An Evolutionary Distinct Nipah Virus N-Glycosylation Site Provides Stability for Receptor Engagement

TL;DR

This study explores how glycosylation affects the Nipah virus's ability to bind to host cells, revealing key residues important for stability and receptor engagement.

Contribution

The first site-specific glycan characterization of the Nipah virus G glycoprotein and identification of critical residues for receptor binding.

Findings

The N481 N-glycan site is not evolutionarily conserved across Nipah species.

The Ser/Thr residue in the N481 sequon is crucial for maintaining G glycoprotein stability and ephrin B2 binding.

Hydrogen bonding networks contribute to G stability and host engagement.

Abstract

Nipah virus is a deadly paramyxovirus with 40 to 75% mortality and >750 cases since 1998. Currently there are no clinically approved vaccines or therapeutics to treat infection. Nipah is an enveloped virus with two surface glycoproteins, the trimeric fusion glycoprotein (F), and the tetrameric attachment glycoprotein (G), which is responsible for cellular attachment via binding to the host ephrin B2/B3 receptor. Glycosylation can substantially affect immunogenicity, receptor binding, and structural conformations for virus glycoproteins, but its effects on Nipah G receptor engagement have not been studied. Here, phylogenetic and mass spectrometry analysis of the Nipah G Malaysia strain reveal how N-glycosylation has evolved since the appearance of the virus in 1998. We discovered that the N481 sequon is not conserved and the threonine/serine in the glycosylation site is critical for maintaining long-range stability of G subunits that facilitates ephrin B2 binding affinity. Together, these data reveal plasticity of N-glycosylation sites across Nipah species and the presence of hydrogen bonding networks that contribute to G stability and host engagement—results that are valuable for understanding virus attachment/entry mechanisms and the rational design of structure-based vaccines.

•First site-specific glycan characterization of the Nipah virus glycoprotein (G) tetramer.•Phylogenetic analysis reveals the N481 N-glycan site is not evolutionary conserved.•The absence of the N481 glycan does not contribute to ephrin B2 receptor binding.•However, the Ser/Thr in the sequon is conserved and maintains Nipah G stability.

First site-specific glycan characterization of the Nipah virus glycoprotein (G) tetramer.

Phylogenetic analysis reveals the N481 N-glycan site is not evolutionary conserved.

The absence of the N481 glycan does not contribute to ephrin B2 receptor binding.

However, the Ser/Thr in the sequon is conserved and maintains Nipah G stability.

Nipah virus is a deadly pathogen and the roles of glycosylation in its infection are poorly understood. Here, we investigated the evolution of the Nipah virus attachment glycoprotein (G) and characterized its site-specific glycosylation. We determined key allosteric residues in a nonconserved N-glycan site that are critical for receptor binding. Our data provide structural and dynamic evidence for Nipah glycoprotein stabilization within the N481 sequon, offering insights for therapeutic development.

Figures

Full text

Nipah virus is a zoonotic paramyxovirus first reported in 1998 following outbreaks of severe encephalitis in Malaysia and Singapore (1, 2, 3), with near yearly outbreaks currently occurring in Bangladesh (4, 5, 6). Nipah disease progression begins with fever, myalgia, nausea, and coughing before rapid onset of encephalitis, seizures, coma, and death (1, 6). The difficulty of Nipah diagnosis in the early stages of infection, coupled with a lack of approved therapeutics or vaccines, contributes to its high mortality rate (40–75%) (5). Consequently, Nipah has been placed on the World Health Organization (WHO) R&D blueprint for emerging pathogens and epidemics (7).

Nipah virus entry and infection are not well understood, and the lack of effective treatment regimens reflects this. Nipah virus harbors two glycosylated surface proteins, the tetrameric attachment glycoprotein (8) (G) that binds to the host receptor ephrin B2 or B3 via its globular head domains (9, 10, 11) and the trimeric fusion glycoprotein (12) (F) that mediates viral fusion (13, 14) (Fig. 1A). Although G does not display neuraminidase activity, its head domains adopt a classical neuraminidase structure, with a six-bladed beta-propeller fold (3, 15, 16). F is a class I fusion protein, activated by cathepsin L and B cleavage in the endosome (17). G binds ephrin B2 and B3 with nanomolar affinity for each head domain (10), and picomolar affinity for the tetramer (18) due to avidity effects. Recent data suggest the extension of a G protein head outward toward the F protein could induce the F conformational changes that lead to fusion pore formation (19). Despite several studies of host cell attachment and entry, there remains a general lack of understanding of structure-function relationships involving G and F glycoproteins, which can inform effective treatment regimens.Fig. 1**Nipah virus evolved a divergent N-glycosylation site.**A, schematic representation of Nipah virus binding to host cells through ephrinB2/B3 interactions with the G protein, leading to fusion through F protein rearrangement. B, overview of Nipah G_M_ domain structure and predicted N-glycans sites. C, distribution of clades across location of isolates (left). Most prevalent clade shown for each country, identified through phylogenetic analysis (Supplemental Fig. S1). clade Ia: yellow, clade Ib: green, clade IIa: purple, clade IIb blue. Sequence alignment of clades and Hendra outgroup for residues 479 to 489 (right). Red amino acids denote a loss of N-glycosylation.

Currently, there are four candidate Nipah vaccines in phase I trials, including a live attenuated (PHV02) (20), mRNA (mRNA-1215) (21), adenoviral vector (ChAdOx-1 NiV) (22) and recombinant protein (HeV-G-sV) (23) vaccines (24). Although these candidates use different routes to elicit an immune response, vaccine-mediated immunity is acquired through exposure to Nipah G and F glycoproteins. Furthermore, the only biotherapeutic currently in clinical trials is the monoclonal antibody m102.4, which targets the Nipah G protein (25). Defining Nipah G and F glycosylation, as well as conserved areas of vulnerability, will aid therapeutic and vaccine development.

The Malaysia strain of Nipah G (G_M_) has 28 predicted N-glycan sites per tetramer, and F (F_M_) has 15 per trimer. N-glycan sites are genetically encoded through N-X-S/T sequons that are subject to both positive and negative selection pressures (26, 27, 28). Despite their prevalence, there is limited research into the roles of glycosylation for Nipah virus protein function (16, 29). Glycosylation of G has been suggested to be important for viral fusion (29, 30, 31), but a site-specific characterization of glycan structure and the associated effects on receptor engagement has not been performed. Here, we examine the effects of Nipah G N-glycosylation on receptor binding using mass spectrometry (glycomics, glycoproteomics and hydrogen deuterium exchange (HDX)), mass photometry (MP) and biolayer interferometry (BLI) with phylogenetic analysis to uncover how a single glycan site governs structural dynamics and stability of a receptor binding complex.

Experimental Procedures

Phylogenetic Analysis of Nipah G Isolates

Eighty-one Nipah G isolates and the Hendra G RefSeq (AAC83193, NP_047112) were exported as fasta sequences from the NCBI Virus database (https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/, accessed 31/03/2025) (Supplemental Tables S1-S3, Supplemental Data File 1). Evolutionary analyses were conducted in MEGA11 (32) and the evolutionary history was inferred by using the Maximum Likelihood method and Jones-Taylor-Thornton (JTT) matrix-based model (33). Initial tree(s) for the heuristic search were obtained automatically by applying neighbour joining and BioNJ algorithms to a matrix of pairwise distances estimated using the JTT model, and then selecting the topology with superior log likelihood value. There were a total of 604 positions in the final dataset (32).

DNA Production and Purification

NiVG 71 to 602, NiVG 183 to 602, ephrin B2 28 to 165 and E. coli biotin ligase (BirA) constructs in pOPIN vectors for mammalian expression were kindly gifted by Raymond Owens and Joseph Thrush (Supplemental Table S4). DNA was transformed into DH5⍺ NEB competent cells (New England Bioscience), and purified at scale using the Qiagen Gigaprep kit, following manufacturers’ protocols. The coding sequence of each plasmid was verified by Sanger sequencing prior to use.

Site-Directed Mutagenesis and 3C Site Cleavage Insertion

Site-directed mutagenesis and insertion of a 3C cleavage site was carried out using PCR with Q5 polymerase via the manufacturer’s instructions (New England Biolabs). Primers were designed so codon mutations were in the forward primer only to generate linear DNA with nonoverlapping primers (Supplemental Table S5). Subsequent linearized plasmids were ligated using KLD enzyme mix following manufacturer’s instructions (New England Biolabs). The 3C cleavage site was inserted by separately linearizing the NiVG 71 to 602 vector backbone and coding sequence using overlapping primers where one contained the 3C sequence (Supplemental Table S6). The linearized insert and vector were purified by gel extraction (Qiagen) and ligated using the In-Fusion enzyme kit (Takara Biosciences) as per the manufacturer’s instructions. Constructs were transformed into chemically competent DH5⍺ cells. DNA was purified and verified as described above.

Protein Expression and Purification

Expi293F(Gibco) suspension cells cultured in Expi293 expression medium (Gibco) were seeded to 1 × 10^6^ cells per mL and placed at 37 °C, 5% CO_2_, with shaking, 24 h prior to transfection. For transfection, a solution amounting to 10% of the final transfection culture volume of OptiMem (Gibco), DNA (1 μg/ml final concentration) and PEI 40 kDa (5.4 μg/ml final concentration) (Polysciences) was added to the culture. Eighteen hours post transfection, valproic acid (5.9 mM final concentration), sodium propionate (6.8 mM final concentration), and glucose (50 mM final concentration) were added (biotin (840 μM final concentration) was also added to cultures containing BirA) before incubating for a further 6 days. Conditioned media was harvested (6000 g, 30 min, 4 °C) and the supernatant loaded onto a HisTrap Excel 1 ml (Cytiva). Proteins were washed in PBS pH 7.4, 20 mM imidazole before eluting with 1 M imidazole in PBS pH 7.4 with a gradient elution of 0 to 100% over 15 column volumes. Ephrin B2 purification was further supplemented with a high salt wash of PBS, pH 7.4, 20 mM imidazole, 300 mM NaCl prior to elution. Fractions containing proteins of interest were pooled and concentrated in 10 kDa (ephrin B2), 30 kDa (Nipah G head and mutants) or 50 kDa (Nipah G ectodomain) molecular weight cutoff (MWCO) polyethersulfone (PES) concentrators (Vivaspin) to 600 μl (4000 g, 4 °C). The Nipah G 3C construct was incubated with hexahistidine tagged HRV 3C protease for 18 h, 4 °C, at a ratio of 10:1 mg Nipah G: 3C. The concentrate was loaded onto a Superdex 200 increase 10/300 Gl (Cytiva) pre-equilibrated and run in PBS pH 7.4 at 0.75 ml/min. For the 3C cleaved Nipah G protein, a 1 ml Histrap Excel was connected in tandem with the size-exclusion chromatography (SEC) column. Desired peak fractions were analyzed via SDS-PAGE and concentrated as described above, and snap frozen in liquid nitrogen before storing at −80 °C. Herein, Nipah G protein which has been 3C cleaved is described as “3C cleaved Nipah G_M_ ectodomain”.

Mass Photometry

MP measurements were carried out on a OneMP (Refeyn Ltd). Sample chambers were assembled by placing CultureWell gaskets (3 mm × 1 mm, Grace Bio-Labs) on glass coverslips (High Precision No. 1.5H, Marienfeld Superior). Eighteen microliters of PBS pH 7.4 was placed in a gasket to find focus, before addition of 2 μl of sample. Data were collected using Acquire MP (2023 R1.1, Refeyn Ltd). Nipah G 71 to 602 was measured at a final concentration of 5 nM with a medium field of view, exposure time = 1.91 ms, frame rate = 509.5 Hz, frame binning = 3, pixel binning = 4. Nipah 183 to 602 was measured at a final concentration of 10 nM with a small field of view, exposure time = 0.95 ms, frame rate = 999.0 Hz, frame binning = 10, pixel binning = 4. Movies were recorded for 60s. Mass calibrations were performed for each field of view using an in-house protein standard, with peaks at 90, 180, 360, and 540 kDa. Data were analyzed in Discover MP (v2023 R1.2, Refeyn Ltd), including gaussian fits and histogram generation. Data were exported and plotted in GraphPad Prism v10.2.3.

HPLC of N-Glycans

Ten micrograms Nipah G_M_ ectodomain was run on a 4 to 12% Bis-Tris SDS PAGE gel (Invitrogen) in biological quintuplicate. Gel bands were extracted and destained in 50 mM ammonium bicarbonate (AmBic) 50% acetonitrile (MeCN), incubated at 37 °C for 30 min and washed in 50 mM AmBic 50% MeCN. Gel bands were dehydrated in 100% MeCN before covering with 700 μl containing peptide:endoglycosidase F (PNGase F) at 12.5 ng/μl (in house). Digests were incubated at 4 °C for 1 h, before moving to 37 °C for a further 18 h.

Following digest, the PNGase F reaction mixture was pooled with a subsequent three washes of gel bands with 100 μl of HPLC grade water (10 min, 37 °C). N-glycans were dried and resuspended in 30 μl of HPLC grade water. In addition, 80 μl of freshly prepared 2-anthranilic acid (2-AA) reaction mix (30 mg/ml 2-anthracillic acid, 45 mg/ml NaBH_3_CN, 4% w/v CH_3_COONa∙3H_2_O, 2% w/v B(OH)3; in MeOH) was added to glycans and incubated at 80 °C for 1 h. Samples were cooled before clean-up with Spe-ed amide 2 cartridges. Samples were loaded with 1 ml 97% ACN, washed twice with 1 ml 95% ACN, then eluted twice with 750 μl HPLC grade water and lyophilized for 18 h.

Labeled glycans were resuspended in 20 μl HPLC grade water, and half of the glycans were further treated with 500 NEB units of endoglycosidase H (Endo-H) in 1X GlycoBuffer 3 (New England Bioscience), before incubating at 37 °C for 22 h. The Endo-H reaction mix was cleaned up with 10 kDa MWCO cartridges (Omega), before drying in a vacuum concentrator and resuspending in HPLC grade water.

MeCN was added to labeled glycans to a final concentration of 70% MeCN. Five microliters of glycans were separated using an ACQUITY UPLC BEH Amide column (1.7 μM, 2.1 x 150 mm) coupled to a 1290 Infinity II UHPLC (Agilent) (Supplemental Table S7). Dextran, Man_5_-GlcNAc_2_, and Man_9_-GlcNAc_2_ glycan standards (Ludger) were run to help identify N-glycan peaks. Glycan peak identifications and quantification can be found in Supplemental File S2.

IM-MS/MS of N-Glycans

Subsequently, 40 μg of Nipah G was run on a 4 to 12% Bis-Tris SDS PAGE gel (Invitrogen) in biological triplicate. Gel bands were extracted and destained in 50 mM AmBic 50% MeCN, incubated at 37 °C for 30 min and washed in 50 mM AmBic 50% MeCN. Gel bands were dehydrated in 100% MeCN before covering with 700 μl 50 mM AmBic containing PNGase F at 12.5 ng/μl (made in-house). Reaction was incubated at 4 °C for 1 h, before moving to 37 °C for a further 18 h. Released N-glycans were desalted using C18 zip tips (Merck), manually overlaid with porous graphite carbon (extracted from a Hypercarb porous graphite carbon column), then dried down and resuspended in 50:50 MeOH:H_2_O.

Approximately 2 μl of N-glycan sample material was injected by static nanoelectrospray into a SYNAPT XS (Waters) using gold-coated borosilicate glass capillaries (prepared in-house). Data acquisition settings are presented in Supplemental Table S8. Data acquisition and processing were carried out using Waters DriftScope (version 2.8) software and MassLynx (version 4.1). Ion mobility mass spectrometry (IM-MS) peak lists can be found in Supplemental File S3, while mass spectrometry (MS) assignments and glycan compositions are listed in Supplemental Tables S9–S11.

N-Linked LFQ Glycoproteomics

Three micrograms of Nipah G_M_ ectodomain was run on a 4 to 12% Bis-Tris SDS PAGE gel (Invitrogen) in biological sextuplicate. Three micrograms of Nipah G_M_ head domain, N481D and T483A mutants were also run on an SDS-PAGE gel in biological triplicate. Gel bands of interest were extracted and destained in 50 mM AmBic 50% MeCN and incubated at 37 °C for 30 min, before washing in 50 mM AmBic 50% MeCN. Gel bands were dehydrated in 100% MeCN before adding 10 mM dithiothreitol in 100 mM AmBic and incubating for 30 min at 56 °C. Gel bands were then dehydrated in 80% MeCN for 10 min at room temperature (RT). Dithiothreitol solution was replaced with 50 mM iodoacetamide in 100 mM AmBic and the immersed gel bands were incubated in the dark for 30 min. Iodoacetamide solution was removed, and gel bands were washed sequentially with 100% MeCN and 50 mM AmBic. Gel bands were dehydrated 100% MeCN before addition of 12.5 ng/μl of the relevant protease (trypsin (Promega), chymotrypsin (Promega), or alpha-lytic (New England Bioscience)) in 50 mM AmBic. Gel bands were incubated with the protease solution at 37 °C for 1 h before replacing the solution with fresh 50 mM AmBic and incubating the bands at 37 °C for a further 18 h. The resulting peptide solution was pooled with sequential washes of 5% (v/v) formic acid (FA) in water and MeCN. Samples were dried in a vacuum concentrator and reconstituted in 2% MeCN, 0.05% TFA for analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS).

Samples were analyzed using an Ultimate 3000 RSLCnano system coupled to an Orbitrap Q Exactive mass spectrometer (Thermo Fisher Scientific). Peptides were loaded onto a 75 μm × 2 cm precolumn and separated on a 75 μm × 15 cm Pepmap C18 analytical column (Thermo Fisher Scientific). Buffer A was 0.1% FA in water and buffer B was 0.1% FA in 80% MeCN with 20% H_2_O. A 100-min linear gradient (Supplemental Table S12) and universal high energy collision induced dissociation (HCD) identification method (Supplemental Table S13) was used.

O-Linked Glycoproteomics

Ten micrograms of 3C cleaved Nipah G_M_ ectodomain was run on a 4 to 12% Bis-Tris SDS PAGE gel (Invitrogen) in biological triplicate. Samples were then prepared for LC-MS/MS analysis in the same way as the N-linked glycoproteomics samples (described above).

Samples were analyzed using an Ultimate 3000 RSLCnano system coupled to an Orbitrap Ascend mass spectrometer (Thermo Fisher Scientific). Peptides were loaded onto a C18 PepMap100 trap column (300 μm × 5 mm, 100 Å, Thermo Fisher Scientific) and separated on an in-house packed Reprosil Gold C18 analytical column (50 μm × 50 cm, 1.9 μm particle size, Dr Maisch). Buffer A was 0.1% FA, 5% dimethyl sulfoxide (DMSO) in water, buffer B was 0.1% FA, 5% DMSO in MeCN and the loading solvent was 0.1% FA in water. A 90-min linear gradient (Supplemental Table S14) and oxonium ion triggered electron-transfer high energy collision induced dissociation (EtHCD) method was used (Supplemental Tables S15 and S16).

Nipah G_M_ Head, N481D and T483A samples were analyzed using an Ultimate 3000 RSLCnano system coupled to an Orbitrap Eclipse mass spectrometer (Thermo Fisher Scientific). Peptides were loaded onto a 75 μm × 2 cm precolumn and separated on a 75 μm × 15 cm Pepmap C18 analytical column (Thermo Fisher Scientific). Buffer A was 0.1% FA in water and buffer B was 0.1% FA in 80% MeCN with 20% H_2_O. A 60-min linear gradient (Supplemental Table S17) and oxonium ion triggered EtHCD (Supplemental Tables S18 and S19) method was used.

Glycoproteomics Data Analysis

Peak lists were generated using Xcalibur (Thermo Fisher Scientific, version 4.0.27.19) and data were analyzed in Byos (Protein Metrics, version 5.4.52). The database queried consisted of the FASTA sequence for the Nipah G protein sequence of interest, alongside Byonics 71 common contaminants and decoys. Digestion was set to RK, FYWML, and TASV for trypsin, chymotrypsin, and alpha-lytic protease digests, respectively, with maximum of two missed cleavages. Fixed modifications were carbamidomethylation (57.02 Da) and variable modifications were methionine oxidation (15.99 Da) and deamidation (N or Q) (0.98 Da). Byos common human N-linked (132 glycans, which covered the structures identified in our IM-MS/MS described below) and O-linked (6 glycans for N-linked analysis, 9 glycans for O-linked analysis) glycan libraries were used. False discovery rate (FDR) was set to 0.01 and mass tolerances for precursor and fragment ions set to 6.00 ppm and 20.00 ppm, respectively. For analysis, a minimum Byos threshold score of 100 was used for O-glycopeptide filtering and 300 for N-glycopeptide filtering, with the exception of peptides covering site N72 for which a score of 100 was used due to limited detection and fragmentation data arising from the hexahistidine tag, leading to lower scoring. Glycopeptides were manually validated, with O-glycan site localization confirmed through c/z and intact b/y fragmentation products, as well as oxonium ion presence. For N-glycan quantification, one peptide sequence per glycosite was chosen as representative for each site (Supplemental Table S20) and the extracted ion chromatogram intensities for each glycopeptide were summed and plotted relative to the total intensity for each glycosite (Supplemental File S4). O-glycopeptide site identifications are listed in Supplemental File S5.

Differential Endoglycosidase Digestion with 18O Labeling

The method by Cao et al (34) was adapted with the following modifications; 5 μg of Nipah G_M_ ectodomain was denatured in 4 M urea, 1% SDS for 1 h at RT. Protein was reduced in 10 mM tris(2-carboxyethyl)phosphine (TCEP) (pH 7.5) for 30 min RT and labeled with 50 mM iodoacetamide in 100 mM AmBic for 30 min in the dark, RT. Protein solution was added to 250 μg of 50:50 mix of Sera-Mag speed bead magnetic carboxylate modified particles (Cytiva). MeCN was added to a final concentration of ∼90% and the solution was incubated for 30 min, 37 °C, with shaking. A magnetic plate was used to immobilize the beads, allowing aspiration of the supernatant. Beads were washed sequentially with 80% EtOH (x2) and 100% MeCN. Subsequently, 0.2 μg of trypsin/LysC (Promega) in 50 mM AmBic was added to the beads, before incubation overnight at 37 °C. The subsequent peptide solution was aspirated and kept, before 5% DMSO H_2_O was added for 2 min, and pooled with the peptides. Half of the samples were further treated with 0.2 μg of GluC (NEB) in 50 mM AmBic for 5.5 h at 37 °C, after which all peptides were dried in a vacuum concentrator. Experiment was carried out in biological triplicate for both trypsin/LysC only and trypsin/LysC + GluC conditions.

Peptides were resuspended in 50 mM ammonium acetate (pH 5.5) (37 °C with 10 min shaking) before addition of 313 NEB units of Endo-H (New England Bioscience) per sample. Samples were incubated for 1 h at 37 °C before drying and sealing. PNGase F (New England Bioscience) was lyophilized and resuspended in 50 mM AmBic in H_2_^18^O before adding to peptides, with a 160 NEB units per sample. Samples were incubated for 30 min 37 °C before heat inactivation (10 min, 100 °C) and drying in a vacuum concentrator. Labeled peptides were reconstituted in 2% MeCN 0.05% TFA and were run on an LC-MS/MS using the same method as for glycoproteomics.

Peak lists were generated by XCalibur (Thermo Fisher Scientific, version 4.0.27.19) and data were analyzed in Proteome Discoverer (Thermo Fisher Scientific, version 3.0). Digestion was set to RK, and RKED for trypsin or trypsin/LysC + GluC protease digests respectively with maximum two missed cleavages. Static modifications were carbamidomethylation (57.02 Da) and dynamic modifications were methionine oxidation (15.99 Da) and ^18^O deamidation (N only) (2.98 Da). Strict FDR was set to 0.01, and mass tolerances for precursor and fragment ions were set to 10 ppm and 0.6 Da, respectively. The library searched consisted of the Nipah G_M_ sequence lacking a signal sequence, supplemented with the CRAPome (35). The same peptide sequences were used for quantification, with the exception of N529 (Supplemental Table S20). This was due to low intensities of the matching peptide (<3000) and subsequent missing values for multiple replicates (Supplemental File S6). A shorter, more abundant, peptide spanning N529 from the trypsin/LysC-GluC digestion was used instead. For quantification, the extracted ion chromatogram intensities for each peptide of that sequence were summed and plotted relative to the total intensity for each glycosite, listed in Supplemental File S6.

Biolayer Interferometry

Octet streptavidin biosensors (Sartorius) were soaked in 200 μl PBS for 30 min prior to each run. Nipah G mutants and ephrin B2 were diluted in PBS with 0.01% Tween 20. Blocking buffer was PBS with 0.01% Tween 20. Runs were carried out in technical quintuplicate using an Octet R8 system (Supplemental Table S21 for run settings). Concentrations of mutants were titrated in a 2-fold dilution series starting at 200, 100, or 44 nM. Timings and concentrations were determined prior such that the association phase would reach steady state, and the dilution series covered both above and below the determined K_D_. Analysis was carried out in Octet Analysis Software, where baselines were subtracted prior to fitting data to a 1:1 binding model. The R_max_ of each curve was fit to a steady state model to give the K_D_ values. Students unpaired, two-tailed t test was used to calculate statistical significance for K_D_ values.

HDX-MS

Nipah G WT head, T483A and N481D were diluted to 5.9 μM, and deuterated PBS (pD 7.4) was added to a deuterium fraction of 88% to initiate the exchange reaction. At selected time intervals (15 s, 150 s, 1500 s, 15,000 s, and 24 h), a fraction of reaction mixture containing the equivalent of 23 pmol of protein sample was withdrawn and quenched with ice-cold buffer containing 45 mM phosphate, 2 M urea and 40 mM TCEP, which dropped the estimated pH value calculated for a deuterated sample (pH/D) to 2.3 and the deuterium fraction to 50%. Quenched samples were kept for 30 s on ice then flash frozen in liquid nitrogen and stored at −80 °C before LC-MS analysis. For WT-ephrin B2 complexing, Nipah G WT head was incubated in a 1:1.2 M ratio with ephrin B2 (28–165), or an equivalent volume of PBS. Deuterium labeling was carried out as specified above, and the equivalent of 46 pmol of protein samples was quenched. Maximally labeled samples (MaxD) were produced by labeling Nipah G WT head, T483A and N481D at 37 °C in the presence of 4 M deuterated urea in D_2_O and 2.4 mM TCEP, keeping the deuterium fraction at 88%. The maximally labeled samples were quenched after 18 h with ice-cold buffer containing 45 mM phosphate, 1.4% FA, in a similar manner as for the other labeled samples and stored in the same fashion. Time points were repeated and recorded in technical triplicate, with the exception of the following time points: WT (mutant analysis) 15 s, WT (ephrin B2 analysis) 1500 s, N481D 150 s, T483A 150 s, which were recorded in duplicate

For LC-MS, samples were quickly thawed and injected into an Acquity UPLC M-Class System with HDX Technology (Waters), on-line digested at 20 °C into a hand packed pepsin column, prior to online deglycosylation with a PNGAse-Rc column (36) (Affipro), which removed N-glycans through enzymatic deamidation of N-glycosylated asparagines into aspartates. Peptides were trapped/desalted with solvent A (0.23% FA in water) for 4 min at 120 μl/min and at 1 °C through an Acquity BEH C18 VanGuard precolumn (1.7 μm, 2.1 mm × 5 mm, Waters). Peptides were eluted from the trap column into an Acquity UPLC BEH C18 analytical column (1.7 μm, 1 mm × 100 mm, Waters) with a 7 min-linear gradient raising from 8 to 35% of solvent B (0.23% FA in MeCN) at a flow rate of 40 μl/min and at 1 °C (Supplemental Table S22 for LC instrument settings). Eluted peptides went through electrospray ionization in positive mode and underwent MS analysis onto a Synapt-G2 Si TWIMS-TOF (Waters) with ion mobility separation enabled (Supplemental Table S23 for IM-MS instrument settings). For peptide identification, nondeuterated protein samples were run with the same method as for deuterated samples, but underwent fragmentation using an MS^E^ method, applying collision energy ramping from 20 to 30 kV (Supplemental Table S24 for MS^E^ instrument settings)

Peak lists were generated by MassLynx (version 4.1, Waters) and MS^E^ runs were analyzed with ProteinLynx Global Server (PLGS) (version 3.0, Waters) and peptides identified in three out of four replicates, with at least 0.2 fragments per amino acid (at least two fragments in total), with mass error below 10 ppm were selected in DynamX (version 3.0, Waters). Using both retention time and drift time, spectra of deuterated peptides were matched to the peptide map and manually annotated to calculate peptide deuterium incorporation. HDX-MS uptake plots and individual uptake values can be found in Supplemental Files S7 and S8, respectively.

Statistical Rationale

For the glycoproteomics experiments, FDR was set to >0.01. Students unpaired, two-tailed t test was used to test for statistical significance between complex glycan containing glycopeptides and Endo-H resistant HPLC glycan species. For testing statistical significance between complex glycan containing glycopeptides and ^18^O labeled peptides at each glycosite, a one-way ANOVA with post hoc Šidák multiple pairwise comparison correction was applied to control for type-I error inflation when conducting multiple pairwise comparisons between paired means. This test was also used for the comparison between ectodomain glycopeptides and head domain glycopeptides. Identical glycopeptides and ^18^O labeled peptides were compared, except for N529, where the more abundant trypsin/LysC-GluC peptide was used for ^18^O labeled peptides (see “Differential endoglycosidase treatment with ^18^O labeling” methods section for more detail). A one-way ANOVA with post hoc Dunnett’s multiple comparison correction was used to test for statistical significance between the K_d_ of Nipah G_M_ Head and glycosite mutants, and between Nipah G_M_ head ± kifunensine. This test and correction were chosen to control for type-I error inflation when conducting pairwise comparisons between a control and multiple experimental means. Statistics and graphing for these analyses were conducted in GraphPad Prism v10.2.3. Statistical reporting for HPLC, proteomics and BLI can be found in Supplemental File S9.

The threshold for a statistically significant difference in HDX between two states was established based on an approach described earlier (37). Briefly, a confidence interval (CI) was calculated based on the average standard deviation (SD) of peptide deuterium uptake values for time points performed in triplicates, according to Equation 1:

$$[eqn]$$

where N is the number of peptides considered, multiplied by the number of time points performed in triplicate. The CI at the significance level of 99% between two states was calculated based on a zero-centered average difference in deuterium content, considering a two-tailed distribution with two degrees of freedom (n = 3) (Equation 2)

$$[eqn]$$

The deuterium uptake of peptides carrying T483A was compared to the correspondent WT peptides by normalization with their respective MaxD uptake values (Equation 3) (38):

$$[eqn]$$

HDX-MS reporting forms, including statistical outcomes, are reported in Supplemental Files S10 and S11.

Results

Phylogenetic Analysis of Nipah G N-Glycosylation

Most research on the G protein has used the primary sequence isolated from the 1998 outbreak (3), the so-called “Malaysia” strain. However, evolutionary analyses have shown Nipah has diverged to a second strain, the Bangladesh strain (39, 40). To identify potential differences in N-glycosylation sites between strains, we aligned 81 complete sequences from the National Center for Biotechnology Information (NCBI) virus database using MEGA11 (Supplemental Figs. S1 and S2), including viruses isolated from four different host species, across five countries. In agreement with other studies, we identified distinct groupings corresponding to the Malaysia and Bangladesh strains (clade I and II, respectively, in our analyses). These clades were further classified into a and b subclades, with IIa consisting of the perennial Bangladesh outbreaks, and IIb the more recent outbreaks in India. The location of sample collection shifted from Malaysia to Bangladesh and India over time (Supplemental Tables S1–S3), where most infections are now located. Notably, we found six of the seven predicted N-glycan sites were conserved between all clades (Fig. 1b); however, a specific mutation (N481D) resulted in the loss of an N-glycan site. This was either introduced during the split from Hendra virus or occurred independently in both clade II and clade Ib isolates (Fig. 1C). This mutation event was also independent of host and country of origin, as there were no differences in host and origin within clades themselves (Supplemental Figs. S1 and S2).

Site-specific Glycosylation of Nipah G

The role of the Nipah G N481 glycan is unknown but it could contribute to receptor binding affinity, infectivity and/or immune detection/evasion. Critically, there is no comprehensive site-specific glycan analysis of Nipah G, which is a first step for exploring glycan structure-function relationships, including N481. We used the soluble ectodomain of Nipah G_M_ (Fig. 2) for site-specific glycan characterization, which was recombinantly expressed and purified from human embryonic kidney (HEK) Expi-293 cells to best mimic human glycosylation in vitro (41). Purified size-exclusion chromatography fractions of Nipah G_M_ ectodomain and head domain (Supplemental Fig. S3, A and B) were measured by MP, showing that soluble G_M_ formed the expected stable tetramer (Supplemental Fig. S3, D and E) (8). Using BLI, we determined soluble G_M_ tetrameric ectodomain bound with picomolar affinity to recombinantly expressed soluble ephrin B2 binding domain (Supplemental Figs. S3C and S4). The G_M_ monomeric head domain interacted with ephrin B2 with nanomolar affinity in a 1:1 binding model (Supplemental Fig. S4B).Fig. 2**Nipah G_M_ ectodomain N-glycosylation.**A, total ion spectrum of released N-glycans by IM-MS scaled to the most abundant ion, Man_5_-GlcNAc_2_ (m/z 1347). B, HPLC of 2-AA labeled N-glycans with peaks labeled as endo-H sensitive (i.e. oligomannose and hybrid) in green and endo-H resistant (i.e. complex) in purple. C, mean abundance of glycan subtype (colored as above). D, glycan subtype specific for each site determined by label-free quantitative (LFQ) glycoproteomics, (n = 6 and n = 5 for N72). Glycans types are labeled green for oligomannose, yellow for hybrid and purple for complex. E, occupancy of N-glycan sites measured by ^18^O labeled proteomics (n = 3, except N592 where n = 2). F, location of glycans on Nipah G tetrameric ectodomain (PDB: 7TXZ and 7TY0) modeled using GLYCAM (the most abundant glycan species per site is shown). IM-MS, ion mobility mass spectrometry; PDB, Protein Data Bank; 2-AA, 2-anthranilic acid; Endo-H, endoglycosidase H.

To identify all N-glycan structures present on G_M_, we performed IM-MS/MS on three biological replicates. A previous study used MALDI-MS to investigate glycosylation, and identified a range of hybrid and complex type N-glycans, in addition to Man_5_GlcNAc_2_ for the monomeric G_M_ head domain (16). However, our data showed a shift toward oligomannose structures with the tetrameric G_M_ ectodomain, with Man_5_-8_GlcNAc_2 (Man_5_-8) present as major ion species (Fig. 2A). Three biantennary complex type glycans were also detected as major ion species (m/z 1183, 1576 and 2077), as well as two bisected complex type glycans (m/z 1779 and 1941) (Supplemental Figs. S5 and S6). Additional complex-type N-glycans, specifically tri- and tetra-antennary species, were identified at low levels from ion-mobility extracted doubly and triply charged species (Supplemental Fig. S5). Oligomeric assembly has been implicated in influencing glycosylation by affecting protein conformation and glycosite accessibility for glycan trimming by mannosidases, and extension by glycosyltransferases (42, 43, 44). Therefore, the abundance of oligomannose glycans is likely due to differences in glycan processing for the G_M_ tetrameric ectodomain versus monomeric head domain, as previously seen for HIV Env (envelope protein) trimers and gp120 monomers (44).

HPLC of 2-AA labeled N-glycans confirmed the relative abundance of G_M_ structures by type (Fig. 2b). Treatment of N-glycans with Endo-H identifies oligomannose and hybrid-type glycans (i.e. Endo-H sensitive) from complex-type structures (i.e. Endo-H resistant). Overall, IM-MS/MS and HPLC data both indicated Endo-H sensitive glycans accounted for just under 50% of the total N-glycan pool (Fig. 2C and Supplemental Fig. S7), with ∼70% present as Man_5_ (32.6 ± 6.1%) Man_7_ (23.8 ± 4.2%) and Man_8_ (14.7 ± 4.1%) (Supplemental File S2). Conversely, multiple peaks were identified as Endo-H resistant glycans across the chromatogram, indicating complex-type N-glycans are structurally diverse, but individually present at lower levels compared to oligomannose structures.

We mapped the site-specific attachment and abundances of N-glycans on G_M_ from six biological replicates using label-free quantification (LFQ) glycoproteomics, (Fig. 2D). Four of the seven N-glycan sites were predominantly complex-type: N72 (72.3 ± 1.9%) and head sites N378 (84.9 ± 4.1%), N417 (80.7 ± 4.1%), and N529 (83.1 ± 6.3%). N378 and N529 glycopeptides were also predominantly fucosylated (82.3% and 82.9%, respectively) and, along with N72, contained extensively sialylated N-glycans (36.6%, 43.0%, and 35.1%, respectively) (Supplemental Fig. S8), consistent with the complex N-glycans identified by IM-MS/MS (Supplemental Fig. S5). N417 was minimally fucosylated (4.7 ± 1.6%) and had notably lower levels of sialylation (19.3 ± 5.6%). N306 had a mixture of complex (56.4 ± 7.6%), oligomannose (36.7 ± 9.6%) and hybrid-type (6.9 ± 2.6%) glycans. The N159 glycan, present in the neck of G_M_, contained primarily oligomannose glycans (61.4 ± 7.4%), while the evolutionary divergent site, N481, was oligomannose rich (74.2 ± 6.6%) and was the only site with negligible amounts of complex N-glycans (<1%). Overall, results between the three methods were in high agreement, with no significant difference between HPLC and glycoproteomics quantifications (Supplemental Fig. S9). We also compared the relative abundance of glycopeptides between the Nipah G_M_ tetrameric ectodomain and the head domain. Sites N306, N378, N417, and N529 in the head domain displayed a significantly greater proportion of complex-type glycans than in the tetrameric ectodomain, ranging from an 11.4% increase in complex type glycans for N378 (ectodomain 80.1% ± 5.6%, head domain 91.5 ± 1.2%) to a 26.1% increase for N306 (ectodomain 56.4 ± 7.6%, head domain 82.5 ± 1.8%). These data, along with our IM-MS/MS data, suggest that the glycan differences between the monomeric head domain and tetrameric ectodomain are a result of G tetramer oligomerization (Supplemental Fig. S10).

To further confirm G_M_ N-glycosylation site occupancy as determined by LFQ glycoproteomics, we employed differential glycopeptide digestion, with sequential treatment of Endo-H, followed by PNGase F digestion in the presence of H_2_^18^O, from three biological replicates (Supplemental Fig. S11) (34). As PNGase F requires a minimum of a chitobiose core, this method enables identification and quantification of glycopeptides through different mass shifts: +203 (+GlcNAc) for glycopeptides containing Endo-H sensitive glycans whose core GlcNAc are not removed by PNGase F, +3 for glycopeptides with Endo-H resistant glycans that undergo enzymatic deamidation by PNGase F in the presence of ^18^O (Asn→Asp(^18^O)), and +0 for unoccupied N-glycan sites. We showed that our PNGase F had negligible activity on the remaining GlcNAc following Endo-H digestion, as kifunensine-treated G protein had >99% Endo-H sensitive peptides, which were not modified by further PNGase F treatment (μ = 99.3%, σ = 0.81). Due to the usage of different proteolytic digest methods (in-gel digestion versus SP3), we were unable to cover site N306 (see further description in the methods section). Five of the six sites were ∼100% occupied, including N481 (μ = 100.0%, σ = 0.03), and only N529 was not fully occupied with 46.9% occupancy observed (σ = 1.8, n = 2) (Fig. 2, E and F). To validate our glycoproteomics data, we compared the relative abundances observed for complex glycopeptides and Endo-H resistant glycopeptides using the LFQ and ^18^O datasets, respectively. We compared peptides with identical sequences, except for N529 (see Experimental procedures) (Supplemental Table S20) and samples from the same biological replicate. There were no significant differences between these two methods (Supplemental Fig. S11B), suggesting ionization efficiency between glycopeptides did not affect quantification among N-glycan types.

Previous research into Hendra virus showed the stalk domain its G protein is highly populated with O-glycans that are important for viral fusion, and some of these sites are conserved with Nipah (30, 31). However, Nipah site-specific O-glycosylation has not been investigated. Using three biological replicates, we identified and localized eight O-glycan sites on Nipah G_M_ and two glycosite regions (Supplemental File S5 and Supplemental Figs. S12 and S13). O-glycans and their sites were assigned using oxonium ion triggered EtHCD, with sufficiency of c/z fragmentation affecting whether a site could be localized to a residue or a region. Seven sites were conserved between Nipah and Hendra (T103, T117, T119, S132, S134, T135, and S137) including the two ST-rich regions S110-T119 and S132-S137 (Supplemental Fig. S12A). The first ST-rich region, S110-T119, located in the lower stalk region between the bottom two heads of Nipah (Supplemental Fig. S12B), contained both short and extended monosialylated and disialylated core-1 and core-2 type O-glycans. In particular, we could localize T117 displaying a single HexNAc, and T119 showing HexNAc, HexNAc_1_Hex_1_, or HexNAc_2_Hex_2_NeuAc_1_ O-glycans. On the other hand, S132-S137, the other ST-rich region sitting on the upper stalk region, presented less extended monosialylated and disialylated core-2 type O glycans than S110-T119. We saw reduced extension among S132 and S137, both displaying a single HexNAc and S132 a HexNAc_1_Hex_1_, while S134 and T135 showed evidence of more extended core-1 and core-2 structures of HexNAc_2_Hex_2_NeuAc_1_ and HexNAc_2_Hex_2_NeuAc_2_. Overall, the lesser global extent of sialylation and extension in S132-S137 suggests that it is less accessible for O-glycosyltransferase enzymes compared to S110-S119. This is expected as S110-T119 resides at the base of the tetrameric ectodomain, where the stalk helices are further apart from one another, while S132-S137 sit above the lower head regions on the stalk, where it is packed more tightly. The other conserved site, T103, which flanks S110-S119, was mainly populated with extended monosialylated/disialylated core-1 and core-2 O-glycans, as was observed for Hendra (31).

The other O-glycosylation site, which was not present on Hendra G, was T149. Only a single HexNAc was present on T149, which is located in the neck domain. The model of the Nipah G_M_ ectodomain, based on the cryo-EM structure (8), has the side chains of T149 facing outward (Supplemental Fig. S12B), and therefore we would potentially expect further extension of these glycans. The lack of extended, larger structures at this site suggests this O-glycan is not accessible to glycosylation enzymes in the same way as the other sites. Our findings reveal that N-glycans are present near the ephrin B2 binding site while O-glycans are localized to the stalk. Based on known structures of Nipah G, they could have direct or allosteric implications for receptor binding.

Receptor Binding of Nipah GM Glycosite Mutants

Glycans attached to viral glycoproteins can influence receptor binding via direct interactions (45) and by altering protein structural conformations (46, 47, 48). For Nipah G protein, the direct biophysical effects of glycan loss upon ephrin B2 binding have not been investigated, with previous studies focusing on phenotypic outcomes (viral fusion and cellular infectivity) (29). To examine binding, we generated individual N-glycosite mutants of G_M_; T308A, S380A, S417A, T483A, and T513A plus the evolutionary mutant N481D (Supplemental Fig. S14). Each N-glycan site was mutated via Ser/Thr to Ala in its respective sequon (N-X-S/T→N-X-A) to prevent potential O-glycosylation of Ser/Thr upon Asn mutation (49). The binding affinity between ephrin B2 and the head domain of each mutant was measured by BLI. The head domain was chosen over the tetramer in order to study the interaction in a 1:1 binding model (Fig. 3 and Supplemental Fig. S15). Furthermore, the picomolar affinity to the tetramer is at the concentration limit for BLI (50) (Supplemental Fig. S4). We did not detect any statistically significant differences in K_d_ values between G_M_ WT and T308A, S380A, S419A, and N481D mutants, indicating that these N-glycans do not directly contribute to ephrin B2 binding. However, a significant 4.7-fold decrease in binding affinity was observed with T531A (K_d_ = 95.5 ± 23.7 nM) compared to the WT (K_d_ = 20.7 ± 8.6 nM). The side chain of T531 was previously suggested to interact through hydrophobic interactions with ephrin B2^9^; therefore, a reduction in affinity for this mutant cannot be solely attributed to N-glycan (or O-glycan) loss. T483A also exhibited a significant 3.7-fold reduction in binding affinity (K_d_ = 76.8 ± 35.6 nM) for ephrin B2 compared to WT. Both the evolutionary mutant N481D and T483A result in N-glycan loss at N481, but T483A elicited 2.1-fold weaker binding than N481D (K_d_ = 36.7 ± 9.7 nM). Affinity for ephrin B2 did not change with kifunensine treated G_M_, indicating that the presence of solely GlcNAc_2_Man_8/9_ glycoforms does not affect binding (51). Furthermore, because the presence of an O-glycan at T483 was not observed for the WT and N481D, O glycosylation, or lack of, at this site did not affect binding (Supplemental File S5). Collectively these data suggest the reduction of ephrin B2 affinity observed for T483A was not due to loss of the associated N481 glycan but was directly related to the amino acid change from threonine to alanine (Fig. 3).Fig. 3Effects of Nipah G_M_ glycan knockout mutants on ephrin B2 binding. Location of glycans on Nipah G_M_ in relation to the ephrin B2 binding interface (top). N-glycan chitobiose or singular GlcNAcs are shown in blue. G_M_ and ephrin B2 are yellow and green, respectively (Image produced in PyMOL using PDB: 2VSM, 7TY0). Box and whisker plots of steady state K_d_ of G_M_ mutants binding to ephrin B2 (bottom). Boxes represent the 25th-75th percentile, with whiskers extending from the minimum to maximum data points. Annotated values are mean K_D_ values ± 1 standard deviation. PDB, Protein Data Bank.

Structural Dynamics of Nipah GM and Ephrin B2 Complex

Loss of binding affinity for T483A implies a structural rearrangement of the binding site involving residues near the N481-X-T483 sequon. We used HDX MS to investigate structural dynamics associated with the N481D and T483A mutations compared to WT G_M_ head domains with and without ephrin B2. We followed the HDX of 139 peptides from 15 s to 24 h, covering 87% of G_M_ head domain sequence and four out of five glycosylation sites, including N481 (Supplemental Fig. S16).

A significant decrease in HDX was measured across overlapping peptides in multiple regions for WT G_M_ head upon ephrin B2 binding (Fig. 4A and Supplemental Fig. S17) mostly in agreement with the cocrystal structure (10) and previously published HDX-MS data for G_M_ ectodomain-ephrin B2 complex (9). Peptide regions that displayed significant HDX changes were localized to the ephrin B2 binding site (Supplemental Fig. S17) and distributed across the oligomerization interface, with amino acids Y231-L234, Q455-S459, E505-Y508, and L526-T538 displaying greatest protection from HDX in the bound G_M_ form. Both Y231-L234 and Q455-S459 do not form direct contacts with ephrin B2 (Supplemental Fig. S18), thus decrease in HDX is indicative of allosteric effects of rigidification in this region.Fig. 4HDX-MS of Nipah G_M_ mutants ± **ephrin B2.**A, differential HDX plot of Nipah G_M_ head compared to N481D, T483A, and plus ephrin B2 (+EB2). Alignment of the Nipah G_M_ structure to the HDX plot (top), with the beta propellers (BP) highlighted. B, normalized uptake plot for peptide T471-X481. C, normalized uptake plot for peptide T471-F496. D, normalized uptake plot for peptide T471-X481 for incubation with ephrin B2. E, normalized uptake plot for peptide T471-F496 for incubation with ephrin B2. F, magnitude of HDX change across the Nipah G_M_ structure upon ephrin B2 binding is superimposed to PDB: 2VSM and 7TY0; peptides T471-X481 (light purple) and T471-F496 (indigo) are highlighted. Ephrin B2 in green, uncovered regions in gray. HDX-MS, hydrogen deuterium exchange-mass spectrometry; PDB, Protein Data Bank.

Compared to the WT, both N481D and T483A mutants exhibited a significant increase in HDX localized to the mutation sites (Fig. 4A and Supplemental Fig. S18, B and C), in the region comprising residues T471-X481 (Fig. 4B). T483A also showed a clear increase in HDX across multiple time points in the longer peptide spanning T471-F496, which was only seen in a single time point for N481D (Fig. 4C). This indicates the effect of deprotection for the T483A mutant is not only localized to the mutation site but is transmitted to the adjacent segment N482-F496 (the region not covered by peptide T471-X481). Interestingly, the receptor-bound complex did not display a significant change in HDX for peptide T471-X481 (Fig. 4D), but did exhibit decreased HDX for T471-F496 (Fig. 4E). This shows protection for residues N482-F496 (Fig. 4F), which is the same region impacted by T483A mutant (and minimally by N481D mutant). Crucially, the residues N482-F496 span a large loop in the ephrin B2 binding site (Fig. 4F, left). As the deuterium uptake of residues T471-F496 increases over time with the unbound WT (Fig. 4E), its slower deuterium exchange indicates the N482-F496 loop has a secondary structure. Ephrin B2 binding results in HDX protection at this loop, and the differences observed in HDX between the two mutants correlates to their different binding affinities for ephrin B2. The T483A mutation has a greater HDX compared to N481D, highlighting that this mutation has a greater impact on the dynamics of this loop—this is further reflected lower affinity for ephrin B2 observed for the T483A mutant compared to N481D, and the WT (Fig. 3). Of the other N-glycan sites covered, peptides spanning the two adjacent glycosites, N306 and N529 displayed lower HDX in the presence of ephrin B2. However, this is likely due to solvent occlusion from the glycan or protein backbone interactions with the receptor, as a loss of N306 through T308 does not change ephrin B2 affinity and both the regions are within the ephrin B2 binding site.

Discussion

Here, we have identified a conserved region of vulnerability in Nipah G involving an N-glycan sequon, which substantially impacts receptor engagement. During Nipah virus evolution, the N481 residue in this sequon evolved to D481, but the following sixteen amino acids, including T483 (S483 in Hendra), are 100% conserved across all clades of Nipah and the primary strain of Hendra (Supplemental Fig. S2). These residues form a structured loop that becomes stabilized upon ephrin B2 binding, with our HDX data revealing stabilization is localized within residues N482-F496. We have also shown that the naturally occurring N481D mutation has minimal impact on the conformational dynamics of this loop and on receptor binding, which in principle does not impact infectivity potential. However, mutating T483 to A483 significantly decreases receptor binding affinity by negatively impacting dynamics of this loop. The crystal structure of the Nipah G_M_-ephrin B2 complex (Protein Data Bank [PDB]: 2VSM) shows the T483 hydroxyl engaging in a hydrogen bond with the carbonyl of the N543 side chain (Fig. 5) and mutation of 483 to alanine would disengage this bonding, leading to conformational rearrangement of the loop and subsequent decrease in ephrin B2 binding affinity. Taken together, our data suggest that the residues spanning 483 to 496 are a good candidate for therapeutic targeting, as many of the other sites displaying increased protection upon receptor engagement are buried within the oligomerization interface, and therefore potentially much less accessible to therapeutic monoclonal antibodies or antivirals.Fig. 5Hydrogen bonding network around N481. N482-F496 is shown in light pink, ephrin B2 is green, T483 is dark magenta. Image produced from PDB: 2VSM (distances are in Å). PDB, Protein Data Bank.

To our knowledge, this is the first site-specific glycan analysis of Nipah G_M_. We identified that most sites are fully occupied with the exception of N529. However, we hypothesize that occupancy of N72 in a full-length G_M_ (i.e. on a Nipah virion) would likely be low due to the close predicted proximity of this residue to the viral membrane preventing entry to the oligosaccharyltransferase (OST) complex active site (52, 53). We found N159 and N481 N-glycans were oligomannose rich, and oligomannose-rich sites have been recently linked to immune escape for enveloped viruses, due to their ability to sample a large conformational space and be recognized by innate immune C-type lectin receptors DC-SIGN, L-SIGN, dectin-2, and langerin (54, 55). The evolutionary variable N481 site had the greatest oligomannose occupancy, and so loss of this glycan could reduce immune pressure and clearance of Nipah. However, this does not correlate with the mortality rates seen for the two Nipah strains, where the mortality rate for the Bangladesh strain, which lacks this N-glycan, is considerably higher (∼70–90%) than the Malaysia strain (∼40%) (56). Previous work suggests that loss of the 481 glycan through N481D increases fusogenicity in a pseudoviral model (29). Therefore, an explanation for the nonconserved glycosite could be in response to increasing viral fusion, suggesting this region could be involved in F-protein engagement.

Data Availability

Glycoproteomics, IM-MS and hydrogen deuterium exchange data are deposited to ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (57) with the dataset identifier PXD062381. HPLC and BLI raw data are deposited at Zenodo (https://doi.org/10.5281/zenodo.15569940). HPLC peak identifications (Supplemental Data File S2), IM-MS peak lists (Supplemental Data File S3) glycoproteomics peptide identifications (Supplemental Data S4–S6) and statistic outputs for HPLC, glycoproteomics and BLI (Supplemental Data File S9) are also attached as Supplementary files. In line with hydrogen deuterium mass spectrometry reporting guidelines (58), HDX-MS uptake plots, uptake values and reporting forms are attached as supplemental data files (Supplemental Data Files 7, 8, 10 and 11). Annotated MS/MS spectra for O-glycan site localization is provided in Supplemental Data File S12. All other data are available upon request.

Conflict of interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests. Weston Struwe is a shareholder of Refeyn Ltd which manufactures mass photometry instrumentation used in this study.