A key region of molecular specificity orchestrates unique ephrin-B1 utilization by Cedar virus

Abstract

The emergent zoonotic henipaviruses, Hendra, and Nipah are responsible for frequent and fatal disease outbreaks in domestic animals and humans. Specificity of henipavirus attachment glycoproteins (G) for highly species-conserved ephrin ligands underpins their broad host range and is associated with systemic and neurological disease pathologies. Here, we demonstrate that Cedar virus (CedV)-a related henipavirus that is ostensibly nonpathogenic-possesses an idiosyncratic entry receptor repertoire that includes the common henipaviral receptor, ephrin-B2, but, distinct from pathogenic henipaviruses, does not include ephrin-B3. Uniquely among known henipaviruses, CedV can use ephrin-B1 for cellular entry. Structural analyses of CedV-G reveal a key region of molecular specificity that directs ephrin-B1 utilization, while preserving a universal mode of ephrin-B2 recognition. The structural and functional insights presented uncover diversity within the known henipavirus receptor repertoire and suggest that only modest structural changes may be required to modulate receptor specificities within this group of lethal human pathogens.

Figure 1.. The CedV-G β-propeller exhibits regions that are structurally conserved with ephrin-tropic HNV-Gs suggestive of a preserved mode of receptor recognition.

(A) The structure of CedV-G is displayed as a cartoon colored from the N- to C terminus (blue to red), with the termini shown as spheres. The approximate extent of each of the six “blades” of the β-propeller is delineated by a colored triangle, labelled β1–β6. The region of highest variability (β6-S2–S3 loop) between the a.s.u. copies (molecule “A” and “B”) of the protein is shown with an inset panel (dashed line). Molecule A is shown in rainbow and the β6-S2–S3 loop of molecule B is shown in black. The protein is displayed in a “top” view (left) and rotated 50° to reveal a side view on which the putative receptor-binding site is depicted with a dashed line. N-linked glycans are shown as sticks and colored according to constituent elements. Asparagine residues from all eight N-linked glycosylation sequons are displayed as sticks. The putative dimerization interface contributed by the β1 and β6 blades is denoted with a solid black line. (B) Domain organization and salient features of CedV-G. The attachment-mediating β-propeller domain, transmembrane, and intra-virion regions are labelled. Putative N-linked glycosylation sites are displayed as pins, with sites occupied in the crystal structure colored black. (A) The extent of the crystallized construct is colored as a rainbow as in (A). (C) Structural comparison of CedV-G and unliganded NiV-G (PDB accession code: 2VWD). Because of the high level of structural similarity between NiV-G and HeV-G (36), an HeV-G comparison is omitted for clarity. RMSD between aligned Cα residues is depicted by both color (in a gradient from blue to red with increased RMSD) and tube width (thin to thick with increased RMSD). Residues that failed to align or exhibited RMSDs greater than 3 Ε were assigned values of 3 Ε. The average RMSD across each blade is displayed next to the respective blade label and the more structurally conserved region of the molecule (β4–β6) is indicated with a dashed line. The NiV-G–ephrin-B2 interface is displayed as a grey shadow superposed onto the structure of CedV-G. (D) Structure-based phylogenetic analysis of paramyxovirus receptor-binding proteins places CedV-G amongst ephrin-using HNV-G proteins. Pairwise distance matrices were calculated with Structural Homology Program (37) and plotted with PHYLIP (38), using the structures of unliganded receptor-binding glycoproteins, where available. The corresponding structures are shown in surface representation with previously characterized receptor-binding surfaces shown in dark grey. The structure of CedV-G is colored according to sequence conservation with NiV-G, identical residues are red and similar residues are pink. Measles virus hemagglutinin (MeV-H) branch is truncated for illustrative purposes. Structures used for the analysis were the Ghanaian bat henipavirus G (GhV-G; PDB 4UF7), Nipah virus G (NiV-G; PDB 2VWD), Hendra virus G (HeV-G; PDB 2X9M), CedV-G, MeV-H (PDB 2RKC), Mòjiāng virus G (MojV-G; PDB 5NOP), human parainfluenza virus 3 hemagglutinin-neuraminidase (HPIV3-HN; PDB 1V3B), Newcastle disease virus HN (PDB 1E8T), parainfluenza virus 5 HN (PIV5-HN; PDB 4JF7), and mumps virus HN (MuV-HN; PDB 5B2C). Branches of the resultant tree are labelled with the calculated evolutionary distances.

Figure S1.. Sequence alignment of ephrin-tropic HNV receptor-binding glycoproteins.

Sequence alignment of the β-propeller domains from CedV-G (NCBI reference sequence: YP_009094086.1), NiV-G (NP_112027.1), HeV-G (NP_047112.2), and GhV-G (YP_009091838.1). Absolutely conserved residues are shown as white on a red background, partially conserved residues are shown in red, and non-conserved residues are shown in black. Residues that interact with ephrin-B1 or ephrin-B2 are denoted with grey (ephrin-B1) or black (ephrin-B2) filled boxes beneath the sequence. Secondary structure elements of CedV-G are shown above the sequence and labelled according to type: β-strands are illustrated with arrows and labelled sequentially according to the β-blade (β1–β6) and strand number (S1–S4); α- and 3₁₀ helices are illustrated as coils and labelled α1–α3 and η1–η3, respectively. Disulphide bonds within CedV-G are labelled beneath the sequence in order of appearance, for example, “1” links cysteine residues 212 and 622. The putatively modified asparagine residues of N-linked glycosylation sequons are outlined with a black box. The blades of the β-propeller are colored in discrete sections, from blue to red. Sequence alignments were determined by MultAlin (80) and plotted using ESPript (81).

Figure S2.. Electron density map features of N-linked glycans.

(A, B) Electron density supported modelling of well-ordered N-linked glycans in the structures of CedV-G (A) and CedV-G–ephrin-B1 (B). The 2Fo–Fc electron density map (blue mesh; contoured to 1σ) are shown for all asparagine residues of N-linked glycosylation sequons and modelled asparagine-linked N-acetylglycosamine moieties, both shown as sticks. The main chain of CedV-G is displayed as a cartoon colored from blue to red (from N- to C terminus) and ephrin-B1 is shown as a dark grey cartoon. Sticks are colored according to constituent elements (nitrogen is blue, oxygen is red, and carbon is colored according the moiety to which it belongs, for example, grey if an ephrin-B1 atom). Glycans are labelled according to the sequence position of their associated asparagine residue.

Figure 2.. Ephrin-B1 and ephrin-B2 bind CedV-G.

(A) Increasing concentrations of soluble ephrin-B–Fc (10⁻¹¹ to 10⁻⁷ M) were added to HNV-G transfected HEK-293T cells and binding (measured as GMFI values) was assessed by flow cytometry using Alexa Fluor 647–labelled anti-human Fc antibodies. Four-parameter dose–response or logistic (4PL) curves were generated by nonlinear regression using GraphPad Prism. Data from GMFI values are displayed as percent of maximal binding, with the maximal binding value at the highest concentration of ligand used set to 100%. The bottom of each 4PL curve was constrained to have a constant value of zero. The reported K_d (dissociation constant) corresponds to the ephrin ligand concentration [sEphrin-B3–Fc], at which 50% maximal binding is achieved. A value of N/A refers to data that could not be fitted unambiguously to a 4PL curve, that is, binding of soluble ephrin was not titratable or saturation could not be achieved even at concentrations up to 100 nM. Values for ephrin-B3–Fc binding to GhV-G are not displayed, as no detectable binding over background was observed. Data shown are the averages of three independent biological replicates ± SE. (B) NiV-F/G (NiVpp), HeV-F/G (HeVpp), GhV-F/G (GhVpp), and CedV-F/G (CedVpp) VSV-ΔG-rLuc pseudotyped viruses were used to infect Vero CCL81 cells in the presence of increasing amounts of soluble ephrin-B1–Fc, ephrin-B2–Fc, and ephrin-B3–Fc fusion proteins (10⁻¹² to 10⁻⁸ M). 4PL curves were generated using GraphPad Prism as above. Data from relative light unit(s) (RLU) values are displayed as percent of maximal infection, defined as the RLU achieved in the presence of media alone, which is set to 100%. The top and bottom of each 4PL curve was constrained to have a constant value of 100 and 0, respectively. Data shown are the averages of three independent biological replicates ± SE.

Figure 3.. Ectopic expression of ephrin-B1 or ephrin-B2 is sufficient to confer CedVpp entry into a non-susceptible cell type.

(A, B, C, D) NiVpp, CedVpp, and BALDpp (VSV pseudotypes bearing no viral glycoprotein) were used to infect (A) CHO-pgsA745 cells (a naturally ephrin-negative cell line) or CHO-pgsA745 cells that stably express (B, C, D) ephrin-B1 (CHO-B1), ephrin-B2 (CHO-B2), or ephrin-B3 (CHO-B3), respectively, over a range of viral inoculum (viral genomes/mL). Entry was measured as described in the legend to Fig 2. The asterisk indicates an RLU value above the maximum limit of detection. RLU were plotted against numbers of viral genome copies per milliliter and fitted to a linear regression (dashed lines) using GraphPad Prism. Data shown are the averages of three independent biological replicates ± SE.

Figure S3.. CedV glycoproteins are efficiently incorporated into VSV particles but display reduced fusogenicity.

(A) (left) Western blot analysis of the HNVpp. A fourfold dilution series of NiVpp, CedVpp, and BALDpp was subjected to SDS–PAGE under reducing conditions. HNV-G and HNV-F proteins were detected using anti-HA and anti-AU1 antibodies, respectively. VSV matrix protein was used as a loading control and detected by a mouse anti-VSV-M monoclonal antibody. Arrows indicate the bands for HNV-G and both cleaved (F₁) and uncleaved (F₀) HNV-F. (right) Incorporation was quantified by western blotting densitometry. To calculate a level of incorporation, HNV-G and HNV-F₁ band intensity was normalized to that of VSV matrix. Data shown are the individual normalized densitometry values for two dilution points within the linear dynamic range in the fourfold series. Horizontal dashes represent the mean from the two points. Statistical significance for the indicated comparisons were evaluated using a two-tailed unpaired t test, n/s denotes no significance. (B) Co-expression of CedV-F and CedV-G induced syncytia formation in U87 cells that were smaller than syncytia formed by NiV-F/-G and HeV-F/-G. CedV-F/G, NiV-F/G, HeV-F/G, or empty vector control were transfected into U87 glioblastoma cells. Images were taken 48 h posttransfection. Blue arrowheads highlight large syncytia from NiV glycoprotein-mediated fusion and red arrows highlight the reduced syncytia generated from CedV glycoprotein-mediated fusion. Scale bar = 30 μm.

Figure 4.. Ephrin-B1 facilitates CedVpp entry into biologically relevant primary HUVECs.

(A) Active transcription of ephrin-B1, ephrin-B2, and ephrin-B3 in primary HUVECs was determined by qPCR. Transcript levels are shown normalized to hypoxanthine phosphoribosyltransferase (HPRT) transcripts. Ephrin-B3 transcript levels were below the limit of detection (asterisk). Data shown are the individual data points from two independent biological replicates. Horizontal dashes represent the mean from the two replicates. (B, C) NiVpp, CedVpp, and VSVpp were used to infect primary HUVECs in the presence of increasing amounts (10⁻⁹ to 10⁻⁷ M) of Fc-tagged (B) soluble NiV-G (sNiV-G–Fc) or (C) soluble Eph-B3 receptor (sEph-B3–Fc). Entry was measured as in Fig 2. Data are shown as percent infection relative to the signal achieved when the viruses are incubated in the presence of media alone. Data shown are the individual data points from two independent biological replicates. Dashed lines connect the means from the duplicate data. Statistical significance for this entry inhibition assay was tested with a two-way ANOVA with Holm-Sidak’s correction for multiple comparisons, n/s denotes no significance, * denotes P < 0.05, ** denotes P < 0.005.

Figure S4.. Electron density map features at a key receptor interaction site.

2Fo–Fc electron density maps (blue mesh; contoured to 1σ) are shown for a key region at the CedV-G–ephrin-B1 interface. The side chains of ephrin-B1 residues comprising the YM motif (Y121 and M122) and the CedV-G residue Y525 are shown as sticks and colored according to constituent elements. The α3–β3-S2 loop (residues 420–427) is shown as a cartoon in the CedV-G–ephrin-B1 structure with the addition of side chains in unliganded CedV-G. CedV-G is colored in a gradient from N- to C terminus (blue to red) and ephrin-B1 is shown in dark grey. The a.s.u. chains to which each panel correspond is indicated in the bottom right. The complex comprising chains I and J exhibited regions of considerably poorer quality electron density than other a.s.u. molecules. Assessment of complexes B–D, A–C, and E–F, and the unliganded CedV-G structure (bottom right panel) enabled confidence in model building and allowed the formulation of structure-guided hypotheses that were tested with functional assays.

Figure 5.. CedV-G binds ephrin-B1 and ephrin-B2 at a conserved and overlapping binding site.

(A) The crystal structure of CedV-G in complex with ephrin-B1 reveals a conserved mode of receptor engagement across ephrin-tropic HNV-G proteins. The receptor-binding domain of CedV-G (colored in a gradient from blue to red, from N- to C terminus) forms a 1:1 complex with the extracellular domain of ephrin-B1 (dark grey). The principal interaction region of ephrin-B1, the “G–H loop,” is displayed as a thick tube for clarity. Modelled N-linked glycan moieties (pink) and the asparagine residues of all putative N-linked glycosylation sequons are shown as sticks. (B) Superposition of NiV-G–ephrin-B2 (PDB: 2VSM) on CedV-G–ephrin-B1. The NiV-G–ephrin-B2 complex is colored in light grey, with NiV-G shown as transparent for clarity. (C) Comparison of bound ephrin-B1 and ephrin-B2 molecules. CedV-G is shown as a white transparent surface with the ephrin-B1 interface colored according to sequence position as in panel (A). Regions of ephrin-B1 (dark grey) and ephrin-B2 (light grey) bound by CedV-G and NiV-G, respectively, are shown as cartoon tubes. β-propeller blades are delineated with triangles that are colored according to sequence position as in panel (A). (D) Ephrin-B1 and ephrin-B2 compete for binding to CedV-G. NiVpp and/or CedVpp were used to infect CHO-B2 cells (middle and right panels) or CHO-B1 cells (left panel), as indicated, in the presence of increasing amounts of soluble ephrin-B1–, ephrin-B2–, and ephrin-B3–Fc (10⁻¹² to 10⁻⁸ M). Entry was measured as in Fig 2. 4PL dose–response curves were generated as in Fig 2 and based on values displayed as percentages of infection with the RLU achieved in the presence of media alone set to 100%. The top of each fit curve was constrained to a have a constant value of 100. Data shown are the individual data points from two independent biological replicates with the fit curves shown in dashed lines. (E) NiV-G and CedV-G compete for binding to ephrin-B2. NiVpp or CedVpp were used to infect CHO-B2 (middle and right panels) or CHO-B1 (left panel) cells in the presence of increasing amounts of soluble Fc-tagged NiV-G (sNiV-G–Fc) (10⁻¹⁰ to 10⁻⁷ M). Data are shown as percent infection relative to the signal achieved when the viruses were incubated in the presence of media alone. Data shown are the averages of three independent biological replicates ± SE.

Figure 6.. Accommodating the YM motif is critical for ephrin-B1 utilization by CedV.

(A) Sequence alignment of the G–H loop region of human B-type ephrins. Absolutely conserved residues are highlighted red, partially conserved residues are colored red, and non-conserved residues are black. Residues that constitute the YM/LW motif are outlined with a black box. Sequences are numbered according to ephrin-B1. Alignments were determined by MultAlin (80) and plotted using ESPript (81). (B) Structure of CedV-G–ephrin-B1. CedV-G is displayed as a white surface with the ephrin-B1 interface colored according to the sequence position (blue to red, N- to C terminus). The principal interacting region of ephrin-B1 is shown as a dark grey cartoon tube, with the side chains of Y121 and M122, the “YM motif,” shown as sticks and colored according to constituent atoms. (C) Detailed view of the YM motif of ephrin-B1 (boxed region in a; middle) and the equivalent view of the LW motif of NiV-G–bound ephrin-B2 (light grey; top). (bottom) Overlay of the NiV-G residue, W504 (red) onto CedV-G–ephrin-B1, demonstrates potential steric overlap between NiV-G and ephrin-B1. The side chain of Y525 in CedV-G is sequestered away from the receptor-binding site. Side chains of key residues are shown as sticks and colored according to constituent atoms. (D) CedV is more tolerant to substitution of the YM/LW motif than NiV. CHO cells expressing both wild-type ephrins (B1 and B2) and mutants with reciprocally exchanged LW/YM motifs (B1LW and B2YM) were infected with NiVpp (top) or CedVpp (bottom). Entry was assessed and quantified as in Fig 2. Data represent the average of quadruplicate measurements ± SE. Statistical significance for the indicated comparisons were evaluated using a two-tailed unpaired t test, ** denotes P < 0.005 and n/s denotes no significance.

Figure S5.. RNA-seq expression data of EFNB1, EFNB2, and EFNB3 in select human tissues.

Expression values are shown in transcripts per million, calculated from a gene model with isoforms collapsed to a single gene. Box plots are shown as median and 25^th and 75^th percentiles; points are displayed as outliers if they are above or below 1.5 times the interquartile range. Data were obtained through the Genotype-Tissue Expression Project (GTExPortal.org) where samples were collected from more than 50 non-disease tissue sites from close to 1,000 individuals. Data for ephrin-B1 (EFNB1, top panel), ephrin-B2 (EFNB2, middle panel), and ephrin-B3 (EFNB3, bottom panel) are shown for the selected tissues listed alphabetically: Arteries (Ao, Aorta; Co; Coronary; Tb, Tibial) | Bladder | Brain Regions (a = Amygdala, b = Anterior cingulate cortex [BA24], c = Caudate [basal ganglia], d = Cerebellar hemisphere, e = Cerebellum, f = Cortex, g = Frontal Cortex [BA9], h = Hippocampus, i = Hypothalamus, j = Nucleus accumbens [basal ganglia], k = Putamen [basal ganglia], l = Spinal cord [cervical c-1], m = Substantia nigra) | Esophagus (Mu = Mucosa, Ms = Muscularis) | Kidney | Liver | Lung | Salivary Gland | Nerve-Tibial | Skin (Ex = Sun exposed; NE = Sun not exposed) | Spleen.