|Home | About | Journals | Submit | Contact Us | Français|
Hendra virus is a negative-sense single-stranded RNA virus within the Paramyxoviridae family which, together with Nipah virus, forms the Henipavirus genus. Infection with bat-borne Hendra virus leads to a disease with high mortality rates in humans. We determined the crystal structure of the unliganded six-bladed β-propeller domain and compared it to the previously reported structure of Hendra virus attachment glycoprotein (HeV-G) in complex with its cellular receptor, ephrin-B2. As observed for the related unliganded Nipah virus structure, there is plasticity in the Glu579-Pro590 and Lys236-Ala245 ephrin-binding loops prior to receptor engagement. These data reveal that henipaviral attachment glycoproteins undergo common structural transitions upon receptor binding and further define the structural template for antihenipaviral drug design. Our analysis also provides experimental evidence for a dimeric arrangement of HeV-G that exhibits striking similarity to those observed in crystal structures of related paramyxovirus receptor-binding glycoproteins. The biological relevance of this dimer is further supported by the positional analysis of glycosylation sites from across the paramyxoviruses. In HeV-G, the sites lie away from the putative dimer interface and remain accessible to α-mannosidase processing on oligomerization. We therefore propose that the overall mode of dimer assembly is conserved for all paramyxoviruses; however, while the geometry of dimerization is rather closely similar for those viruses that bind flexible glycan receptors, significant (up to 60°) and different reconfigurations of the subunit packing (associated with a significant decrease in the size of the dimer interface) have accompanied the independent switching to high-affinity protein receptor binding in Hendra and measles viruses.
The zoonotic Hendra virus (HeV) uses the flying fox as a natural host and is highly virulent in humans and horses. Reported cases of HeV transmission in humans have, to date, been sporadic and restricted to Australia (23, 32, 42). HeV was originally discovered in 1994 following the infection and subsequent death of a horse trainer and numerous horses (42, 49). Infection was characterized by the swift onset (7 to 10 days) of acute respiratory disease with clinical symptoms including fever, dizziness, hypotension, and respiratory illness. HeV is closely related to Nipah virus (NiV) which, in its largest outbreak, was responsible for the death of approximately 105 people in Malaysia (46). HeV and NiV (referred to as HNV) collectively form the Henipavirus genus which belongs to the Paramyxoviridae family of single-stranded, negative-sense RNA viruses. Putative new members of this genus have been recently identified in African bats (19).
HeV contains attachment (HeV-G) and fusion (HeV-F) membrane glycoproteins which extend from the viral envelope and are required for efficient entry (20). HeV and NiV exhibit a broad tissue tropism which correlates with the use of the widely expressed cell surface glycoproteins ephrin-B2 (EFNB2) and ephrin-B3 (EFNB3) as functional receptors for HeV-G and NiV-G (referred to as HNV-G). Oligomerization of HeV-G is also critical for productive attachment (1, 4). Despite the shared underlying molecular architecture of the attachment glycoprotein across the paramyxoviruses, there is marked divergence in their cellular receptors. Newcastle disease virus (NDV) and parainfluenza viruses (PIVs) attach to cell surface sialic acid (neuraminic acid), a negatively charged nine-carbon saccharide that is located at the nonreducing termini of glycolipids and N- and O-linked glycans of glycoproteins. In contrast, morbilliviruses such as measles virus (MV), canine distemper virus, and rinderpest virus have been shown to require surface lymphocyte activation molecule (SLAM; CD150) (51, 52) or the complement regulator CD46 for viral attachment (18, 44, 48). Structure-based phylogenetic analysis of available paramyxovirus structures shows that NiV-G and HeV-G are structurally more similar to attachment glycoproteins of sialic acid-binding viruses such as NDV and PIVs than to that of MV (7). We have suggested that the protein-binding capacities of henipaviruses and morbilliviruses have evolved independently, indicating an innate propensity for viruses to acquire novel protein receptor specificity, a characteristic that presents a natural route for the emergence of new pathogenic viruses (7).
HeV-G and NiV-G are closely related by sequence (81% identity) and structure (7). This similarity is underscored by the observation that vaccination of cats with recombinant HeV-G can protect against challenge with NiV (39). The similarity is evident in the crystal structures of NiV-G and HeV-G in complex with EFNB2, which both display a common binding mode (7). Knowledge of the molecular determinants of HeV attachment and fusion is of value for the rational development of immunotherapeutic and antiviral reagents with which to target these viruses. Such a structure-based approach to drug design has led to the development of active antivirals against influenza (2). However, in contrast to the rigid carbohydrate binding cleft of influenza neuraminidase, crystallographic analysis of the unbound NiV-G revealed significant plasticity in many of the EFNB2-binding loops which extend from the β-propeller scaffold (10). This information redefines the structural template for rational antiviral drug design against the HNV-G family of viral glycoproteins. Here, we have sought to further refine this template by identifying structural characteristics of HeV attachment.
We have crystallized and solved the structure of the globular six-bladed β-propeller domain of HeV-G. Analysis of this structure reveals that, as previously observed for the closely related NiV-G, significant conformational changes occur in the EFNB2 and EFNB3 receptor binding loops between unliganded and receptor-bound structures (10). Additionally, based on the packing of identical HeV-G molecules in the crystal asymmetric unit, we provide a model for the dimeric component of the native oligomeric assembly of the attachment glycoprotein. This model is consistent with oligomeric structures from other paramyxovirus attachment glycoproteins, and we propose that it may represent a component of the biological assembly. Furthermore, we observe that the angle of association between monomeric subunits is much more conserved between glycan-binding hemagglutinin-neuraminidases (HN) than the protein-binding HeV and MV attachment glycoproteins. As a result, we suggest that the acquisition of protein-binding functionality by paramyxoviruses may require structural adaptation, which perturbs the conserved dimeric packing. Together these data support the hypothesis of a globally conserved mode of oligomerization among the paramyxoviruses.
Full-length soluble ectodomain of HeV-G (HeV-Gecto; residues 71 to 604, GenBank accession number NC_001906; cDNA synthesized by GeneArt, Regensburg, Germany) and the β-propeller domain (HeV-Gβ; residues 183 to 604) were cloned into the pHLsec vector and transiently expressed in human embryonic kidney (HEK) 293T cells (ATCC CRL-1573) and purified as previously described (3, 7-10, 14). Transient transfections were performed using 2 mg DNA per liter cell culture. When used, the α-mannosidase inhibitor kifunensine (Cayman Europe, Tallinn, Estonia) was added at the stage of transfection to a concentration of 5 μM (11, 14). Glycoprotein was purified by immobilized metal affinity chromatography and then treated with Flavobacterium meningosepticum endoglycosidase (endo) F1 (75 μg mg−1 protein; 12 h; 21°C) which hydrolyzes the glycosidic linkage of the di-N-acetlychitobiose core of oligomannose-type N-linked glycans. Following deglycosylation, proteins were purified by size exclusion chromatography (SEC) using a Superdex 200 10/30 column (Amersham, United Kingdom) in 150 mM NaCl, 10 mM Tris (pH 8.0) buffer. Typical pure glycoprotein yields per liter of cell culture were approximately 1.0 mg for glycosylated HeV-Gecto and 0.75 mg of deglycosylated HeV-Gβ.
Crystals were grown by sitting drop vapor diffusion using 100 nL protein plus 100 nL precipitant as described previously (54). HeV-G crystals grew at 4°C (5.8 mg ml−1) in 30% PEG 8000 (wt/vol), 0.2 M ammonium sulfate, and 0.1 M sodium cacodylate buffer (pH 6.5) after 49 days. Crystals were flash frozen by immersion in a cryoprotectant containing 25% (vol/vol) glycerol followed by rapid transfer to a gaseous nitrogen stream. Data were collected at Beamline ID23-2 at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. Images were integrated and scaled using the programs HKL2000 and Scalepack (45). Details are presented in Table Table1.1. The structure of HeV-G was solved by molecular replacement using the program Phaser (38) with unbound NiV-G (Protein Data Bank [PDB] accession number 2VWD) as the search model. Four molecules were identified in the asymmetric unit. Structure refinement was performed using Refmac 5 in the CCP4 suite and included iterative restrained refinement with TLS modeling and tight main chain and loose side chain noncrystallographic symmetry restraints between the four HeV-G molecules in the asymmetric unit (43, 56). The program COOT was used for manual rebuilding (22), and Molprobity was used to validate the model (17). Structural alignments were performed with the program SHP (50), and molecular graphics images were generated using PyMOL (DeLano Scientific, San Carlos, CA).
Glycans were released enzymatically following the method of Küster et al. (34). Coomassie blue-stained SDS-PAGE bands containing approximately 10 μg of target HeV glycoproteins were excised, washed with 20 mM NaHCO3, pH 7.0, and dried in a vacuum centrifuge before rehydration with 30 μl of 30 mM NaHCO3 (pH 7.0) containing 100 units ml−1 of protein N-glycanase F (PNGase F) (Prozyme, San Leandro, CA). After incubation for 12 h at 37°C, the enzymatically released N-linked glycans were eluted with water and the sample was passed through a 0.45-μm-pore-size filter (Millex-LH, hydrophobic polytetrafluoroethylene). Aqueous solutions of the glycans were cleaned with a Nafion 117 membrane (5).
Positive ion matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectra were recorded with a Shimazu AXIMA performance MALDI-TOF/TOF mass spectrometer fitted with delayed extraction and a nitrogen laser (337 nm). Samples were prepared by adding 0.5 μl of an aqueous solution of the glycans to the matrix solution (0.3 μl of a saturated solution of 2,5-dihydroxybenzoic acid in acetonitrile) on the stainless steel target plate and allowing it to dry at room temperature. The sample/matrix mixture was then recrystallized from ethanol. Further samples were examined after removal of sialic acids by heating at 80°C for 1 h with 1% acetic acid.
Electrospray mass spectrometry (MS) was performed with either a Waters quadrupole-time of flight (Q-TOF) Ultima Global instrument or a Q-TOF model 1 spectrometer (Waters MS Technologies, Manchester, United Kingdom) in negative-ion mode. Samples in 1:1 (vol/vol) methanol/water were infused through Proxeon nanospray capillaries (Proxeon Biosystems, Odense, Denmark). The ion source conditions were as follows: temperature, 120°C; nitrogen flow, 50 liters h−1; infusion needle potential, 1.1 kV; cone voltage, 100 V; RF-1 voltage, 150 V (Ultima Global instrument). Spectra (2-s scans) were acquired with a digitization rate of 4 GHz and accumulated until a satisfactory signal-to-noise ratio had been obtained. For MS/MS data acquisition, the parent ion was selected at a low resolution (a mass window of about 5 m/z) to allow transmission of isotope peaks and fragmented with argon at a pressure of 0.5 mBar. The voltage on the collision cell was adjusted to give an even distribution of fragment ions across the mass scale. Typical values were 80 to 120 V. Other voltages were as recommended by the manufacturer. Instrument control, data acquisition, and processing were performed with MassLynx (Waters) software version 4.0 (Q-TOF 1) or 4.1 (Ultima Global). Analysis of fragmentation spectra was performed as previously described (25-28, 30).
Coordinates and structure factors have been deposited in the Protein Data Bank under accession number 2X9M.
The full-length soluble region of HeV-G (HeV-Gecto; residues 71 to 604) and the globular six-bladed β-propeller domain (HeV-Gβ; residues 183 to 604) of HeV-G were transiently expressed in HEK 293T cells. The glycoproteins contained a C-terminal hexahistidine tag to facilitate metal affinity purification. The HeV-G β-propeller domain elutes as a monomeric species on size exclusion chromatography (Fig. 1A and B). In contrast, HeV-Gecto, containing the N-terminal stalk region, was a mixture of putative dimeric and tetrameric states (Fig. 1C and D). These observations are consistent with a previous analysis of HeV-Gecto by Bossart et al. (6) and studies of related PIV3 and PIV5 hemagglutinin-neuraminidases, which have demonstrated that the β-propeller domain alone does not display oligomerization in solution-state measurements (35, 58, 59). Furthermore, negative-stain electron microscopic studies revealed similar oligomeric and monomeric species for the full-length ectodomain and isolated β-propeller domain of PIV5-HN, respectively (58). These observations also mirror our previous analysis of the β-propeller domain of NiV-G (NiV-Gβ) by analytical ultracentrifugation in which only monomeric species were detected (7). In contrast, equivalent analysis of the full-length ectodomain of NiV-G revealed putative tetrameric species (7).
The glycosylation of HeV-Gβ was manipulated to promote crystallogenesis (11). HeV-Gβ was expressed in the presence of the α-mannosidase inhibitor, kifunensine. Kifunensine inhibits both endoplasmic reticulum (ER) α-mannosidase I and the Golgi α-mannosidases IA to C and consequently prevents processing of the nascent glycoproteins (15, 21). This inhibition traps the glycans as immature oligomannose-type compounds and renders them sensitive to cleavage by endo F1 (11). Mass spectrometric analysis of glycosylation on equivalently expressed NiV-Gβ confirmed that the glycans were mainly Man9GlcNAc2, although lower levels of the Man5-8GlcNAc2 were also detected (10). Following endo F1 treatment, size exclusion liquid chromatography and SDS-PAGE analysis (Fig. 1A and B) suggested an approximate 15-kDa decrease in apparent molecular mass. This indicates that the recombinant oligomannose glycoform of HeV-Gβ is composed of approximately 20% carbohydrate, consistent with the presence of five predicted N-linked glycosylation sequons.
Crystals were generated of deglycosylated HeV-Gβ, and the structure was determined to a resolution of 2.9 Å. The structure was solved by molecular replacement using the program Phaser (38) with the unbound structure of NiV-Gβ (PDB accession number 2VWD) as the search model.
The crystallographic asymmetric unit contains four HeV-G molecules. As previously reported, HeV-G forms a six-bladed β-propeller fold (7). Each blade (β1 to β6) is composed of four antiparallel β-strands (S1 to S4), and the blades are assembled in a toroidal arrangement around a central axis (Fig. (Fig.2A).2A). The ephrin-binding site is located across one face of the β-propeller fold perpendicular to the central axis (Fig. (Fig.2A).2A). The globular β-propeller fold is stabilized by seven intrachain disulfide bonds, and the N and C termini extend in the same direction. The overall topology and disulfide bond pattern of the β-propeller is conserved with NiV-G (7, 10, 57).
The four molecules identified in the crystallographic asymmetric unit are very similar to each other and superpose with very small deviations (chains A to D; 0.3-Å root mean square [RMS] deviation in equivalent Cα position over 418 residues). In contrast to the previously reported structure of HeV-G in complex with EFNB2, the main chain atoms could be modeled into continuous electron density from residue Asn186 to Glu603. The electron density is improved for the ephrin-binding face (Tyr581 to Asp585) (Fig. 2B and C, region 1) and in the region of interaction between molecules within the asymmetric unit (Ser204 to Arg212) (Fig. 2B and C, region 2).
Analyses of NiV-G alone and in complex with human cell surface receptors, EFNB2 and EFNB3, revealed significant rearrangements in the EFN-binding loops upon receptor recognition (10, 57), and a comparison of the unliganded HeV-G structures determined here with those in complex with EFNB2 also reveals significant plasticity around the Glu579-Pro590 loop (Fig. (Fig.2C,2C, region 1). The observed conformational change in this region upon receptor engagement is consistent with an induced-fit mechanism for receptor binding, further emphasizing that the structures of both bound and apo forms of viral attachment glycoproteins must be considered when used as templates for structure-based design of antiviral drugs.
Superposition of unbound HeV-G and NiV-G structures shows that the receptor binding domains of the two glycoproteins are very similar (1.1-Å RMS deviation in equivalent Cα position over 410 residues) (Fig. 2B and C, region 3). Nevertheless, there are significant structural differences in the Glu579-Pro590 EFN-binding loops. In the unbound HeV-G, this loop is elongated and extends toward the periphery of the β-propeller, while in NiV-G this region forms two antiparallel β-strands joined by a short loop (Fig. (Fig.2B,2B, region 1). These different conformations result in a variance in position between equivalent Cα atoms of up to 8 Å (Fig. (Fig.2B,2B, region 1). However, these regions participate in crystal packing in both HeV-G and NiV-G structures, and there is a potential that the lattice environments may contribute to their different conformations (although within the HeV-G crystal there are four molecules in the crystallographic asymmetric unit which, by definition, make different contacts). Therefore, the two very distinct conformational states exhibited by the two viral proteins may not be strongly characteristic of the individual viruses but rather reflect sampling of the wider conformational space available to these proteins prior to attachment.
A second region of HeV-G that exhibits significant conformational differences to NiV-G occurs between residues Leu202 through Val215 (Fig. (Fig.2B2B and and2C,2C, region 2), where equivalent Cα atoms deviate by up to 9 Å. This part of the structure lies away from the EFN-binding face and is positioned on the β6 and β1 blades of the β-propeller. The two different conformations may be explained by packing interactions which occur between each of the two pairs of adjacent HeV-G molecules in the asymmetric unit. These interactions occlude over 800 Å2 (33) of surface from each HeV-G molecule and result in the generation of a small, two-stranded antiparallel β-sheet (Thr206-Pro208) (Fig. (Fig.2B2B and and2C,2C, region 2) not observed in any of the previously reported receptor-bound or unbound HNV-G structures. However, among all HNV-G structures, this region exhibits conformational plasticity (Fig. 2B and C, region 2), high B factors, or in some cases is completely disordered. The intersubunit interactions may therefore be required to order this region.
Although the identification of oligomeric states within crystal structures cannot formally prove the mode of oligomerization occurring within the intact virus, such packing interactions can suggest a potential mode of oligomerization. The shared dimeric packing interactions observed in the paramyxovirus attachment glycoproteins are illustrated in Fig. Fig.3.3. The two dimers observed in the HeV-G crystal structure, although unlikely to represent a biological tetramer, are highly similar (0.2-Å RMS deviation of equivalent Cα positions over the 835 residues of the dimer). The mode of oligomerization for the putative dimers of HeV-G agrees with previous predictions (10, 58, 59) and is strikingly similar to the interactions observed for several other paramyxovirus attachment glycoproteins, including NDV-HN (13), PIV3-HN (35), PIV5-HN (59), and MV-H (12, 31, 48) (Fig. 3B and D), strongly supporting the hypothesis that the mode of dimerization of attachment glycoproteins is conserved between the henipa-, avula-, respiro-, rubula-, and morbilliviruses. Although the total buried surface is less than half of that observed in NDV-HN, PIV3-HN, and PIV5-HN structures (average buried surface of approximately 1,800 Å2) (Table (Table2),2), the location of the dimeric HeV-G association and the overall surface complementarity (36) of the dimer interface is comparable (average dimer surface complementarity score of 0.63 and 0.67 for HeV-G and other paramyxovirus-HN structures, respectively) (Table (Table2).2). Analysis of the relative orientations of the two propeller domains in each dimer using SHP (50) shows that for all five viruses there is a very close to exact 2-fold relationship between the two subunits; however, the relative orientations nevertheless differ significantly (Fig. 3B and C). Thus, the three sugar binding attachment proteins (from NDV, PIV3, and PIV5) are closely similar, but if we superpose one of the two subunits on a subunit of the HeV-G dimer then rotations of between 31° and 42° are required to superpose the second on the matching HeV-G subunit (Table (Table2).2). MV-H is even more of an outlier, lying some 63° away from the HeV orientation and no closer to the sialic acid binders (Fig. (Fig.3D;3D; Table Table2).2). We note also that the glycan binders form dimers with much more contact surface than the protein binders (Table (Table2).2). Our observations confirm those of Yuan et al. (58), who noted some flexibility in the subunit orientation of PIV5 but suggest that further rearrangements have evolved to correctly accommodate the more bulky, and perhaps more rigidly presented, protein receptors.
The orientation of the two HeV-G subunits in the dimer is such that when an EFNB2 molecule is docked onto the EFN-binding site of each propeller it does not preclude the formation of the HeV-G dimer. Furthermore, the N-terminal stalk regions would extend in similar directions away from the EFN-binding face, consistent with interstalk disulfide bonding which has been reported to be responsible for HeV-G dimerization and tetramerization (4). While the N-terminal stalk domains of paramyxovirus attachment glycoproteins drive oligomerization (4, 7, 37, 41, 47, 53, 58), homodimers and homotetramers have also been observed in crystallographic analyses of isolated β-propeller domains. Packing interactions in the crystal structures of NDV-HN, PIV3-HN, and MV-H reveal dimeric and tetrameric arrangements (12, 13, 31, 35, 48). These observations demonstrated that the detection of oligomeric states within the crystal lattice is not necessarily precluded by the absence of the stalk domain.
A tetramer composed of two dimers was also observed in the structure of PIV5-HN (59). The glycoprotein used in that study contained the oligomerization-inducing stalk domain. Although this region was disordered in the structure, its presence in the construct used for crystallization suggests that the mode of oligomerization of the globular, receptor binding domains resemble that occurring on the virion. As the dimers observed in the crystal structure of PIV5-HN resemble those of NDV-HN, PIV3-HN, MV-H, and HeV-G dimers which lack the stalk (Fig. (Fig.3),3), it is likely that the mode of dimerization observed in the latter are similarly biologically relevant. However, in no NiV-G or HeV-G crystal structure is there any evidence for tetramer (dimer-of-dimers) formation, as was observed in the crystals of NDV-HN (60) and PIV5-HN (59). Additionally, we note that no putative dimers were observed in the crystal structure of NiV-G alone, or the HeV-G-EFNB2, NiV-G-EFNB2, and NiV-G-EFNB3 complexes (7, 10, 57), indicating that weak interdomain interactions are not observed in all crystals.
Analysis of the superposed crystal structures of HN from PIV5, NDV, and PIV3 (59) led Yuan et al. to predict the oligomerization interface. It was assumed that the mode of oligomerization would be conserved among paramyxovirus attachment glycoproteins and that these regions would exhibit low divergence in structure. We previously widened this analysis by including the crystal structure of NiV-G and extended the structural analysis by considering the locations of N-linked glycosylation sequons from across the Paramyxoviridae family onto the β-propeller fold (10). Glycosylation sites are not usually found at oligomerization interfaces. Mapping the glycosylation sequons from across the Paramyxoviridae family onto the β-propeller fold of unliganded NiV-G did not obscure the putative oligomerization interface (10). We have applied this analysis to the structure of the HeV-G monomer reported herein (Fig. (Fig.3A).3A). A total of 24 positions of N-linked glycosylation sequons were detected upon sequence alignment across the β-propeller domains of the attachment glycoproteins of NiV, HeV, PIV1, PIV2, PIV3, PIV5, Sendai virus, and NDV (Fig. (Fig.3B).3B). Mapping of these glycan sites onto the structure of our crystallographic dimer of HeV-G, and onto the dimers of NDV-HN (13), PIV3-HN (35), and PIV5-HN (59), reveal that they are located away from the dimer interface (Fig. (Fig.3B).3B). This same analysis was applied to the more distantly related MV-H (7, 12, 31, 48) and shows a similar arrangement of glycans sitting away from the dimer interface (Fig. (Fig.3D);3D); however, two glycans native to MV-H (Asn200 and Asn215) are located in closer proximity to the MV-H dimer interface but do not interfere with the dimeric interface. This provides further support for the physiological relevance of the dimer-of-dimers model of oligomer formation in which the surfaces of the β1- and β6-propellers of HNV-G form one dimerization interface.
The analysis of oligomerization interfaces by consideration of the positions of the glycosylation sequons rests on the assumption that large N-linked glycans are unlikely to be located at the oligomeric interface of the attachment glycoprotein. However, glycans can occur at subunit interfaces of glycoproteins and immobilized glycans at dimer interfaces have been detected upon crystallographic analysis (24). Immobilized glycans that exhibit significant interactions with a protein surface are usually protected from α-mannosidase processing and remain as oligomannose-type glycans (16). Therefore, to assess the validity of our use of glycosylation sequons to identify putative oligomerization interfaces, we performed MALDI-TOF and electrospray-MS analysis on the N-linked glycans of HeV-Gecto and HeV-Gβ to determine if any glycan was protected from glycosylation processing upon oligomerization (Fig. (Fig.3;3; Table Table3).3). In addition to compositional analysis by MALDI-TOF MS (Fig. (Fig.4;4; Table Table3),3), glycans were structurally analyzed by ion collision-induced dissociation (CID) MS/MS, and the full assignments of the spectra are presented in the supplemental material.
The location of the N-linked glycosylation sites are conserved between HeV-Gβ and NiV-Gβ. MALDI-MS analysis of the HeV-Gβ glycans showed very similar profiles to those previously reported for NiV-Gβ (10) and other viral glycoproteins (9) recombinantly expressed in HEK 293T cells using the pHLsec vector (3). This indicates that no pronounced protein-directed glycosylation processing was occurring. However, MALDI-TOF MS analysis of HeV-Gecto demonstrated that the specific abundances of the complex-type glycans were altered. Specifically, levels of galactosylation were reduced, suggesting that oligomerization had influenced the accessibility of the N-linked glycans (Fig. (Fig.4;4; Table Table3).3). The sensitivity of β1→4 galactosyltransferase to the structural environment of glycosylation sites has been observed in the processing of IgG glycans. Interdomain glycans of the IgG Fc domain exhibit partial galactosylation, whereas glycosylation occurring in the complementarity-determining regions of the Fab exhibit higher levels of galactosylation (40). Interestingly, alanine-scanning mutagenesis of the stalk region of HeV-G was shown to influence glycan processing, as assessed by mobility on SDS-PAGE (4). These mutations did not seem to disrupt oligomerization of HeV-G but led to disruption of functional interactions with coexpressed HeV-F (4). Overall, these previously reported mutagenesis effects combined with our MS data suggest that glycan processing of HeV-G is sensitive to its oligomerization state and functional assembly.
Despite the reduction in glycosyltransferase processing of HeV-Gecto, only trace levels of oligomannose-type glycans were detected, confirming that all glycan sites are accessible to α-mannosidase processing when expressed as an oligomer. These data confirm that glycosylation sites on HNV-G are unlikely to significantly contribute to the buried surface area of the oligomerization interface, further supporting our hypothesis that the oligomerization interfaces are predicted by the paucity of glycosylation sequons.
Our crystallographic analysis of henipaviral attachment glycoproteins has demonstrated plasticity in many regions of the ephrin-binding loops. Understanding the conformations explored by unliganded viral attachment glycoproteins will inform the design of templates for rational drug discovery. The modest success of structure-based drug design in the identification of compounds that specifically disrupt target protein-protein interactions has been suggested to be, to some extent, explained by the limited conformational space defined by the template (55). We anticipate that knowledge of the structures of both unliganded and receptor-bound conformations will be necessary for more robust de novo computational screening of small-molecule libraries and for structure-based optimization of lead antiviral compounds.
From the detailed crystallographic analysis of isolated henipavirus attachment glycoproteins to the low-resolution electron microscopic analysis of intact virions, there is a paucity of information of the architecture of the attachment and fusion assembly. In accord with Yuan et al. (58), we have suggested a mode of oligomerization of HeV-G based upon the packing interactions within our crystal structure as well as the interactions of related paramyxovirus structures but find that this association is significantly reconfigured for those viruses which bind protein receptors. This analysis has supported the notion that information on low-affinity interactions can be gleaned by the analysis of modes of crystal packing. Understanding the intermolecular interactions of the attachment glycoproteins is likely to be a prerequisite for a full structural description of viral attachment and fusion.
We are grateful for the help of W. Lu for the tissue culture, K. Harlos for data collection, and the staff of Beamline ID23.2 at the European Synchrotron Radiation Facility for assistance. We thank C. N. Scanlan for access to the Shimazu AXIMA performance MALDI-TOF/TOF mass spectrometer funded by the International AIDS Vaccine Initiative (IAVI).
D.I.S. is an MRC Professor of Structural Biology, E.Y.J. is a Cancer Research UK Principal Research Fellow, T.A.B. is a Sir Henry Wellcome Postdoctoral Fellow, and M.C. is a Fellow of Oriel College, Oxford, United Kingdom.
This work was funded by the Wellcome Trust (075491/Z/04), Medical Research Council, Cancer Research UK, the Oxford Glycobiology Institute, and Spine2 Complexes (FP6-RTD-031220).
Published ahead of print on 7 April 2010.
†Supplemental material for this article may be found at http://jvi.asm.org/.