On the Stability of Parainfluenza Virus 5 F Proteins

Abstract

The crystal structure of the F protein (prefusion form) of the paramyxovirus parainfluenza virus 5 (PIV5) WR isolate was determined. We investigated the basis by which point mutations affect fusion in PIV5 isolates W3A and WR, which differ by two residues in the F ectodomain. The P22 stabilizing site acts through a local conformational change and a hydrophobic pocket interaction, whereas the S443 destabilizing site appears sensitive to both conformational effects and amino acid charge/polarity changes.

TEXT

Paramyxoviruses, the cause of many important human and animal diseases, constitute a large family of enveloped negative-stranded RNA viruses. For viral infection, enveloped viruses need a means of fusing their viral membrane with a target cell membrane. Most paramyxoviruses achieve membrane fusion by the coordinated action of a tetrameric attachment protein (variably called HN, H, or G) and a trimeric fusion protein (F). While the tetrameric attachment protein tethers the virion to the target cell by binding sialic acid or protein moieties on target cells, the trimeric fusion protein physically merges the viral envelope with a host cell membrane at neutral pH. F initially folds to a metastable prefusion conformation that is activated by cleavage during cellular trafficking. Upon target cell binding, the attachment protein, comprised of a globular head anchored to the membrane by a helical tetrameric stalk, triggers an irreversible ATP-independent refolding event in F that promotes the fusion of viral and host cell membranes (reviewed in references 1 and 2).

The current “stalk exposure” model for the activation of F (3) posits that the binding of target cell receptors by the attachment protein globular heads reveals an F-activating region in the attachment protein stalk. The attachment protein's F-activating region then interacts with an exposed hydrophobic loop in the IgG-like domain of F to provide the energy needed to promote F refolding and membrane fusion (4, 5). This energy barrier to the activation of F can also be overcome by increased temperature (6–10) or by the introduction of destabilizing point mutations (11–13). The paramyxovirus parainfluenza virus 5 (PIV5) WR isolate differs in sequence from the W3A isolate by only three residues: the L22, P443, and A516 in WR are P22, S443, and V516 in W3A (14). Mutation of the WR F residues into the W3A F backbone reveals that mutation P22L stabilizes F, resulting in a hypofusogenic phenotype, whereas mutation S443P destabilizes F, resulting in a hyperfusogenic phenotype (11). As previously noted, WR F alone does not significantly deviate in fusion phenotype from that of W3A (11).

To understand the structural basis of stabilization and destabilization by residues 22 and 443 (with residue 516 not considered because it is in the F cytoplasmic tail), we determined the atomic structure of soluble WR PIV5 F-GCNt and compared it to the previously determined structures of cleaved and uncleaved prefusion W3A F-GCNt (PDB identification no. 4GIP and 2B9B, respectively) (15, 16). The P22L and S443P mutations (Fig. 1A) were made in the W3A F-GCNt backbone to yield WR F-GCNt; the protein was expressed and purified as described previously (15) and analyzed on a silver-stained gel and by immunoblotting (Fig. 1B). Electron microscopy of F confirmed it was predominantly in the prefusion conformation (Fig. 1C and D) (6). The atomic structure was determined to a final resolution of 2.98 Å by molecular replacement with the atomic structure of W3A PIV5 F-GCNt. The model refinement statistics and details are shown in Table 1. In WR F, L22 causes the N terminus to rotate such that two hydrophobic residues, L20 and L22, face inwards toward a hydrophobic pocket and D21, which points inwards in W3A, rotates away from the globular head (Fig. 1F). We hypothesize that the stabilizing effects of P22L are due to an increased interaction of the N terminus with a hydrophobic pocket.

FIG 1.

The atomic model of the F protein of the WR isolate of parainfluenza virus 5. (A) Schematic diagram of the domain organization of PIV5 F-GCNt. Domains I, II, and III are colored in yellow, red, and magenta, respectively. The hydrophobic fusion peptide (FP) is colored in light pink. The locations of P22 and S443, which are mutated to a leucine and a proline, respectively, in the WR F-GCNt construct, are highlighted. (B) Western blot (WB) and a silver-stained (SS) SDS-PAGE gel showing the protein purity post-affinity column purification. (C and D) Electron micrographs of negatively stained WR F-GCNt and W3A F-GCNt, respectively, showing comparable percentages of prefusion trimers in each sample. (E) A cartoon representation of the WR F-GCNt atomic structure colored as in panel A. (F and G) Zoomed-in regions of panel E showing the 2F_o − F_c electron density map at 1 σ around residues 22 and 443, respectively. The W3A F-GCNt model, in cyan, was aligned to the WR F-GCNt model, in gray, in PyMOL to highlight the changes between the two structures.

TABLE 1.

Data collection and refinement statistics for WR F-GCNt

Parameter	Result for WR F-GCNta
Data collection
Source	Advanced Photon Source
Wavelength (Å)	1.07808
Space group	P212121
Unit cell dimensions
a (Å)	82.528
b (Å)	155.523
c (Å)	157.260
α (°)	90
β (°)	90
γ (°)	90
Resolution range (Å)	35.0–2.98 (3.11–2.98)
Rmerge (%)	12.2 (61.0)
I/σ〈I〉	14.64 (2.15)
Completeness (%)	98.4 (90.8)
Redundancy	5.1 (4.8)
Refinement
Resolution (Å)	34.85–2.98
No. of reflections used (work/free)	38,970/1,995
Rwork/Rfree (%)	23.04/28.13
No. of:
Residues/atoms	1,435/1,0736
Ordered waters modeled	0
Solvent content (%)	61.44
Avg B-factors (Å2)	83.4
RMSDb
Bond length (Å)	0.011
Bond angle (°)	1.317
Ramachandran plot statistics (%)
Residues in preferred regions	92.61
Residues in allowed regions	6.26

^aValues in parentheses represent the highest-resolution shell.

^bRMSD, root mean square deviation.

The structural change in WR F introduced by P443 is subtle relative to the structure of W3A F. WR-P443 resulted in a slight lateral shift of the backbone in the extended HRB linker, most likely due to disruption of the hydrogen bond between W3A-S443 and D448 (dotted turquoise line in Fig. 1G). The lack of a significant structural change does not explain the dramatic hyperfusogenic phenotype of W3A mutant S443P and its lower temperature for triggering fusion (11), suggesting that this mutation acts by lowering the activation energy barrier.

Conservative and nonconservative mutations were introduced into the W3A F backbone at residues 22 and 443 to probe the mechanistic basis of the P22L and S443P phenotypes (Fig. 2). Cell surface expression of the mutants was determined by flow cytometry using the prefusion conformation-specific monoclonal antibody (MAb) F1a (17). All the mutants tested, with the exception of the S443D and S443T mutants, were well expressed at the cell surface, suggesting that the proteins are properly folded (Fig. 2A). Fusion activity of the mutants was measured by luciferase reporter assay (Fig. 2B) or syncytium assay in BHK-21 cells (Fig. 2C).

FIG 2.

Mutagenesis analysis for residues P22 and S443. (A) Prefusion F protein expressed on the surface of transfected 293T cells was detected using the prefusion specific monoclonal F1a antibody and flow cytometry. Mean fluorescence intensity (MFI) values were calculated and normalized against wild-type W3A F. The experiments were done in triplicate, and the error bars represent ±1 standard deviation. (B) F protein fusion activity quantified using a luciferase reporter system and normalized against wild-type W3A F plus HN fusion activity. P22 mutations are ordered according to increasing side-chain hydrophobicity, and S443 mutations are ordered by side-chain molecular weight. PIV5 HN was transfected in equal amounts with each construct, with the exception of the multiple-cloning site (MCS) vector negative control. The experiments were done in triplicate, and the error bars represent ± 1 standard deviation. (C) Monitoring F protein activity through syncytium formation of transfected BHK-21 cells. Cells were fixed and stained 18 h posttransfection. “MCS,” “HN only,” and “F only” are negative controls, while “F+HN” is the W3A wild-type F and HN positive control.

The phenotype of the P22 mutations suggests that both local peptide conformation and a size-dependent hydrophobic interaction are important for F stabilization. None of the P22 mutations tested showed elevated levels of membrane fusion, including the P22G mutation, for which no side-chain interaction would be present. Whereas the P22G and P22A mutants showed moderate degrees of fusion, the P22L, P22F, and P22W mutants (Fig. 2B and C) showed minimal levels of fusion. These data are consistent with the larger hydrophobic side chains interacting with a hydrophobic pocket to stabilize the prefusion F conformation. Overall, mutation of P22 showed a fusion phenotype resistant to significant destabilization as mutations P22H and P22N also led to reduced fusion.

Residue S443 participates in a hydrogen-bonding network at the base of the W3A F globular head, interacting with the charged residue D448 (15). It was hypothesized that the disruption of this network would lead to destabilization of the prefusion state and a hyperfusogenic phenotype. However, the S443G mutation, which would disrupt the H-bond network, behaves similarly to W3A F, indicating that conformational effects of the proline substitution may influence F stability at this position. S443 is sensitive to steric and electrostatic potential changes. The electrostatic potential modulating S443H, S443N, and S443D mutations all resulted in significantly hyperfusogenic phenotypes. Given the conservative nature of mutant S443T, the low level of surface expression and relatively high fusion activity were surprising. These data suggest that even minor changes at residue 443 can significantly alter F stability. In contrast, the S443L mutation resulted in a reduction in fusion activity, which could be due to stabilization of prefusion F or perhaps to unfavorable hydrophobic interactions occurring along the F refolding pathway.

Overall, the 443 site appears more sensitive to amino acid changes than the P22 site. Here, we investigated the molecular basis by which single point mutations cause changes in fusion phenotype by comparison of the atomic structures of the PIV5 WR and W3A isolates. The P22 stabilizing site appears to act through a local conformational change and a hydrophobic pocket interaction, while the S443 destabilizing site appears sensitive to both conformational effects and amino acid charge/polarity changes.

Protein structure accession number.

The atomic coordinates and structure factors have been deposited in the Protein Data Bank (www.pdb.org) under PDB accession no. 4WSG.