Close this search box.

Inter-coat protein loading of active ingredients into Tobacco mild green mosaic virus through partial dissociation and reassembly of the virion – Scientific Reports

We investigated the assembly/disassembly phase diagram of TMGMV and determined conditions amenable to AI entrapment. The primary goal was to achieve “breathing” without full disassembly. First, we focused on pH-induced structural changes and AI infusion. In this approach the process required extensive parameter optimization, including pH (7 to 8), incubation time (2 to 24 h), protein concentration (100 to 500 AI equivalences per CP), as well as the AI addition intervals (one feed vs. daily increments). The latter was a critical parameter: bulk additions led to severe aggregation and insolubility—likely as a result from the hydrophobic AI binding to the nanoparticle surface promoting interparticle association and aggregation. We determined best results were obtained when the AI was added in daily increments over 10 days; this assured particle stability and AI loading (see below). In brief, 10 equivalences of AI per CP were fed daily at pH 7.5 and then left stirring overnight; this process was repeated for 10 days. After completion, excess reagents were removed by spin filtration, samples reconstituted in buffer, and stored at 4 °C until further processing.

DMSO was used to disrupt inter-coat protein interactions as a second approach. For some AI, in particular those that are highly hydrophobic (e.g. IVM and FLP), this was favorable. The benefits of this approach are two-fold: increased solubility of AI leads to a higher effective concentration to drive infusion and the cosolvent prevents AI precipitation which interferes with the infusion process. To further improve on this process, the TMGMV preparations were subjected to magnetic stirring and fed from the top of the tube, preventing any short-term spikes in AI concentration that may promote precipitation. For ivermectin and fluopyram, both compounds exhibited immediate precipitation under the ‘pH approach’. In stark contrast, the ‘DMSO approach’ demonstrated to be feasible with no visible aggregates. Additionally, the increased agitation (continuous stirring), bond disrupting effect of DMSO, and higher concentrations of AI in solution—all factors should drive AI infusion more quickly helping to maintain the TMGMV nanoparticle structure avoiding aggregation. The optimized procedure is shown in Figure S1 and described in methods.

TEM imaging was performed and rod shaped virions were observed (Fig. 2). Most intriguing was the notable structural change upon AI-loading: AI-laden TMGMV appeared swollen and the structural transitions suggested AI entrapment. To gain insights into the degree of structural changes, quantitative TEM image analysis was performed comparing native vs. the AI-laden TMGMV (n = 30). Based on the negative stain, native virions were 15.7 nm in width in average (± 1.9 nm), which is an underestimation to native TMGMV of 18 nm. This could be due to the uranyl acetate negative staining yielding a heavy shadow on the borders of the virion; this artifact may lead to the apparent reduced width. While there was no statistical significance between native TMGMV and controls that underwent both “breathing” methodologies with no AI added, for each AI added the increment in comparison to native TMGMV was observed in Fig. 2. FLP- and IVM-loaded TMGMV displayed a maximum width of 18 or 23 nm, respectively. In other words, fluopyram loading resulted in a 14% width increase and ivermectin-loading resulted in a 46% width increase compared to native TMGMV (p-value =  < 0.00001). Meanwhile, CTD and RIF loading produced 22–27 nm rods, significantly thicker than native TMGMV; the width increase for clothianidin and rifampicin was 65% and 73%, respectively, compared to native TMGMV (p-value =  < 0.00001).

Figure 2
figure 2

TEM and image analysis of the non-covalently conjugated viral nanoparticles. ****p-value is < .00001, n = 30. CTD = clothianidin, FLP = fluopyram, IVM = ivermectin, RIF = rifampicin.

Next, to observe any possible alterations in the secondary structure of TMGMV after exposing the particles to breathing and infusion, CD was performed (Fig. 3). The effects of structural motifs on CD are additive and can be challenging to deconvolute46. Rather, the differences between spectra of treatment groups can signify whether or not structural changes occurred. The most intense signal for protein or virus CD is around 205–220 nm, which represents the sum of contributions from alpha helices, beta sheets, and random coil. The shift of the global minimum from 208 to 220 nm suggests a larger contribution from aggregation behavior or alpha helical content than beta sheets in both, the pH and DMSO samples. Other than that, CD indicated no changes in structure, and this is as expected. In this protocol, CPs are dissociated yet not denatured—thus, there is no expectation for the CP structure itself to change. Rather the supramolecular assembly is altered with inter-coat protein interactions likely distorted through the AI cargo. In the range of 208–220 nm, the AIs tested all showed similar molar ellipticity profiles, suggesting there were no differences in the secondary structure. In the near UV range (250–320 nm), where RNA is the highest contributor to these spectra, a similar trend holds with the AIs, showing each shares a general shape of signal profile. These results suggest that both the pH and DMSO approaches for breathing do not significantly alter the secondary structure of TMGMV. In summary, the modest pH elevation and relatively low volume fraction of DMSO used for these breathing experiments were not expected to alter the secondary structure of the virion or its CP, but rather the tertiary structure of the nucleoprotein assembly to allow inter-coat protein loading47,48.

Figure 3
figure 3

Circular dichroism spectra for TMGMV samples loaded with AI as indicated (control = native TMGMV). CTD = clothianidin, FLP = fluopyram, IVM = ivermectin, RIF = rifampicin.

To further verify structural integrity of the AI-laden TMGMV particles post processing and purification, SEC was performed. SEC measurements showed no significant difference between native and AI-laden TMGMV for any AI showing the typical elution profile from the Superose6 Increase column with elution at  ~ 10 mL and an A260:280 ratio of 1.2, indicative of intact TMGMV, where 260 nm denotes RNA absorption and 280 nm protein absorption (Figure S2). However, it is of note that SEC does not provide the resolution to show differences in width or length of TMGMV nanoparticle formulations.

In addition to changes in the width of the AI-laden TMGMV, we noted differences in the length of the nucleoprotein assemblies with or without AI. Overall TMGMV samples are heterogeneous in length; several disks were apparent indicating partial disassembly and breakage. Also, blob-like structures were observed in the IVM sample. These are believed to be IVM aggregates of precipitated of the highly hydrophobic IVM (Fig. 2). Nanoparticle length was measured through image analysis using ImageJ software. Overall, AI-laden samples were shorter compared to treated (but not AI-laden) or untreated TMGMV: the pH treatment showed slightly less breakage, having a higher distribution of lengths, compared to the DMSO treatment (Fig. 4). DMSO treatment showed the majority of the TMGMV below 100 nm, and this was independent of AI or loading method.

Figure 4
figure 4

Comparison of the length of virions after AI infusion. Image analysis from transmission electron microscopy: (A) pH breathing methodology; (B) DMSO-induced breathing methodology. CTD = clothianidin, FLP = fluopyram, IVM = ivermectin, RIF = rifampicin.

Using HPLC, we proceeded to quantify the TMGMV-loaded AI (Table 1). Using the pH-based method for infusion, CTD and RIF showed successful loading after 10 days of batch loading the AI in solution, achieving  ~ 1108 molecules per virion for clothianidin and 738 molecules per virion for rifampicin. In the cases of fluopyram and ivermectin, a large amount of precipitation was observed when the pH method was used for AI loading; this likely stripped the virus from solution and made the AI inaccessible for diffusion into the virus. Therefore, there was negligible loading of FLP or IVM per TMGMV (see Table 1). When comparing these results to the TEM micrographs and changes in the aspect ratio of the AI-laden TMGMV, the amount of AI per virion and changes in morphology appear to correlate: the more AI is loaded the larger the width increase. CTD and RIF resulted in the significant width increases (> 65% over native TMGMV), while FLP only resulted in  < 15% width increase. IVM may be an outlier: while IVM-loading was not apparent, significant changes in morphology were observed and this may be attributed to challenges in extraction of IVM or molecular properties of IVM that permanently distort the structure of TMGMV without permanent loading of the AI.

Table 1 Quantification of AI in samples by HPLC using the pH and DMSO infusion techniques.

Using the DMSO method, the AI loading efficiency for FLP and IVM was improved. FLP loading resulted in 186 moieties per virion, which is a  ~ 12-fold increase compared to the pH method. A similar result was observed for IVM, with a  ~ 21-fold increase allowing  > 60 molecules to be loaded per virion. However, for agriculture applications, the IVM loading is considered low and would require further optimization (for example, we recently demonstrated an alternate and efficient loading strategy for IVM by use of spherical protein nanoparticles49). Interestingly for CTD, the DMSO method was slightly less effective with 995 moieties per virion loaded (a 10% decrease over the pH method), and RIF-loading has a modest increase aching 1.5-fold higher loading reaching 1104 molecules per virion.

Lastly, to probe robustness of the methods, we also performed loading of a chemotherapeutic (DOX) and near-infrared fluorophore (Cy5) with utility in nanomedicine (drug delivery and imaging). In each case, swelling of the TMGMV diameter was observed by TEM while SEC confirmed intactness of the assembly and co-localization of the fluorescent cargo with the TMGMV carrier. Both DOX and Cy5 are fluorescent molecules, there loading was quantified using absorbance spectra and their extinction coefficients resulting in  ~ 615 DOX per TMGMV and  ~ 80 Cy5 per TMGMV (Figure S3).

Together, we developed a pH and solvent-based approach to load various AIs into TMGMV nanorods. AI-infused TMGMV particles remain intact yet are swollen as a function of the AI. Data indicates that for some compounds the DMSO method may be favored. There is room for further optimization and tailoring the infusion period, cosolvent choice and concentration, mixing rates, and increments of AI as well as its concentration added.

Soil mobility studies indicated that there is a similar distribution of the non-treated and treated (non-loaded) TMGMV throughout the soil (Figure S4, original blots/gels are presented in Figure S5). These suggest that there is no alteration of the properties of the virus and its ability to move through soil. Assays performed with treated and loaded (DOX or Cy5) TMGMV, it could be observed that the cargo remained with the virus and got distributed through the soil column similarly as non-loaded TMGMV (Figure S4, original blots/gels are presented in Figure S5). These indicates that the creation of these pockets favors the interaction of the hydrophobic cargo and the virus particle, allowing it to travel further into the soil, instead of it being very exposed to the surface and having it interact strongly with the soil.

To gain a better understanding of the molecular properties that lead to inter-coat protein loading of Ais, we compared their aqueous and organic partition coefficients (logP), molecular weights, and surface charge distributions. A summary of these properties can be found in Table 2 and Fig. 5. IVM and FLP had the highest logP values of 4.4 and 3.33, respectively. CTD and RIF had values of 1.3 and 2.4, respectively50,51,52. The values for IVM and FLP indicate the molecules are water insoluble, which matches well to our observations in the loading experiments. This could explain why the same effective morphology changes using these Ais was achieved within 1 day using DMSO versus 10 days for the pH approach, as the effective concentration of AI in solution was much higher. Another factor to consider regarding infusion efficiency is their size. It is possible that larger molecules could have steric hindrance when entering the spaces between coat proteins during these measurements. When we analyzed the changes in particle width using both approaches compared to their molecular weights, we see clothianidin (249.68 Da) had the largest change in width and FLP (396.71 Da) had the smallest change in width. RIF (822.94 Da) and IVM (875.1 Da) had intermediate values for changes in width. There is no clear trend in this set based on molecular weight, so AI size does not seem to be the limiting factor for loading using this approach.

Table 2 Comparison of the Ais aqueous and organic partition coefficients (logP) and molecular weights.
Figure 5
figure 5

Surface charge distribution of AI molecules. A ball and stick model of the Ais with a space-filling mesh surface surrounding them. Atoms are colored so grey is carbon, white is hydrogen, blue is nitrogen, red is oxygen, yellow is sulfur, chlorine is green, and chartreuse is fluorine. The surface mesh is colorless if there is no electrostatic potential, blue if positive, and red if negative. The conformations of the molecules are energy minimized.

Beyond the range of small molecules, steric hindrance would be expected to dominate. From the electron density plots of the Ais, we observed ivermectin and rifampicin have large regions with no charge and small regions of small charge that are largely separated, creating mildly amphiphilic molecules. On the contrary, FLP and CTD are much smaller and have a higher surface area of charge. Because TMGMV is zwitterionic in nature but also contains many hydrophobic interfaces, it is challenging to isolate the predicted changes in morphology to a single physicochemical interaction. The amphiphilic, charged, compact, and flexible structure of clothianidin may all work together to alter the morphology of TMGMV.

To gain some insight into how the Ais interact with the coat protein surface, molecular docking experiments of TMGMV CP (PDB: 1VTM) and the four Ais were conducted. In the analysis, the top 20 docking conformations were analyzed for their binding energy and the residues involved in stabilizing the AI. These data do not suggest the Ais are proper ligands for TMGMV CP, but rather identify putative residues that may be implicated in inter-coat protein loading. True ligand interactions have been reported for heats of binding greater than 8 kcal mol−1, while a majority of these interactions fall within 3–8 kcal mol−153,54. Table S1 summarizes the regions of binding, their function for TMGMV, and the residues specifically identified to stabilize the AIs. Figure 6 shows examples of docked AIs on TMGMV and the implicated residues, and Figure S6 shows the heats of binding for each conformation as calculated by Autodock 4. From the simulated docking, we observe that of the 20 best binding sites on TMGMV CP, all 4 AIs have many sites that are likely inaccessible. Depending on the mechanism of separation of TMGMV CPs (between CPs versus between disks), there are up to 10 accessible sites for IVM, 8 for RIF, 11 for FLP, and 5 for CTD. The binding energy distribution in Figure S6 shows rifampicin has the highest heats of binding to the surface, followed by ivermectin, then FLP and CTD. Despite a high number of potential binding sites, IVM is a very large molecule and would require a high degree of separation of CPs to intercalate into the virion. Its relatively high affinity may manifest in transient surface binding which can disrupt inter-coat protein bonds, explaining the widening of TMGMV in the presence of IVM. Ultimately, the IVM is not detectable during quantification, suggesting it does not stay bound to TMGMV. RIF has the highest heat of binding to TMGMV CP, loads well onto TMGMV, and induces morphological changes on TMGMV. It shows improved loading in the presence of DMSO compared to the pH approach, suggesting the structural changes induced by DMSO allow this relatively large molecule access to binding sites. Despite having 11 potential binding sites, FLP also had some of the lowest heats of binding and had the highest affinities for the inner channel. Because this molecule is insoluble and relatively small, it may preferentially partition to the inner channel than to load between the CPs. CTD had 5 accessible sites on the exterior according to the docking model, but also had some of the lowest binding energies. However, its relatively small size and surface charge distribution may have aided in its binding and disruption of structure between TMGMV CPs. CTD demonstrated some of the highest loading by HPLC and largest differences in virion width, suggesting the properties of this molecule make it well suited for this approach. With more robust docking analysis and a larger library of small molecules to load between TMGMV CPs, it may be possible to pinpoint molecular properties of the AIs and individual residues of TMGMV CP that are implicated in these binding events.

Figure 6
figure 6

Docking of AI to TMGMV. TMGMV coat protein (PDB: 1VTM) where red arrows indicate the location of nucleotides (A). A space filling model of a putative binding site for RIF (B), IVM (C), FLP (D), and CTD (E). Red and blue spheres indicate negative and positive electrostatic contributions, respectively. The three-letter amino acid abbreviation are followed by the residue number.