Crystal structure and solution state of the C-terminal head region of the narmovirus receptor binding protein

ABSTRACT Increased viral surveillance has led to the isolation and identification of numerous uncharacterized paramyxoviruses, rapidly expanding our understanding of paramyxoviral diversity beyond the bounds of known genera. Despite this diversity, a key feature that unites paramyxoviruses is the presence of a receptor-binding protein (RBP), which facilitates host-cell attachment and plays a fundamental role in determining host range. Here, we study the RBP presented on the surface of rodent-borne paramyxoviruses Mossman and Nariva (MosV and NarV, respectively), viruses that constitute founding members of the recently defined Narmovirus genus within the Paramyxoviridae family. Crystallographic analysis of the C-terminal head region of the dimeric MosV and NarV RBPs demonstrates that while these glycoproteins retain the canonical six-bladed β-propeller fold found in other paramyxoviral RBPs, they lack the structural motifs associated with established paramyxovirus host-cell receptor entry pathways. Consistent with MosV-RBP and NarV-RBP undergoing a distinct entry pathway from other characterized paramyxoviruses, structure-based phylogenetic analysis demonstrates that these six-bladed β-propeller head domains form a singular structural class that is distinct from other paramyxoviral RBPs. Additionally, using an integrated crystallographic and small-angle X-ray scattering analysis, we confirm that MosV-RBP and NarV-RBP form homodimeric arrangements that are distinct from those adopted by other paramyxovirus RBPs. Altogether, this investigation provides a molecular-level blueprint of the narmovirus RBP that broadens our understanding of the structural space and functional diversity available to paramyxovirus RBPs. IMPORTANCE Genetically diverse paramyxoviruses are united in their presentation of a receptor-binding protein (RBP), which works in concert with the fusion protein to facilitate host-cell entry. The C-terminal head region of the paramyxoviral RBP, a primary determinant of host-cell tropism and inter-species transmission potential, forms structurally distinct classes dependent upon protein and glycan receptor specificity. Here, we reveal the architecture of the C-terminal head region of the RBPs from Nariva virus (NarV) and Mossman virus (MosV), two archetypal rodent-borne paramyxoviruses within the recently established genus Narmovirus, family Paramyxoviridae. Our analysis reveals that while narmoviruses retain the general architectural features associated with paramyxoviral RBPs, namely, a six-bladed β-propeller fold, they lack the structural motifs associated with known receptor-mediated host-cell entry pathways. This investigation indicates that the RBPs of narmoviruses exhibit pathobiological features that are distinct from those of other paramyxoviruses.

Jeilongvirus, and Salemvirus (1)(2)(3)(4)(5).This update to paramyxovirus taxonomy has allowed two previously termed "orphan paramyxoviruses" (6,7), Nariva virus (NarV) and Mossman virus (MosV), to become founding members of the Narmovirus genus.NarV was isolated on four separate occasions in the early 1960s from forest rodents in Trinidad and Tobago (8,9) and MosV was isolated in the 1970s from the pooled organs of wild rats originating from two separate locations in Queensland, Australia (10,11).The "bank vole virus" (BaVV) (12), which was discovered in Russia, and a narmovirus (UKMa K4D) prevalent in approximately 4% of field voles studied in Cheshire and Leicestershire, UK, have been more recently identified and putatively added to this newly established genus (13).Little is known about the disease burden imposed by these pathogens upon wild animal reservoirs, nor the capacity of these viruses to be transmitted to non-native host species, a characteristic common among many paramyxoviruses (14).
A fundamental component of the paramyxovirus life cycle, in both infection of native host reservoirs and during spillover into new host-species, is the ability of the virus to productively interact with host-cell surface receptors during host-cell entry (7).This process is facilitated by a virus envelope-displayed receptor-binding protein (RBP), which is presented on the paramyxovirus surface as a dimer-of-dimers (15).Each protomer of the RBP consists of an N-terminal intraviral (IV) region, transmembrane (TM) domain, an α-helical stalk, and a C-terminal six-bladed β-propeller head domain, which mediates the interaction with the cognate host-cell receptor (15)(16)(17)(18).The initial interaction between an RBP and receptor precedes virus internalization and fusion glycoprotein (F)-mediated merger of the virus and host-cell membranes (19,20).
Although crystallized constructs of the C-terminal six-bladed β-propeller head of paramyxovirus RBPs have been predominantly observed as monomers or homodimers (17,(21)(22)(23)(24)(25)(26)(27)(28)(29), the full-length protein forms dimer-of-dimers, which assemble through the formation a four-helix bundle (4HB) in the stalk region (17,20,26).Independent of the receptor utilized, receptor recognition allosterically triggers class-1 type rearrangements of the fusion (F) glycoprotein, via residues encoded in the stalk region of the RBP, in a process that merges virus and host-cell membranes (30).The molecular basis of this process is unknown and has led to the "clamp" and "provocateur" models, where the F glycoprotein is shielded from premature activation or triggered by the paramyxoviral RBP, respectively (31).
All known paramyxovirus RBP stalk regions share structural features, including the 4HB and a flexible linker region that connects to the receptor binding head domain.However, despite these commonalities, the exact sites of F activation may likely differ given that the length of the stalk region in HN-type RBPs is shorter than those of the protein binding H/G-type RBPs.In vitro, the process of fusion activation can be decoupled from receptor binding, where studies utilizing truncated paramyxovirus RBPs, that do not present the β-propeller head, found that the stalk region alone was capable of promoting fusion (20,(32)(33)(34)(35)(36).However, it was also revealed that in the case of MeV, that reverse engineered virus bearing headless RBP had severely impaired growth in mammalian cell culture, likely due to premature F triggering which could lead to a loss of infectivity (35).Similarly, virions presenting headless NiV-RBP were unable to enter cells due to premature triggering of F to an irreversible post-fusion state (36).
Reflective of paramyxovirus RBPs displaying diverse receptor-binding functionality, the primary amino acid sequences of these proteins are highly variable (7).For example, pairwise comparison reveals that although MosV-RBP and NarV-RBP cluster together within a phylogenetic tree of RBP sequences (7), their primary sequences exhibit only ~30% identity.Here, we sought to clarify the structural relationship of MosV-RBP and NarV-RBP glycoproteins with characterized paramyxoviral RBPs.X-ray crystallographic analysis of these RBPs to 1.6 Å and 2.1 Å resolution, respectively, reveals a distinctive six-bladed β-propeller architecture that lacks receptor recognition features associated with H, G, and HN RBPs.Our observed dissimilarities support a model whereby narmoviruses likely use a receptor distinct from known canonical paramyxovi rus receptors.
To assess the potential for narmoviral glycoproteins to interact with cells, we measured binding of NarV-RBP β and MosV-RBP β constructs that encoded an N-terminal Fc recognition tag, to Vero cells (African green monkey), LA-4 (mouse lung adenoma) cells, and CHO-pgsA745 cells by flow cytometry.Fc-MosV-RBP β bound to all cell types, and to a lesser extent Fc-NarV-RBP β bound to Vero and LA-4 cells (Fig. S1B).We selected CHO-pgsA745, LA-4, and Vero cell lines as they have been commonly used as a benchmark in the characterization of virus-host interactions for other paramyxoviruses (40,(46)(47)(48)(49)(50)(51)(52).Indeed, due to its well-characterized interaction with Vero and LA-4 cells, and lack of interaction with CHO-pgsA745 cells, Fc-tagged Nipah virus-RBP β-propeller (Fc-NiV-G RBP β ) (40,46,53,54), was used as both a positive and negative control in this cell binding assay.These data suggest that both Vero and LA-4 cells likely display cell-surface receptor(s) recognized by NarV and MosV.Additionally, CHO-pgsA745 cells may also display cell-surface receptor(s) targeted by MosV virus only.

Narmoviral RBPs are dimeric in solution and in the crystal
Size exclusion analysis of purified NarV-RBP β and MosV-RBP β revealed that the two proteins form putative dimers in solution (Fig. S2A and C).Furthermore, the asymmet ric unit of both NarV-RBP β and MosV-RBP β structures consists of two near-identical β-propeller head domains, where each protomer forms an extensive protein−protein interface between the first (β1) and sixth (β6) blades of the β-propeller (Fig. 3).As Structural Homology Program (61), was used to calculate evolutionary distance matrices by means of pairwise superposition of RBP structures.The resultant matrices were used to plot an unrooted tree in PHYLIP (62).RBP surfaces are represented with cognate receptor binding sites colored (red), and receptors presented either as yellow ribbon (protein) or spheres (carbohydrate).Calculated evolutionary distances are indicated beside the branches.calculated by the "Proteins Interfaces Surfaces Assemblies" server (64), the association between MosV-RBP protomers occludes ~2,070 Å 2 of solvent accessible surface area, and is stabilized by 16 hydrogen bonds and three salt bridges.The interface between NarV-RBP protomers is similarly substantial, with an occluded surface area of ~2,160 Å 2 stabilized by 19 hydrogen bonds and three salt bridges.NarV-RBP β and MosV-RBP β exhibit the greatest level of structural conservation with each other in the region of this oligomeric interface with respect to the rest of the molecule (Fig. 2B), suggestive that these blades of the narmoviral β-propeller are structurally constrained to promote a similar mode of overall assembly.
Interestingly, the mode of MosV-RBP and NarV-RBP dimerization contrasts that observed in previously reported H, HN, and G-type RBP structures.Indeed, although all reported homodimeric RBP structures utilize the first and/or sixth blades (β1 and β6, respectively) of the β-propeller, the narmovirus RBP utilizes a different angle of association and level of buried surface to previously reported paramyxovirus RBP dimers.Indeed, while the HN-type RBP dimers typically assemble with a 60° angle of association between protomers, with ~1,790 Å 2 of buried surface (29), the MosV-RBP and NarV-RBP protomers are organized approximately ~90° relative to each other (Fig. 3).This mode of dimerization also contrasts morbillivirus MeV-H (117° association angle and ~1,080 Å 2 buried surface), henipavirus HeV-G RBP (40° association angle and ~880 Å 2 buried surface) structures, and pararubulavirus SosV-RBP (31° association angle and 1,340 Å 2 buried surface) (29) (Fig. S3), providing further evidence of the structural independence of MosV-RBP and NarV-RBP from structurally characterized paramyxoviral RBPs.We also note that the observed mode of MosV-RBP β and NarV-RBP β dimerization is in agreement with the location of N-linked glycosylation sequons (NXS/T where X ≠ P).Indeed, although MosV-RBP β and NarV-RBP β lack the high level of glycosylation inherent to most paramyxoviral proteins (63,(65)(66)(67)(68)(69)(70)(71), each only presenting one predicted sequon on the RBP head domain, the locality of these sequons is in line with the hypothesis that glycosylation is not expected to be occluded within protein−protein interfaces.Electron density corresponding to a single N-acetylglucosamine (GlcNAc) linked to residue Asn319 of MosV-RBP β (Fig. 3) was visible in the crystal structure.In contrast, the putative glycosylation site in the NarV-RBP β (Asn575) crystal structure was not observed.However, given that treatment of NarV-RBP β with endoglycosidase F1 resulted in the reduction of molecular mass (Fig. S2D), it seems likely that the GlcNAc at Asn575 is occupied, yet mobile in the crystal.Beyond the construct boundaries for our structural analysis, MosV-RBP β and NarV-RBP β encode one and two N-linked glycosylation sequons in their stalk regions, respectively (Fig. 1).Similarly located N-linked glycans have been shown to have a role in fusion regulation in other paramyxoviruses (72)(73)(74).
To determine the oligomeric state of glycosylated NarV-RBP β and MosV-RBP β in solution, we performed size-exclusion chromatography-coupled small-angle X-ray scattering (SEC-SAXS) on the glycosylated proteins (Table S3; Fig. S4 to S6).Under dilute conditions, SAXS measures the shape and size of macromolecular particles in solution (75).SAXS measurements suggest both proteins are compact, globular proteins in solution (Fig. S6A).For each protein, fitting of the monomeric RBP β structures to either SEC-SAXS data sets was exceptionally poor (χ 2 = 80.0 and χ 2 = 105.6 for NarV-RBP β and MosV-RBP β, respectively), suggesting that their monomeric forms do not represent the solution state (Fig. 4).In addition, model-independent analysis of the respective pair-distance distribution functions (Fig. S6B) for NarV-RBP β and MosV-RBP β shows a shoulder at ~1/2 maximum height at 60-70 Å suggesting a dimeric form of the protein.
Significant improvements in the fits to the SAXS curves were observed when using the crystallographic dimeric forms (χ 2 = 1.95 and χ 2 = 4.50 for NarV-RBP β and MosV-RBP β , respectively).However, the crystallographic models are incomplete, as electron density for the N-and C-terminal residues and GlcNAc 2 Man 9 glycans (derived by kifunensine treatment) were not observed.To complete the models, we used simulated annealing molecular dynamics (SA-MD) simulations with torsion angle restraints and additional hydrogen-bond and distance restraints derived from the respective crystal structures.The high-temperature simulated annealing was cycled repeatedly producing ~1,000 sampled conformations of each RBP β protein.Using this SA-MD approach, a set of best-fitting models was identified, which demonstrated an expected flexibility across the glycans and disordered termini, including the NarV-RBP β extended C-terminus with unknown function (Fig. 4).Using the updated dimeric forms of the structures, the fits to the SAXS curves were improved to χ 2 = 0.65 and χ 2 = 1.28, indicative that the derived models represent the solution state of NarV-RBP β and MosV-RBP β , respectively.The resulting MosV-RBP β and NarV-RBP β dimers retained angular differences in subunit association that were in line with the values of the dimeric crystal structures, 81° and 89°, respectively (Fig. S6C and S6D).Additionally, the MosV-RBP β and NarV-RBP β best-fit structures generated by high-temperature simulated annealing, largely retained the extensive homodimeric interfaces found in both crystal structures.Post-MD, MosV-RBP β and NarV-RBP β retained ~1,370 Å 2 and ~1,880-1,960 Å 2 (calculated from three of the best-fitting NarV-RBP MD models) of the previously calculated ~2,070 Å 2 and ~2,170 Å 2 buried surface area, respectively.These results are summarized in Table S3.This analysis indicates that the RBP β dimerization occurs in solution, further supporting the hypothe sis that our structurally observed homodimeric narmovirus RBP β organization resembles a biologically relevant assembly (Fig. 4).Furthermore, this analysis also supports the intrinsic flexibility of surface-exposed loop regions that exhibit high RMSD values upon overlay of NarV-RBP β and MosV-RBP β crystal structures (Fig. 2B).In toto, this integrated structure, solution state analysis, and molecular dynamics approach indicate that our structurally observed homodimeric interfaces of narmovirus RBPs are unlikely to be features that are solely specific to crystallization.

DISCUSSION
Viral genome sequencing within wildlife reservoirs has provided glimpses into the extensive genetic diversity that exists within the Paramyxoviridae, allowing expansion of the number of genera within the family (4).Rodent-borne paramyxoviruses from the genus Narmovirus exemplify this diversity, where the founding members, NarV and MosV (from Trinidad and Tobago and Australia, respectively), have been putatively joined by recently identified narmoviruses from the UK (13) and Russia (12).Indeed, the growth of this genus reflects our increasing appreciation of the broad geographic range assumed by this group of pathogens.However, despite this surprising prevalence, little is known about narmovirus pathobiology, host-range, or inter-species transmission potential, features that are expected to be modulated, at least in part, by the RBP (1,7).
Here, we take an initial step to address this paucity of knowledge through the structural determination of the C-terminal head region of the RBP from NarV and MosV.Our previous structure-based classification analyses have revealed that paramyxovirus RBP head domains that group within the same structural class recognize the same or similar receptors (38).When applied to the structural data presented here, this approach (Fig. 2D) reveals that the head domain of narmovirus RBDs bear little structural similarity with paramyxoviruses with known receptors, suggestive of unique receptor utilization.Sendai virus (SeV) (NC_001552.1).The seven conserved sialidase residues (45) and hexapeptide motif (78) are labeled according to residue and blade location (45) and annotated above alignments.
While narmoviral receptor(s) are yet to be identified, our cell binding data show that Fc-MosV-RBP β and Fc-NarV-RBP β putatively interact with both LA-4 and Vero cells, and that Fc-MosV-RBP β may also interact with CHO-pgsA745 cells, albeit at low levels (Fig. S1B).The observed level of Fc-tagged narmoviral RBP β binding was less than that of Fc-NiV-RBP β binding to LA-4 and Vero cells.These data indicate that LA-4 and Vero cells may present a receptor(s) on their surface which interacts with MosV and NarV, and CHO-pgsA745 cells a receptor(s) that interacts with MosV.However, further studies are necessary to clarify whether this level of binding is functionally relevant to support the productive entry of native virions and whether the studied cell types are permissive to MosV and NarV infection.To assess if binding equates to cell entry, as observed for NiV (39,43,54,79), fusion and cell entry assays with native narmoviral virions or narmovirus pseudoviruses require development.Such assays are essential for future investigations focused on identifying host-cell receptors and augmenting our understanding of the determinants of narmovirus tropism.
Similar to other paramyxoviruses, full-length and ectodomain constructs of MosV-RBP and NarV-RBP are expected to oligomerize as a tetrameric, dimer-of-dimers organization.This oligomerization is likely driven by hydrophobic residues within the stalk region and disulfide bonding (17,20,58).Furthermore, MosV-RBP β and NarV-RBP β are expected to be flexibly linked to the stalk region, where, similar to other paramyxoviruses, this conformational plasticity may facilitate fusion activation (32).However, the exact organization of narmoviral RBPs, especially with respect to the cognate receptor(s) and/or F, requires further investigation.Nonetheless, these combined observations support a model whereby NarV and MosV undergo a process of host-cell attachment that is distinct from characterized H, HN, and G RBP-bearing paramyxoviruses.
In sum, our NarV-RBP β and MosV-RBP β structures provide molecular-level blueprints that support the independent functional and structural classification of narmovirus RBPs from other paramyxoviruses (4).This work offers a platform for future investiga tions focused on assessing and rationalizing the pathobiological characteristics and the receptor(s) utilized by narmoviruses for host-cell entry.While the threat that NarV, MosV, and other narmoviruses pose to human health and animal husbandry remains unknown, by defining the RBP architecture assumed by this group of viruses, this work renders us better prepared to understand and respond to pathogenic narmoviruses, if they emerge.
Protein was produced by transient transfection of human embryonic kidney 293T cells, and secreted protein was harvested after 72 h incubation at 37°C, 5% CO 2 .For crystallization and SAXS analysis, protein was produced in the presence of 5 µM kifunensine (55) and purified using immobilized metal-affinity chromatography.Prior to application on the HisTrap HP (Cytvia) column, cell supernatant was exchanged into 10 mM Tris (pH 8.0), 150 mM NaCl, and concentrated using an ÄKTA Flux diafiltration system (Cytvia).His-tagged protein was eluted using 250 mM imidazole.Subsequently, if required for crystallization, N-linked sugars were cleaved at the di-N-acetylchitebiose core using endoglycosidase F1 (EndoF1) (10 µg/mg protein, 12 h, 21°C).SEC was performed in 10 mM Tris pH 8.0, 150 mM NaCl buffer using a Superdex 200 10/30 column (Cytvia).For cell-binding analysis, cell supernatants were exchanged into 20 mM sodium phosphate pH 7.0 buffer prior to affinity purification on a HiTrap Protein G HP (Cytvia) column.Fc-tagged protein was eluted by washing the column into 0.1 M glycine-HCl pH 2.7 before immediate neutralization with 60 µL of 1 M Tris-HCl pH 9.0 per mL of eluate.Subsequently, Fc-tagged protein was purified and exchanged into 20 mM sodium phosphate pH 7.0 buffer using SEC on a Superdex 200 10/30 column (Cytvia).

Cell-binding studies
Vero cells were maintained in Dulbecco's modified Eagle medium (DMEM) with 10% heat-inactivated (HI) fetal bovine serum (FBS).CHO pgsA745 hamster ovary cells were maintained in DMEM/F12 medium supplemented with 10% HI FBS.LA-4 cells, a mouse lung epithelial cell line, were obtained from the American Type Culture Collection (ATCC) and maintained in Ham's F12K medium supplemented with 15% HI FBS.Cultured cells were collected with 10 mM EDTA, then incubated for 1 h with soluble, Fc-tagged receptor-binding protein, which was produced as described above.Cells were then washed twice in 2% FBS in Dulbecco's phosphate-buffered saline (DPBS), stained with secondary anti-Fc-allophycocyanin (APC) antibody at a 1:2,000 dilution and washed twice again.Cells were subsequently fixed in 2% paraformaldehyde and resuspended in 2% FBS in DPBS prior to flow cytometry (Guava easyCyte).For flow cytometry experi ments, approximately 3,000 events were captured per condition for Cho pgsA745 and Vero cells and approximately 2,000 events were captured per condition for LA-4 cells.Cho pgsA745 was previously described to display no binding to NiV-RBP, so it served as a negative control for these experiments (54).Each cell line was stained with condition using only secondary anti-Fc-APC to serve as a background control for the respective cell line.FlowJo software was subsequently used to analyse the data by first gating for live cells then single cells prior to determining the geometric mean fluorescent intensity for the gated population.

Crystallization and structure determination
MosV-RBP β crystals were grown using the nanoliter-scale sitting-drop vapor-diffusion method at room temperature, using 100 nL protein (5.5 mg/mL) and 100 nL reservoir (81).Crystals grew in a precipitant containing 0.2 M L-arginine, 0.1 M Tris pH 7.8, 8% poly-γ-glutamic acid (PGA)-LM, 6% dextran sulfate and were immersed in 20% glycerol prior to cryo-cooling by plunging into liquid nitrogen.Initial X-ray diffraction data were collected at a wavelength of 0.9795 Å on beamline I03, Diamond Light Source (DLS).Reflections were processed to a resolution of 2.75 Å using the xia2 package (82) (Table S1).For experimental phasing, crystals from this same condition were soaked in a solution containing potassium tetrachloroplatinate (K 2 PtCl 4 ) for 3 h, prior to cryo-cooling with a 20% glycerol solution containing K 2 PtCl 4 diluted in precipitant.Crystals were exposed to a wavelength consistent with the platinum LIII absorption edge (λ = 1.072Å) at beamline I04, DLS.The isomorphous differences between the native data set and the derivatized data set, in addition to the anomalous signal derived from platinum, enabled subsequent phase determination using SIRAS using the Autosol wizard (57).A further MosV-RBP β data set was collected on crystals that grew in the following precipitant mixture: 0.2 M potassium bromide, 0.2 M potassium thiocynate, 0.1 M sodium cacodylate pH 6.5, 3% PGA-LM, and 20% (wt/vol) polyethylene glycol (PEG) 550 MME.Crystals were harvested and cryo-cooled in a 20% glycerol solution diluted with precipitant.Data were collected at a wavelength of 0.9795 Å, on beamline I04, DLS, and reflections were processed to 1.62 Å using the xia2 package (82) (Table S1).Molecular replacement using PHASER (83) with the initial MosV-RBP model was utilized to solve the higher-resolution data set.
NarV-RBP β crystals were grown using nanoliter-scale sitting-drop vapor-diffusion at room temperature, using 100 nL protein (4.5 mg/mL) and 100 nL reservoir (81).Small crystals grew in a precipitant containing 0.2 M magnesium chloride, 0.1 M HEPES pH 7.5, 25% PEG 3350, and 10% PEG 400.Crystals were optimized by seeding into a lower concentration solution of NarV-RBP β (84).Crystals were pulverized, utilizing a seed bead (Hampton Research, USA) (85), and dispensed onto a plate prepared with the original precipitant mix and NarV-RBP β at a concentration of 3 mg/mL, yielding a crystal suitable for X-ray data collection (84).Larger crystals formed following seeding, enabling harvesting.A crystal was immersed in 20% glycerol prior to cryo-cooling by immersion into liquid nitrogen.Data were collected at a wavelength of 0.9795 Å on beamline I04, DLS, and reflections were processed to 2.07 Å using the xia2 package (82) (Table S1).A partially refined model of MosV-RBP was used to solve the structure of NarV-RBP β by molecular replacement with PHASER (83).For all models, building and structure refinement were iteratively performed using the programs COOT and Phenix.Refine, respectively (Table S2) (86,87).Non-crystallographic symmetry restraints were employed throughout, and translation-libration-screw parameters were employed for later rounds of refinement.Models were validated using the Molprobity server (88).

SAXS with inline high-performance liquid chromatography
SAXS was used to characterize the solution state of glycosylated MosV-RBP β and NarV-RBP β (Table S3).Data were collected on beamline B21, at the DLS, configured to measure across the scattering vector range 0.0032 Å −1 < q < 0.38 Å −1 .For high-perform ance liquid chromatography mode, a 45 µL sample at a concentration of 5 mg/mL was loaded onto a Superdex 200 PC 3.2/30 (Cytvia).Buffer TBS was washed over at a rate of 0.075 mL/min.The SAXS instrument was coupled directly with in-line SEC with exposures collected every 2 s (89).SEC-SAXS profiles corresponding to a single chromatographic separation were analyzed with the program, ScÅtter, where peak and background selection and data reduction (www.bioisis.net)were performed to produce a single SAXS curve for each protein sample.

Molecular dynamics with SAXS
Molecular dynamics simulations were performed with the crystallography and NMR systems (CNS) program CNSsolve version 1.3 (http://cns-online.org/v1.3/).Missing Nand C-terminal tails were added back using the generate_seq.inp,generate.inp,and model_anneal.inpscript from CNS. Crystallographic structures for each RBP served as templates to derive NOE like distance restraints that folded a starting extended polypeptide chain into the respective, folded, crystallographic monomeric RBP.Scale factors for the NOE energy term and molecular dynamics time steps were adjusted down to minimize large energy terms in the gradient descent.For each refold, greater than 20 models were produced from independent random starts.The model with the lowest energy term (maximized NOE-like distance restraints) served as the base template for further model building.Base templates were then uploaded to the GLYCAM-Web server (https://glycam.org) to add the GlcNAc 2 Man 9 glycans.The glycosylated monomer was duplicated and superimposed onto the crystallographic subunits to produce a full-length, glycosylated RBP dimer for each virus.The completed models were then used with model_anneal.inpfor SA, torsion angle MD to sample conformation space.Sampling was performed using randomly selected distance restraints (~3% non-hydro gen, backbone, and C-beta distance pairs) derived from the crystallographic models.Residues with B-factors greater than two times the average were excluded from selection, which coincided with residues in the loops forming the rim of the β-propeller.Additional intersubunit hydrogen bond restraints were derived between the subunit interfaces to act as rigid restraints.The upper and lower limits for each restraint was set by the Shannon resolution (π/q max ).Several rounds of SA-MD produced a set of refolded models that varied slightly in the conformation of the sugars, loops and N-and C-termini.SAXS profiles were calculated for MosV-RBP β and NarV-RBP β structures using the fast X-ray scattering (FoXS) server (76), and matched to SAXS experimental profiles.Determination of the success of the fit was demonstrated by assessment of the χ 2 statistic as well as the FOXS parameters c 1 and c 2 , with c 2 <4 (c 2 >4 suggests over-fitting).

Structure-based phylogenetic analysis
For structural phylogenetic analysis, RBP six-bladed β-propeller monomers were prepared by removal of water molecules, ligands, and protein residues outside of the canonical fold.Structures were analyzed with the Structural Homology Program (61,90).Pairwise evolutionary distance matrices were used to generate an un-rooted phyloge netic tree in PHYLIP (62).

Dimer angle analysis
UCSF Chimera was utilized to analyze the relative angles of monomers within the paramyxoviral dimers (91).Planes representing the top faces of the monomer subunits were constructed based upon conserved stretches of paramyxoviral RBP sequence using the "Define plane functionality, " with the angle between the monomers of a dimer calculated using these planes.the gene diagram.(B) Soluble receptor binding protein (RBP) binding to LA-4, Vero, and CHO pgsA745 cell lines.The indicated concentration of soluble, Fc-tagged RBP was incubated with cells, then stained with APC-tagged anti-Fc secondary antibody and subjected to flow cytometry as described in the methods.Adjusted geometric mean fluorescence intensity (GMFI) was calculated as a product of the percent APC positive cells and the GMFI of the APC positive cells.Shown are the results of two replicates for each cell line with error bars representing the standard error of the mean.MosV-RBP β showed moderate binding and NarV-RBP β showed low level binding to both LA-4 and Vero cells.Additionally, MosV-RBP β showed a low level of binding to the CHO pgsA745 cells.The decreased binding affinity of NarV-RBP β could be attributed to the shorter construct length.Fig. S2 (mBio01391-23_s0002.tif).Size exclusion chromatograms of MosV-RBP β and NarV-RBP β .(A) MosV-RBP β and (B) NarV-RBP β expressed transiently in the presence of kifunensine, pre-(blue) and post-deglycosylation (grey).Samples were run across a Superdex™ 200 10/30 column (Cytvia) and compared with gel filtration standards (orange; BioRad).The expected size of the monomeric and dimeric MosV-RBP β was 49 kDa and 98 kDa, respectively.The size standards eluted from the column in the following order: thyroglobulin (670 kDa), β-globulin (158 kDa), ovalbumin (44 kDa), myoglobu lin (17 kDa), vitamin B12 (1.35 kDa).The calculated molecular weight of monomeric MosV-RBP β is 49 kDa, the species observed through SEC analysis (44 kDa−158 kDa) would be consistent with a dimeric arrangement of MosV-RBP (98 kDa).The expected molecular weight of NarV-RBP, is 53 kDa, however the species observed with SEC analysis is much larger, displaying a size consistent with the 158 kDa gel filtration size standard.This is larger than the expected size of the monomer, consistent with hypothesis that NarV-RBP forms a stable higher order oligomer in solution (dimer 106 kDa; trimer 159 kDa).Prior to crystallization the protein was treated with EndoF1 to remove high-man nose glycoforms.SDS-PAGE analysis of the purified (C) MosV-RBP β and (D) NarV-RBP β pre-and post-deglycosylation, shows a shift following treatment with EndoF1.0), orange, and Rg, cyan, estimated from the Guinier region for each subtracted frame.For a single concentration measurement made over several frames, radiation damage will be observed as an increase in I(0) and Rg.For SEC-SAXS, I(0) should change with the concentration of the particle during elution.(E) Log 10 intensity plot of subtracted and merged SAXS frames.Black represents averaged buffer frames subtracted from averaged sampled frames.Cyan represents median of the buffer frames subtracted from the averaged sample frames.Poor buffer subtraction leads to a displacement between the two curves at high-q.Fig. S5 (mBio01391-23_s0005.tif).NarV-RBP β SEC-SAXS summary.(A) SEC-SAXS Signal Plot.Each point is the integrated area of the ratio of the sample SAXS curve to the estimated background (gray).Averaged background horizontal dashed line.(B) Overlay of subtracted SAXS curves used in final averaged curve (frames 418 to 437).Each frame is scaled to peak.(C) Durbin-Watson and Shapiro-Wilks tests examining the distribution of the residuals between two frames.In this case, comparisons are made in reference to the first frame.Radiation damage or lack of similarity can be observed as a trend in either statistic across the frame set.Likewise, similarity is demonstrated by a random distribu tion of the statistics.(D) Double Y plot with I(0), orange, and Rg, cyan, estimated from the Guinier region for each subtracted frame.For a single concentration measurement made over several frames, radiation damage will be observed as an increase in I(0) and Rg.For SEC-SAXS, I(0) should change with the concentration of the particle during elution.(E) Log 10 intensity plot of subtracted and merged SAXS frames.Black represents averaged buffer frames subtracted from averaged sampled frames.Cyan represents median of the buffer frames subtracted from the averaged sample frames.Poor buffer subtraction leads to a displacement between the two curves at high-q.Fig. S6 (mBio01391-23_s0006.tif).SAXS Summary plots for MosV-RBP β and NarV-RBP β and alignment of narmovirus β-propeller structures pre-and post-molecular dynamics (MD).(A) Dimensionless Kratky Plot.Crosshairs denote Guinier-Kratky plot (peak position for an ideal globular particle).Convergence to baseline and peak position supports MosV (violet) and NarV (cyan) RBP β-propellers are compact but not ideally globular parti cles.(B) Pair-distance distribution function from indirect Fourier transform of datasets described in table S3).Plots were prepared using the program ScÅtter (www.bioisis.net).MosV-RBP β (C) and NarV-RBP β (D) post-MD dimers were aligned to a single protomer of the crystallographic dimer.For this analysis only the b-propellers were aligned, with the GlcNAc2Man9 glycans and additional residues at the termini being omitted.The MosV-RBP β and NarV-RBP β structures are shown in cartoon representation, with the crystallographic dimers coloured gray, and the post-MD dimers coloured violet and cyan, respectively.Fig. S7 (mBio01391-23_s0007.tif).The surface of MosV-RBP is likely unable to accommo date the G-H loop of B-type ephrin ligands.The surface of NiV-RBP (left) (white), with the G-H loop of ephrinB2 shown in cartoon representation (yellow) (2VSM) [2].To the right MosV-RBP (blue) is shown in surface representation with the G-H loop docked into the putative binding site, revealing that the loop would be unlikely to bind due to the lack of the necessary pocket.Supplemental tables (mBio01391-23_s0008.pdf).Tables S1 to S3

FIG 3
FIG 3 MosV-RBP β and NarV-RBP β present homodimeric interfaces.The crystallographic MosV-RBP β and NarV-RBP β dimers are shown in cartoon representation (gray).The angle of association, calculated using UCSF Chimera, is shown above the dimer.Blue and red spheres are shown at the positions of the N-and C-termini, respectively.For MosV-RBP, 2Fo-Fc electron density is shown at the position of glycan Asn319.

FIG 4
FIG4 SAXS analysis validates the organization of MosV-RBP β and NarV-RBP β as dimers in solution.Experimental MosV-RBP (A) and NarV-RBP (B) SAXS data (gray circles) compared to theoretical SAXS curves calculated from crystallographically observed monomeric (dark gray line) and, dimeric (black line) oligomeric states.In both cases, crystallographic structure determination did not reveal electron densities for N-, C-termini, and the putative GlcNAc 2 Man 9 glycans that were present during the SAXS experiment.The SAXS data sets fit better to completed structures (addition of modelled N-and C-termini, and glycosylations, red) than the crystal structures.Application of high-temperature, simulated annealing molecular dynamics (SA-MD) of completed, dimeric structures (purple and cyan for, MosV-RBP β and NarV-RBP β , respectively) improved fitting further.Log intensity (I(q)) is plotted against the Porod invariant (Q)(76).χ 2 values are shown and were generated by calculating the fit between structural models (i.e., "monomer, " "crystal dimer, " "starting all atom glycosylated dimer, " and "best post-MD glycosylated dimer") and the experimental (SAXS) data.The best χ 2 values were obtained following MD of glycosylated, all atom MosV-RBP (χ 2 = 1.28) and NarV-RBP (χ 2 = 0.65) dimers.Inset plots (right) show the residuals calculated against the final models.

FIG 5
FIG 5 MosV-RBP and NarV-RBP lack motifs associated with known modes of paramyxovirus entry.(A) NiV-RBP (PDB ID 2VSM) (2) is shown as a surface with ephrinB2 shown as a yellow ribbon and the cognate receptor-binding footprint colored gray and outlined in yellow (far left).Ghana virus (GhV) RBP (second left; 4UF7), MosV-RBP (second right), and NarV-RBP (far right) are shown in surface representation with residues colored according to sequence conservation with NiV-RBP.Identical residues are colored red and equivalent pink.(B) MeV-RBP (PDB ID 3ALZ) (22) is shown in surface represen tation with SLAM-F1 (green) and nectin-4 (cyan) shown in ribbon representation.The cognate joint receptor-binding footprint is colored gray (far left) and outlined according to the SLAM-F1 (green) and nectin-4 (cyan) binding footprints.As no CDV-RBP structure is currently available, sequence conservation of CDV-RBP (second left) with MeV-RBP is mapped onto the surface of MeV-RBP.MosV-RBP (second right) and NarV-RBP (far right) are shown in surface representation with residues colored as above, according to sequence conservation with MeV-RBP.(C) Alignment of the RBP amino acid sequences from MosV (NP_958054.1),NarV (YP_006347588.1),BaVV (ATW63189.1),Newcastle disease virus (NDV) (YP_009512963.1),AMPV-2 (AQQ11616.1),AMPV-3 (AWU68199.1),mumps virus (MuV) (MG460606.1),and Fig. S3 (mBio01391-23_s0003.tif).Representative paramyxoviral RBP homodimeric interfaces.The crystallographic dimers of NDV-RBP β (1E8V), SosV-RBP β , HeV-RBP β (2X9M) and MeV-RBP β (3INB) are shown in cartoon representation (grey).The angle of associa tion, calculated using UCSF Chimera, is shown above the dimer.Fig. S4 (mBio01391-23_s0004.tif).MosV-RBP β SEC-SAXS summary.(A) SEC-SAXS Signal Plot.Each point is the integrated area of the ratio of the sample SAXS curve to the estimated background (gray).Averaged background horizontal dashed line.(B) Overlay of subtracted SAXS curves used in final averaged curve (frames 418 to 437).Each frame is scaled to peak.(C) Durbin-Watson and Shapiro-Wilks tests examining the distribution of the residuals between two frames.In this case, comparisons are made in reference to the first frame.Radiation damage or lack of similarity can be observed as a trend in either statistic across the frame set.Likewise, similarity is demonstrated by a random distribu tion of the statistics.(D) Double Y plot with I(