). Although X-ray structures of the six-helix bundle of many type I fusion proteins have been determined, more complete postfusion ectodomain structures are known only for PIV3 F, NDV F, and RSV F (21–25). Furthermore, structures of the prefusion conformation of type I fusion proteins have been solved only for influenza virus HA, PIV5 F, and Ebola virus GP (20, 26–28). The atomic structures of both uncleaved and protease-cleaved prefusion forms are available only for influenza virus HA (26, 28). The before and after cleavage HA structures are largely superimposable, except for residues near the protease cleavage site that compose a surface loop. The structures yield valuable information that helps explain observations regarding the protease recognition site and provides insight into the acid lability of HA after cleavage activation (28).

Earlier work using antisera to peptides derived from the F sequence suggested considerable change in antibody reactivity occurring on cleavage of F0 to F1,F2 (29). Here we present the crystal structure of the cleaved, prefusion form of the soluble ectodomain of PIV5 F. Similar to influenza virus HA, the paramyxovirus F uncleaved and cleaved structures are largely superimposable, except for the residues composing and surrounding the protease recognition site. Unlike HA, there is no concerted movement or burying of the N-terminal residues of the FP. However, because PIV5 F is triggered for fusion by its receptor binding protein HN at neutral pH, there is no need for an HA-like mechanism for priming sensitivity to low pH.

Results and Discussion

Expression and Crystallization of the Prefusion PIV5 F-GCNt Protein in its Cleaved Form.

PIV5 F is a type I transmembrane GP with a 19-residue signal sequence, a large (465-residue) ectodomain, a C-terminal transmembrane anchor, and a short cytoplasmic tail. F is synthesized as a precursor (F0) that must be cleaved for F to be biologically active. Proteolytic cleavage of F0 to form the disulfide-linked chains F1 and F2 generates an N terminus on F1 that contains the hydrophobic FP. WT PIV5 (strain W3A) F contains five arginine (Arg) residues at the cleavage site, and cleavage occurs intracellularly in the trans Golgi network by furin or a furin-like protease during transport of F to the cell surface (1). The atomic structure of uncleaved prefusion F, a mutant F containing three Arg residues at the cleavage site (residues Arg-98, -99, and -100, with Arg-101 and -102 deleted), indicates that the cleavage site residues form a small surface-exposed loop (20).

To express a soluble ectodomain of F0 protein suitable for crystallization, we used an F protein (F-GCNt) that contains three Arg residues at the cleavage site and a coiled-coil trimerization domain fused in a helical frame with the C-terminal heptad repeat B (HRB) region in place of the transmembrane anchor and cytoplasmic tail (20). This trimerization domain is necessary for stabilizing the PIV5 F-soluble protein in its prefusion form (20, 22) (Fig. 1A).

Fig. 1.

Biochemical and structural characterization of PIV5 F-GCNt. (A) The major F-GCNt domains are highlighted by color. Residue ranges are also indicated. (B) SDS/PAGE analysis under reducing conditions of uncleaved (F0) and trypsin-cleaved (F1 + F2) PIV5 F-GCNt. The gel indicates the completeness of protease cleavage and purity of protein used for crystallization trials. The F1-GCNt and F2 fragments are the expected sizes, indicating that the trypsin cleavage is specific for the multibasic protease cleavage/activation site. (C and D) Electron microscopy of cleaved prefusion (C) and postfusion (D) FGCNt reveals characteristic “tree-like” and “golf tee” structures, respectively. Heat was used as a surrogate for HN activation to convert F-GCNt to the postfusion conformation, which assembles into rosettes. Arrowheads indicate molecules viewed from the side and in C point to the boundary between the treetop and trunk. (E) Cartoon representation of the crystal structure of the cleaved prefusion F-GCNt trimer viewed from the side and colored by domains as in A. Arrowheads indicate the locations of the termini resulting from protease cleavage. (F) Surface representation of the trimer oriented as in E with color-coded chains (A, yellow; B, red; C, orange) and the hydrophobic FP highlighted in blue. On cleavage, the FP remained sandwiched between DII and DIII of adjacent chains as in the uncleaved structure. The GCNt domain was omitted from the model because of insufficient electron density.

F-GCNt was expressed from insect cells using a recombinant baculovirus expression system, secreted into serum-free medium, and affinity-purified via a 6× His tag that was appended to the C terminus of the GCNt domain. Exogenous trypsin was used to cleave F-GCNt to allow temporal control and optimization of the cleavage protocol, thereby facilitating a homogenous population of cleaved prefusion F-GCNt. After inactivation of trypsin with protease inhibitors and repurification of F-CGNt by nickel chromatography (Ni-NTA), protein gel analysis (SDS/PAGE) indicated the protein was cleaved into F1-GCNt and F2 and was >95% pure (Fig. 1B). Electron microscopy revealed that the cleaved F-GCNt was stable and remained in the prefusion form when stored at 4 °C for days, as demonstrated by the characteristic “tree-like” morphology (Fig. 1C). However, heating the sample to 60 °C for 10 min (a surrogate for HN activation) converted cleaved F to the postfusion conformation, with its characteristic “golf tee” morphology with rosette formation (30) (Fig. 1D).

Structure of Cleaved Prefusion PIV5 F-GCNt.

The cleaved prefusion PIV5 F-GCNt crystallized in the space group C2 and diffracted X-rays anisotropically up to 2.0–3.0 Å resolution (Table S1). The structure was solved by molecular replacement using the uncleaved prefusion PIV5 F-GCNt structure [Protein Data Bank (PDB) ID code: 2B9B], and a single trimer was found in the asymmetric unit.

Each F-GCNt monomer consists of four domains—DI, DII, DIII, and HRB regions—along with connecting loops. The three monomers are highly interconnected, and together DI, DII, and DIII form a globular head, whereas HRB forms a trimeric coiled-coil stalk (Fig. 1 E and F). The overall shape of the molecule resembles a “tree” and is remarkably similar to the uncleaved prefusion PIV5 F-GCNt structure (20), which superimposes with an all-atom rmsd of 0.473 Å over 1,096 atoms (Fig. S1). Notably, most residues of the FP remain stationary after protease cleavage, likely because many of the FP residues are buried at the interchain boundary where DII and DIII interact (Fig. 1F and Fig. S2).

The model contains all residues of the secreted protein N-terminal to the GCNt domain for all three chains. The GCNt region exhibited weak electron density, and attempts to fit or build GCNt into the available density increased the R_free value slightly; thus, GCNt was omitted from the final model.

An omit map was generated by deleting residues near the cleavage site (Arg-88 to -107) of the uncleaved F model used for molecular replacement. A strong difference in electron density was detected for the missing residues up to residue 97 before the cleavage site and starting with residue 106 after the cleavage site in each monomer. Missing residues were built into the cleaved model independently for each monomer, and noncrystallographic symmetry restraints were not used during refinement. Electron density is relatively weak for the residues immediately adjacent to the cleavage site (Arg-98 to -100 and -103 to -105; Arg-101 and -102 are deleted from the construct). However, in each case, the R_free value increased when these residues were separately deleted from the final model. Thus, these residues were included in the final model.

The relatively high B-factors and weak electron density of residues bracketing the cleavage sites show that these residues can adopt multiple conformations after cleavage. Indeed, the model differs slightly near the cleavage sites for each chain (Fig. 2 A, C, D, and E). A portion of heptad repeat C (HRC) near the cleavage loop is shown in Fig. 2B to demonstrate the agreement between the final model and the electron density map (2mFo-DFc) for this region.

Fig. 2.

Cleavage site in the cleaved prefusion PIV5 F-GCNt structure. (A) Overlay of chains A (magenta), B (yellow), and C (blue) of the cleaved F-GCNt structure zoomed-in on the protease cleavage site. The F103 side chain, at the N terminus of F1 after cleavage, is shown as sticks, and the C-terminal alpha carbon of F2 is shown as a sphere for each chain. (B) Ribbon diagram overlay of chain A from the uncleaved (green) and cleaved (magenta) structures zoomed-in as in A. The cleaved structure is shown as a trimer making DII from the adjacent chain C (light red) visible. F103 side chains are shown as sticks. (C) Detailed view of interactions involving F103 of chain A (magenta). An intermolecular hydrogen bond is apparent between the carbonyl carbon of F103 and the NH1 nitrogen of R419 in DII (light red) of chain C. F103 of chain A is further stabilized by several intermolecular hydrophobic interactions with DII of chain C, including interactions with Q389, M388, and S419, as well as an intramolecular hydrophobic interaction with V106. (D) Detailed view of interactions involving F103 of chain B (yellow). Here intermolecular interactions with DII of the adjacent chain are minimal. Instead, F103 packs against nearby residues in chain B, including T97, R98, A104, and V106. The carbonyl carbon of F103 forms an intramolecular hydrogen bond with the nitrogen of G105, possibly shared with the nitrogen of V106. (E) Detailed view of interactions involving F103 of chain C (blue). F103 packs against the globular structure, making hydrophobic contact with P96, V107, V125, V128, and K129. A water-mediated hydrogen bond is observed between the NH2-terminal nitrogen and the NZ nitrogen of K129. (F) A portion of the model from the cleaved prefusion PIV5 F-GCNt crystal structure near the protease cleavage site showing the final fit to the 2mFo-DFc electron density map.

Comparison of the Cleaved and Uncleaved Prefusion PIV5-FGCNt Structures.

After cleavage of F-GCNt, the surface loop containing the cleavage site relaxes relative to the more kinked conformation in the uncleaved structure (Fig. 2B). This action directs the Arg residues immediately preceding the cleavage site away from the globular structure. Before cleavage, F103 is buried against the globular head in a hydrophobic patch comprising I95, P96, V106, V125, V128, and K129 from the same chain; however, after cleavage, F103 of chain A moves outward to form hydrophobic intramolecular interactions with V106 as well as intermolecular interactions with Q384, M388, and S412, and a hydrogen bond with R419 of an adjacent chain (chain C; Fig. 2C). F103 of chain B is also projected away from the globular structure; however, its side chain forms intramolecular interactions with T97, R98, A104, and V106 (Fig. 2D). In contrast to chains A and B, in chain C, F103 packs into the same hydrophobic patch as in the uncleaved structure, albeit with a different orientation (Fig. 2E).

The calculated surface electrostatic values of PIV5 F-GCNt reveals several interesting features. The stalk domain is almost entirely negatively charged. Another large negative patch lies at the boundary where the FP interacts with the HRC helix. There is also a large positively charged patch formed by surface residues in DIII, the FP, and the heptad repeat A helix at the boundary where these intersect (Fig. 3 A and C). Although subtle differences are observable in the electrostatic surfaces when comparing the uncleaved and cleaved structures, the differences likely are not significant, but rather represent small variations among the structures. Importantly, except for exposure of the newly generated termini of F2/F1-GCNt, the conformational changes occurring near the cleavage site after protease cleavage exhibit no substantial net burying or exposure of hydrophobicity or charge (Fig. 3).

Fig. 3.

Electrostatic surface renderings of uncleaved and cleaved PIV5 F-GCNt. (A) Side view of uncleaved PIV5 F-GCNt trimer with electrostatic surface shown. (B) Close-up view of the furin cleavage loop (chain A) with a cartoon representation (gray) visible through the partially transparent surface. The F103 side chains are shown as sticks. (C and D) Similar views of the cleaved structure as shown in A and B. Electrostatic solvent-accessible surfaces were calculated at a salt concentration of 0.15 M using the Adaptive Poisson–Bolzmann Solver and PDB2PQR software (v1.3) via the CHARM algorithm and mapped onto the vdw surface in Pymol (v1.3). The electrostatic potential ranges from −2 (red) to +2 (blue) kT/e.

PIV5 F Hydrophobic FP Residues Are Buried Before Cleavage and Do Not Undergo a Concerted Conformation Change on Cleavage Activation.

Influenza virus HA and Ebola virus GP have many similarities to the F protein of paramyxoviruses in that these proteins initially fold to a metastable prefusion form, require proteolytic activation to become fusogenic, and mediate virus–cell membrane fusion through dramatic conformational rearrangements. These proteins have several important differences, however, including the fact that HA and GP bind cellular receptors directly (31, 32), whereas most F proteins are not thought to engage receptors. In addition HA and GP are triggered to cause fusion in the low pH environment found in the lumen of endosome/lysosome compartments (33, 34). In contrast, F is activated by a separate receptor binding protein (HN, H, or G) at neutral pH.

The atomic structures of uncleaved and cleaved influenza HA provide a structural basis for understanding the pH sensitivity and proteolytic activation of the cleaved prefusion form of HA (26, 28). These structures are superimposable except for the 19 residues composing and adjacent to the protease cleavage site. In uncleaved HA, these residues form a surface-exposed loop (Fig. 4A and Fig. S2A). On cleavage, residues 1–10 of the newly formed N terminus of HA2 bind in a nearby cavity and project into the interior of the trimer (Fig. 4B and Fig. S2B). Protonation of these buried residues is thought to help destabilize the prefusion structure, inducing the conformational rearrangement to the postfusion form. In addition, the HA structures reveal that interactions between the core structure and residues in the cleavage loop may reduce the accessibility of proteases and limit virus pathogenicity, particularly when the cleavage loop is short. Strains with longer cleavage loops, especially those with multibasic residues recognized by ubiquitous proteases, are more pathogenic (reviewed in ref. 35).

Fig. 4.

Comparison of FPs in prefusion PIV5 F-GCNt, influenza HA, and Ebola virus GP crystal structures. (A and B) Top-down view of ribbon diagrams of the uncleaved and cleaved influenza HA crystal structures, respectively (PDB ID codes: 1HA0 and 2HMG) (28, 44) [Drawn from PDB ID: 1HA0 (45).] (C and D) A similar view of uncleaved and cleaved PIV5 F crystal structures, respectively (PDB ID code: 2B9B) (20). In A–D, the residues that move after protease cleavage are highlighted. Residues N-terminal to the cleavage site are highlighted in blue, and residues C-terminal to the cleavage site are shown in red. Additional residues composing the FP are shown in orange, and all FP residues have side chains, shown as sticks. (E) A similar view of cleaved prefusion Ebola GP (PDB ID code: 3CSY) (27). The entire fusion loop of Ebola virus is highlighted in orange with FP residue sidechains shown as sticks. (F) Close-up view of Ebola virus GP FP shown packed against the surface of an adjacent chain in the trimer. Color-coding is as in E. (G) Close-up view of PIV5 F FP buried at the interface between adjacent chains of the trimer. Color-coding is as in B, except the entire FP is colored orange. The C-alpha carbons at each terminus are shown as spheres, colored yellow for the new N- and C-termini created by protease cleavage (B, D, E, and G only). Ovals illustrate the area of FP.

The uncleaved and cleaved prefusion structures of PIV5 show major differences between influenza virus HA and paramyxovirus F cleavage activation. The cleavage site of PIV5 (W3A) is the C-terminal side of the R-R-R-R-R↓ recognition sequence that lies in a surface loop at the junction between the HRC and the FP (Fig. 4C and Fig. S2C). For PIV5 F-GCNt, all three Arg residues in the F cleavage site mutant are surface-exposed and protease-accessible, and all five Arg residues are assumed to be surface-exposed in WT F. After cleavage of F-GCNt, the distance between the newly created N- and C-termini of F1 and F2 is an average of 9.9 Å (Cα−Cα distance), much shorter than that for influenza virus HA1 and HA2 (22.2 Å). Furthermore, there is no concerted movement of the N- and C-termini after cleavage, no burying of exposed hydrophobic residues of the FP, and no setting of a pH-sensitive trigger (Fig. 4 C and D and Fig. S2 C and D). These observations are consistent with the findings that the FP of PIV5 F-GCNt is already mostly buried in the prefusion uncleaved structure and that PIV5 fusion is triggered by HN at neutral pH.

The crystal structure of Ebola virus GP in its furin-cleaved (GP1/GP2) prefusion form was solved in complex with a neutralizing antibody (27). Unlike paramyxovirus F and influenza HA, the hydrophobic FP resides in an internal fusion loop rather than at the N- terminus of GP2 following furin cleavage. Like PIV5 F, the FP residues of Ebola virus GP pack against an adjacent monomer within a trimer, although the Ebola FP is substantially more exposed on the exterior of the trimer surface (Fig. 4 E and F and Fig. S2E). However, unlike PIV5 F, in which FP residues are deeply buried at the interface between chains, GP FP residues bind the surface of the adjacent chain. Although additional cleavage of GP by cathepsins to remove the mucin-like domain is necessary for GP activation (36), which occurs in the low-pH environment of the endosome/lysosome, evidence suggests that an additional unidentified trigger is required to initiate refolding to the postfusion conformation (34, 37). It is interesting, however, that of the three examples of FP packing in the prefusion conformation discussed, PIV5 FP residues are mostly solvent-inaccessible (Figs. 1F and 4G and Fig. S2F), and for the fusion proteins described, PIV5 F is the only one triggered at neutral pH.

Conclusions

Here we present the crystal structure of the cleaved prefusion form of PIV5 F-GCNt, the stabilized ectodomain of the F protein. This structure, together with the previously reported structures of uncleaved prefusion F-GCNt and the postfusion structures of hPIV3 F, NDV F, and RSV F, provide a detailed high-resolution view of the various static forms of paramyxovirus fusion proteins (Fig. 5). The cleaved prefusion structures of PIV5 F, influenza HA, and Ebola virus GP exhibit different paradigms of cleavage activation for type I fusion proteins. For Ebola virus GP, the fusion activation trigger remains unknown (34, 37). For influenza HA, the details of the structure help explain how the length of the cleavage loop is a determinant of pathogenicity and how burying of hydrophobic residues helps set a low pH trigger. For PIV5 F, fusion occurs at neutral pH on activation by the HN receptor binding protein, and the hydrophobic residues that compose the FP are mostly prepacked in the uncleaved structure and remain so in the cleaved form.

Fig. 5.

Structure of distinct conformational states of the F protein ectodomain. Side-view ribbon diagrams of prefusion PIV5 F-GCNt protein in its uncleaved (PDB ID code: 2B9B) and protease-cleaved states, as well as the uncleaved postfusion structure of the hPIV3 F ectodomain (PDB ID code: 1ZTM). Arrowheads indicate the position of protease cleavage sites in the prefusion structures for two of the three chains (the third is hidden from view). Arrows indicate the progression of F through its various conformational states. An exception for the uncleaved postfusion hPIV3 F structure is noted by square brackets as cleavage activation is a biological requirement for fusion activity, although not for refolding of the majority of the F ectodomain into the postfusion form. The hPIV3 F protein lacked a GCNt trimerization domain, and it folded directly into a postfusion conformation even in the absence of protease cleavage. Each structure is color-coded by domain (DI, yellow; DII, red; DIII, magenta; HRB, blue).

Although the conformational changes are not as dramatic as those observed for HA, the changes occurring on protease cleavage of F are not benign. Earlier findings using F-specific antipeptide sera were interpreted as indicating an extensive conformational change on cleavage (29), but in retrospect this conclusion might have been affected by the possible conversion of cleaved F into postfusion F on detergent solubilization. In previous work, we found that the anti-PIV5 F mAb F1a recognizes membrane-bound cleaved PIV5 F better than membrane-bound uncleaved F (29, 38). Subtle structural changes near the cleavage site might be important for interaction with HN and in triggering of F.

Materials and Methods

Cells.

Hi5 insect cells were maintained in Express 5 serum-free medium (Gibco) supplemented with 10% GlutaMax (Gibco). Sf9 insect cell lines (for generating baculovirus stocks) were maintained in SF900 II medium (Invitrogen) containing 10% FBS.

Cloning and Mutagenesis.

The cloning and expression of PIV5 F-GCNt, which includes ectodomain residues Arg-20 to -100 and -103 to -477 (with two Arg residues in the protease recognition sequence deleted to prevent intracellular cleavage) fused to the GCNt trimerization domain, have been described previously (20). However, to optimize protein expression, the construct used in this study was subcloned into pBACgus-11, and the factor Xa cleavage site between the GCNt domain and the His tag was replaced with a single Ala residue.

Protein Expression, Digestion, and Purification.

A recombinant baculovirus that expresses F-GCNt was generated using the BacVector-2000 Transfection Kit (Novagen). Sf9 cells were used for the production of virus stocks. The virus titer was determined by plaque assay. High Five cells (Invitrogen) were infected at a multiplicity of infection of five plaque-forming units. At 65 h postinfection, the medium was harvested, and soluble F protein was purified by affinity chromatography using Ni-NTA resin (Qiagen). Purified F protein was treated with tosyl phenylalanyl chloromethyl ketone-treated trypsin (Worthington) at a ratio of 1 unit of trypsin per 10 μg of protein. After a 1-h incubation at 4 °C, the reaction was quenched with a molar excess of soybean trypsin inhibitor (Worthington). The cleaved protein was again purified with Ni-NTA resin and then dialyzed against 50 mM Tris and 100 mM NaCl (pH 7.4).

Cleaved F Protein Crystallization.

The purified cleaved F protein was heated to 40 °C for 40 min to reduce the amount of soluble aggregates. The crystals were initially obtained by the hanging-drop vapor diffusion method using a Mosquito robot (TTP LabTech) at Northwestern University’s High-Throughput Analysis Laboratory using a PEG/Ion screen (Hampton). The optimized crystals were grown at 20 °C by the sitting-drop vapor diffusion method with an equal-volume mixture of protein (9.6 mg/mL) and precipitant [0.15 M ammonium citrate dibasic (pH 5.5) and 15% wt/vol PEG 3350]. Crystals were harvested and transferred to the precipitant plus 20% glycerol for flash-freezing in liquid nitrogen.

Data Collection, Structure Determination, and Refinement.

A native dataset (400 frames spanning 240 degrees) was collected at the Life Sciences Collaborative Access Team beamline at the Argonne National Laboratory Advanced Photon Source. The data were processed to 2.0 Å using the HKL-2000 program package (HKL Research) (39); however, because of anisotropy, the data were subsequently submitted to the UCLA Molecular Biology Institute’s Diffraction Anisotropy Server for ellipsoidal truncation and anisotropic scaling (40). Using a cutoff of Fσ ≥3, the server recommended resolution limits with ellipsoid dimensions of 1/3.0, 1/2.0, and 1/2.2 Å⁻¹ along the three principle axes a*, b*, and c*, respectively. The server also applied an isotropic B-factor of −22.75 Ų. Electron density maps generated with the anisotropically processed data appeared to be of noticeably higher resolution and contained more features than those generated by processing the data to 2.55 Å in HKL-2000 [completeness = 99.9% (100%), redundancy = 5.1 (4.9), R_merge (linear) = 0.118 (0.765), R_merge (square) = 0.084 (0.532)] with no anisotropic scaling.

The uncleaved prefusion PIV5 F-GCNt structure (PDB ID code: 2B9B) was used as a search model for molecular replacement to determine initial phases in the C2 space group. Three monomers comprising the biological trimer were found in the asymmetric unit (rotation function z-score = 46.7, translation function z-score = 80.8, log-likelihood gain = 8,781). Molecular replacement with postfusion F models was not successful. Model building, structure refinement, and validation were performed with Coot (41), PHENIX Refine (42), and MolProbity (43), respectively. The use of translation, libration, screw motion (TLS) parameters (as recommended by PHENIX) and individual B-factor restraints during late stages of refinement helped lower the R_free value; however, the use of noncrystallographic symmetry restraints increased the R_free value. Data collection and final refinement statistics are provided in Table S1. Atomic coordinates and structure factors have been deposited in the PDB (www.pdb.org).