Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2022 Aug 24;126(36):6802–6810. doi: 10.1021/acs.jpcb.2c03830

Viral Size Modulates Sendai Virus Binding to Cholesterol-Stabilized Receptor Nanoclusters

Amy Lam 1, Daniel S Yuan 1, Samir H Ahmed 1, Robert J Rawle 1,*
PMCID: PMC9484459  PMID: 36001793

Abstract

graphic file with name jp2c03830_0005.jpg

Binding to the host membrane is the initial infection step for animal viruses. Sendai virus (SeV), the model respirovirus studied here, utilizes sialic-acid-conjugated glycoproteins and glycolipids as receptors for binding. In a previous report studying single virus binding to supported lipid bilayers (SLBs), we found a puzzling mechanistic difference between the binding of SeV and influenza A virus (strain X31, IAVX31). Both viruses use similar receptors and exhibit similar cooperative binding behavior, but whereas IAVX31 binding was altered by SLB cholesterol concentration, which can stabilize receptor nanoclusters, SeV was not. Here, we propose that differences in viral size distributions can explain this discrepancy; viral size could alter the number of virus–receptor interactions in the contact area and, therefore, the sensitivity to receptor nanoclusters. To test this, we compared the dependence of SeV binding on SLB cholesterol concentration between size-filtered and unfiltered SeV. At high receptor density, the unfiltered virus showed little dependence, but the size-filtered virus exhibited a linear cholesterol dependence, similar to IAVX31. However, at low receptor densities, the unfiltered virus did exhibit a cholesterol dependence, indicating that receptor nanoclusters enhance viral binding only when the number of potential virus–receptor interactions is small enough. We also studied the influence of viral size and receptor nanoclusters on viral mobility following binding. Whereas differences in viral size greatly influenced mobility, the effect of receptor nanoclusters on mobility was small. Together, our results highlight the mechanistic salience of both the distribution of viral sizes and the lateral distribution of receptors in a viral infection.

1. Introduction

Virus binding to a receptor or attachment factor on the host cell plasma membrane is the initial step of infection for all animal viruses. As such, viruses have evolved the ability to recognize receptors on cells which are permissive for infection, and evolutionary pressures carefully balance the virus-receptor binding affinity and other factors which could influence this critical step.1,2

Sendai virus (SeV), formally murine respirovirus, is the model virus used in this report to study biophysical questions of viral binding. SeV is the prototypical member of the respiroviruses, a genus of the Paramyxoviridae family which includes human parainfluenza viruses 1 and 3.3,4 As with other closely related paramyxoviruses, SeV utilizes sialic acid-conjugated glycoproteins and glycolipids as receptors for cell entry, and these are bound by the viral HN receptor binding protein, embedded in the viral envelope.5−8 Receptor binding is the trigger for viral membrane fusion, which is catalyzed by the viral F protein, allosterically triggered by the HN protein following receptor binding.9

Previously, we reported a platform to study single virus binding of SeV to glass supported lipid bilayers (SLBs), using sialic-acid-terminated gangliosides in the SLB as receptors for viral binding.10 Importantly, these measurements were carried out under environmental conditions which minimize membrane fusion so that viral binding could be studied in isolation. Using that platform, we demonstrated that SeV exhibited cooperative binding behavior with respect to receptor density in the SLB. Interestingly, this cooperative binding behavior was quantitatively quite similar to that of influenza A virus (IAVX31, strain X-31, A/Aichi/68, H3N2) which can also utilize gangliosides as receptors for binding, but which otherwise has many functional and structural differences from SeV.11 For both SeV and IAVX31, this cooperative binding behavior was consistent with many weak binding interactions being necessary for stable viral binding; in other words, avidity rather than affinity dominates the binding interaction.

The previous influenza study11 had also reported that a ∼2.3-fold increase in binding was observed with an increasing concentration of cholesterol from 0 to 40% in the target SLB, with the density of ganglioside GD1a receptor held constant. Using molecular dynamics (MD) simulations, the authors explained this increase in binding as due to the stabilization of nanoclusters of GD1a receptors, which were stabilized by the higher concentrations of cholesterol and which could serve as hotspots for viral binding. Although the MD simulation results, including the proposed mechanism of cholesterol-mediated stabilization, were particular to the lipid compositions used in that report, such clustering is consistent with experimentally observed self-aggregation and clustering of gangliosides in model lipid membranes,12−17 as well as with the observation of nanodomains of sialic acid in cell membranes, shown to be important for influenza binding.18 In contrast, in our recent SeV report,10 we surprisingly observed that SeV exhibited no sensitivity to cholesterol in the target SLB; binding to SLBs containing 2% GD1a was identical both at low or high cholesterol concentration in the target, even though otherwise identical SLB lipid compositions were used as in the previous influenza study.11

This discrepancy was puzzling. How could two viruses that bind to the same viral receptor (GD1a) and that exhibit very similar cooperative binding behavior have such a markedly different response to cholesterol (and to receptor nanoclusters) in the target membrane? One hypothesis was that a difference in virus size might explain the different response, as SeV is on average larger than IAVX31. Electron microscopy data has reported SeV outer diameters ranging up to 400 nm, mean 165 nm,10,19 whereas IAVX31 only ranges up to 170 nm, mean 120 nm.20 Larger viral particles would presumably have a larger contact area between virus and target, increasing the number of glycoprotein–receptor interactions. This could render the virus less sensitive to receptor nanoclusters, as a sufficient number of glycoprotein–receptor contacts to sustain binding might still be achieved in their absence. A simple back of the envelope calculation illustrates this. For example, at 2% GD1a in the target SLB, a small virus with a contact area of 15 nm in diameter would encounter on average only five GD1a receptors in the virus–target interface. By contrast, a larger virus with a 50 nm diameter contact area would encounter on average 56 receptors (see SI for details on this calculation). Given that multiple low affinity glycoprotein–receptor binding interactions are needed for stable virus binding, differences in viral size might explain the discrepancy in the two data sets.

In this report, therefore, we explore the hypothesis that viral size (more accurately, virus-target contact area) can modulate the sensitivity of virus binding to receptor nanoclusters, stabilized by increasing concentrations of cholesterol. To examine this hypothesis, we generated a smaller size distribution of SeV by size-filtering and compared the single virus binding behavior of the size-filtered virus with unfiltered virus. We observed that whereas the unfiltered virus exhibited no sensitivity toward SLB cholesterol concentration at 2% GD1a, the size-filtered virus exhibited a very similar dependence to cholesterol as influenza A virus. We also observed that at lower receptor densities in the target SLB, unfiltered SeV did demonstrate a cholesterol sensitivity consistent with the idea that sensitivity to receptor nanoclusters for the larger virions becomes apparent only with fewer available receptors on average in the contact interface. We also explored the influence of virus size and SLB cholesterol concentration on viral mobility via “rolling” diffusion following binding, finding that whereas viral size yielded a large difference in viral rolling diffusion behavior, the presence of receptor nanoclusters stabilized by cholesterol only marginally altered viral diffusion for the size-filtered virus. Taken together, our results underscore the physical importance of the distribution of viral sizes as an evolutionary feature relevant to receptor binding and infection.21

2. Experimental Methods

2.1. Materials

Dioleoylphosphatidylethanolamine (DOPE), palmitoyloleoylphosphatidylcholine (POPC), and cholesterol (Chol), as well as ganglioside GD1a (from porcine brain), were purchased from Avanti Polar Lipids (Alabaster, AL). Oregon Green-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (OG-DHPE) and Texas Red-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (TR-DHPE) were obtained from Thermo Fisher Scientific (Waltham, MA, U.S.A.). Polydimethylsiloxane (PDMS) Sylgard 184 elastomer base and curing agent were purchased from Ellsworth Adhesives (Germantown, WI, U.S.A.). Sendai virus (purified Sendai Cantell Strain, egg-grown, batch 960216) was obtained from Charles River Laboratories (Wilmington, MA, U.S.A.) and handled according to a BSL-2 protocol at Williams College. Chloroform, methanol, and buffer salts were obtained from Fisher Scientific (Pittsburgh, PA) and Sigma-Aldrich (St. Louis, MO, U.S.A.). Formvar/carbon-coated square mesh grids (Cu, 200 Mesh, SB) were purchased from Electron Microscopy Sciences (Hatfield, PA, U.S.A.). Mouse anti-HN (1A6) IgG2a antibody was purchased from Kerafast Inc. (Boston, MA) and produced in the laboratory of Prof. Benhur Lee (Mt. Sinai). Characterization of the 1A6 antibody has been reported previously.10,22

2.2. Buffer Definitions

Reaction buffer (RB) = 10 mM NaH2PO4, 90 mM sodium citrate, 150 mM NaCl, pH 7.4. HEPES Buffer (HB) = 20 mM HEPES, 150 mM NaCl, pH 7.2. All buffers were filtered at 0.22 μm pore size.

2.3. Fluorescence Microscopy

Fluorescence microscopy was carried out using a Zeiss Axio Observer 3 microscope with a 63× oil immersion objective, NA = 1.4 (Carl Zeiss Microscopy, LLC., White Plains, NY), a Lumencor Spectra III, LED Light Engine, and a Hamamatsu ORCA Flash 4.0 V2 Digital CMOS camera (Hamamatsu Photonics K.K., Hamamatsu City, Japan). The microscope was operated using μmanager software.23 Additional microscopy settings are given in the Supporting Information.

2.4. Sendai Virus Labeling and Size-Filtering

For reliable single virus measurements, proper virus labeling and validation that labeling does not perturb the viral behavior are critical.10,24 Sendai virus labeling and validation were performed essentially as described previously, using dye concentrations shown not to perturb viral binding activity.10 Briefly, 4 μL of TR-DHPE (0.75 g/L in ethanol) was mixed with 240 μL of HB buffer to create a dye–buffer mixture. A total of 15 μL of Sendai virus (2 mg/mL viral protein) was mixed well with 60 μL of the dye–buffer mixture and incubated at room temperature for 2 h. To remove unincorporated dye, 1300 μL of HB was added, and the solution was centrifuged at 21K × g at 4 °C for 50 min. The pellet was resuspended in 100 μL of HEPES buffer, and the supernatant was discarded. To prepare the size-filtered virus for binding assays, at least 30 μL of fluorescently labeled Sendai virus was passed through a 0.22 μm low-binding, small volume syringe filter (4 mm Durapore PVDF Membrane, Millipore, Cork, Ireland). Total viral protein content by BCA assay was used to assess viral loss due to filtering; in typical viral preps, the viral suspension before filtering contained ∼2× higher total protein content than after filtering.

2.5. Single Virus Binding Measurements to SLBs

As reported previously,10 single virus binding assays were performed to SLBs inside microfluidic devices. Methods for SLB formation and construction of microfluidic flow cells have been described previously,10,11,25 and are briefly elaborated in the Supporting Information. The microfluidic device was a PDMS flow cell plasma bonded to a glass coverslip, and consisted of 2 parallel channels (channel dimensions = 2.5 mm × 13 mm × 70 μm) with inlet and outlet holes (2.5 mm diameter) for sample exchange. The volume of each channel in the assembled device was ∼4 μL. SLBs were formed in each channel by the vesicle fusion method.26,27 The lipid composition of all SLBs was 20 mol % DOPE, 0–30% cholesterol (as indicated), 0.05% Oregon Green-DHPE, 0.5–2% GD1a (as indicated), and the remaining amount (47.95–77.95%) POPC.

Following SLB formation, 4 μL of labeled virus (typical concentration = 0.1 nM as estimated by viral protein concentration10) was added to the flow cell channel and pulled through by pipet. Virus was incubated in the channel for 15 min at room temperature. The flow cell was then rinsed with ∼1 mL RB by a Fusion 200 syringe pump (Chemyx Inc., Stafford, TX, U.S.A.) at 0.8 mL/min. Images of bound virions and SLB were then collected in the center region of the flow cell, typically in ≥10 areas. The number of viral spots in each image were quantified using custom-built Matlab scripts,10,28,29 which automatically exclude large or irregularly shaped spots. The version of these scripts used for this report is available at https://github.com/rawlelab/SendaiBindingAnalysis.

2.6. Immunofluorescence (IF) Measurements

For IF measurements of SeV bound nonspecifically to the glass coverslip, 4 μL of labeled virus sample was added to a freshly prepared microfluidic flow cell, and incubated for 30 min at room temperature. The flow cell was rinsed by syringe pump with ∼1 mL of HB, and then 4 μL of 30 g/L bovine serum albumin was introduced and incubated for 15 min. For IF measurements of SeV bound to SLBs, virus binding to SLBs was accomplished as above.

For both types of samples, 4 μL of 20 μg/mL anti-HN (1A6) was then introduced into the flow cell, and incubated for 15 min. The flow cell was rinsed with ∼2 mL of HB and then 4 μL of goat antimouse IgG-Alexa 488 at 2 μg/mL was introduced and incubated for 15 min. The flow cell was rinsed again with ∼3 mL of HB, and then sequential images of TR-labeled viral particles and Alexa 488-antibody were observed by fluorescence microscopy. Co-localization was quantified using the same previously published Matlab scripts described above.

2.7. Single Particle Tracking (SPT) and Analysis

For single particle tracking, virus binding to SLBs was performed as above, and then fluorescence micrographs of the bound virus were taken every 5 s for 25 min. Single particle tracking analysis was accomplished using the TrackMate30 plug-in of Fiji,31 largely as described previously.10 Briefly, single particle localization in each image was carried out using a Laplacian of Gaussian (LoG) detector, σ = 3.0 μm. The HyperStack displayer was then used for manual editing and filtering of spots by mean intensity (>420–445) or spot quality (∼10). Next, tracks of single particles over time were compiled using the simple Linear Assignment Problem tracker with linking max distance = 8.0 μm, gap-closing max distance = 8.0 μm, and gap-closing max frame gap = 3.0 frames. Tracks were filtered by the number of time points per track (>100–200 points per track) and manual editing was then used to weed out erroneous tracks (typically those incorrectly assigned to jump back-and-forth between two adjacent immobile particles). Mean squared displacement (MSD) versus Δt curves were then compiled and were fit to a linear model to extract the diffusion coefficients using a published algorithm to determine the optimal number of points to include in the fit.32 The Matlab code of our implementation of this algorithm is available at https://github.com/rawlelab/SendaiBindingAnalysis. The lower limit of detectable diffusion (1.3 × 10–5 μm2/s) was determined by calculating the average “diffusion coefficient” of immobile virions bound directly to a glass surface, using the analysis above.

2.8. Negative Stain Transmission Electron Microscopy (TEM)

To assess the effects of size-filtering by negative stain TEM, a purified SeV sample as received from the manufacturer was split into two aliquots, one was size-filtered and the other was not. Size-filtering was accomplished using a 0.22 μm syringe filter as described above. SeV samples were then prepared for negative-stain TEM using the side blotting method,33,34 as previously described.10 Briefly, 5 μL of SeV sample was pipetted onto a Formvar/carbon-coated square mesh grid, and adsorbed for 1 min. The grid was then rinsed 3x by touching the coated side of the grid to a 50 μL drop of ultrapure water on clean parafilm, with excess liquid blotted onto filter paper. The samples were then stained 2X for 30 s with 50 μL of a 1% uranyl acetate solution, with excess stain blotted onto filter paper. Stained grids were air-dried for at least 1 h, and then stored at room temperature in the dark. Samples were later imaged with a JEOL JEM-1400 Plus transmission electron microscope operating at an acceleration voltage of 120 kV, and equipped with a Gatan Orius SC1000A charge-coupled device (CCD). Outer diameters of the pleiomorphic, roughly spherical SeV particles were measured at the widest diameter using the measurement tool of the Fiji software.31

2.9. Dynamic Light Scattering (DLS)

DLS was used to validate trends in the TEM data and assess the extent of viral clumping. A purified SeV sample as received from the manufacturer was split into two aliquots. Each was diluted to ∼500 μg/mL viral protein using HEPES buffer; one was subsequently size-filtered at 0.22 μm pore size as above, the other was not. Additional HEPES buffer was added to the size-filtered virus to compensate for volume loss during filtration and achieve the minimum volume requirements for analysis. All HEPES buffer used for dilution was filtered at 0.22 μm pore size, and centrifuged briefly using a tabletop mini-centrifuge to ensure the removal of any large dust particles. A total of 75 μL of each sample was analyzed in a sealed cuvette using a DynaPro NanoStar DLS (Wyatt Technologies, Santa Barbara, CA) with a temperature control set at 20 °C. A total of 10 acquisitions were averaged and analyzed using the Hollow Sphere model in the DYNAMICS software package (Wyatt Technologies).

3. Results and Discussion

To determine whether viral size could modulate the sensitivity of Sendai virus binding to supported lipid bilayers with different concentrations of cholesterol, we first produced and characterized virus with a smaller size distribution. These results are discussed below.

3.1. Size-Filtering of SeV Produces a Substantially Shifted Size Distribution

As we and others have previously reported,10,19 Sendai virus particles are observed by negative stain TEM to be roughly spherical, pleomorphic virions, with a reasonably broad size distribution–diameters ranging from ∼80 nm up to 400 nm or more. To produce a smaller size distribution, we therefore size-filtered SeV using a standard 0.22 μm low-binding syringe filter, and characterized the resulting particles by negative stain TEM. We observed that size-filtered virions were morphologically similar to unfiltered virions; both were observed to be roughly spherical and pleomorphic. However, the size distribution of the filtered virus had shifted substantially, even below the nominal pore size cutoff for the syringe filter (Figure 1). The difference between the unfiltered virus distribution (Figure 1A) and the size-filtered virus distribution (Figure 1B) was statistically significant (p < 2.5 × 10–7, Kolmogorov-Smirnov two-sided test), as was the difference between the size-filtered virus and the unfiltered virus distribution which had been mathematically truncated at the nominal cutoff value (lighter blue bars in Figure 1A, p < 0.005). This indicates that size-filtering does not merely remove particles above the pore size cutoff, but particles which approach the pore size are increasingly likely to be filtered out.

Figure 1.

Figure 1

Open in a new tab

Distribution of viral particle sizes of (A) unfiltered and (B) size-filtered Sendai virus as measured by negative stain TEM. Sendai virus was size-filtered using a 0.22 μm low binding syringe filter, and particle outer diameters were measured by negative-stain TEM. Dashed line indicates the nominal filter pore size; dark blue bars in panel A indicate the fraction above the nominal filter pore size. Inset shows descriptive statistics for each condition.

Consistent with that explanation, we observed a large ∼2× reduction in the viral protein concentration determined by BCA assay before and after filtration, much larger than would be expected merely by removing the fraction of viral particles above the filter size cutoff (darker blue bars in Figure 1A). This reduction in total protein concentration suggested that some fraction of the virus may be clustered prior to filtration, and that those clustered particles are also removed during size-filtering. Indeed, clusters of viral particles were occasionally observed in the TEM images of unfiltered virus (see example in Figure S1), but rarely in the images of size-filtered virus.

To observe the extent of clustering, as well as to rule out sample preparation artifacts from the negative stain process, we also performed DLS analysis of size-filtered and unfiltered SeV (Figure S2). As expected, the average hydrodynamic diameter measured by DLS intensity was larger than the dried particle diameter measured by negative stain TEM for comparable samples, but the trends were consistent, the distribution of size-filtered virus was substantially smaller than for unfiltered virus (mean diameter ≈190 nm vs 290 nm for size-filtered and unfiltered virus, respectively). Additionally, the distribution of unfiltered SeV was multimodal, with a secondary peak at much larger mean diameter (several μm), consistent with viral clustering.

Coincidentally, the size distribution of the size-filtered SeV measured by TEM was reasonably close to that of IAVX31 determined by electron microscopy in a prior report.20 The mean outer diameter of IAVX31 particles was 120 nm and that of size-filtered SeV particles was 129 nm, compared to unfiltered SeV particles with a mean outer diameter of 167 nm.

It is also important to note that the observed size distributions of single virions indicate that except for the largest particles at the tail end of the distribution, all virions are below the diffraction limit for fluorescence microscopy. Thus, in the fluorescence microscopy measurements described below, size information on the observed viruses cannot be determined.

3.2. Size-Filtering of SeV Minimally Perturbs Viral Binding Behavior

Size-filtering has the potential to perturb viral proteins or other aspects of the virus that would not be apparent by TEM imaging. Such perturbations could impair normal viral binding behavior. To determine whether viral binding behavior had been perturbed by size-filtering, we performed standard validation measurements.

First, we used our single virus binding platform10 to determine whether size-filtered SeV showed similar binding sensitivity toward the GD1a receptor in SLBs as did unfiltered SeV. We observed very similar binding behavior for both unfiltered and size-filtered virus, with very little binding to SLBs without receptor relative to those with 2% GD1a (Figure 2A,B). We also performed a side-by-side comparison of the binding activity of size-filtered vs unfiltered virus to 2% GD1a SLBs, matching the total viral protein concentration in each measurement (Figure S3). Under these conditions, increased binding of size-filtered virus is to be expected, as matched protein concentrations between the conditions will yield a higher viral particle concentration for the size-filtered virus given its smaller size distribution. And indeed, we observed that size-filtered virus exhibited substantially higher binding compared to unfiltered virus. These results also suggest that the larger viral clusters removed by size-filtering may be binding-inactive (Figure S2). In either case, the results indicate that the size-filtered virus continues to exhibit robust binding activity.

Figure 2.

Figure 2

Open in a new tab

Size-filtering SeV minimally perturbs viral binding behavior. Sendai virus was size-filtered using a 0.22 μm low binding syringe filter. (A) and (B) depict the relative number of virions bound to SLBs with 10% cholesterol and either 2% GD1a or no receptor for (A) unfiltered and (B) filtered SeV, respectively. Binding is calculated relative to 2% GD1a for each panel. Error bars are ± standard error of ≥3 sample replicates, and ≥10 separate image locations within each sample, with propagated relative error as described.10 (C) shows the fraction of bound spots with positive immunofluorescence (IF) signal after labeling by anti-HN (1A6) antibody and Alexa 488-labeled secondary antibody. Particles were bound either to SLBs with 2% GD1a and 10% cholesterol (2% GD1a SLB), or were attached nonspecifically to the glass coverslip instead of an SLB, followed by surface passivation by 30 g/L bovine serum albumin (no SLB). Colocalization between Alexa 488 and the Texas Red-DHPE membrane label was used to determine whether a particle was IF-positive. Error bars are ± standard error of ≥2 sample replicates, with ≥10 separate image locations in each sample.

Finally, we performed immunofluorescence (IF) measurements of size-filtered and unfiltered viruses, observing the fraction of labeled viral particles which were bound by an HN-specific antibody (1A6). Previously, we demonstrated that the percentage of particles that are IF-positive is substantially enriched upon binding to receptors in the target SLB, relative to particles bound nonspecifically to a glass coverslip.10 These results indicated that other particles, such as extracellular vesicles, quasi-viral particles, and virions with conformationally inactive epitopes, are copurified with active viral particles and labeled with the TR membrane dye, but do not bind well to ganglioside receptors in an SLB. Such purification challenges are not unique to Sendai virus, and this highlights the need for careful IF validation of single virus experiments. We therefore compared IF measurements of size-filtered SeV and unfiltered SeV (Figure 2C). Similar results were observed in both cases; ∼30–40% of nonspecifically bound particles were IF-positive and ∼90% of particles were IF-positive following receptor binding.

Taken together, these validation measurements indicate that filtering has produced little or no perturbation in the viral binding behavior of size-filtered SeV relative to the unfiltered virus.

3.3. Size-Filtered Virus Shows Binding Sensitivity toward Cholesterol in SLB

To determine whether a difference in viral size could produce a different binding sensitivity to cholesterol concentration in the target membrane, we compared viral binding between size-filtered and unfiltered SeV to SLBs with cholesterol concentrations ranging from 0% to 30%, keeping the GD1a receptor concentration constant at 2% (Figure 3). We observed that whereas unfiltered SeV exhibited little binding dependence on SLB cholesterol concentration, size-filtered SeV displayed a linear relationship, with relative binding increasing ∼2× from 0 to 30% cholesterol in the SLB.

Figure 3.

Figure 3

Open in a new tab

Size-filtered SeV exhibits cholesterol-dependent binding, similar to influenza A X-31 virus. Single virus binding experiments to SLBs with 2% GD1a and cholesterol concentrations ranging from 0% to 30% were performed using SeV that was either size-filtered (orange) or unfiltered (blue). Shown is the relative number of bound virions, calculated relative to 0% cholesterol for each virus. IAVX31 data (green) is replotted from ref (11). Whereas unfiltered SeV exhibited no cholesterol sensitive binding, size-filtered SeV showed a linear dependence on SLB cholesterol concentration similar to IAVX31. Dashed lines are linear fits to each data set. SeV error bars are ± standard error of ≥3 sample replicates, and ≥10 separate image locations within each sample, with a propagated relative error as described.10

As discussed in the Introduction, molecular dynamics modeling has indicated that increasing SLB cholesterol concentration in the lipid compositions used here can stabilize nanoclusters of receptors.11 According to the thermodynamic model developed in that report, cholesterol stabilizes ganglioside clustering by ordering lipid tails, thereby decreasing the entropic penalty of association, otherwise favored by hydrogen bonding between sugar head groups.12 In the context of multiple, cooperative virus-receptor binding events being required for stable viral binding, these receptor nanoclusters can therefore serve as hotspots for virus binding. With this in mind, the results in Figure 3 support the hypothesis that the size of the incoming viral particle directly influences whether these receptor nanoclusters will or will not stabilize viral binding. As hypothesized, larger SeV particles would have a larger contact area between virus and target, increasing the number of HN-receptor interactions. This would render the virus less sensitive to receptor nanoclusters, as a sufficient number of HN-receptor contacts to sustain binding might still be achieved in their absence.

Given this, it seems contradictory at first glance that such little cholesterol sensitivity is observed for unfiltered SeV in Figure 3, even though some fraction of the unfiltered virions are presumably small enough to exhibit cholesterol sensitivity. If, however, the “cholesterol sensitive” viruses are a small enough fraction of the virions that bind to the SLB, then the observed extent of cholesterol sensitivity of the entire population of the unfiltered virions would be difficult to measure above experimental noise. Only when the larger virions are removed by size-filtering and the proportion of “cholesterol sensitive” viruses is substantially enriched does the cholesterol sensitive behavior become apparent. From our data, it is not clear what size of virus constitutes the “cholesterol sensitive” fraction, but for illustration, ∼40% of the size-filtered viruses observed by TEM have an outer diameter <110 nm, whereas only ∼15% of the unfiltered virus size distribution is <110 nm (Figure 1). Additionally, the existence of viral clusters suggested by both the TEM and DLS data (see Section 3.1 above) could reduce the fraction of “cholesterol sensitive” viruses even further. This could occur for two reasons. (1) Clustering could inactivate some “cholesterol sensitive” viruses; this likely occurs for the larger clusters identified in the DLS data (Figure S2), which are large enough to have been easily identified by fluorescence microscopy, but which were rarely observed to bind, and those rare events were excluded from analysis. (2) The binding behavior of a cluster that includes “cholesterol sensitive” viruses would resemble that of a larger “cholesterol insensitive” virion. This seems a likely explanation for some fraction of the smaller clusters that fall below or near the diffraction limit and that would not have been distinguishable from single virions by fluorescence microscopy.

Interestingly, the results for the size-filtered SeV in Figure 3 are remarkably similar to those previously observed for IAVX31, which yielded a ∼2.3-fold linear increase in binding from 0 to 40% cholesterol in the target SLB.11 Given the smaller size, on average, for IAVX31 compared to unfiltered SeV, and the more comparable size between IAVX31 and size-filtered SeV, this further underscores the importance of viral size as an important factor in determining sensitivity of binding to target cholesterol concentration.

3.4. Unfiltered Virus Shows Binding Sensitivity toward Cholesterol in SLB at Low Receptor Density

If binding sensitivity to cholesterol (and to receptor nanoclusters) is indeed governed by the number of HN–receptor interactions that occur in the virus–target contact interface, it should be possible to manipulate the viral sensitivity to cholesterol either by altering the viral size (i.e., contact area) as above or by lowering the density of receptors in the SLB. We therefore asked whether unfiltered SeV would exhibit a sensitivity to SLB cholesterol concentration at low receptor density, where fewer HN–receptor contacts occur on average. Under these conditions, stable binding is less readily achieved even with a larger contact area, and therefore the presence or absence of receptor nanoclusters may influence the likelihood of stable binding even for larger viruses. Previously, we demonstrated that SeV exhibits a cooperative binding dependence with respect to receptor density, and for GD1a, maximum binding occurs at ≥2 mol % in the SLB.10 Therefore, to test this hypothesis, we measured viral binding of unfiltered SeV at 0% and 30% SLB cholesterol concentration, with GD1a receptor concentrations ranging from 0.5% to 2% (Figure 4A–D), corresponding to 0.2× to 1× maximum viral binding, see ref (10). We then compared the fold increase in viral binding at 30% cholesterol versus 0% cholesterol across the different receptor concentrations (Figure 4E).

Figure 4.

Figure 4

Open in a new tab

Unfiltered SeV exhibits cholesterol-dependent binding at low receptor densities. Unfiltered SeV was bound to SLBs with either 0% or 30% cholesterol and GD1a receptor densities varying from 0.5% to 2%. (A)–(D) shows numbers of bound virus per field of view for representative measurements at 0.5%, 0.75%, 1%, and 2% GD1a concentration, respectively. Error bars are ± standard error of the mean of ≥3 sample replicates, with sample averages calculated from ≥10 separate image locations within each sample. (E) depicts the fold increase of bound virions at 30% cholesterol versus 0% cholesterol for each GD1a concentration. Dashed line is drawn at fold increase = 1, indicating no sensitivity of binding to SLB cholesterol concentration. At ≤ 1% GD1a, unfiltered virus exhibited cholesterol-dependent binding (fold increase > 1), which was not apparent at 2% GD1a. Error bars are ±standard error of ≥3 sample replicates and ≥10 separate image locations within each sample, with propagated relative error as described.10

We observed that unfiltered SeV did exhibit a sensitivity to SLB cholesterol concentration at lower receptor densities, and that the sensitivity diminished and ultimately disappeared as the receptor density increased (Figure 4E). At 0.5% and 0.75% GD1a, large differences (>3×) in binding were observed between SLBs with 0% or 30% cholesterol. At 1% GD1a, only a small difference was observed (<2×), and at 2% GD1a the difference had disappeared. Below 0.5% GD1a, very few viruses bound to either 0% or 30% cholesterol SLBs, and so a comparison did not prove meaningful.

Together, this data provides further support for the mechanistic explanation that cholesterol concentration can stabilize the formation of receptor nanoclusters, which in turn can stabilize viral binding. However, this stabilization of viral binding by receptor nanoclusters only becomes apparent when the number of potential receptor interactions at the virus–target interface is small enough, either because the contact area itself is physically small (such as is the case for size-filtered SeV at 2% GD1a in Figure 3), or because the average density of receptors in the target is small enough (such as is the case for unfiltered SeV at lower concentrations of GD1a in Figure 4).

3.5. Influence of Viral Size and SLB Cholesterol Concentration on Viral Mobility Following Binding

Viral contact area with the target membrane, the presence and stability of receptor nanoclusters and the number of glycoprotein–receptor interactions, all have the potential to influence not only viral binding, but also the diffusive mobility of the virus following binding, another important feature of the infection process.1 Previously, we presented evidence that SeV exhibits diffusion on SLBs following a rolling mechanism,10 where stochastic binding and unbinding of individual, weak HN-receptor binding interactions leads to Brownian diffusion as the virus “rolls” on the SLB. To determine the influence that viral contact area and the stability of receptor nanoclusters can exert on this rolling diffusion, we measured the diffusion of size-filtered versus unfiltered SeV at low (10%) and high (30%) SLB cholesterol concentrations, at 2% GD1a. Virus diffusion was measured by single particle tracking, and histograms of diffusion coefficients were compiled from hundreds of tracked particles (Figure S4). As we have reported previously, we observed that all histograms were tailed, with a substantial fraction of particles immobile (D < 1.3 × 10–5 μm2/s, the lower limit of detectable diffusion in our measurement).

As might be expected, we observed that size-filtered virus exhibited significantly faster diffusion than unfiltered virus (Table 1). Median diffusion coefficients of the size-filtered virus at both low and high SLB cholesterol concentrations were 7–13× larger than for the unfiltered virus (p < 0.0002 or lower for all comparisons, determined by bootstrap resampling of each distribution). This supports the reasonable conclusion that as the virus-target contact area decreases, the rolling mobility increases, presumably due to fewer HN–receptor interactions. Similarly, the fraction of immobile particles was significantly larger for the unfiltered virus than for the size-filtered virus (∼0.3 vs ∼0.2, p < 0.007 or lower for all comparisons, by bootstrap resampling). This suggests that at least some of the immobile particles are viruses bound to a large enough number of receptors such that no diffusion is observed, even if individual HN are stochastically binding and unbinding at any given time. As the viral size (and virus-target contact area) increases on average, the number of such immobile particles would be expected to increase.

Table 1. Mobility of Size-Filtered and Unfiltered SeV at Different SLB Cholesterol Concentrations.

virus preparation SLB chol conc (mol %) median Da(μm2/s) fraction immobilea,b sample size (N)
unfiltered 10 5 × 10–4 ± 2 × 10–4 0.30 ± 0.03 277
unfiltered 30 5.5 × 10–4 ± 9 × 10–5 0.26 ± 0.02 873
size-filtered 10 6 × 10–3 ± 1 × 10–3 0.18 ± 0.02 250
size-filtered 30 3.8 × 10–3 ± 5 × 10–4 0.18 ± 0.02 589
Open in a new tab
a

Error values are ± standard deviation, determined by bootstrap resampling of the distribution of diffusion coefficients (10000 bootstraps).

b

Immobile particles were those with D < 1.3 × 10–5 μm2/s, the lower limit of detectable diffusion in our measurement.

In comparing between the unfiltered virus bound to SLBs with low vs high cholesterol concentration (Table 1), we observed little difference between either the median diffusion coefficients or the fraction of immobile particles (pmedian = 0.85 and pfraction_mmobile = 0.20, respectively, by bootstrap resampling). These diffusion results are consistent with the similar mechanistic explanation for the binding results for the unfiltered virus at 2% GD1a (see Figure 3 above), namely, the number of HN–receptor interactions is large enough that the increased stability of receptor nanoclusters at high cholesterol concentration does not strongly influence either the binding behavior or the resulting mobility.

On the other hand, in comparing between size-filtered virus at low versus high SLB cholesterol concentration (Table 1), a small ∼1.7× difference was observed between the median diffusion coefficients (pmedian = 0.07, by bootstrap resampling). This is suggestive that the mobility of these (on average) smaller viruses are more sensitive to the presence and stability of receptor nanoclusters. As receptor nanoclusters become stabilized at high SLB cholesterol concentration, the number of HN–receptor contacts becomes high enough to inhibit rolling diffusion somewhat. However, it is difficult to infer too much mechanistic information from this comparison, as it is likely that different size distributions of virus are bound at the different cholesterol concentrations. At high cholesterol concentration, the binding of viruses at the smaller end of the distribution may be more readily achieved, and these smaller viruses would be expected to be more mobile than their larger counterparts, as described above. As a population, this would compensate to some degree for the decreased mobility of the bound viruses due to receptor nanocluster stabilization.

Finally, when taken together, these mobility results support a general conclusion quite consistent with the binding results reported above: differences in viral size can influence whether or not viral mobility is sensitive to the presence and stability of receptor nanoclusters in the SLB.

4. Conclusions

In this report, we studied the role that virus size and number of HN–receptor contacts in the virus-target interface can play in altering the sensitivity of Sendai virus binding to receptor nanoclusters, which are stabilized by increasing concentrations of cholesterol in the target membrane as proposed by previous MD simulation results.11 To study the influence of virus size, we produced and characterized a smaller size distribution of SeV by size-filtering, and demonstrated that size-filtering did not substantially perturb the viral binding activity. Using the size-filtered virus, we found that whereas unfiltered SeV showed little binding sensitivity to SLB cholesterol concentration at 2% GD1a, size-filtered SeV exhibited a linear dependence of binding on SLB cholesterol concentration, quite similar to that previously reported for the X-31 strain of influenza A virus.11 This data supported the mechanistic hypothesis that sensitivity of viral binding to receptor nanoclusters is directly related to the number of HN-receptor contacts in the virus-target interface, which can be changed by altering the viral size. Further supporting this hypothesis, we demonstrated that unfiltered virus could exhibit a dependence of binding on SLB cholesterol concentration, but only at low (<1%) GD1a densities, presumably where the average number of receptors in the virus-target contact interface became low enough such that the presence and stability of receptor nanoclusters could modulate virus binding. We also studied the effects of virus size and SLB cholesterol concentration on viral mobility following binding, which we had previously identified as occurring by a “rolling” mechanism.10 We found that viral mobility was strongly dependent on virus size, and whereas the mobility of unfiltered virus remained unchanged at low or high SLB cholesterol concentration, size-filtered virus exhibited a small sensitivity to SLB cholesterol concentration, becoming moderately slower at higher cholesterol concentration, where receptor nanoclusters would be stabilized.

Together, our results reinforce the hypothesis, proposed in a prior report,11 that target membrane cholesterol concentration can influence viral binding via the stabilization of receptor nanoclusters that can serve as hotspots for binding. However, our results indicate that this stabilization of viral binding by receptor nanoclusters only becomes apparent when the number of potential receptor interactions at the virus-target interface is small enough, either because the contact area itself is physically small (such as is the case for size-filtered SeV at 2% GD1a) or because the density of receptors in the target is low enough (such as is the case for unfiltered SeV at lower concentrations of GD1a). As the likelihood of viral binding directly influences the ultimate probability of infection, these results underscore the importance of the target membrane cholesterol composition, lateral distribution of receptors, and the distribution of viral sizes as salient features to consider in studying and understanding viral infection.

Acknowledgments

The authors thank Peter Kasson (University of Virginia) and an anonymous seminar audience member at Amherst College for helpful discussions and ideas. The authors thank Abraham Park (Williams College) for help analyzing TEM images, Nancy Piatczyc (Williams College) for TEM instrument support, and Sarah Goh (Williams College) for sharing DLS instrumentation. R.J.R. acknowledges financial support from Williams College and NIH Grant R15AI171754. A.L. was supported by a Roche and Gomez Student Research Fellowship at Williams College.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcb.2c03830.

  • Supporting experimental methods, figures, and references. Representative TEM images of Sendai virus, DLS viral size distributions, diffusion coefficient distributions (PDF)

Author Contributions

A.L. designed experiments, collected and analyzed data, and helped write the manuscript. D.Y. and S.A. collected and analyzed DLS data, analyzed TEM data, and helped write the manuscript. R.J.R. designed experiments, analyzed data, acquired project funding, and wrote the manuscript.

The authors declare no competing financial interest.

Special Issue

Published as part of The Journal of Physical Chemistry virtual special issue “Steven G. Boxer Festschrift”.

Supplementary Material

jp2c03830_si_001.pdf (5.8MB, pdf)

References

  1. Boulant S.; Stanifer M.; Lozach P.-Y. Dynamics of Virus-Receptor Interactions in Virus Binding, Signaling, and Endocytosis. Viruses 2015, 7 (6), 2794–2815. 10.3390/v7062747. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Maginnis M. S. Virus-Receptor Interactions: The Key to Cellular Invasion. J. Mol. Biol. 2018, 430 (17), 2590–2611. 10.1016/j.jmb.2018.06.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Faísca P.; Desmecht D. Sendai Virus, the Mouse Parainfluenza Type 1: A Longstanding Pathogen That Remains up-to-Date. Res. Vet. Sci. 2007, 82 (1), 115–125. 10.1016/j.rvsc.2006.03.009. [DOI] [PubMed] [Google Scholar]
  4. Nakanishi M.; Otsu M. Development of Sendai Virus Vectors and Their Potential Applications in Gene Therapy and Regenerative Medicine. Curr. Gene Ther 2012, 12 (5), 410–416. 10.2174/156652312802762518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Markwell M. A.; Paulson J. C. Sendai Virus Utilizes Specific Sialyloligosaccharides as Host Cell Receptor Determinants. Proc. Natl. Acad. Sci. U. S. A 1980, 77 (10), 5693–5697. 10.1073/pnas.77.10.5693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Holmgren J.; Svennerholm L.; Elwing H.; Fredman P.; Strannegård O. Sendai Virus Receptor: Proposed Recognition Structure Based on Binding to Plastic-Adsorbed Gangliosides. Proc. Natl. Acad. Sci. U. S. A 1980, 77 (4), 1947–1950. 10.1073/pnas.77.4.1947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Kvellestad A.; Dannevig B. H.; Falk K. Isolation and Partial Characterization of a Novel Paramyxovirus from the Gills of Diseased Seawater-Reared Atlantic Salmon (Salmo Salar L.). J. Gen. Virol 2003, 84 (8), 2179–2189. 10.1099/vir.0.18962-0. [DOI] [PubMed] [Google Scholar]
  8. Suzuki Y.; Suzuki T.; Matsunaga M.; Matsumoto M. Gangliosides as Paramyxovirus Receptor. Structural Requirement of Sialo-Oligosaccharides in Receptors for Hemagglutinating Virus of Japan (Sendai Virus) and Newcastle Disease Virus. J. Biochem. (Tokyo) 1985, 97 (4), 1189–1199. 10.1093/oxfordjournals.jbchem.a135164. [DOI] [PubMed] [Google Scholar]
  9. Azarm K. D.; Lee B. Differential Features of Fusion Activation within the Paramyxoviridae. Viruses 2020, 12 (2), 161. 10.3390/v12020161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Lam A.; Kirkland O. O.; Anderson P. F.; Seetharaman N.; Vujovic D.; Thibault P. A.; Azarm K. D.; Lee B.; Rawle R. J. Single-Virus Assay Reveals Membrane Determinants and Mechanistic Features of Sendai Virus Binding. Biophys. J. 2022, 121 (6), 956–965. 10.1016/j.bpj.2022.02.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Goronzy I. N.; Rawle R. J.; Boxer S. G.; Kasson P. M. Cholesterol Enhances Influenza Binding Avidity by Controlling Nanoscale Receptor Clustering. Chem. Sci. 2018, 9 (8), 2340–2347. 10.1039/C7SC03236F. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Frey S. L.; Lee K. Y. C. Number of Sialic Acid Residues in Ganglioside Headgroup Affects Interactions with Neighboring Lipids. Biophys. J. 2013, 105 (6), 1421–1431. 10.1016/j.bpj.2013.07.051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Frey S. L.; Chi E. Y.; Arratia C.; Majewski J.; Kjaer K.; Lee K. Y. C. Condensing and Fluidizing Effects of Ganglioside GM1 on Phospholipid Films. Biophys. J. 2008, 94 (8), 3047–3064. 10.1529/biophysj.107.119990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Šachl R.; Amaro M.; Aydogan G.; Koukalová A.; Mikhalyov I. I.; Boldyrev I. A.; Humpolíčková J.; Hof M. On Multivalent Receptor Activity of GM1 in Cholesterol Containing Membranes. Biochim. Biophys. Acta 2015, 1853 (4), 850–857. 10.1016/j.bbamcr.2014.07.016. [DOI] [PubMed] [Google Scholar]
  15. Shi J.; Yang T.; Kataoka S.; Zhang Y.; Diaz A. J.; Cremer P. S. GM1 Clustering Inhibits Cholera Toxin Binding in Supported Phospholipid Membranes. J. Am. Chem. Soc. 2007, 129 (18), 5954–5961. 10.1021/ja069375w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Sarmento M. J.; Ricardo J. C.; Amaro M.; Šachl R. Organization of Gangliosides into Membrane Nanodomains. FEBS Lett. 2020, 594 (22), 3668–3697. 10.1002/1873-3468.13871. [DOI] [PubMed] [Google Scholar]
  17. Fricke N.; Dimova R. GM1 Softens POPC Membranes and Induces the Formation of Micron-Sized Domains. Biophys. J. 2016, 111 (9), 1935–1945. 10.1016/j.bpj.2016.09.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Sieben C.; Sezgin E.; Eggeling C.; Manley S. Influenza A Viruses Use Multivalent Sialic Acid Clusters for Cell Binding and Receptor Activation. PLoS Pathog 2020, 16 (7), e1008656 10.1371/journal.ppat.1008656. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Loney C.; Mottet-Osman G.; Roux L.; Bhella D. Paramyxovirus Ultrastructure and Genome Packaging: Cryo-Electron Tomography of Sendai Virus. J. Virol 2009, 83 (16), 8191–8197. 10.1128/JVI.00693-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Harris A.; Cardone G.; Winkler D. C.; Heymann J. B.; Brecher M.; White J. M.; Steven A. C. Influenza Virus Pleiomorphy Characterized by Cryoelectron Tomography. Proc. Natl. Acad. Sci. 2006, 103 (50), 19123–19127. 10.1073/pnas.0607614103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Edwards K. F.; Steward G. F.; Schvarcz C. R. Making Sense of Virus Size and the Tradeoffs Shaping Viral Fitness. Ecol. Lett. 2021, 24 (2), 363–373. 10.1111/ele.13630. [DOI] [PubMed] [Google Scholar]
  22. Genoyer E.; Kulej K.; Hung C. T.; Thibault P. A.; Azarm K.; Takimoto T.; Garcia B. A.; Lee B.; Lakdawala S.; Weitzman M. D.; Lopez C. B. The Viral Polymerase Complex Mediates the Interaction of Viral Ribonucleoprotein Complexes with Recycling Endosomes during Sendai Virus Assembly. mBio 2020, 11 (4), na. 10.1128/mBio.02028-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Edelstein A. D.; Tsuchida M. A.; Amodaj N.; Pinkard H.; Vale R. D.; Stuurman N. Advanced Methods of Microscope Control Using MManager Software. J. Biol. Methods 2014, 1 (2), e10 10.14440/jbm.2014.36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Rawle R. J.; Villamil Giraldo A. M.; Boxer S. G.; Kasson P. M. Detecting and Controlling Dye Effects in Single-Virus Fusion Experiments. Biophys. J. 2019, 117 (3), 445–452. 10.1016/j.bpj.2019.06.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Rawle R. J.; Boxer S. G.; Kasson P. M. Disentangling Viral Membrane Fusion from Receptor Binding Using Synthetic DNA-Lipid Conjugates. Biophys. J. 2016, 111 (1), 123–131. 10.1016/j.bpj.2016.05.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Sackmann E. Supported Membranes: Scientific and Practical Applications. Science 1996, 271 (5245), 43–48. 10.1126/science.271.5245.43. [DOI] [PubMed] [Google Scholar]
  27. Watts T. H.; Brian A. A.; Kappler J. W.; Marrack P.; McConnell H. M. Antigen Presentation by Supported Planar Membranes Containing Affinity-Purified I-Ad. Proc. Natl. Acad. Sci. U. S. A 1984, 81 (23), 7564–7568. 10.1073/pnas.81.23.7564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Rawle R. J.; Boxer S. G.; Kasson P. M. Disentangling Viral Membrane Fusion from Receptor Binding Using Synthetic DNA-Lipid Conjugates. Biophys. J. 2016, 111 (1), 123–131. 10.1016/j.bpj.2016.05.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Rawle R. J.; Webster E. R.; Jelen M.; Kasson P. M.; Boxer S. G. PH Dependence of Zika Membrane Fusion Kinetics Reveals an Off-Pathway State. ACS Cent. Sci. 2018, 4 (11), 1503–1510. 10.1021/acscentsci.8b00494. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Tinevez J.-Y.; Perry N.; Schindelin J.; Hoopes G. M.; Reynolds G. D.; Laplantine E.; Bednarek S. Y.; Shorte S. L.; Eliceiri K. W. TrackMate: An Open and Extensible Platform for Single-Particle Tracking. Methods San Diego Calif 2017, 115, 80–90. 10.1016/j.ymeth.2016.09.016. [DOI] [PubMed] [Google Scholar]
  31. Schindelin J.; Arganda-Carreras I.; Frise E.; Kaynig V.; Longair M.; Pietzsch T.; Preibisch S.; Rueden C.; Saalfeld S.; Schmid B.; et al. Fiji: An Open-Source Platform for Biological-Image Analysis. Nat. Methods 2012, 9 (7), 676–682. 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Michalet X. Mean Square Displacement Analysis of Single-Particle Trajectories with Localization Error: Brownian Motion in an Isotropic Medium. Phys. Rev. E 2010, 82 (4), 041914. 10.1103/PhysRevE.82.041914. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ohi M.; Li Y.; Cheng Y.; Walz T. Negative Staining and Image Classification - Powerful Tools in Modern Electron Microscopy. Biol. Proced. Online 2004, 6, 23–34. 10.1251/bpo70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Scarff C. A.; Fuller M. J. G.; Thompson R. F.; Iadaza M. G. Variations on Negative Stain Electron Microscopy Methods: Tools for Tackling Challenging Systems. J. Vis. Exp 2018, 132, 57199. 10.3791/57199. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

jp2c03830_si_001.pdf (5.8MB, pdf)

Articles from The Journal of Physical Chemistry. B are provided here courtesy of American Chemical Society

RESOURCES