pH Stability and Disassembly Mechanism of Wild-Type Simian Virus 40

Abstract

Virus are remarkable self-assembled nanobiomaterial-based machines, exposed to a wide range of pH values. Extreme pH values can induce dramatic structural changes, critical for the function of the virus nanoparticles including assembly and genome uncoating. Tuning cargo - capsid interactions is essential for designing viral-based delivery systems. Here we show how pH controls the structure and activity of wild-type simian virus 40 (wtSV40) and the interplay between its cargo and capsid. Using cryo-TEM and solution X-ray scattering, we found that wtSV40 was stable between pH 5.5 and 9, and only slightly swelled with increasing pH. At pH 3, the particles aggregated, while capsid protein pentamers continued to coat the virus cargo but lost their positional correlations. Infectivity was only partly lost after the particles had been returned to pH 7. At pH 10 or higher, the particles were unstable, lost their infectivity, and disassembled. Using time-resolved experiments we discovered that disassembly began by swelling of the particles, poking a hole in the capsid through which the genetic cargo escaped, and followed by a slight shrinking of the capsids and complete disassembly. These findings provide insight into the fundamental intermolecular forces, essential for SV40 function, and for designing virus-based nanobiomaterials, including delivery systems and antiviral drugs.

1. Introduction

Viruses are fascinating self-assembled sophisticated nanobiomaterial-based machines. The cargo of a virus contains the genetic information and interacts with the viral capsid in various ways. The cargo may facilitate capsid assembly¹, stabilize the assembled virus², play a role in virus disassembly, and is the critical moiety of the virus life cycle and propagation. The interactions between the capsid and its cargo are therefore crucial for understanding the function of viruses and for designing viral-based nanotechnology platforms^3,4. During their life-cycle, viruses may operate under a wide range of pH values. The pH controls the charged state of the amino acids⁵ and DNA⁶, and therefore may profoundly affect the interactions between virus capsid proteins^7–9 as well as between the capsid proteins and the encapsulated genetic biomaterial. Resolving the details of virus assembly and disassembly, however, is experimentally challenging because of their transient character, immense number of possible intermediate states, and their structural complexity and heterogeneity.

SV40 is a member of the polyomavirus family. The interest in SV40 is growing with the increasing number of identified human polyomaviruses¹⁰. The structure of SV40 capsid, composed of three viral encoded proteins (VP1, VP2 and VP3), was resolved by cryo-TEM¹¹ and crystallography^12,13. SV40 is a 48.4 nm T = 7 icosahedron, encapsidating a circular double-stranded (ds) DNA genome of 5,243 base pairs (bp)¹⁴. The DNA is compacted by histone octamers, forming a minichromosomal structure. Five VP1 monomers assemble into a tightly bound pentamer (VP1₅) during mRNA translation¹⁵, whereas 72 pentamers form the outer capsid shell. A molecule of VP2 or VP3 is inserted in each VP1₅, forming a capsomere, connected to the minichromosome via the DNA binding sites of VP1 and VP2/3. The surface of SV40 presents charged, polar and hydrophobic residues¹⁶, which re spond to changes in pH. At pH 7 the capsid surface is acidic¹⁴. Capsomere-capsomere, capsomere-DNA, DNA-histones, and the hydrogen bonds of the dsDNA, control the virus response to different pH values.

SV40 binds to GM1 ganglioside receptors and enters cells by a caveola-mediated endocytosis^17–19. The particle is then translo cated to the endoplasmic reticulum (ER) via the endosomic pathway^20,21, where it disassembles²², and the genome enters the nucleus via VP₂ viroporin^23,24. Viruses may be used for gene delivery or fabricated into functional nanobiomaterials, devices, nanocapsules, biosensors, or nanoreactors^25–33. SV40 developed a special mechanism for nuclear entry of its DNA^24,34, which does not require cell division, providing an advantage over retroviral vectors for certain gene delivery applications. In vitro assembly systems based on VP1 pentamers of SV40 show remarkable flexibility for encapsulation of different cargoes^3,4,35. VP1 based virus-like particles (VLPs) were shown to assemble around flexible charged polymers including short and long ssRNA^1,36, ssDNA and the less flexible dsDNA molecules^1,37, and polystyrene sulfonate³⁵. The type of cargo and solution conditions (pH, ionic strength, reducing potential) direct the assembly process toward different capsid symmetries¹. wtSV40 was also shown to create 3D micro crystals when multivalent cations were added to the solution¹⁴. This rich and tunable assembly behavior with the relative stability of the particles makes SV40 a good nanobiomaterial candidate to be used as a delivery vehicle. To understand the basic interactions and to capture their effect on the assembly and stability of wtSV40, high resolution structural data at high temporal resolution are needed. Recently, our lab has developed solution X-ray scattering analysis tools^38–43 that can capture detailed structural information about components in complex mixtures. In the present work, we have used a combination of cryo-TEM, time-resolved cryo-TEM, Zeta potential, solution small angle X-ray scattering (SAXS), time-resolved SAXS (TR-SAXS), and virus titration to characterize the structural changes and infectivity of wtSV40 as a function of pH. SAXS is a highly reliable bulk method with msec temporal resolution, and is extremely sensitive to small changes in particle dimensions. Cryo-TEM and time-resolved cryo-TEM provided direct visualization of particle morphology and their pH-triggered structural changes at specific time points during the process. Our results have demonstrated that SV40 is resistant to pH changes over a wide range of values (between 5.5 and 9). At pH 3, however, the capsid proteins lost their positional correlations in the capsids, the capsids deformed and aggregated. Upon retuning to pH 7, the structural changes were reversible and infectivity was only partly lost. At pH ≥10, SV40 particles swelled, deformed, their genome escaped from the capsid, and then the capsid completely disassembled (Figure 1). These findings are important for understanding the properties of SV40 during its life-cycle and as a biomaterial for encapsulation and delivery of drugs.

Fig. 1.

Illustration of the mechanism of wtSV40 disassembly, induced by a pH jump to 10.7 (using carbonate buffer), as captured by time-resolved SAXS and time-resolved cryo-TEM. There are four phases in the process: the virus swells, its cargo escapes and the capsid shrinks back close to its original size, the virus partially disassembles, and then completely disintegrates to its subunits.

2. Results

2.1. pH Stability of SV40

Cryo-TEM images and SAXS measurements (Figure 2, and Figure S1 and S2 in Section 1 of the Supplementary Information, SI) show that SV40 was rather stable and underwent only small structural changes between pH 5.5 and 9. Taking pH 7 as the reference point (at which the capsid diameter was 48.4±0.2 nm^13,14), the shifts in the SAXS oscillations shown in Figure 2a suggest that as the pH was increased above 7 the capsid radius monotonically increased. The radius of the virus particles was accurately determined by fitting the SAXS curves to a low resolution model of concentric spherical shells with a smoothly varying radial electron density profile (see Section 3 in the SI, Figure S4, and Table S1). The analysis shows that the average external virus diameter increased from 48.4±0.2 to 49.2±0.2 nm between pH 5.5 and 9 (Figure 2b). A similar change was also observed by cryo-TEM images taken at pH 7.5 and 9, where the measured average diameters were 46.7±1.2 nm and 47.5±1.2 nm, respectively (Figure 2c and d). The small differences between the crystallography, SAXS and cryo-TEM, may be caused by differences in sample preparation and data analysis protocols (see Materials and Methods and Section 3 in the SI). Changing the pH to values lower than 7 did not lead to further decrease in capsid diameter, most likely because of the limited capsid compressibility (as the virus particles could not collapse into themselves). In contrast to shrinkage, swelling is unlimited and proceeds to disassembly as seen at high pH (see below).

Fig. 2.

SAXS and cryo-TEM stability analysis of wtSV40 at different pH values. a. Azimuthally-integrated background-subtracted small-angle X-ray scattering (SAXS) curves at pH values between 5.5 and 9. Scattering intensities are plotted as a function of q, the magnitude of scattering vector $\overrightarrow{q}$. The curves are shifted for clarity of presentation. The broken vertical lines serve as reference points to the first three minima of the scattering curve at pH 7. For higher pH values the minimum shifted to the left, consistent with swelling of the virus particles. b. The outer radius of the concentric spherical shell models that best fit the SAXS data as a function of pH (see Table S1). In these models the radial electron density of the virus is smoothly varying and represented by a sum of hyperbolic tangent functions (Equation S1). As was previously shown^44,45 the low resolution representation (up to q ~ 0.6nm⁻¹) is sufficient to accurately extract the radius of the particles. c. Cryo-TEM images of capsids at pH 9 and 7.5. Scale bars are 200 nm. d. Capsid diameter histograms based on cryo-TEM images as in (c.) The mean diameters and their standard deviations are indicated in the figure in nanometer units as a function of pH.

Figure 3 shows that when the pH of the solution was changed to 3 the virus particles did not maintain their regular shape. At this pH value the virus particles became more compact and aggregated, making it difficult to determine their size via cryo-TEM or SAXS (Figure 3a and b). At pH 3, the oscillations in the SAXS curve between q = 0.9 and 1.3 nm⁻¹, associated with the arrangements of the capsid subunits on a curved surface, disappeared.The disappearance of the oscillations is consistent with a reduction of positional correlations between subunits in the curved capsid (inset to Figure 3a, black curve)⁴⁵. This phenomenon was also observed by cryo-TEM images at pH 3, in which the particles lost the discrete character, observed at higher pH values (Figure 3b).

Fig. 3.

wtSV40 at extreme pH values. a. SAXS curves at pH 3, 10,10.7, and the saline reference. At pH 10.7 (red curve) no oscillations were observed because the virus particles disassembled. At pH 10 the oscillations are slightly shifted to the left (swelling) whereas at pH 3 the minimum shifted to the right (shrinking). The high q oscillations disappeared at pH 3 (inset), suggesting the loss of positional correlations of capsid pentamers b. Cryo-TEM images of wtSV40 at pH 10 and 3. At pH 10, empty (indicated by white arrows) and full (black arrows) particles were observed. The broken rectangle was magnified (image in the center) to better show the minichromosome present outside the particles (yellow arrows). At pH 3 the virus particles formed large aggregates, lost the discrete arrangement of pentamers in the capsid, and the high q oscillations in the SAXS curve disappeared (black curve at the inset to Figure 3a). Scale-bars are 200 nm. More images are presented in Figure S1. c. Electrophoretic mobility of wtSV40 as a function of pH. Each data point and error bars are based on three repeated measurements. As the pH increased the electrophoretic mobility changed from net positive to net negative. Virus surface charge neutrality is expected at about pH 5. d. Diameter distributions of full (left) and empty (right) capsids at pH 10, based on cryo-TEM images as shown in b.

When the pH was raised to 10, using carbonate buffer (Figure 3a,b), cryo-TEM images showed coexistence of empty and full capsids with average diameters of 48.5±2.0 nm and 47.8±1.5 nm, respectively. Distinction between full and empty particles is explained in Section 2 in the SI and in Figure S3, which is showing empty wtSV40 particles, extracted from CV1-PD cells⁴⁵. Empty capsids resulted from either a genome that escaped from the capsid, as seen by the DNA molecules outside the capsids (Figure 3b middle image, and Figure S1 and S2), or partial disassembly of the virus^46,47. To analyze how stable the interactions were, the samples were measured at elevated temperatures (Figure S5 and S6). Viruses that were kept at pH 5 remained stable even at 55 °C, whereas at pH 10, the particles disassembled already at 40 °C. These results imply that at pH 10, even complete particles were held together by relatively weak interactions. A decrease in the melting temperature of cowpea chlorotic mottle virus (CCMV) was also observed when the pH was raised from 4.5 to 8⁴⁸.

When the pH of the solution was further raised to 10.7, which is higher than the pK_a value of free lysine in solution, a complete disassembly reaction occurred (Figure 3a, red curve). Ly-sine residues play an important role in stabilizing the nucleosomal structure of the minichromosome⁴⁹. In addition, the DNA-binding domain at the N-terminal end of VP1 contains 7 lysines and the respective domain at the C-terminus of VP2/3 contains 4 lysines^50,51. Furthermore, at pH 10.7, the capsomere-capsomere interactions were also too weak and the virus disassembled to its components.

Apart from pH, the SV40 capsid is also stabilized by calcium ions and disulfide bonds^15,44,52. Calcium ions clamp the intra calating carboxy-terminal arms of the pentamers that hold the icosahedron together¹³, and their chelation has been shown to destabilize the capsid^44,53–55.The carbonate buffer could have therefore weakened the capsomere-capsomere interactions also through calcium chelation. To isolate calcium chelation from the high pH effect of the buffer, we performed additional experiments at high pH values using N-cyclohexyl-3-aminopropanesulfonic acid (CAPS) buffer.

Figure 4 and S7 show the effect of CAPS buffer at pH 10.7 on the structure of SV40. Figure 4a compares the scattering from SV40 in unbuffered saline solution with its scattering after 1 h incubation in 50 mM CAPS buffered saline at pH 10.7. At pH 10.7, the absolute SAXS intensity at low q decreased, indicating a decrease in the average mass of the particles. The broken vertical lines in Figure 4a indicate that the first two local minima of the dark yellow curve (from SV40 at pH 10.7) were shifted towards lower q values whereas the position of the minima at higher q values did not change (compared with SV40 in unbuffered saline). This change in the SAXS profiles cannot be attributed to a larger average radius because the oscillations at higher q values, which are sensitive to the capsid radius, remained unchanged^40,56.

Fig. 4.

Cargo escape at pH 10.7. a. SAXS curve of SV40 in unbuffered saline (blue curve), in CAPS buffer at pH 10.7 (dark yellow curve), and in the carbonate buffer at pH 10.7 (orange curve). The vertical broken lines indicate the position of the minima in the blue curve. b. Cryo-TEM images of empty and full virus particles observed following addition of CAPS buffer at pH 10.7 to the virus solution. The images also show that the DNA escaped from the SV40 capsids (indicated by arrows). Samples were prepared about 40 min after mixing about $\raisebox{1ex}{$\text{mg}$}\!\left/ \!\raisebox{-1ex}{$\text{ml}$}\right.$ SV40 in 150mM NaCl with 0.4 M CAPS buffer at pH 10.7, to get a final buffer concentration of 50 mM CAPS. Scale-bars are 100 nm. More images are presented in Figure S7. c. Modeling of the SAXS curves from panel a. The left panel shows the fitting of a hybrid model (see Section 10 in the SI) of SV40 to the SAXS measurement of SV40 in unbuffered saline (blue curve in panel a.). The right panel shows the best fitted model to the signal of SV40 at pH 10.7 (dark yellow curve in panela.). The model included three states: the hybrid model of SV40, a model of partial empty particle, and the disassembled state. The disassembly state was taken from the measured signal of the complete disassembled virus at pH 10.7 using the carbonate buffer (see red curve in Figure 3). The inset shows the mass fractions (Equation 2) of the three models that best fit the data. The measured intensities, fitted models, and measurement errors are given by blue symbols, red curve, and gray bars, respectively. The parameters of the hybrid models and the calculation methods are given in Section 10 in the SI and Equation S4. d. Distribution of capsid diameters at pH 10.7, based on the cryo-TEM images.

Cryo-TEM images taken under the same solution conditions (Figure 4b) show both empty and full particles. Within the resolution of our cryo-TEM measurements and analysis, the average characteristic size of the two populations was rather similar. We attribute the lower stability of particles in the carbonate buffer (compared with the CAPS buffer) to the chelation of the calcium ions by the carbonate ions^12,13,44,52. The relative stability of the capsid shell in CAPS buffer facilitated capturing the escape of the DNA from the capsid (Figure 4b and S7). The cargo appeared to escape (Figure 4b and S7b) through a local hole in the capsid, presumably created by a combination of the high internal pressure, due to destabilization of the double helix⁶, dissociation of the minichromosome, and weakening of capsomere-capsomere interaction.

To quantify the effect of pH 10.7 in the CAPS buffer, we fitted our SAXS data to a higher resolution hybrid model that included the atomic structure of the capsid (see Section 10 in the SI and Equation S4). The left panel in Figure 4c shows the best fitted hybrid model to wtSV40 in unbuffered saline solution. The hybrid model included the hydrated atomic structure of the T = 7 capsid (PDB ID 1SVA) after adding the hydrogen atoms to the PDB, using MolProbity server⁵⁷. The radial swelling factor, f_r, compared with the original PDB structure was 1.0075 (see Equation S4). The radius of the inner uniform sphere, representing the minichromosome, was r = 18.7nm and its electron density contrast with respect to the solvent, was Δρ_inner = 52^e/nm³. The model assumed uniform minichromosome electron density distribution inside the capsid. As the position and shape of the oscillations matched the data, the hybrid model, computed by our analysis software D+ (https://scholars.huji.ac.il/uriraviv/software/d-software)³⁹, provided a good approximation to the scattering curve from SV40. We attribute the slightly lower intensity of the model compared with the experimental signal, between q = 0.3 and 0.8 nm⁻¹, to the discrete character of the minichromosome structure⁴⁵. As the nucleosomes were confined to the inner capsid diameter and their size was about 10 nm, their positional correlations contributed to the measured scattering signal at that q range.

Neither SAXS nor cryo-TEM showed a significant difference in the size of empty and full particles (Figure 3, 4, S7, and S8). We therefore modeled the empty particles by the same structure of the full particle, after reducing Δρ_inner from 52 to 16^e/nm³. The basic assumption in this model was that the contribution of the DNA, which escaped from the capsid, to the scattering intensity was negligible since its electron density no longer correlated with the electron density of the empty particle (Figure S7 and Section 6 in SI). Figure S8 shows the differences in the modeled scattering intensity between the full and partially empty SV40 particles, revealing changes that are similar to the experimental signals (Figure 4a).

To fit the scattering intensity from the solution of SV40 in CAPS at pH 10.7, we used three models: the wtSV40 model (calibrated against wtSV40 data), the empty particles model (with the reduced Δρ_inner value), and the complete disassembled state, measured at pH 10.7 in the carbonate buffer (Figure 3a, red curve). The disassembled state was added because disassembly also occurred in the CAPS buffer (Figure S7b), albeit to a lesser extent. The calculated three-state model that best fit the data is shown at the right panel of Figure 4c, where the inset shows the mass fractions of the three states. The results show that about half of the total mass comprised partially empty particles, presumably because of the DNA escape.

2.2. Kinetics of pH-Triggered Virus Disassembly

TR-SAXS data (Figure 5) and time-resolved cryo-TEM images (Figure 6 and Figure S12 in Section 8 in the SI) were recorded during the disassembly process, triggered increasing the pH to 10.7, using the carbonate buffer. The background subtracted TR-SAXS data can be divided into four phases along the disassembly process. In phase I (during the first 30 or 40 sec, measured using a stopped-flow setup), the intensity at q → 0, I₀, was nearly constant (Figure 5e) and the oscillations of the scattering intensity shifted to lower q values (Figure 5a). These features show that in phase I the virus particles swelled, while keeping a nearly constant average mass. The X-ray result is consistent with cryo-TEM images, taken 30 sec after the shift to pH 10.7 in the carbonate buffer (Figure 6a and b), showing larger SV40 particles and few capsids that remained intact but lost their icosahedral shape. The mean capsid diameter at the end of phase I was 55.2 ±3.9 nm (Figure 6b). The wide particle size distribution is attributable to irregular capsid swelling and deformation.

Fig. 5.

Time-resolved SAXS (TR-SAXS) measurements of the disassembly of wtSV40 induced by a pH jump from 7 to 10.7, using the carbonate buffer. Panels a.-d. show the azimuthally-integrated background-subtracted TR-SAXS curves of wtSV40 during disassembly at pH 10.7. a. wtSV40 was mixed in a stopped-flow setup at ESRF ID02 beamline, with carbonate buffer at pH 10.7 (see Section 12 in the SI). Three representative SAXS curves along the first 42 sec of disassembly are shown. The fixed intensity at small scattering angles and the shift in the oscillation toward smaller q values are consistent with swelling of the capsid during the initial disassembly phase. b. to d. The experiment was repeated, using a flow-through setup and disassembly was recorded between 90 and 780 sec following the onset of disassembly. b. The second phase in the disassembly process (90–255 sec) is characterized by a decrease in the intensity at low q and a shift in the position of the first minimum towards lower q values as shown in the inset. c. SAXS curves at the third phase (270–480 sec) where a two states kinetics is consistent with the observed three isosbestic points in the time-resolved spectra, indicated by the three vertical lines. d. The final disassembly phase (465–780 sec) where gradual decrease in the scattering intensity at low q values is observed. e. Time evolution of the scattering intensity, I₀, at q → 0, using a stopped-flow setup. We marked the swelling phase during the first 42 sec, at which there was essentially no change (within the scatter) in the value of the intensity. f. Time evolution of I₀ and the intensity at the q value of the first minima (q_min) as measured in the flow-cell experiment (panels b. to d.). The complete data set is shown in Figure S15–S18.

Fig. 6.

Time-resolved cryo-TEM images during the disassembly of SV40 at pH 10.7, in the carbonate buffer. a. 30 sec following the transfer to pH10.7. b. Histogram of particles diameter 30 sec after disassembly (in red), are based on images as shown in a., and in saline buffered at pH 7 (in blue). Images taken 200 sec and 760 sec following incubation at pH 10.7 are shown in (c.) and (d.), respectively. Scale bars are 100 nm

At longer incubation times at pH 10.7 in the carbonate buffer (measured using a flow cell setup), I₀ decreased while only the first minimum in the scattering curve, at q ≈ 0.17, shifted with time to lower q values (phase II in Figure 5b and f). This change is reminiscent of the scattering curve in the CAPS buffer at pH 10.7 (Figure 4), suggesting that during phase II the DNA escaped from the virus particles. During the third phase, three isosbestic points were observed (marked by vertical lines in the inset to Figure 5c), suggesting that the distribution of intermediates was dominated by two principal states³⁶. The SAXS curve at the end of the third phase (blue curve in Figure S13 in Section 9 in the SI) did not resemble the oscillatory signal expected from a capsid-like struc ture. Figure 6d and S12 show that at this point, the solution, contained mostly capsid fragments, disassembled minichromosome, and probably free subunits (pentamers, DNA and histone octamers). These fragments appear to have disintegrated into their subunits in the final phase of the process as seen in Figure 5d (see also orange curve in Figure S13 or red curve in Figure 3a for the scattering curve of the final dissociated state of the assembly reaction). It is interesting to note that after the DNA escaped, the empty SV40 capsids disassembly path (Phases III and IV) is somewhat similar to that of CCMV empty capsids⁵⁸.

2.3. Structural and Functional Reversibility

We then examined the reversibility of the structural changes and biology, measured as virus infectivity, in response to pH shift (see Section 7 in the SI). SV40 was first incubated for 1.5 h at pH 3 or 10, the two extreme pH values in which complete particles were still observed. The particles were then dialyzed against a solution of 150 mM NaCl at pH 7, as explained in Materials and Methods. SAXS curves, measured after incubation at pH 3 and returning to pH 7 (Figure S9a, blue curve), nearly reproduced the saline control sample at pH 7 (Figure S9a, black curve). Cryo-TEM micro-graphs and the corresponding diameter distribution (Figure S9c and e) showed that the average capsid diameter was 47.7±1.1 nm. These data indicate that the structural changes seen at pH 3 were nearly reversible (Figure 3). To test whether capsid structural reversibility correlates with biological activity, the infectivity level of the virus was examined by measuring the virus titer, as explained in Section 7 in the SI. Titer measurements, determined by the number of T-antigen expressing cells following infection, using fluorescence-activated cell sorting (FACS), required cell infection and virus dilution under physiological conditions. Infectivity level was therefore examined only at pH 7.4. The measurements are explained and shown in Section 7 of the SI and in Figure S10 and S11. The data show that ~ 50% of the viral infectivity was maintained following 90 min incubation at pH 3, compared with the reference sample, kept under physiological conditions (Figure S10b and S11).

In contrast, ~ 80% of the infectivity was lost during incubation for 90 min in carbonate buffer at pH 10 (Figure S9b and d). This finding is accounted for by the irreversible partial loss of DNA, observed by SAXS and cryo-TEM (Figure S9) following the incubation at pH 10. It appears that the irreversible release of the DNA during the incubation at pH 10 did not lead to significant disruption of the particles. Empty particles were only slightly deformed under cryo-TEM (Figure S9b), probably because of the deformation of the capsid at the point of DNA escape were only partially reversible in the absence of the DNA scaffold⁵⁹. After returning to pH 7 the average diameter of particles that maintained their cargo was 47.7±1.1 nm (Figure S9d), similar to that of the particles that were incubated at pH 3 and returned to pH 7 (Figure S9e).

3. Discussion

Compared to other spherical viruses, SV40 has a more complex electrostatic interactions and pH sensitivity⁶⁰. These properties are attributable to the patchy, nonuniform surface charge distribution, having a net negative external surface charge and a net positive internal surface charge^14,45,60, and the variability in its more flexible capsomere-capsomere interactions, mediated by the flexible C-arms. In addition, the electrostatic interactions between the histones and the dsDNA are affected by the protonation state of the positively charged amino acids. The physical properties of the encapsidated minichromosome may therefore dramatically change with pH. These properties play a critical role in the virus life cycle. During cell entry, many viruses undergo structural changes when they reach the late, acidic endosomes on their passage through the endosomal pathway. SV40, however, is stable at acidic pH and reaches the ER intact, unexposed to the lysosomal and/or proteosomal degradation. This property is consistent with the findings that SV40 evades adaptive immunity^61,62, pre sumably by avoiding VP1 antigen presentations at the cell surface during its cell recognition, entry, and infection cycle.

The observed swelling with increasing pH (Figure 2) can be explained by the higher charge density at the viral surface at elevated pH as detected by electrophoretic mobility measurements (Figure 3c). The net negative charge at a high pH is most likely because of deprotonation of the capsid amino acids and the nucleosomes that condense the DNA. The higher charge density is leading to stronger capsomere-capsomere electrostatic repulsion, expansion of the minichromosome, leading to higher internal pressure, and weaker interactions between capsomeres. Swelling at alkaline solutions was previously observed for other viruses. CCMV swells upon increasing the pH from 5 to 7 due to loss of contacts within the protein shell leading to its softening^48,63.

The gastro-intenstinal Norovirus-like particles gradually become softer and increase in size with increasing pH because of weakening of capsomere-capsomere interactions⁶⁴. This type of pH response allows capsids to withstand highly acidic environments (as in the stomach) and become infective under weakly basic conditions (as in the ileum). In contrast, influenza virus enters host cells by endocytosis. The low pH of endosomes triggers its structural changes that mediate fusion of the viral and endosomal membranes⁶⁵.

As indicated by electrophoretic mobility measurements (Figure 3c) at pH 3 and 4 the net outer surface charge of SV40 became positive, leading to aggregation by the citrate buffer trivalent anions. Citrate buffer was also shown to promote aggregation of poliovirus and reovirus⁶⁶. The citrate anions could bridge positively charged amino acids belonging to different pentamers across different capsids. Bridging between pentamers within the same capsid may have caused the pentamers to change their characteristic positional correlations. The intrinsic interaction between pentamers could have also changed with pH, as observed in studies about the assembly of SV40 particles around DNA^37,67.

At a high pH, partial deprotonation of the lysine amino-acids may have loosened the DNA-histone interactions^68,69, and the interaction of the minichromosome with the inner capsid wall, through the DNA binding domains of VP1 and VP2/3. The minichromosomal structure of the dsDNA enables compaction at natural pH of the otherwise stiff dsDNA^45,70.

We suggest that at high pH, weakening of the condensing interactions led to the increase in the effective persistent length of the DNA inside the capsid that further led to the escape of the dsDNA with or without the histones. The DNA escaped through one or few capsomere vacancies, formed by the high internal pressure in the capsid, and the weakening of capsomere-capsomere and DNA-capsid stabilization. A rather similar pH dependence was also shown for Triatoma virus, which is a ssRNA virus¹⁶. In SV40 at pH 10, despite the escape of DNA and the loss of DNA-capsid stabilization, the capsomere-capsomere interactions were still strong enough to keep some of the capsids intact or only slightly deformed, while others lost their DNA.

To obtain the mass fraction of the different principal states during the disassembly process we fitted the TR-SAXS data to a set of five principal structural models, representing the disassembly pathway. The expected scattering intensity curves of the first three principal states included the initial wtSV40 model, the swollen SV40 structure, and the partially empty SV40, following the DNA escape. The three curves were computed by our home developed software D+ (https://scholars.huji.ac.il/uriraviv/book/d-0)^38,39, using the hybrid model, described in Section 10 in the SI (Equation S4). The last two principal states, the disassembled state at the end of phase III and the final state of the process (Figure S13) contained a variety of components that could not be reliably modeled, hence the most relevant experimental scattering curves represented the final two principal states of the process.

The fitted scattering curves of the entire disassembly process are shown in Figure 7 (representative curves) and Figure S15–S18 (entire data set), where for the clarity of presentation, the curves were vertically shifted. Phase I was fitted by a first order kinetic reaction [wtSV40] $\stackrel{k}{\to }$ [Swollen SV40] where the initial state model was calibrated by the wtSV40 in saline solution and the swollen SV40 structure was determined by the scaling swelling factor, f_r (see Equation S4 in Section 10 in the SI) that minimized the c² function of the entire time series of phase I. The best fitted value of f_r was 1.07, corresponding to an increase of1.3 nm in the average radius of the SV40 particles. The swelling factor evaluates an average conformation. Time-resolved cryo-TEM images revealed the degree of polydispersity in the particle diameter (Figure 6 and S12). The characteristic time scale of the swelling process was found to be ${\tau }_{\text{swelling }}=\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$k$}\right.=24\text{\hspace{0.17em}sec}$.

Fig. 7.

Analysis of the disassembly process during the four phases, shown in Figure 5. TR-SAXS data were fitted to a linear combination of the five principle states that describe the disassembly process. The modeled scattering curve of wtSV40, swollen SV40, and partially empty capsid were generated using D+ program³⁹ (see Section 10 in the SI). The last two principle disassembly states included the final signal measured after 780 sec (the completely disassembled state) and the last signal of the third phase, where a two states kinetics was observed (swollen SV40 and the partially disassembled state). The scattering intensities of the computed principle states and the parameters of each model are found in Section 10 in the SI. The measured scattering intensity, the measurement standard deviation, and the best fitted model are shown as blue symbols, grey bars and red curves, respectively. Grey Roman numerals correspond to the different disassembly phases presented in Figure 5a–d. The inset of phase I shows the radius of the particles as a function of time. The complete data set and fitting are shown in Figure S15–S18 in Section 11 of the SI.

The slower phases of the disassembly process were measured using a flow-cell setup starting from 90 sec following the pH jump and up to 760 sec. To fit these data we used a non-negative least-square algorithm (see Section 11 in the SI) to determine the distribution of mass fractions, among the five principal states, as a function of time. The resulting mass fractions are shown in Figure 8. Linear combinations of the principal models adequately fitted our TR-SAXS data. During the second phase of the disassembly process it can be seen that the oscillations in the models were stronger than the measured signals. This deviation resulted from the fact that during the second phase, the dominant principal states were the swollen and partially empty particles. These particles had a relatively wide distribution of sizes and some degree of deformation (Figure 6). Our models, however, represented only the average characteristic conformations. During the second phase, approximately 100 sec following the addition of the carbonate buffer, our fit shows that empty particles (purple symbols in Figure 8) accumulated in the solution. Our fitting procedure suggests that the diameter of the empty particles was, on average, slightly smaller than the diameter of the swollen conformation as the best fitted swelling factor, f_r, was 1.03 (compared with 1.07 for the maximally swollen SV40 particles). The 0.5 nm radius difference suggests that the inner pressure was relaxed following the escape of the DNA from the capsid, leading to the decrease in the size of the particles before the capsid disassembled.

Fig. 8.

Mass fractions of the different principle states as a function of time, derived from the fitting to the TR-SAXS data (Figure 7 and Figure S15–S18). Grey vertical lines and numbers corresponds to the different disassembly phases presented in Figure 5a–d and 7.

The CAPS buffer equilibrium state (Figure 4) resembled an intermediate scattering curve in the carbonate-induced disassembly (Figure 5), suggesting that the particles swelled until the capsids failed to hold their cargo. Cryo-TEM images (Figure 6) show that following swelling, a hole was poked at a certain point in the capsid through which the circular dsDNA escaped. The breaking point could have been at one of the relatively weaker capsomere-capsomere contacts (refer to as the γ–γ contacts)¹³. The CAPS buffer results at pH 10.7 were rather similar to the carbonate buffer results at pH 10 (Figure 3 and 4). The similarity reveals that even at pH 10.7, the capsomere-capsomere interactions were sufficiently strong to partially stabilize the capsid, and hold the internal pressure exerted by the minichromosome. The average diameter of the capsids that remained intact increased from 46.7 ±1.2 to 48.5±1.3 nm (based on cryo-TEM images shown in Figure 2d and 4d). Carbonate/bicarbonate buffer tends to interact with calcium ion. We propose that calcium ions that stabilize the capsid could interact with the buffer and precipitate as CaCO₃ (K_sp of CaCO₃ is ~ 10⁻⁹). The capsid shell could have therefore been less resilient to deformations, hence the DNA escaped and particles readily disassembled. Figure 1 illustrates the disassembly process as revealed by our findings.

Our data show that structural reversibility correlates with preservation of infectivity. It could be that at pH 3 small irreversible structural changes occurred and influenced the infection cycle but were below the SAXS and cryo-TEM resolutions (of about 1 nm). These changes, however, are minor compared with the deformations and DNA escape seen at pH 10, allowing the pH 3 incubation to have a smaller effect on the infectivity of the virus particles. The structural reversibility results suggest that the stability range of wtSV40 virus may be broaden to include pH values between 3 and 9.

4. Conclusions

Our study demonstrated that between pH 5 and 9 wtSV40 was stable and only minor reversible increase in its size took place with increasing pH, making SV40 an interesting flexible and stable nanobiomaterial for a number of applications such as gene or drug delivery, or for antigen display for vaccine applications. At pH 3, the virus aggregated but remained stable and infective upon returning to pH 7. The virus, however, was unstable under highly basic conditions and lost its infectivity. At pH 10.7 (carbonate buffer), we discovered the mechanism underlying the high pH disassembly of the virus. We found that the nanoparticles initially swelled followed by the creation of a hole in the capsid, presumably because of inside pressure exerted by the minichromosome. This hole allowed the escape of the cargo and the capsids shrank back close to their original size. The empty capsids disassembled into fragments, which then completely disintegrated into the virus subunits. Similar pathway is likely to be discovered in other members of the polyomaviruses family. Our study provides insight into the fundamental intermolecular interactions between the genetic cargo and the capsid proteins, essential for design ing antiviral drugs, virus-based delivery systems, and functional nanobiomaterials. The approaches used here could be applied to study other viruses and complex self-assembled nanostructures.

5. Materials and Methods

5.1. wtSV40 purification

wtSV40 was purified as previously described¹⁴. Briefly, CV1-PD cells were infected by wtSV40 at a multiplicity of infection (moi) of 0.05. After an incubation period of 5 to 6 days, wtSV40 was harvested by the di-detergent method⁷¹. Ultracentrifugation at 40,000 rpm and 4 °C for 24 h in CsCl generated two bands of different densities that separated the virus from other particles. The lower band (density of 1.34 g/cm³) contained wtSV40, whereas the upper band (density of 1.3 g/cm³) contained empty capsids⁴⁵. Both bands were collected and analyzed using SDS-polyacrylamide gel electrophoresis (Novex WedgeWell 4–12% Tris-Glycine) with Coomassie Brilliant Blue staining (Instant blue stain, Exedeon). wtSV40 samples were dialyzed against 0.9 wt% NaCl solution. Empty capsids were dialyzed against 0.5 M NaCl solution. Dialysis was preformed at 4 °C using GeBa Mini and Midi dialysis tubes (Gene Bio Application Ltd. Cat. No. D070–6). Both particles were then stored at 4 °C.

5.2. Cryo-TEM

Sample preparation for cryogenic transmission electron microscopy (cryo-TEM) was performed as in our previous work⁷². Briefly, samples were deposited onto a 300 mesh Lacey grid (Ted Pella Ltd.), blotted and vitrified with Vitrobot Mark IV (FEI Co.). Micrographs were obtained at −177 °C using a Tecnai G2 Spirit Twin T-12 TEM (FEI Co.) at 120 kV acceleration voltage, in a low-dose mode. Images were recorded on a 4K × 4K FEI Eagle CCD camera at defocus values between 1.5 and 3 μm. Disassembly ki netic experiments were obtained by measuring the time passing between mixing wtSV40 in saline solution with carbonate buffer (CAPS buffer pH 10.7, see Section 12 in SI) to a final concentration of 50 mM and the moment at which the grid was plunged into the liquid ethane. 30 sec was the shortest recordable time point with our setup. Capsids were characterized by their diameter. If a capsid assumed an oval shape (as in pH 10.7), its long axis was reported as its diameter.

5.3. Electrophoretic mobility measurements

Electrophoretic mobility (EM) measurements were performed using Zetasizer nano ZSP (Malvern) that uses phase analysis light scattering (PALS) for the determination of EM. Purified wtSV40 solution was dialyzed against 20 mM NaCl to reduce the conductivity of the solution. SV40 virus solutions at different pH values were prepared by mixing 30 μL of SV40 solution with 3.4 μL of 0.5 M buffer at the desired pH. The measurements were carried out using the diffusion barrier technique to minimize sample degradation near the electrode positions. In this technique, a standard 1 mL zeta cell (1070) was filled with the buffer solution of each sample and then 33.4 μL of the virus sample was pipetted directly into the bottom of the cell. All measurements were performed at 25 °C. Measurement protocol was changed according to the conductivity of each sample. At pH 3 and 9, we performed the automated protocol that uses a longer application of the steady field at 50 and 150 V, respectively. At pH 4 and 5.5 the measurements were carried out using the monomodal analysis method, which uses the fast field reversal only to minimize degradation of the sample and electrodes. The applied voltages were 50 and 150 V for pH 4 and 5.5, respectively. For this analysis method the result of the measurement is only the mean electrophoretic mobility. The reported values and standard deviations for each condition were based on three measurements.

5.3.1. Sample preparation for SAXS measurements

The results in Figure 2a. and 3a., correspond to samples that were prepared by mixing wtSV40 solution at concentration of ~1.5 mg/ml of capsid proteins and 150 mM NaCl with a stock solution containing 150 mM NaCl and 0.5 M of the relevant buffer at the desired pH (see Section 12 in SI)). Mixing was done at volume ratio of 9:1 to a final 50 mM buffer solution. Sample volumes were between 30 and 40 μL. The samples were incubated on ice for approximately 1 h before measurements were performed. The sample for the SAXS experiment, shown in Figure 4, was obtained by mixing wtSV40 solution at capsid protein concentration of ~1mg/mL at 150 mM NaCl with 0.4 M CAPS buffer at pH10.7 to a final buffer concentration of 50 mM. The solution was then equilibrated at ambient room temperature for ~1 h before it was measured. Stopped-flow experiment shown in Figure 5a. was perform by mixing a solution containing ~1mg/mL wtSV40 at 150 mM NaCl with a solution containing 150 mM NaCl and 150 mM carbonate buffer at pH 10.7 (see Section 12 in SI). The mixing was done at volume ratio of 2:1 to a final carbonate buffer concentration of 50 mM. The mixing was done in the stopped-flow setup at 25 °C. The first measurement was taken 0.1 sec following the mixing phase. TR-SAXS experiments shown in Figure 5b.–d. were performed by mixing ~ 1mg/mL wtSV40 solution at 150 mM NaCl with 0.5 M carbonate buffer at pH 10.9, at a volume ratio of 9:1, to give a final buffer concentration of 50 mM. Immediately following mixing, the sample was injected into the measuring capillary flow-cell, which had been pre-equilibrated at 25 °C. The first measurement was taken 90 sec following the mixing.

5.4. SAXS measurements

Solution small angle X-ray scattering (SAXS) measurements, shown in Figure 2a. and 3a., were performed at the SWING beamline (headed by J. Perez) in Soleil synchrotron (Gif-sur-YVETTE). Detailed experimental description of this setup was provided elsewhere⁷³. SAXS measurements were performed at 10 keV with sample to detector distance of 2.44 m, resulting in a q range of 0.045–4.64 nm⁻¹. Each sample was exposed for 20 frames while flowing through the quartz capillary measuring chamber and the exposure time per frame was 1 sec. SAXS measurements, presented in Figure 3a were performed at sample to detector distance of 1.377 m (resulted in a q range of 0.081–8.99 nm⁻¹), and the exposure time per frame was 0.5 sec. The SAXS data presented in Figure 4 was measured at the P12 EMBL BioSAXS Beamline (headed by D. Svergun) in PETRA III (DESY, Hamburg), using wavelength of 0.124 nm, an automated sample changer setup, and PILATUS 2M (DECTRIS) detector⁷⁴. The sample to detector distance was 2 m, leading to a q range of 0.025– 5 nm⁻¹. Between 30 and 35 μL were injected into the measurement cell. Between 30 to 35 frames were measured while the sample was flowing through the capillary, using an exposure time of 50 msec per frame. Before and after each sample a background measurement containing the same buffer condition was taken, averaged, and subtracted from the averaged sample signal. Time-resolved SAXS (TR-SAXS) measurements of the wtSV40 disassembly reaction at 25 °C, shown in Figure 5 and 7, were performed at ID02 beamline (headed by T. Narayanan) in ESRF, Grenoble⁷⁵. Stopped-flow experiment shown in Figure 5a. was performed in a stopped-flow setup (BioLogic SFM 400)⁷⁶, and observed over the initial minute. Prior to mixing, wtSV40 solution at concentration of ~1mg/mL at 150 mM NaCl and a solution containing 150 mM NaCl and 150 mM carbonate buffer at pH 10.7 were kept in two separated syringes in cooled (5 °C) reservoir. To initialize disassembly 150 μL of the wtSV40 solution and 75 μL of the carbonate buffer were mixed and pushed into the measuring quartz capillary that had been pre-equilibrated at 25 °C. The mixing phase took 35 msec and the first measurement was taken 0.1 sec after the mixing was over. For each mixing reaction we took 15 frames with an exposure time of 20 msec per frame. Background measurements were taken using the same setup by mixing the carbonate buffer with 150 mM NaCl. The flow-through capillary setup of ID02 was used to follow the slower part of the disassembly reaction (Figure 5b.–d.). The measurement capillary was temperature controlled^75,77 at [25]°C. 72 μL of wtSV40 and 8 μL of 0.5 mM carbonate buffer at pH 10.7 were manually mixed and then injected into the flow-cell capillary. The first frame was taken 90 sec following mixing. The frames between 90 and 780 sec were taken at 15 sec intervals, while the sample was flowing through the capillary. The exposure time per frame was 50 msec. The wavelength of the incident beam for all the time-resolved measurements was 0.995 nm and a FReLoN 16M Kodak CCD detector was used to record the scattered intensity at ID02. Background buffer measurements under identical conditions were taken before and after each sample. Scattering 2D patterns were normalized to the intensity of the transmitted beam, and azimuthally averaged to yield the scattering intensity curve as a function of the magnitude of the scattering vector, q⁷⁸. Background scattering curves were averaged, and the averaged background signal was subtracted from each sample frame to give the final background-subtracted scattering intensity curves of the assembly reactions, as explained in our earlier papers^{14,36,38,42,79}.

5.4.1. SAXS data analysis

The scattering intensity from non-interacting virus particles is:

$$I_{\text{virus }}(q)=\langle |FF(\overrightarrow{q}){|}^2\rangle =\langle {\left|-r_0{\int }_V\Delta \rho (\overrightarrow{r})\text{\hspace{0.17em}exp\hspace{0.17em}}\left\{-i\overrightarrow{q}\cdot \overrightarrow{r}\right\}d\overrightarrow{r}\right|}^2\rangle ,$$ (1)

where the form factor, FF, is the Fourier transform of the[vextendsingle] electron density contrast of the virus with respect to the solvent, $\Delta \rho (\overrightarrow{r})\cdot \overrightarrow{q}$ is the elastic momentum transfer vector. The brackets 〈…〉 represent average over time and virus orientations in the solution. To determining the outer radius of SV40 and to follow the change of the radius at varying pH values we used a spherical, low resolution model, to represent $\Delta \rho (\overrightarrow{r})$ (see Section 3 in the SI). To follow the structural changes in wtSV40 beyond the average outer radius, we extended the resolution of the models to a higher q range, where the details of the capsid structure contributed to the SAXS curve (see Section 10 in the SI). The SAXS intensity curves of polydispersed samples were modeled (I_model) as a linear combination of the contributing components, I_j(q), weighted by their mass fraction x_j:

$$I_{\text{model }}(q)=\sum _{\text{j}}{x}_{\text{j}}{I}_{\text{j}}(q)$$

where

$$x_{\text{j}}=m_j/{\Sigma }_j{m}_j$$ (2)

and mj is the total mass of component j in the solution.