Characterization and functional analysis of N-linked glycosylation on the Hendra virus attachment glycoprotein

Abstract

Background

Hendra virus (HeV) is a bat-adapted zoonotic henipavirus belonging to the Paramyxoviridae family. It is classified as a biosafety level 4 (BSL-4) pathogen owing to its broad host range and high fatality rate. Currently, no vaccines or therapeutics are approved for human use. Viral entry is mediated by the attachment (G) and fusion (F) glycoproteins; the heavily glycosylated G protein is responsible for receptor binding.

Methods

The extracellular domain of HeV-G was expressed in Expi293F cells and its glycosylation sites and glycan composition were identified by mass spectrometry. A series of functional assays—including viral entry, receptor binding, cell-cell membrane fusion, antibody neutralization and immunogenicity—were performed to delineate the role of each N-linked glycosylation site.

Results

Glycan profiling of HeV-G identified seven N-linked and multiple O-linked glycosylation sites, revealing that the stalk residues (N72, N159) predominantly carry high-mannose glycans, whereas the head-domain N-glycan sites (N306, N378, N417, N481, N529) are primarily modified with complex glycans. Notably, among the head-domain sites, N481 also harbors a substantial proportion of high-mannose glycans. Functional assays revealed that removal of N-glycans at N159, N306 and N417 markedly reduced membrane fusion. The N159 residue is a key site for fusion triggering, and its function is tolerant to specific amino acid substitutions, which may contribute to stabilizing or facilitating the conformational cascade required for F protein-mediated membrane fusion. The N529Q mutant specifically decreased EB3 binding by 2.6-fold, correlating with reduced infectivity. Binding assays with neutralizing antibodies showed that most N-glycan deletions had negligible effects, except that N159Q and N481Q reduced affinity to nAH1.3. Immunization studies in mice demonstrated that N-glycans had minimal impact on humoral immunity, with only minor site-specific differences.

Conclusion

These findings provide a comprehensive characterization of HeV-G glycosylation, reveal site-specific roles of N-glycans in viral entry, receptor binding and membrane fusion, and offer new insights for vaccine and antibody development.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12985-026-03095-4.

Background

Hendra virus (HeV), a member of the genus Henipavirus in the family Paramyxoviridae, was first discovered during an outbreak in Australia in 1994 [1]. The natural reservoir hosts are fruit bats belonging to the genus Pteropus [2]. On occasion, the virus spills over from bats to horses, and subsequently to humans, where it can cause rapidly progressive respiratory or neurological disease with high case fatality rates [3]. More recently, in 2021, surveillance in horses revealed a novel variant—designated Hendra genotype 2 (HeV-g2)—adding to the known genetic diversity of the species [4–6]. The Henipavirus genus also includes Nipah virus (NiV) [7], Ghana virus (GhV) [8], Cedar virus (CedV) [9] and Angavokely virus (AngV) [10]. Of these, HeV and NiV stand out as biosafety level 4 (BSL-4) pathogens due to their unusually broad host tropism and high lethality.

Vaccines and postexposure treatments are currently the two main strategies against HeV infection [11]. In Australia, a recombinant soluble attachment glycoprotein (sG) subunit vaccine, Equivac^® HeV, has been approved for horses in November 2012 [12, 13] and a human HeV-sG vaccine formulation is undergoing Phase I clinical evaluation [14]. Furthermore, the fully human monoclonal antibody m102.4, which targets the G glycoprotein, has displayed safety and pharmacological efficacy in Phase I trials [15]. But to date, there are no approved vaccines or therapeutics for prevention and treatment against HeV infection.

Henipavirus entry depends on the coordinated activity of the attachment glycoprotein (G) and the fusion glycoprotein (F). The G protein interacts with the host cell receptors ephrin-B2 (EB2) and ephrin-B3 (EB3), triggering conformational rearrangements in the F protein, thereby facilitating membrane fusion [16–19]. Within the C-terminal stalk of G, residues 159–163 form a structural and functional bridge linking the head and stalk domains, contributing to proper oligomerization, receptor-induced conformational changes and F-triggering [20]. In addition, a series of conserved isoleucine residues and cysteines in the stalk domain are also critical for conformational stability and fusion activation [21, 22]. The mobility of this stalk segment appears critical: engineered disulfide bonds can strengthen G–F interactions but at the expense of fusion efficiency [23]. Intriguingly, even full-length ectodomain constructs have not succeeded in elucidating the stalk in cryo-electron microscopy (cryo-EM) studies [24], thereby leaving its conformational dynamics and F-triggering determinants largely unresolved.

Structurally, HeV-G is a type II transmembrane oligomer composed of an N-terminal cytoplasmic tail, a single transmembrane domain, a long alpha-helical stalk region and a C-terminal beta-propeller globular head [25, 26]. HeV-G is heavily glycosylated and several N-linked glycosylation sites have been identified in different expression systems [27, 28]. For example, seven N-linked glycosylation sites were determined to be occupied in the HeLa cells, while only four sites (N159, N378, N417 and N481) were found to be utilized in HEK293 cells [29]. Glycans in the stalk region, such as at N159, as well as O-linked modifications, have been implicated in regulating fusion and viral entry [30–32]. However, the precise, site-specific contributions of these glycans—including how they modulate receptor binding, fusion triggering, and immune recognition—have not been systematically dissected yet.

In light of the critical role that G protein plays in attachment, fusion, and viral entry, coupled with the lack of structural information for the stalk region, a detailed analysis of its glycosylation landscape is essential. In this work, we expressed the extracellular domain of HeV-G in Expi293F cells and identified occupied N- and O-linked glycosylation sites along with their respective glycan compositions through mass spectrometry. We subsequently generated targeted mutants at each N-glycosylation site to assess their effects on viral entry, receptor affinity, cell-cell membrane fusion, recognition by neutralizing antibodies and humoral immune responses in mice. This study provides a detailed functional map of HeV-G glycosylation and reveals site-specific roles—most notably highlighting the dual functional significance of N159 in triggering fusion and facilitating viral entry. Furthermore, it offers valuable insights to guide the rational design of vaccines and therapeutics against henipaviruses.

Methods

Expression plasmids

HeV-G (GenBank ID: 1446471), HeV-F (GenBank ID: 1446467), NiV-Bd-G (GenBank ID: AAY43916) and NiV-Bd-F (GenBank ID: AAY43915) were cloned into the pcDNA3.1 vector, respectively. Asparagine (Asn, N) to glutamine (Gln, Q) mutations were introduced at each experimentally identified N-glycosylation site at residues N72, N159, N306, N378, N417, N481 and N529 in the HeV-G plasmid. Besides, N159A, N159E, N159K, N159L, N159M, N159P and N159R mutants based on structural stability were constructed by site-directed mutagenesis. All the constructs were verified by DNA sequencing (genecfps.com, Jiangsu, China). HIV backbone vector pNL4-3-Luc-R-E used for pseudovirus packaging was purchased from HonorGene (HG-VMH1160).

Cell lines

HEK293T, BHK-21 and BSR-T7/5 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% penicillin–streptomycin and 2 mM L-glutamine. BSR-T7/5 cells derived from BHK-21 cells that have been engineered to stably express T7 RNA polymerase. Expi293F cells were cultured in Expi293 Expression Medium (Gibco) in a 37℃ incubator with 8% CO₂.

Protein expression and purification

Codon-optimized HeV-G (residues 72–604) and N-glycosylation site mutations and NiV-Bd-G (residues 72–602) containing an N-terminal signal peptide, an N-terminal Strep-tag Ⅱ and GGGGS linker were cloned into pcDNA3.1 vector, with the codon optimization performed for Homo sapiens. The plasmid was transiently transfected into Expi293F cells with ExpiFectamine 293 transfection kit (Gibco) and Opti-MEM Ⅰ reduced serum medium (Gibco). Transfected cells were incubated on an orbital shaker (37℃, 8% CO₂) in Expi293 Expression Medium (Gibco). Five days post-transfection, the culture supernatant was harvested and clarified by centrifugation (3500 g for 15 min at 4℃). And then the protein supernatants were filtered through a 0.45 μm filter and purified using an AKTA (GE Healthcare) by 5 mL StrepTrap XT (Cytiva) with binding buffer (100 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, pH 8.0) and elution buffer (50 mM biotin in binding buffer). Elution fraction was concentrated in centrifugal filters and replaced with PBS buffer. Protein was analyzed by Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE).

Recombinant IgG expression and purification

The Ephrin-B2 (residues 27–229) and Ephrin-B3 (residues 28–227) receptors, along with the heavy and light chain variable regions of four previously reported neutralizing antibodies m102.4, HENV-26, HENV-32 and nAH1.3 were inserted into the pcDNA3.1 vector in the human antibody IgG1 format. The IgGs were expressed in Expi293F cells as described above. After being clarified and filtered, the receptor and antibody supernatants were purified by 5 ml Protein A HP (Cytiva) following the manufacturer’s instruction.

Glycosidase digestion

N-linked glycans on glycoprotein samples (10 µg) were digested with 2500 units of recombinant PNGase F (New England Biolabs) for 16 h at 37 °C under non-denaturing reaction conditions. This enzyme cleaves between the innermost N-acetylglucosamine (GlcNAc) and asparagine residues. For rapid N-glycan removal, a separate aliquot of glycoprotein (10 µg) was incubated with non-reducing format Rapid PNGaseF (New England Biolabs, 1x concentration) at 50 °C for 10 min according to the manufacturer’s protocol. O-linked glycans were removed by incubating samples (10 µg) with a combination of O-glycosidase (250 U) and neuraminidase (250 U) (New England Biolabs) for 16 h at 37 °C under non-denaturing conditions. Neuraminidase cleaves terminal sialic acids, allowing O-glycosidase to subsequently hydrolyze core 1 and core 3 O-glycans. In addition, a more comprehensive deglycosylation was performed using the Protein Deglycosylation Mix II (New England Biolabs; 1× concentration, as per manufacturer’s instructions) for 16 h at 37 °C under non-denaturing conditions with buffer 1. This enzyme mixture removes all N-glycans, short O-glycans and extended O-glycans in a stepwise manner. The extent of deglycosylation was assessed by SDS-PAGE followed by Periodic acid–Schiff (PAS) staining.

Mass spectrometry

Samples were reduced and alkylated according to previous methods [29]. Briefly, 25 µg HeV-G glycoprotein were subjected to SDS-PAGE and the corresponding gel strips were cut. Subsequently, the gel particles were decolorized using a solution consisting of acetonitrile (ACN) and 50 mM NH₄HCO₃ (1:1), followed by dehydration with 100% ACN. And then 10 mM DTT was added to reduce disulfide bonds in a water bath maintained at 56 °C for 1 h. After the solution was discarded, the samples were incubated with 55 mM iodoacetamide (IAM) for 1 h at room temperature in the dark. Reduced and alkylated samples were treated with trypsin and chymotrypsin to generate peptides of reduced molecular size, which are suitable for subsequent LC-MS/MS analysis and sequencing.

Samples were chromatographically separated on a Vanquish Neo UHPLC system (Thermo Fisher Scientific) combined with the ReproSil-Pur 120 C18-AQ column (3 μm, 120 Å, 100 μm i.d. × 180 mm) (Dr. Maisch, Germany) using a linear gradient elution from 4% solvent B to 95% solvent B over 131 min with a flow rate of 0.6 µL/min. The mobile phases consisted of solvent A (0.1% formic acid in water) and solvent B (20% solvent A-80% acetonitrile). Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific) were conducted on the acquisition of raw data with a spray voltage of 2.2 kV and a capillary temperature of 270 ℃. Full MS spectra were collected with a mass range from 300 to 1800 and followed by a standard data-dependent MS/MS acquisition mode.

Analysis of mass spectra and database searching

The raw mass spectrometric data files were analyzed and searched against target protein database based on the species of samples using Byonic, as described previously [33, 34]. The parameters were set as follows: the protein modifications were carbamidomethyl (C) (fixed), oxidation (M) (variable), acetyl (N-term) (variable), N-glycan 309 mammalian no sodium.txt @ NGlycan (variable), O-glycans 78 mammalian.txt @ OGlycan (variable); the enzyme specificity was set to Chymotrypsin and Trypsin; the maximum number of missed cleavages were set to 3; the precursor ion mass tolerance was set to 20 ppm, and MS/MS tolerance was 0.02 Da. Only high confident identified peptides were chosen for downstream protein identification analysis.

Pseudovirus generation

The pseudovirus was generated by co-transfecting HEK293T cells with the HIV backbone plasmid pNL4-3-Luc-R-E-, together with full-length HeV-G or NiV-Bd-G and a truncated form of HeV-F or NiV-Bd-F (where the cytoplasmic tail was shortened to retain only five C-terminal amino acids, based on an established protocol [35]), at a ratio of 12:2:1 using the Lipofectamine 3000 transfection kit (Invitrogen). Virus-containing supernatants were collected at 48 h and 60 h post-transfection, clarified by passage through a 0.45 μm filter to remove the cell debris and aliquoted for neutralization assays. All pseudovirus stocks were stored at − 80 °C to avoid multiple freeze-thaw cycles.

Viral entry

Pseudotyped viruses used for viral entry assays were generated using the same co-transfection protocol described above, with plasmids encoding full-length HeV-G (or its mutants) and truncated HeV-F at a ratio of 12:2:1 using Lipofectamine 3000 (Invitrogen). Cells transfected with the HIV backbone plasmid pNL4-3-Luc-R-E- alone served as a negative control, and processed identically. At 48 h and 60 h post-transfection, supernatants containing pseudovirions were clarified by centrifugation at 500 × g for 5 min to remove cell debris and potential aggregates before the infection step.

HEK293T cells were seeded at 3 × 10⁴ cells per well and inoculated with 50 µL of clarified pseudovirus supernatant. After 48 h of incubation at 37 °C with 5% CO₂, the supernatant was then gently aspirated and cells were lysed with 30 µL/well of Luciferase Cell Culture Lysis 5× Reagent (Promega; diluted to 1×) for 15 min at room temperature by shaking. A 20 µL aliquot of the lysate was transferred to a white 96-well plate, and luciferase activity was measured using a GloMax Navigator microplate luminometer (Promega) with the Luciferase Assay System (Promega). Results are expressed as relative light units (RLU). All experiments were performed with at least three independent replicates, each consisting of six technical repeats.

Syncytium formation assay

BSR-T7/5 cells were transfected under one of four conditions: (i) pcDNA3.1-HeV-F and pcDNA3.1-HeV-G (or its mutants) at a 1:1 ratio (test group); (ii) pcDNA3.1-HeV-G alone (specificity control); (iii) pcDNA3.1-HeV-F alone (specificity control); or (iv) no transfection (background control). Separately, BHK-21 cells were transfected with the pT7-Luc plasmid. At 6 h post-transfection, both cells were trypsinized, mixed together at a density of 4 × 10⁵ cells/well in 96-well plates with six technical replicates, and co-cultured for 24 h at 37℃ under 5% CO₂. Cells were then fixed with 4% paraformaldehyde and nuclei were stained with Hoechst 33,342 trihydrochloride for 10 min. DAPI microscope images were acquired with the BioTek Cytation 1 cell imaging multimode reader (Agilent) with a 10× objective.

Cell-cell fusion firefly luciferase reporter gene assay

Effector BSR-T7/5 cells were transfected with pcDNA3.1-HeV-F together with pcDNA3.1-HeV-G or mutants at a 1:1 ratio. Control groups included effector cells transfected with HeV-G alone, HeV-F alone, or none. Target BHK-21 cells were transfected with pT7-Luc plasmid using the Lipofectamine 3000 reagent (Invitrogen) as described above. At 6 h post-transfection, cells were trypsinized, combined at a density of 4 × 10⁵ cells/well in 96-well plates (six technical replicates per condition), and incubated for 24 h (37 °C, 5% CO₂). Luciferase activity was measured to quantify the cell-cell fusion as described above.

Biolayer interferometry assay

The kinetics of binding of receptors to HeV-G and HeV-G mutants were determined on an Octet-R8 device (Sartorius). Briefly, 10 µg/mL receptor EB2 or EB3 was loaded onto Octet ProA Biosensors (Sartorius) for 200 s. After a brief baseline step in PBST buffer (PBS with 0.05% Tween 20), two-fold serial dilutions of samples starting at 62.5 nM for EB2 or starting at 250 nM for EB3 were associated to biosensor tips for 300 s and then biosensors were transferred into PBST buffer to allow a 300 s dissociation step. Finally, the biosensors were regenerated with 0.1 M glycine (pH 2.0). Association and dissociation steps were aligned to reference wells to account for dissociation of antigen from the biosensor tip. Octet analysis studio v.13.0 software was used for curve-fitting with a 1:1 binding model to calculate equilibrium dissociation constant values.

Surface plasmon resonance assay

The binding affinity of neutralizing antibodies (m102.4, HENV-26, HENV-32, nAH1.3) to HeV-G and its N-glycosylated mutants was analyzed by surface plasmon resonance (SPR) on a Biacore T200 instrument (Cytiva) using a multi-cycle kinetics approach. First, antibodies were diluted in HBS-EP+ buffer (m102.4: 3 µg/mL; HENV-26: 3 µg/mL; HENV-32: 0.5 µg/mL; nAH1.3: 5 µg/mL) and immobilized on a Protein A chip at 10 µL/min for 120 s. Then two-fold serially diluted antigen (50 nM to 3.125 nM) was injected over the surface at 30 µL/min, with a 120 s association and 500 s dissociation phase per cycle. The chip surface was regenerated after each cycle by injecting Glycine 1.5 (Cytiva) for 180 s to remove bound analyte. Finally, the sensorgrams were globally fitted to a 1:1 binding model using Biacore T200 evaluation software (Cytiva) to calculate the equilibrium dissociation constant (KD).

Animal experiment

Specific pathogen-free (SPF) BALB/c mice (6–8 weeks old, female) were purchased from Vital River Laboratories (Charles River) and randomly allocated into nine groups, with each group containing six mice. Each mouse was immunized intramuscularly on day 0 (prime) and day 14 (boost) with 100 µL of a vaccine formulation containing 10 µg of antigen (HeV-G or N-glycosylated mutants) adsorbed to 50 µg of alhydrogel adjuvant (2%, InvivoGen); Control group mice received an equivalent volume of PBS adsorbed to the same amount of adjuvant. Serum samples were collected on days 0, 14, and 28. At the experimental endpoint (day 28), all mice were anesthetized with 2% isoflurane and humanely euthanized by cervical dislocation following the final blood collection. All animal experiments have been approved and were carried out in accordance with the guide for the Care and Use of Laboratory Animals (IACUC-20230438YZA-3R).

Enzyme-linked immunosorbent assay

The specific binding antibody (IgG) was detected by Enzyme-linked immunosorbent assay (ELISA). 96-well plates (Corning) were coated with 100 µL HeV-G or NiV-Bd-G at 1 µg/mL in coating buffer (0.1 M carbonate, pH 9.6) and left overnight at 4 ℃. After 3 times washes with PBST (PBS with 0.2% Tween 20), the plates were blocked with 150 µL blocking buffer (2% BSA in PBST) at 37℃ for 1 h. After blocking, serum samples were serially diluted and incubated at 37℃ for 1 h. Plates were washed three times before adding HRP conjugated goat anti-mouse IgG (1:20000 dilution, Abcam) at 37℃ for 45 min. Finally, the plates were washed three times and colored with 100 µL/well TMB substrate solution (Solarbio) for 5 min at room temperature in the dark. Substrate reactions were stopped by adding 50 µL/well stop solution (Solarbio) and the absorbance was measured at 450 nm and 630 nm. Pre-immune sera (Day 0) from all mice were included on each assay plate to establish the background optical density (OD). A predefined cutoff value was set as 2.1 -fold the mean OD of these day 0 sera, with a minimum threshold of 0.21 enforced to eliminate false-positive signals. The antibody endpoint titer was defined as the highest dilution that yielded an OD value at or above this cutoff, thereby ensuring they only specific antigen-binding activity was quantified.

Pseudovirus neutralization assay

A relative light unit (RLU) of approximately 1 × 10⁶ was used for viral dilution in pseudovirus neutralization assay. Mouse serum samples were inactivated by heating at 56 ℃ for 30 min and diluted in a three-fold gradient in 96-well plates. For the neutralizing antibodies, the starting concentration was 1 µg/mL with a three-fold serially dilution. Next, gradient serum was mixed with 50 µL/well pseudovirus and incubated for 1 h at 37℃, 5% CO₂. And then HEK293T cells were added into the 96-well microplates. After incubating for 48 h, luciferase activity was detected with substrate as mentioned above. The half-maximal inhibitory concentration (IC₅₀) was calculated using nonlinear regression model with GraphPad Prism software.

Statistical analysis

The results of viral entry, cell-cell membrane fusion, ELISA and pseudovirus neutralization are represented as mean ± standard deviation (SD). All the data were analyzed using GraphPad Prism version 9.5. Unpaired t-tests were used to compare the effect of HeV-G N-glycans on viral entry, cell-cell membrane fusion and antigenicity with wild type HeV-G. Statistical significance is denoted as follows: ns, not significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Results

HeV-G is highly glycosylated predominantly with N-glycans

Sequence alignment of the HeV-G and HeV-g2 G proteins showed high amino-acid identity (Supplementary Fig. 1). Consistently, NetNGlyc 1.0 predicted nine identical N-linked glycosylation sequons in both proteins (Fig. 1A). The close sequence relatedness and shared glycosylation pattern led us to select HeV-G as a representative model for subsequent experiments.

Fig. 1.

Predominant N-glycosylation and limited O-glycosylation in HeV-G. (A) NetNGlyc 1.0 prediction of N-glycosylation sites in HeV-G sequence, which applies artificial neural networks to evaluate the sequence context of Asn-Xaa-Ser/Thr sequons. (B) SDS-PAGE analysis of HeV-G under non-reducing conditions following treatment with the indicated glycosidases. (C) Periodic acid–Schiff (PAS) staining of treated HeV-G samples following deglycosylation. Lane 1: Marker; Lane 2: 5 µg Untreated HeV-G; Lane 3: 5 µg HeV-G + PNGase F; Lane 4: 5 µg HeV-G + Rapid PNGase F; Lane 5: 5 µg HeV-G + O-Glycosidase (OG) and Neuraminidase (NA); Lane 6: 5 µg HeV-G + Deglycosylation Mix Ⅱ; Lane 7: Horseradish Peroxidase (HRP, positive control); Lane 8: Soybean Trypsin Inhibitor (STI, negative control). The distinct bands correspond to the HeV-G monomer, dimer and tetramer, while the band below 40 kDa represents PNGase F

To investigate the glycosylation profiles, the HeV-G extracellular domain (72-604aa) expressed in Expi293F cells was purified and digested with different endoglycosidases. The mobility shifts on SDS-PAGE under non-reducing conditions were used to assess the extent of deglycosylation. As shown in Fig. 1B, digestion with PNGase F and the Deglycosylation Mix Ⅱ resulted in faster migration compared with untreated HeV-G. In particular, significant changes were observed following treatment with rapid PNGase F (non-reducing format), which indicated a decrease in the apparent molecular weight by approximately 10–20 kDa. This finding suggests the presence of substantial N-glycosylation modification. The slight shift observed after O-glycosidase and neuraminidase digestion, compared to the untreated control, indicates a lower abundance of O-glycans.

In addition, periodic acid–Schiff (PAS) staining was used to confirm the presence of glycan moieties on HeV-G. Periodic acid oxidizes vicinal diols on glycoproteins to dialdehydes, which react with Schiff’s reagent to produce a magenta color with a light pink background. The appearance of the magenta color suggested the presence of carbohydrates. As illustrated in Fig. 1C, magenta bands corresponding to HeV-G digested with PNGase-F or Deglycosylation Mix Ⅱ appeared lighter than those of untreated protein. Horseradish peroxidase and soybean trypsin inhibitor were used as positive and negative control, respectively. Notably, treatment with Rapid PNGase F (non-reducing format) almost completely abolished the magenta color, consistent with extensive N-glycan removal. The residual pale pink reflected a lower degree of O-glycosylation modification. Likewise, O-glycosidase and neuraminidase treatment produced a similar color intensity to the untreated group, indicating the presence of relatively few O-glycans. Collectively, these results demonstrate that HeV-G is highly glycosylated with more N-glycans and fewer O-glycans.

Mapping of HeV-G glycosylation sites and glycan composition

Following confirmation of glycosylation and prediction of N-linked glycosylation sites in the HeV-G protein, liquid chromatography–tandem mass spectrometry (LC-MS) was employed to determine the actual glycosylation sites and characterize glycan composition. The results showed that seven N-linked glycosylation sites, along with multiple O-linked glycosylation sites, were confirmed to be glycosylated (Fig. 2A). The identified glycosylation sites and their corresponding glycosylation efficiencies, as determined from glycopeptides generated by trypsin and chymotrypsin digestion, are listed in supplementary material Table Ⅰ.

Fig. 2.

Identification of glycosylation sites and characterization of N-linked glycan composition by LC-MS. (A) Glycosylation sites and corresponding glycosylation efficiency. (B) Composition and relative abundance of site-specific N-glycans. The bar graphs indicate the relative proportion of high-mannose glycans (green; M5-M8: Man₅GlcNAc₂ to Man₈GlcNAc₂), complex glycans (blue) and hybrid glycans (red) at each site. Accompanying pie charts summarize the overall distribution of these glycan types for each site. Due to glycan heterogeneity, only glycan compositions representing ≥ 5% of the total relative abundance are presented

Further analysis of N-glycan composition revealed a clear regional distribution pattern. Sites N72 and N159, located in the stalk domain, were predominantly modified with high-mannose glycans, varying from five to eight mannose residues. In contrast, N-glycosylation sites in the head domain (N306, N378, N417, N481, and N529) were mainly occupied by complex-type glycans. Notably, the N481 site, although situated in the head domain, exhibited a unique glycan profile characterized by a relatively high proportion of both high-mannose and complex glycans. Specifically, the proportions of high-mannose glycans at N72, N159, and N481 were 35.8%, 55.4%, and 36.5%, respectively. Meanwhile, the complex-type glycans content at the head-domain sites (N306, N378, N417, N481, and N529) all exceeded 43.8%, with the N417 site showing the highest proportion of complex glycans at approximately 84% (Fig. 2B). In addition, a smaller population of hybrid-type glycans were detected at certain sites. Due to glycan heterogeneity, only glycan compositions representing ≥ 5% of the total abundance are presented in supplementary material Table Ⅱ.

Effect of HeV-G N-glycan on viral entry

To determine the role of N-glycans in viral entry, we generated a panel of N-linked glycosylation site mutations and assessed their infectivity using a high-efficiency pseudovirus system (Fig. 3A). Pseudoviruses harvested at 48 h (Fig. 3B) or 60 h (Fig. 3C) post-transfection were used to infect HEK293T cells and infection efficiency was measured by relative luciferase activity at 48 h post-infection. At both 48 h and 60 h post-transfection, mutants N159Q, N306Q, N417Q and N529Q exhibited reduced infectivity compared with HeV-G. At 48 h as shown in Fig. 3B, relative light units (RLU) of these mutants decreased by about 32.5-fold, 2.0-fold, 2.7-fold and 2.0-fold, respectively. At 60 h as illustrated in Fig. 3C, the reductions were approximately 38.8-fold, 1.5-fold, 2.6-fold and 2.1-fold, individually. The most pronounced defect was consistently observed for N159Q, which displayed the lowest luminescence signal levels, indicating a substantial impairment in viral entry and single-cycle infectivity. In contrast, the more modest reductions for N306Q, N417Q, and N529Q suggest that glycans at these positions may play supportive rather than essential roles.

Fig. 3.

N159 is essential for HeV entry. (A) Schematic workflow of pseudovirus production and infection assay. (B and C) Effects of HeV-G N‑linked glycosylation site mutations on viral entry efficiency in HEK293T cells. Relative infection efficiency of HeV-G N-glycosylated site substitutions was quantified luciferase activity at 48 h post-infection. Pseudoviruses were harvested at 48–60 h post-transfection, as indicated. (D) Impact of mutations on HeV-G protein stability. FoldX 4.0 was used to assess the effect of mutations on HeV-G protein stability by calculating free energy changes (ΔΔG). Positive ΔΔG values indicate reduced protein stability, whereas negative values suggest improved protein stability. (E and F) Effect of N159 substitutions on viral entry efficiency in HEK293T cells. Pseudoviruses harvested at 48 h (E) or 60 h (F) post-transfection were used to infect cells, with luciferase activity measured at 48 h post-infection. Experiments were performed with three independent biological replicates, each consisting of six technical replicates per condition. Data are shown as mean ± SD. Statistical significance was determined by unpaired t-test with comparisons made against the wild type control. ns, not significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001

Although asparagine-to-glutamine (N→Q) substitutions are generally considered structurally conservative due to side-chain similarity, the remarkable infectivity loss for N159Q indicated that this residue is functionally unique. To evaluate whether this phenotype correlated with local structural stability, we performed FoldX 4.0 calculations to estimate the stability changes (ΔΔG) associated with the N159 substitutions. Most substitutions were predicted to be destabilizing (positive ΔΔG), whereas N159E and N159P showed the smallest ΔΔG changes (Fig. 3D). Consistent with the computational prediction, N159E displayed partially preserved entry efficiency compared to other N159 mutants, though it remained significantly impaired relative to wild type (Fig. 3E and F). These results demonstrate the functional importance of N159 in HeV entry, highlighting that its critical role lies in preserving the local structural stability of HeV-G protein, rather than merely providing specific chemical properties or maintaining simple side-chain conservation.

Impact of HeV-G N-glycan on receptor binding and cell-cell membrane fusion

HeV infection of host cells requires the concerted action of two envelope glycoproteins: the attachment protein (G), which recognizes and binds to the receptors ephrin-B2 (EB2) and ephrin-B3 (EB3), and the fusion protein (F), which mediates membrane fusion. To examine whether N-glycan modifications influence G protein function, the binding affinities of wild-type HeV-G and its N-glycan mutants to EB2 and EB3 were measured using biolayer interferometry (BLI). As displayed in Table 1, most N-glycan mutants bound EB2 and EB3 with comparable affinities to the wild-type protein, except for the N529Q mutant, which displayed a 2.6-fold decrease in binding to EB3 compared to HeV-G wild type (Fig. 4A and B). Detailed receptor-binding kinetics for wild-type and mutant proteins are provided in Supplementary Fig. 2A and 2B. Overall, most N-glycosylation sites had little impact on receptor binding, suggesting that the reduced infectivity was not due to impaired interactions between G protein and receptors.

Table 1.

KD determination between HeV-G N-glycan mutants and receptors by BLI

KD (nM)	HeV-G
Receptors	W.T.	N72Q	N159Q	N306Q	N378Q	N417Q	N481Q	N529Q
EB2	2.065	2.012	2.515	2.119	2.867	2.390	3.489	2.911
EB3	4.030	4.718	4.779	4.826	1.270	1.345	5.141	10.62

Fig. 4.

N159 is a vital amino acid site for membrane fusion. (A and B) Effects of HeV-G N-glycosylated site mutants on binding affinity to EB2 (A) and EB3 (B) as determined by by BLI. (C and D) Effects of HeV-G N-glycosylated site mutants on cell-cell membrane fusion, evaluated through syncytium formation assay (C) and quantitative firefly luciferase reporter gene assay (D). Effector BSR-T7/5 cells were transfected with plasmids encoding HeV-F and HeV-G (or its N-glycosylation site mutants) at a plasmid mass ratio of 1:1); control groups included HeV-G alone (Neg-G), HeV-F alone (Neg-F), or no transfection (Neg-cell). Target BHK-21 cells were transfected with the pT7-Luc reporter plasmid as described above. At 6 h post-transfection, effector and target cells were separately trypsinized, then mixed at a density of 4 × 10⁵ cells/well in 96-well plates, and co-cultured for 24 h. Nuclei were stained with Hoechst 33342 after fixation with 4% paraformaldehyde. Images were acquired using a BioTek Cytation 1 imaging system with a 10× objective. Scale bar: 300 μm. Luciferase activity was detected to quantify the cell-cell fusion efficiency. Experiments were performed with three independent biological replicates, each containing six technical replicates. Data are presented as mean ± SD. Statistical significance was determined by unpaired t-test compared to wild type. ns, not significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001

We next evaluated whether these substitutions influence G-F-mediated cell–cell membrane fusion, a process essential for viral entry and the formation of multinucleated syncytia, which are hallmarks of HeV and NiV pathogenesis. Fusion activity was first assessed qualitatively using a syncytium formation assay and then quantified by a firefly luciferase-based cell-cell fusion assay. In the qualitative experiment (Fig. 4C), the N306Q and N417Q mutants formed fewer syncytia than wild type HeV-G and F, whereas N159Q mutant failed to form syncytia.

Quantitative results (Fig. 4D) showed that most N-glycan mutants displayed reduced fusion activity relative to wild type, with the exception of N481Q and N529Q mutants. After background subtraction and normalization (WT = 100%, negative control = 0%), N159Q was indistinguishable from baseline (~ 14%), indicating near-complete loss of fusion, whereas N306Q and N417Q retained approximately 56% and 44% of WT activity, respectively. These findings, consistent with the qualitative observations, indicate that removal of N-glycans at positions N159, N306, or N417 markedly impairs membrane fusion. Given that these mutants did not substantially affect receptor binding, their reduced infection efficiency is most likely attributed to a functional defect in the membrane fusion function.

Finally, we assessed the impact of structural stability–based N159 substitutions on membrane fusion. As illustrated in Supplementary Fig. 3A and 3B, N159P substitution had no significant effect on fusion and N159E caused only a mild reduction in fusion activity compared to wild type, whereas other substitutions resulted in a complete loss of fusion activity. While a series of mutations did not conclusively implicate the involvement of N-glycans in cell-cell membrane fusion, the findings indicated residue N159 serves as a key amino acid site responsible for activating F protein to initiate membrane fusion.

Impact of HeV-G N-glycan on antibody binding and neutralization

The G glycoprotein plays a crucial role in receptor recognition and triggering of F protein-mediated membrane fusion process, making it an attractive target for neutralizing antibodies. To determine whether N-linked glycans influence antibody recognition, four broadly neutralizing monoclonal antibodies (mAbs) targeting distinct epitopes [25, 36–38] were evaluated.

Pseudovirus neutralization assays demonstrated potent cross-neutralizing activity for all four mAbs, with IC₅₀ values ranging from 8 to 120 ng/mL (Supplementary Fig. 4A and 4B). Surface plasmon resonance (SPR) analysis revealed that most N-glycan mutants maintained comparable binding affinities to m102.4, HENV-26 and HENV-32 relative to wild-type G (Table 2; Supplementary Fig. 5). In contrast, N159Q and N481Q mutants exhibited marked reductions in affinity for nAH1.3 by 5.5-fold and 4.3-fold, respectively (Fig. 5A-D). Moreover, N159E mutant resulted in a 4.9-fold reduction in affinity for nAH1.3 (Supplementary Fig. 6A and 6B).

Table 2.

Interactions between HeV-G N-glycan mutants and four mAbs by SPR

KD (nM)	HeV-G
mAbs	W.T.	N72Q	N159Q	N306Q	N378Q	N417Q	N481Q	N529Q
m102.4	0.540	0.591	0.952	0.680	0.788	0.793	1.212	1.156
HENV-26	0.518	0.240	0.256	0.400	0.519	0.283	0.235	0.694
HENV-32	0.148	0.150	0.135	0.140	0.143	0.140	0.222	0.165
nAH1.3	2.537	5.484	13.930	5.085	6.965	6.163	10.940	4.167

Fig. 5.

N159 may be a key potential neutralizing epitope residue. (A-D) The impact of HeV-G N-linked glycans on the binding affinity to mAbs was assessed by SPR. Mutations that enhance, moderately decrease and substantially decrease the affinity to mAbs are indicated in orange, blue and green, respectively. The cutoff value of greatly decreasing affinity is set as a 3-fold change in KD value relative to wild type, which is marked as gray. (E and F) Analysis of the membrane fusion inhibitory activity of mAbs. BSR-T7/5 cells were transfected with plasmids encoding HeV-F and HeV-G (1:1); control groups included cells transfected with HeV-G alone (Neg-G), HeV-F alone (Neg-F), or none (Neg-cell). BHK-21 cells were transfected with the pT7-Luc reporter plasmid. At 6 h post-transfection, both cells were trypsinized and mixed together at 4 × 10⁵ cells/well. mAbs were then added to a final concentration of 50 µg/mL, followed by a 24-hour co-culture period. Cells were fixed and nuclei were stained with Hoechst 33,342. Representative DAPI images (E) were acquired to visualize syncytium using a BioTek Cytation 1 imager (10× objective; scale bar: 300 μm). Luciferase activity was determined to quantify the inhibitory effect (F). Experiments were performed with three biological replicates and six technical replicates per condition. Data are presented as mean ± SD. Statistical significance was determined by unpaired t-test compared to wild type. ns, not significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001

Functional assays showed that nAH1.3, along with HENV-32, completely abolished syncytium formation at 50 µg/mL, whereas receptor-competing antibodies m102.4 and HENV-26 had no effect on fusion (Fig. 5E and F). These findings suggest that nAH1.3 may neutralize henipaviruses by blocking the F-triggering step of membrane fusion, with N159 potentially contributing to its epitope.

Role of HeV-G N-glycans on immunogenicity

Protein glycosylation can modulate immunogenicity by masking epitopes [33, 39–41]. To evaluate the effect of N-glycans on humoral responses, BALB/c mice were immunized twice with Alhydrogel-adjuvanted G protein antigens (Fig. 6A). ELISA results showed that most N-glycan mutants elicited HeV-G-specific IgG titers comparable to those induced by wild type at both day 14 and day 28 (Fig. 6B and D), except for N417Q mutant, which resulted in significantly lower IgG titers. Cross-binding to NiV-Bd-G was generally similar across all mutants. Although N306Q mutant induced higher titers and N481Q mutant yielded lower titers following the prime dose, these significant differences were no longer evident after the boost immunization (Fig. 6C and E).

Fig. 6.

N-glycans on HeV-G have relatively minor effects on humoral immunity in mice. (A) Schematic illustrationdiagram of mice immunization protocol. (B-E) The titers of serum IgG antibodies against HeV-G and NiV-Bd-G were measured by ELISA following the prime (Day 14) and boost (Day 28) immunizations. (F and G) The serum neutralizing antibody titers against pseudotyped HeV and NiV-Bd at Day 28 post-immunizatio were determined by pseudovirus neutralization assay. For panels (B) to (G), each experimental group consisted of 6 mice. Data are presented as mean ± SD. Statistical significance was determined by unpaired t-test compared to wild type. ns, not significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001

Pseudovirus neutralization assays indicated that N72Q and N306Q mutants induced higher HeV-neutralizing titers than wild type, while N72Q and N529Q mutants exhibited enhanced cross-neutralizing titers against NiV-Bd (Fig. 6F and G). Other mutants produced neutralizing responses comparable to those of the wild type. Overall, these results demonstrate that N-glycans on G have a minimal impact on antibody titers or breadth.

Discussion

This work provides a detailed analysis of the glycosylation profile of the Hendra virus (HeV) G glycoprotein and its functional implications for viral entry, membrane fusion, receptor binding, antibody recognition, and immunogenicity. Mass spectrometry confirmed seven occupied N-linked sites and multiple O-linked sites, with a distinct regional pattern: high-mannose glycans were concentrated in the stalk (N72 and N159), whereas complex-type glycans predominated in the globular head. Such a distribution may contribute to maintaining sufficient flexibility in the stalk region to transmit conformational signals to the fusion protein (F), while the head is optimized for stable receptor engagement and immune evasion.

Studies on Hendra virus (HeV) [30] and Nipah virus (NiV) G proteins [32] showed that six N-glycosylation sites are occupied (one in the stalk, five in the head) using an indirect, mutagenesis-based approach involving site-directed mutation of predicted N-X-S/T sequons followed by electrophoretic mobility shift assays. In contrast, we employed mass spectrometry (MS)–based glycoproteomics to directly identify each glycosylation site and its associated glycan structure, providing the most detailed and definitive site-specific glycosylation map for HeV-G protein to date. Prior findings that N-glycan removal alters neutralization sensitivity and fusion phenotypes remain entirely valid. Our comprehensive glycosylation map now provides a complete set of specific sites that can be investigated to deconvolute those effects.

Functional characterization of O-glycans on the G protein stalk domain revealed that O-glycans modulate G protein conformation and its receptor-induced structural rearrangements, which are required to trigger F protein-mediated membrane fusion [31]. Moreover, the absence of specific O-glycans disrupted G–F interactions and significantly impaired F protein incorporation and processing in pseudotyped virions, thereby altering viral entry. These findings collectively establish O-glycans as key regulators of glycoprotein function, fusion efficiency, and virion maturation—distinct from the roles of N-glycans—thus paving the way for future studies on O-glycans in other paramyxoviruses.

In our study, functional analysis showed that removing N-glycans at site 159, 306, 417 or 529 resulted in pronounced loss of pseudovirus infectivity. Receptor binding was largely unaffected by most glycan deletions, except for N529Q, which displayed reduced affinity for ephrin-B3. The interaction interface between HeV-G and EB2/EB3 has been well characterized [27, 42, 43], with EB2/EB3 positioned on the open upper surface of the HeV-G β-propeller and its G–H loop inserting into the central funnel-shaped cavity. In the crystal structure of the HeV-G/EB2 complex [29], the N-glycan at position 529 lies close to charged residues adjacent to the EB2-binding site, which may help stabilize receptor engagement. This structural context likely explains the reduced affinity of the N529Q mutant and its modest decrease in viral infectivity.

Further assays showed that removing N-glycans at position 159, 306 or 417 impaired cell–cell fusion, with the N159Q producing the most severe defect. These findings suggest that, at many sites, N-glycans are not indispensable for receptor engagement but may contribute to stabilize the functional interaction between the G and F glycoproteins, ensuring the complex remains poised to trigger membrane fusion upon receptor engagement. N159 emerges as a critical site in this context: mutagenesis indicated that the residue itself, in addition to its glycan, is important for triggering F-mediated fusion. Positioned between C158 and C162, which can form an intramolecular disulfide bond, N159 may participate in conformational coupling between the stalk and head domains or interact directly with the F protein, influencing the long-range rearrangements required for membrane fusion. Definitive structural confirmation of this mechanism will require high-resolution information on the stalk region, which is currently lacking. We also observed partial compensatory effects from certain substitutions at site 159, suggesting local structural context modulates its contribution to fusion triggering. Further structural studies will be required to clarify in detail how glycosylation sites and the stalk region contribute to G–F interactions.

It is widely accepted that binding of G glycoprotein to cellular receptors is the trigger for F glycoprotein to undergo fusion cascade (reviewed in [44, 45]). While G glycoprotein requires receptor binding to trigger fusion, the G-F interaction remains receptor-independent, as evidenced by their co-precipitate without receptor engagement [46–48]. The divergence between receptor binding and fusion efficiency, seen for example with N529Q, underscores that these processes are not fully interchangeable. While receptor attachment is a prerequisite for entry, the execution of membrane fusion appears to depend on additional structural dynamics, potentially governed by glycans or amino acid residues located outside the receptor-binding interface.

Prior structural mapping indicates that the nAH1.3 epitope is located in the head domain of G, distal to the ephrin-B2 receptor-binding site, and that the N481 glycan is proximal to the discontinuous epitope, supporting a direct glycan–antibody interaction [25]. Under our experimental conditions, the N159Q and N481Q mutations selectively impaired nAH1.3 binding, while showing minimal effects on the binding of m102.4, HENV‑26 and HENV‑32. Functionally, nAH1.3 and HENV‑32 suppressed syncytium formation, whereas m102.4 and HENV‑26 individually showed no detectable effect on membrane fusion. Our results are consistent with the view that local glycosylation can influence nAH1.3 recognition and its fusion-inhibitory activity. Further structural and functional analyses across defined glycoforms and conformational states will be needed to delineate how specific glycosylation sites and the stalk region modulate nAH1.3 recognition and G–F interactions. The absence of stalk-specific antibodies among the current set of well-characterized mAbs [37, 38, 49–51] suggests that this region may be inherently less immunogenic, perhaps due to its conformation or partial occlusion. This likely reflects the immunodominance of head-domain epitopes, which can mask modest changes in glycan shielding.

In immunized mice, most glycan mutants elicited antibody responses similar to the wild type, and in some cases slightly higher. The observed minor variations in specific antibody levels could stem from two key factors: First, individual glycans alone may not be sufficient to substantially alter the overall immune response. Second, methodological limitations in our detection approach may have prevented precise characterization of antibody distribution and composition. Furthermore, while the live-virus neutralization assay remains the gold standard, its requirement for BSL-4 containment facilities presents significant practical constraints.

Acknowledging the well-known challenge of cell type-specific glycosylation in glycobiology, the glycosylation patterns likely differ between Expi293F (used for protein production) and BHK-21 cells (used in syncytium formation assay and cell-cell fusion firefly luciferase reporter gene assay). It is probable that subtle differences in glycan processing (sialylation, fucosylation or branching) exist. The choice of different cell lines was driven by the specific technical requirements of each assay (e.g., protein yield, viral entry and fusion efficiency), which is a reasonable and common practice. The primary focus of our study was to determine the functional role of the presence or absence of N-glycans at specific sites, rather than to compare the fine structural heterogeneity of the glycans. This was directly tested through N-to-Q mutations, which assess the functional consequence of removing the glycan entirely. Therefore, our key findings on the necessity of the glycan are robust and transcend the potential subtle variations in glycan structure that may arise from the use of different cell lines.

While this study clarifies the role of individual glycosylation sites, efforts to generate and characterize mutants lacking multiple or all N‑glycans were hindered by severely impaired protein expression (Supplementary Fig. 7A and 7B). This limitation reinforces the collective contribution of N-glycosylation to the proper folding and biosynthesis of the G protein. Our systematic single‑site mutagenesis approach employed here provides a clear and interpretable experimental foundation. Based on these findings, future work could employ glycosylation inhibitors under controlled conditions or strategically designed multi-site mutants to investigate potential cooperative interactions within the glycan network.

Conclusion

Taken together, the results indicate that while most N-linked glycans on HeV-G have limited influence on receptor binding or antibody recognition, certain sites—especially N159—play a central role in regulating membrane fusion and viral entry. The high-mannose glycan at N159 is likely to stabilize the conformational transitions that underlie F protein activation. Given the unique position of N159 in the stalk region, further structural and functional dissection of this region will be critical for understanding the fusion mechanism and for evaluating its potential as a target for antiviral antibodies or vaccine design.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HeV

Hendra virus

G

Attachment glycoprotein

F

Fusion glycoprotein

BSL-4

Biosafety level 4

HeV-g2

Hendra genotype 2

NiV

Nipah virus

GhV

Ghana virus

CedV

Cedar virus

AngV

Angavokely virus

EB2

Ephrin-B2

EB3

Ephrin-B3

DMEM

Dulbecco’s Modified Eagle’s Medium

FBS

Fetal bovine serum

SDS-PAGE

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis

PAS

Periodic acid–Schiff

CAN

Acetonitrile

DTT

D​​ithiothreitol

IAM

​​I​​odo​​a​​cet​​a​​mide

RLU

Relative light units

BLI

Biolayer interferometry

SPR

Surface plasmon resonance

ELISA

Enzyme-linked immunosorbent assay

IC₅₀

half-maximal inhibitory concentration

LC-MS

Liquid chromatography–tandem mass spectrometry

pNL4-3-Luc

pNL4-3-Luc.R-E-

ΔΔG

Change in free energy

Author contributions

Experiments: Yamei Feng, Yaohui Li, Zhang Zhang, Yue Zhang. Figure preparation: Yamei Feng and Yaohui Li. Original draft writing: Yamei Feng. Methodology: Yamei Feng and Yaohui Li. Supervision: Xiaodong Zai, Yilong Yang, Ying Yin, Jun Zhang, Junjie Xu. Final revision and editing: Jun Zhang and Junjie Xu.

Funding

This work was supported by the National Natural Science Foundation of China (32200762).

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

The animal study protocol was approved and carried out in accordance with the guide for the Care and Use of Laboratory Animals (IACUC-20230438YZA-3R).

Competing interests

The authors declare no competing interests.