Kinetic analysis of paramyxovirus-sialoglycan receptor interactions reveals virion motility

Abstract

Many viruses initiate infection by binding to sialoglycan receptors at the cell surface. Binding to such receptors comes at a cost, however, as the sheer abundance of sialoglycans e.g. in mucus, may immobilize virions to non-functional decoy receptors. As a solution, sialoglycan-binding as well as sialoglycan-cleavage activities are often present in these viruses, which for paramyxoviruses are combined in the hemagglutinin-neuraminidase (HN) protein. The dynamic interactions of sialoglycan-binding paramyxoviruses with their receptors are thought to be key determinants of species tropism, replication and pathogenesis. Here we used biolayer interferometry to perform kinetic analyses of receptor interactions of animal and human paramyxoviruses (Newcastle disease virus, Sendai virus, and human parainfluenza virus 3). We show that these viruses display strikingly different receptor interaction dynamics, which correlated with their receptor-binding and -cleavage activities and the presence of a second sialic acid binding site. Virion binding was followed by sialidase-driven release, during which virions cleaved sialoglycans until a virus-specific density was reached, which was largely independent of virion concentration. Sialidase-driven virion release was furthermore shown to be a cooperative process and to be affected by pH. We propose that paramyxoviruses display sialidase-driven virion motility on a receptor-coated surface, until a threshold receptor density is reached at which virions start to dissociate. Similar motility has previously been observed for influenza viruses and is likely to also apply to sialoglycan-interacting embecoviruses. Analysis of the balance between receptor-binding and -cleavage increases our understanding of host species tropism determinants and zoonotic potential of viruses.

Fig 1. Characterization of virus-receptor interactions using classical techniques.

(A) Hemagglutination assay. Serial two-fold dilutions of NDV, hPIV3 and SeV (starting concentration 2x10¹⁰ virus particles/ml as determined by nanoparticle tracking analysis [NTA]) were incubated with human or chicken erythrocytes at 4°C for 2h. A red erythrocyte pellet indicates the absence of hemagglutination. Red circles indicate the first dilution at which no hemagglutination was observed. (B) Sialidase activity assay. Sialidase activity of serial two-fold dilutions of NDV, hPIV3 and SeV (starting concentration 2x10¹⁰ virus particles/ml) was assessed in triplicate at 37°C using the fluorogenic substrate MUNANA at pH 5.6 or 7.0. Means and standard deviations are graphed. The fluorescence generated from MUNANA cleavage was measured using a plate reader (in relative fluorescence units [RFU]).

Fig 2. Binding of NDV, hPIV3 and SeV to different sialoglycans.

(A) Schematic representation of the interaction between virus particles and synthetic biotinylated glycans. (B) Biotinylated sialoglycans used in this study are: 3’S(LN)₃: Neu5Acα2-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAc, 6’S(LN)₃: NeuAcα2-6Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAc; and 3’S(LN)₂-PAA: Neu5Acα2-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAc on a polyacrylamide backbone. Binding curves were generated for NDV (C), hPIV3 (D), SeV (E) and H5N1 (F) at 10¹⁰ particles/ml.

Fig 3. Analysis of NDV, hPIV3, and SeV binding and sialidase activity using lectin binding.

(A) Schematic representation of virus binding followed by the lectin binding assay. MAL I specifically binds to the remaining Neu5Acα2-3Galβ1-4GlcNAc glycotope, while ECA prefers to bind the Galβ1-4GlcNAc glycotope present after desialylation. (B) Virus binding curves were generated similarly as described in the legend to Fig 2 using 3’S(LN)₂-PAA. (C) Binding of MAL I after incubation of the sensors with the indicated viruses. MAL I binding in the absence of NA activity (no NA) or after treatment of the sensors with Arthrobacter ureafaciensl NA (AUNA) is also shown. (D) Similar as for (C), but now using ECA. Low binding levels of MAL I and high binding levels of ECA correspond with high levels of desialylation.

Fig 4. Interaction of NDV, hPIV3, and Sendai virus with 3’S(LN)₂-PAA in the presence of BCX2798.

(A) Structure of NDV (PDB 1usr) and hPIV3 (PDB 1v3c) HN dimer is shown in a surface representation. Catalytic site (Site I) and the 2SBS (Site II) occupied with Sia are indicated. Virus binding curves were generated similarly as described in the legend to Fig 2 using 3’S(LN)₂-PAA for NDV(B), hPIV3(C), and SeV(D) in the absence or presence of 1mM BCX2798. (E) Binding of NDV to 3’S(LN)₃, 6’S(LN)₃ and 3’S(LN)₂-PAA in the presence of BCX2798.

Fig 5. NDV, hPIV3, and SeV interaction with 3’SLN, 3’S(LN)₂, or 3’S(LN)₃.

(A) Schematic representation of biotinylated 3’SLN, 3’S(LN)₂,and 3’S(LN)₃ glycan structure. (B) Analysis of NDV’s ability to bind to 3’SLN, 3’S(LN)₂, or 3’S(LN)₃ in the presence or absence of BCX2798. Analysis of the ability of hPIV3 (C) and SeV (D) binding ability to 3’SLN, 3’S(LN)₂, or 3’S(LN)₃. Virus binding curves were generated similarly as described in the legend to Fig 2.

Fig 6. Characterization of NDV and SeV binding kinetics through BLI.

(A) Sensors were loaded with different densities of 3’S(LN)₃ receptor by incubation with different concentrations of receptor. Fractional receptor loading levels were obtained by normalizing receptor binding signals to the signal of sensors containing saturating receptor levels (e.g. sensor 1). Subsequently, sensors were incubated with SeV (B) or with NDV in the presence or absence of BCX2798 (C). The initial binding rate, corresponding to the steepness of the tangent at the beginning of the binding curves was determined form graphs B and C and normalized to the maximal, saturating initial binding rate for each condition, resulting in a fractional initial binding rate. (D) The fractional initial binding rates for SeV, NDV and NDV+BCX2798 were plotted against the fractional receptor density for two independent experiments. Using Graphpad prism software, EC₅₀ values and 95% confidence intervals (CI) were determined by nonlinear regression analysis (sigmoidal 4PL, R² values are indicated). Significant differences between EC₅₀ values (one-way ANOVA with Tukey’s multiple comparisons test; *, p<0.05; **, p<0.01) are indicated.

Fig 7. Virion concentration-dependent interaction with receptor-coated surfaces.

Streptavidin sensors loaded to saturation with biotinylated 3’S(LN)₃ (A) or 3’S(LN)₂-PAA (B) were incubated with SeV at different concentrations for 100 mins. (C) The observed initial binding rates (nm/min) of SeV were plotted against relative virus concentration (100% corresponds to 2.0x10¹⁰particles/ml) and linear regression analysis was performed. (D) The area under the curve (AUC; from start to peak) was determined at those virus concentrations that displayed peak binding values within this time frame. Means and standard deviations of two independent replicate experiments using different virus stocks are shown.

Fig 8. Characterization of receptor-surface modification by viruses using MAL I and ECA lectins.

Different concentrations of SeV, NDV or hPIV3 virus particles were allowed to interact with (A) 3’S(LN)₂-PAA- or (B) 3’S(LN)₃- loaded sensors for 30 min (NDV and SeV) or 1h (hPIV3), followed by sensor regeneration in 10mM Tris/Glycine buffer (pH2). Presence of remaining sialoglycans was probed by analyzing binding with MAL I (left Y-axes; blue data) or ECA (right Y axes; green data).The highest virus concentration used (100%) corresponds to 2x10¹⁰ particles/ml.

Fig 9. Sialidase-(in)dependent NDV virion release.

The streptavidin sensors were loaded to saturation with 3’S(LN)₂-PAA. (A) NDV virions were associated with 3’S(LN)₂-PAA in presence of BCX2798. (B and C) After association to 3’S(LN)₂-PAA, sensors were dipped into PBS to allow NDV virions to dissociate from the sensor. Panels C is the zoom-in of B in the first 10 seconds. (D) Different concentrations of NDV were associated with 3’S(LN)₂-PAA receptors in the presence of BCX2798. (E) Subsequently, sensors were moved to PBS in the absence or presence of BCX2798 to monitor the sialidase-driven virion release. (F) Relative sialidase-dependent dissociation curves of NDV virions. Virion dissociation observed in (E) was normalized to the virion binding level observed in (D).

Fig 10. pH effect in SeV and NDV association and dissociation to 3’S(LN)₂-PAA.

(A) 3’S(LN)₂-PAA-loaded sensors were incubated with SeV at pH5.6 or 7.0 for 30 min. (B) 3’S(LN)₂-PAA-loaded sensors were incubated at pH 5.6 or 7.0 with SeV after which sensors were incubated at pH 5.6 or 7.0 in the absence of free virions. (C) 3’S(LN)₂-PAA-loaded sensors were incubated with NDV at pH5.6 or 7.0 in the absence or presence of BCX2798. (D) 3’S(LN)₂-PAA-loaded sensors were incubated with NDV at pH5.6 or 7.0 in the presence of BCX2798. Subsequently, sensors incubated in the absence of BCX2798, allowing sialidase-driven virion dissociation from the sensor.

Fig 11. Model for sialidase-driven paramyxovirus particle motility on a receptor-coated surface.

Multiple low affinity HN-Sia interactions collectively cause high-avidity binding of paramyxovirions to receptor-coated surfaces. The individual HN-Sia interactions occur via the catalytic site and if present via a 2SBS. These interactions are highly dynamic resulting from low affinity and from cleavage of the sialoglycan receptor. Highly dynamic HN-sialoglycan interactions combined with receptor-destroying activity results in “burnt-bridge”/lawnmower motion [–54], which is directed by successive cleavage of surface-bound sialoglycans, promoting motion towards unvisited substrate.