Introduction

Preventative therapeutics such as vaccination, including campaigns against polio, smallpox, and measles, has been demonstrated to be vitally important for prevention of disease and reducing burden on the global health care system1. In 2010, the World Health Organization (WHO) estimated infectious disease was responsible for 15 million deaths2. Most, if not all vaccine products require a form of refrigerated storage and transport, known as the cold chain, to prevent loss of efficacy. Products such as mRNA-LNPs and some viral vectored vaccines require a −70 °C cold chain which is a significant logistical and economic burden preventing dissemination of critical therapeutics to hard-to-reach places and populations. It has been estimated up to 80% of the cost of vaccination programs are related to cold chain requirements3. Another concern for public health is the need to build both static and rotating stockpiles of vaccines for outbreaks, which is also heavily dependent on stable vaccine products4.

A widely used method for improving vaccine stability is the addition of stabilizing excipients and removal of water, thus decreasing the mobility of molecules and preventing water-based degradation mechanisms5. Solid state forms of biopharmaceuticals are more thermal stable and less prone to degradation by both chemical and physical mechanism5,6,7. Two main theories have been postulated to explain this phenomenon: vitrification and water replacement theory and have been described in depth7. The most widely utilized drying method used commercially is lyophilization, where liquid samples are frozen, and water is removed by sublimation8,9. However, some biopharmaceuticals are sensitive to ice crystals, and cold denaturation which can cause significant loss of activity even with cryoprotectants10. Another widely used technology for drying material is spray drying. This technique does not create the stress related to freezing and is performed as a bulk process improving scalability8. However, the spray drying process imparts heat and shear stresses onto the biomaterial, leading to potential for greater process losses9. While not commercially utilized yet, vacuum and foam drying technologies have been demonstrated as viable methods of drying biopharmaceuticals11. While differentiation between foam and vacuum drying has not been well delineated, foam drying has been described as the intentional formation of a foam while vacuum drying does not11. Both techniques have several advantages over lyophilization and spray drying as they are less energy intensive, impart less stress on the biopharmaceutical during dehydration, and have been demonstrated to give greater stability to biologically active organisms such as probiotic bacteria11.

One viral vector demonstrated to have high therapeutic value is recombinant Vesicular stomatitis virus (rVSV), an enveloped negative strand RNA virus from the Rhabdoviridae family12. rVSV was engineered to recombinantly express heterologous antigens to present to the host immune system and has shown efficacy for therapeutic cancer treatments as well as a vaccine delivery platform13,14,15. Several key advantages of rVSV have been characterized as a vaccination vector, including relative ease of high titer production, ability to express and present heterologous antigens on the virion surface, low seroprevalence in human population, low pathogenicity, and ability to induce humoral and cellular immune responses12. rVSV vaccines also develop a protective immunity towards their targeted disease quickly, between 7–10 days post administration13. However, pseudotyping rVSV with heterologous G-proteins severely decreases the manufacturability of the vector, increasing the value of even modest improvements in stabilization16. The best example of rVSV as a vaccine vector is the Ebola vaccine ERVEBO, where the VSV-G protein was pseudotyped with the Zaire Ebola glycoprotein17. ERVEBO has shown tremendous efficacy, with 100% efficacy in phase III trials and another study in the Democratic Republic of Congo having ~ 97.5% efficacy in 94,000 individuals18. More recently, ERVEBO has been demonstrated to decrease the risk of death by 50% even if administered after the patient has been infected19. This leads to the potential of the VSV platform as a therapeutic vaccine for other diseases (reviewed in20). A significant drawback of this vaccine is the requirement of storage and transportation at −70 °C. Once thawed, the maximal storage is 14 days at 2–8 °C or 4 h above 9 °C21. This limitation severely reduces the widespread dissemination of the vaccine in areas of Western Africa where Ebola outbreaks have occurred: regions with a greater need and less availability of cold chain infrastructure.

Previous work demonstrated thermal stability of therapeutic viral vaccine by air drying in pullulan and trehalose-based formulations22. Two viral vaccines, Herpes Simplex Virus (HSV)-2 and Influenza A virus showed stability and retained efficacy for over 2 months at 40 °C when dried in the carbohydrate mixture. Additional work by Toniolo et al. demonstrated rVSV can be formulated with trehalose and dried with spray drying technology to provide some short-term thermal stability23. Trehalose is a disaccharide and widely used as cryoprotectant or stabilizing agent whereas pullulan is a long chain polysaccharide involved in desiccation resistance (isolated from the fungus Aureobasidium pullulans). Both carbohydrates are readily available, inexpensive, and FDA approved22.

In this work we demonstrate the application and optimization of foam dry methodology for the thermal stabilization of rVSV. Several different drying schedules and excipients were assayed, and long-term stability of formulated and dried rVSV was tested at 4 °C, 25 °C, and 37 °C. Freeze drying methodology was demonstrated to be significantly inferior to vacuum drying, and an optimal two stage dry schedule starting at 4 °C and ending at 25 °C over 24 h was identified. We also identified a key excipient, serum albumin, is required for improving rVSV thermal stability. Finally, this work identified by X-ray diffraction (XRD) divalent cations in dried formulations under nitrogen backfilling can cause amorphous material to become crystalline, which is detrimental to long-term stabilization of rVSV at 37 °C. Overall, this work describes a significant advancement for rVSV thermal stability relative to previously reported literature.

Results

Air drying does not provide stability to VSV; freeze drying with trehalose alone does not stabilize, but pullulan and trehalose provides improved protection

To the best of our knowledge, the best thermal stability data published to date demonstrated a 4-log decrease in VSV titer after spray drying and thermal treatment at 37 °C for 7 days when formulated with trehalose23. Previous work demonstrated thermal stability of viral vaccines by air drying with pullulan and trehalose22. To assess if the air-dried methodology (Table S1) could stabilize VSV, we formulated the virus in two different formulations and tested process loss and thermal stability at 37 °C (Fig. S1). The formulations utilized in22,24 (Table 1; formulation F1) had complete loss of titer during the drying process, and formulation F2 had high process loss of 1.78 ± 0.04 Log PFU. After 14-day incubation at 37 °C, a total PFU loss for VSV in F2 was 3.5 ± 0.17 log, making it clear air drying was not a viable strategy for stabilizing VSV.

Table 1 Summary of formulation components and concentrations used in this study. For powdered components, the percentage is calculated as w/v.
Full size table

Our next experiment was to understand if a standard freeze-drying schedule (Table S1) with formulation F3 or trehalose alone (F4) could outperform the published data (Fig. 1). Visual observation of the dried film revealed F2, and F5 films had cake-like appearances, whereas F3 (higher concentration of pullulan and trehalose) and F4 led to a collapsed cake structure (Fig. 1A). For the three formulations with pullulan and trehalose, no significant difference was observed for process loss, and total log titer loss after 7 days at 37 °C (Fig. 1B). Interestingly, trehalose alone (F4) performed the worst, with the highest process loss and a complete loss of detectable plaques after 7 days at 37 °C. Despite the improvement of 1 log titer in stability of the freeze dry methodology after 7 days at 37 °C relative to spray drying, we did not view ~ 3 log titer loss of formulated VSV as acceptable, and therefore transitioned to exploring vacuum drying methodologies for further process improvement.

Figure 1
figure 1

Stability of VSV dried by freeze dry methodology. (A) Cake morphology after formulation and drying. Cake-like structures were observed for both F2 and F5, whereas collapsed cakes were observed for F3 and F4. (B) VSV stability for process loss and thermal challenge over 7 days. (C) Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

Vacuum drying at 25 °C has high process loss with formulation F7, but no thermal stability loss over 14-day incubation at 37 °C

Unlike freeze drying, vacuum drying does not induce stresses such as freezing and/or ice crystals potentially damaging the VSV lipid membrane. We next assessed if drying VSV in formulation F7/F8 at room temperature with a vacuum pump could improve process loss and thermal stabilization (Fig. 2). Samples were dried for 48 h at room temperature under a vacuum pressure of 1e−4 mBar (Table S1). Visual observation of VSV dried in pullulan and/or trehalose (F7/F8) had a bubbly foam architecture, whereas VSV dried in buffer alone (F6) did not yield a film (Fig. 2A). All three conditions had significant process loss of VSV titer due to drying, with formulation F6 providing no recoverable plaques after 7 days incubated at 37 °C (Fig. 2B/C). Interestingly, both F7 and F8 films did not show significant additional thermal challenge loss over two weeks at 37 °C. This observation of VSV being thermally stable once dried in a film led us to explore different mechanisms to improve the drying process with additional excipients and improvements to the drying schedule.

Figure 2
figure 2

Stability of VSV dried by Vacuum foam dry methodology. (A) Cake morphology after formulation and drying. No film was observed for F6, whereas bubbly foam was observed for F7/F8. (B) VSV stability for process loss and thermal challenge over 14 days. Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

Addition of serum albumin significantly improves stability of VSV in a lot-to-lot dependent manner

Several experiments were conducted testing different pH’s (6.8–8.1), concentration of Tris buffer (between 10-50 mM), and excipients including hydroxyectoine, ectoine, β-cyclodextran, PEG 200, PEG 4000, PEG 6000, Histidine, Glutamic acid, and increased gelatin concentrations with no improvement in VSV stabilization. We also observed no difference in process loss or thermal stability of vacuum dried VSV between 24 and 48 h dry schedules.

We observed the addition of serum albumin, specifically bovine serum albumin (BSA), improved stability of VSV in a lot-to-lot dependent manner (Fig. 3). Two different lots of purified VSV had significantly different total protein content as determined by Bradford assay both pre and post dialysis into the formulation buffer despite no significant difference in the titer of the viral stock (3e10 vs 4e10 PFU/mL; Fig. 3A). We hypothesize the difference is due to cell culture carryover from the purification protocol. Analysis of the process loss of the conditions tested revealed VSV lot 1 (low total protein) had significantly more process loss in the absence of BSA (F9) compared to the addition of 0.5% (Fig. 3B). VSV lot 2 (high total protein) showed no difference in process loss with (F10) or without BSA (F9). For the samples with the two lots mixed at a 1:1 ratio, no significant difference was observed between formulation F9 and F10 for process loss. After 18 days incubation at 37 °C, the VSV lot 1 samples had ~ 2 log difference between +/− BSA, the Lot 2 samples had no difference, and the mixed group was improved by ~ 0.7 logs with the addition of BSA. VSV manufactured in GMP processes are typically done in FBS-free conditions, resulting in less protein carryover during the purification of the viral vector. These results led to the conclusion that the addition of serum albumin, and in this case BSA, improves stability of VSV in low total protein conditions.

Figure 3
figure 3

Role of BSA in stabilizing VSV dried by Vacuum foam dry methodology. (A) Protein quantification of VSV lots before and after dialysis into formulation buffer. (B) VSV stability for process loss and thermal challenge over 14 days at 37 °C. (C) Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

Long term stability of VSV in F2 formulation at 37 °C demonstrates stable titer after initial loss

To determine the long-term thermal stability of the F2 formulation for VSV, we employed an AdVantage Pro drying system for greater control of the foam drying schedule. As described in previous literature, most viral vectors are formulated and dried with sorbitol and gelatin, which we employed as a control formulation25. Samples were dried for 24 h at 25 °C at a pressure of 16 uBar, manually stoppered and crimped (Fig. 4). F2 formulation dried samples were observed to be a bubbly film, whereas the gelatin/sorbitol samples were a flatter and more uniform glassy film (Fig. 4A). Both formulations had similar process loss (Fig. 4B/C). After 1 week incubation at 37 °C, no viable VSV plaques were observed for the F11 samples and were confirmed for week 2 and 4 samples. Conversely, F2 samples showed a total log loss of 2.2 for 1 and 2-week samples, and 3.0 ± 0.19 log loss at 4 weeks at 37 °C. Interestingly, we observed from weeks 4 to 27 a smaller additional log loss of VSV titer, resulting in a total loss at 27 weeks of 4.27 ± 0.07 log VSV at 37 °C. Modelling of the VSV in formulation F2 with a one phase decay non-linear regression model results in a R2 value of 0.94 and a 95% confidence interval of the plateau between 3.84–4.60 log PFU loss. These data suggest the trehalose and pullulan-based formulations, once stable, have limited additional loss of VSV titer.

Figure 4
figure 4

Long term stability of VSV dried by Vacuum foam dry methodology. (A) Film morphology. (B) VSV stability for process loss and thermal stability over 27 weeks at 37 °C. Solid line represents one phase decay regression model with dotted lines for 95% confidence interval. (C) Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

Two stage temperature drying schedule improves process loss and short-term stability at 37 °C

We next sought to improve the drying schedule of foam drying the PT formulation to improve VSV stability. Several different drying schedules were assayed and compared with the F2 formulated VSV (Fig. S2). Of the conditions tested, the two worst total log loss by day 14 were the freeze dry schedule (schedule 4) and drying at 4 °C under 16 μBar pressure (schedule 5). The schedule that had the least amount of process loss was a 24-h dry schedule (Table S1). Comparable total log losses were observed with alternative schedules, but we decided to continue optimizing the VSV stability work with the 4/25 °C schedule due to the lowest observed process loss.

Long-term stability with 4/25 °C dry schedule has less titer loss with formulation F2 compared to drying at 25 °C alone at 37 °C thermal challenge; significantly less PFU loss at 4 °C

Using the optimized drying schedule, we assayed the long-term stability for 32 weeks of our formulations at 4 °C, 25 °C, and 37 °C (Fig. 5). We compared two formulations, F2 and F12, dried over 24 h with 4/25 °C dry schedule which were manually stoppered and crimped (Fig. 5A). Both formulations had comparable process losses and both statistically lower than F2 dried with 25 °C alone dry schedule (P = 0.001 and P = 0.002 respectively). For the thermal stability test at 4 °C (Fig. 5B), the titer loss for both formulations increased to 1.23 ± 0.07 by 8 weeks, and at the end of 32 weeks we observed total titer loss for F2 of 1.82 ± 0.13 log and for F12 of 1.93 ± 0.18 log (P = 0.29). Modelling with a one phase decay non-linear regression model results in a R2 value of 0.93 for both F2 and F11. The calculated plateau for each formulation at 4 °C was 1.87 and 2.12 log PFU loss for F2 and F11 respectively.

Figure 5
figure 5

Log term stability of VSV formulated with F2 or F12 and dried with 4/25 °C dry schedule at 3 different thermal challenge temperatures. (A) Dried film morphology. (B) VSV stability at 4 °C thermal challenge. (C) VSV stability at 25 °C thermal challenge. (D) VSV stability at 37 °C thermal challenge. For panels (B–D), Solid line represents one phase decay regression model with dotted lines for 95% confidence interval. (E) Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

For samples incubated at 25 °C (Fig. 5C), both formulations had significant titer loss up to week 8 with F2 outperforming F12 for total PFU loss (P = 0.0021). Interestingly, for both formulations only a slight increase in loss was observed between week 8 and week 32 samples. At the conclusion of the experiment at 25 °C, a total loss was observed for F2 was less than the loss observed for formulation F12 (P = 0.0001). Modelling with a one phase decay non-linear regression model results in a R2 value of 0.94 for F2 and 0.98 for F11. The calculated plateau for each formulation at 25 °C was 2.40 and 3.00 log PFU for F2 and F11 respectively.

A similar trend occurred for samples incubated at 37 °C (Fig. 5D). By 8 weeks, the F2 formulation total PFU loss was 2.88 ± 0.10 log and F12 was 3.33 ± 0.28 log, with the F2 dried with 4/25 °C outperforming both the F12 formulation and F2 dried at 25 °C (P = 0.0012 and P = 0.0018 respectively). After 32 weeks at 37 °C, the F2 formulation total loss was 3.13 ± 0.10 log but significantly more loss was observed for F12 (4.20 ± 0.16 log; P < 0.0001). For the 25 °C dry schedule, formulation F2 total log loss was 4.3 ± 0.07 at week 27, like the loss observed for F12 at 32 weeks. Modelling with a one phase decay non-linear regression model results in an R2 value of 0.96 for F2 and 0.88 for F11. The calculated plateau for each formulation at 37 °C was 2.94 and 3.69 log PFU for F2 and F11 respectively. The calculated plateau for F11 was significantly less log loss compared to the data collected, suggesting additional factors may be involved resulting in greater VSV titer loss in this formulation.

Overall, no significant difference was observed between the two formulation’s stability time course at 4 °C, but the F2 outperformed the F12 at 25 °C and 37 °C with the 4/25 °C dry schedule as well as outperforming the 25 °C only dry schedule.

Backfilling vials of dried formulation F2 with dry nitrogen gas causes crystallization to occur in film decreasing viability of VSV

For commercial dried products, it is commonplace to backfill samples with an inert gas like nitrogen to improve stability by reducing chemical instabilities6. Over several experimental tests with nitrogen backfilling, we had significantly greater loss of VSV titer compared to the data demonstrated in both Figs. 4 and 5. To understand if the nitrogen backfilling was affecting the structure of the dried film, XRD analysis was performed on a thermal treated time course at 37 °C (Fig. 6). All conditions assayed were completely amorphous on day 0 before incubation at 37 °C (Fig. S3). After 7 days incubations we detected 2.2% crystallinity in the F2 films when backfilled with nitrogen gas, but no crystallinity when backfilled with atmospheric air. The crystallinity percentage decreased by 14 days incubation to 0.6% in the nitrogen filled vials. According to the XRD trace profiles, we hypothesized calcium sulfate crystals are the causative material for the observed crystallinity. Samples comprised of formulation F13 were completely amorphous for the time course independent of backfilling gas. An interesting observation of samples with UV inactivated VSV had a constant lower crystallinity from Day 0 to Day 14 of the experiment (~ 0.2%) in the F2 formulation with nitrogen backfilling but completely amorphous with F13. These results led to the conclusion in the absence of moisture in the headspace gas, crystallization can occur with the ions present in the formulation, and this stress decreases the stability of VSV.

Figure 6
figure 6

XRD analysis of dried films over 14 days incubated at 37 °C. Percent crystallinity was calculated by determining the area under the curve (AuC) of amorphous and crystalline material and dividing values.

Full size image

Nitrogen backfilled vials with formulation F13 improves serum free manufactured VSV stability over atmospheric backfilled formulation F2 modestly at 4 °C and significantly at 25 °C and 37 °C

To determine if a combination of nitrogen backfilled vials and removing all ions from the formulation would improve the VSV stability, we dried two formulations, F2 and F13 with the 4/25 °C dry schedule and backfilled with dried Nitrogen gas (Fig. 7). For this experiment, VSV was manufactured in a serum free media to simulate a GMP process26. Both dried formulations had a similar bubbly dried appearance (Fig. 7A) characteristic of a foam dried material. Both formulations had nearly identical process loss. After a 20-week incubation period, differences were observed between the formulations dependent on the incubation temperature. At 4 °C, no statistical difference in titer loss was observed between the nitrogen backfilled F13 and F2 as well as the atmospheric backfilled F2 (one-way ANOVA; P = 0.67). Modelling with a one phase decay non-linear regression model results in an R2 value of 0.91 for F13 and 0.98 for F2. The calculated plateau for each formulation at 4 °C was different, with a predicted 1.29 log PFU loss for F13, and 2.46 log PFU loss for F2. Samples incubated at 25 °C demonstrated nitrogen backfilled F13 and atmosphere backfilled F2 statistically outperformed nitrogen backfilled F2, but no difference between the two conditions (one-way ANOVA; P = 0.0008, P = 0.008, and P = 0.99 respectively). Modelling with a one phase decay non-linear regression model results in an R2 value of 0.91 for F2 and 0.92 for F13. The calculated plateau for each formulation at 25 °C was 3.10 log PFU loss for F2, and 2.38 log PFU loss for F13. However, at 37 °C, the best performing formulation after 20-week incubation was the atmosphere backfilled F2 compared to nitrogen backfilled F13 and nitrogen backfilled F2 which did not have any detectable PFU (one-way ANOVA P < 0.0001). Modelling with a one phase decay non-linear regression model results in a R2 value of 0.96 for F2 and for F13. The calculated plateau for each formulation at 37 °C was not able to be calculated for F2 and was 3.91 log PFU loss for F13. Overall, these data demonstrate that similar levels of stability can be achieved with nitrogen backfilling of F13 formulation to the atmospherically filled F2 formulation at 4 °C and 25 °C, but atmosphere backfilled formulation F2 outperforms nitrogen backfilled F13 at 37 °C.

Figure 7
figure 7

Stability time course at 3 different incubation temperatures of VSV dried with 4/25 °C dry schedule and backfilled with dried N2 gas. (A) Dried film morphology. (B) VSV stability at 4 °C thermal challenge. (C) VSV stability at 25 °C thermal challenge. (D) VSV stability at 37 °C thermal challenge. For panels (B–D), Solid line represents one phase decay regression model with dotted lines for 95% confidence interval. (E) Calculated means and standard deviation (SD) for each data point of accumulated Log PFU loss for each condition. Each data point is collected from duplicate serial dilution plating of biological duplicate vials.

Full size image

Discussion

The purpose of this study was to determine if a combination of two carbohydrates, pullulan, and trehalose, could be utilized to thermally stabilize an inherently unstable but therapeutically important viral vector VSV. Utilizing both air drying and freeze-drying methodologies did not yield satisfactory process or thermal stability losses for VSV, with losses of greater than 3 log PFU after 7-day incubation at 37 °C. An improvement in stabilization was observed under vacuum drying conditions at room temperature, but the process loss was still problematic. We delved into which additional excipients could improve the stability of VSV during the drying process and initial two weeks of thermal stability at 37 °C. While most excipients assayed showed no improvement in PFU recovery, one key excipient identified was bovine serum albumin (BSA). It appears VSV is more sensitive to drying in the absence of additional proteins in the formulation, as we observed in a lot-to-lot dependent manner. Further work will explore the potential improvements in stability from different serum albumins in a dried carbohydrate foam format.

An exploration of different drying schedules with constant low pressure revealed an optimal two stage method, starting at 4 °C followed by a second stage at 25 °C. Our hypothesis is since purified VSV is inherently thermally unstable above ~ 10 °C in solution (Dr. Lichty personal communication and unpublished observations), maintaining a VSV tolerable temperature until the water has been removed and the film has formed improves the process loss observed. A second drying stage at elevated temperatures is required to remove additional moisture from the film, improving the thermal stability of the formulation. Additional optimization of the drying schedule is ongoing to further improve the process losses observed to date.

For the three long-term stability experiments described, we observed improvements with each optimization. Process loss was improved by optimizing the drying schedule from single 25 °C temperature to a two-stage process. This change in drying schedule also yielded a long-term thermal stability at 37 °C improvement of ~ 1.3 log PFU after 27 weeks incubation. Changing the formulation to remove salt ions and backfilling the vial headspace has not shown any improvement in loss relative to formulation F2 with atmosphere backfilling, but significant improvement relative to F2 with Nitrogen backfilled vials after 12 weeks at 37 °C. This experiment is ongoing, and the data collected will be informative. For all three long term experiments in this work, we observe the vast majority of VSV titer loss during the first 6–8 weeks of incubation at 37 °C. While an additional 1 log loss is observed for 25 °C dry schedule from week 8 to week 27 (Fig. 4), VSV dried with formulation F2 and the 4/25 °C dry schedule had no statistical change in titer from week 8 until the completion of the experiment at week 32 (Fig. 6; P = 0.257). This observed plateau also occurs at 25 °C, at a lower loss of VSV titer. For F2 formulated VSV incubated at 4 °C, we do not observe the plateau in the data collected. We hypothesize since the stability of formulated VSV does not follow Arrhenius kinetics, different degradation mechanisms are occurring, and further investigation is warranted to elucidate the mechanisms at play. For the strategy of vaccine stockpiling for future disease outbreaks, being able to demonstrate a thermally stable plateau is a significant breakthrough for the VSV vaccine platform.

It is widely accepted that oxygen and moisture in the vial headspace cause decrease in the stability of the pharmaceutical agent27. Our initial characterization of the stability of VSV in dried pullulan and trehalose-based films were done with manual stoppering and crimping, resulting in atmospheric gas in the vial headspace. Contrary to our expectation, in transitioning our drying strategy to a more commercial based approach of backfilling with dried nitrogen gas with in-system stoppering, we observed significantly higher VSV titer losses at 37 °C (Fig. 7). Analysis by XRD of the dried formulation under different temperatures, timepoints, and backfilling gases revealed the presence of crystalline species (Fig. 6). Crystallinity is only observed at 37 °C in the presence of nitrogen backfilling with metal ions present with all other conditions being completely amorphous films (Fig. S3). While not as prevalent, a modest degree of crystallinity is detected when UV-inactivated VSV is present in the sample. Our hypothesis is atmospheric humidity has sufficient moisture to prevent crystallization, whereas dried nitrogen gas has low excess moisture which promotes crystal formation. Crystallinity has been previously demonstrated to decrease the stability of several classes of pharmaceutical molecules6,7. However, by removing all free ions from the formulation and backfilling with dried nitrogen gas, stability of VSV at elevated temperatures was improved.

In summary, we have demonstrated by optimizing a pullulan and trehalose-based formulation, drying schedule, and vial backfilling, we have identified conditions to thermally stabilize rVSV vial vector. To the best of our knowledge, these results are a significant improvement on any other described thermal stabilization methodology for rVSV. Future work will focus on improving our initial titer loss, as well as demonstrating vaccine efficacy with in vivo vaccination trials.

Materials and methods

Materials

Pullulan and trehalose SG dried powders were kindly provided by Nagase Viita Co., Ltd. Purified VSV serotype Indiana, with a ΔM51 mutation in the matrix gene (renders it interferon inducing and attenuated) stocks were manufactured and provided by Dr. Brian Lichty’s research group (McMaster University) as previously described28,29. Stock buffers Tris (1M pH 7.2; Teknova Cat No: T1072), 1 M CaCl2 (Sigma Cat No: 21115-100 mL), 1 M MgSO4 (Sigma; Cat No: M3409-100 mL), 5 M NaCl (VWR; E529-500 mL), and dry powders of Bovine Serum Albumin (Sigma; A9418-10G) and Gelatin (Criterion, C7921) were diluted and resuspended as described in Table 1.

VSV dialysis and formulation

Manufactured 100 μL VSV stock aliquots were stored at − 80 °C in 4% sucrose. After gently thawing on ice, viral particles were buffer exchanged into the formulation buffer using Zeba Spin Desalting columns 7K MWCO (Thermo Scientific). Viral particles were mixed at a 1:9 ratio with the formulation buffer containing Pullulan and/or Trehalose and 100 μL was aliquoted into 2 mL vials. Samples were subsequently dried either using a vacuum pump and desiccator or the SP VirTis AdVantage Pro Freeze dryer system. If required, after completion of the dry schedule on the SP VirTis AdVantage Pro, vials were backfilled with dried Nitrogen gas and stoppered in system.

Residual moisture analysis with KF titrator

Residual moisture of the dried VSV formulation was determined by KF titration with a Mettler-Toledo KF Titrator C10S system. The mass of the dried formulated VSV sample was calculated from pre-tared vials, 1 mL of Hydranal (Honeywell; Cat No 34724-1L) was added to reconstitute film and reweighed to calculate mass of solvent. To determine background moisture content of Hydranal buffer, ~ 0.6 mL of Hydranal stock was injected into the KF Titrator to determine PPM. The PPM of the dried film was determined by injecting 0.4 mL of reconstituted sample into the KF titrator. The residual moisture percentage of the dried film was calculated from the weight of the dried film, PPM of Hydranal blank, and the PPM calculated for the injected sample (Doan, 2022). Each formulation was analyzed in technical duplicates of biological replicates.

VSV plaque assay for viable PFU quantification

Vero cells (derived from African green monkey kidney epithelial cells) were employed in 6 cm petri dishes to quantify PFU of VSV. The day prior to infection, Vero cells grown in 150 mm plates to 80–90% confluency were harvested using 3 mL of 1 × Trypsin (Gibco, Life Technologies 15400-054) and incubated for 3 min at 37 °C. Cells were resuspended, quantified, diluted to a concentration of 1.4 × 106 cells/5 mL in completed media (α-MEM), plated in 6 cm dishes, and incubated overnight at 37 °C/5% CO2.

The next day, an overlay solution (1:1 ratio 1% Agar to completed media) was prepared and stored at 42 °C. Dried formulated VSV samples were reconstituted with 1 mL MQH20 with light vortexing for 15 s prior to dilution in completed media. 100 μL of diluted VSV solution was pipetted onto Vero cells in 6 cm petri dishes for infection and rocked 3 × for 15 min (total 45 min) at 37 °C. Add 5 mL of overlay solution and allow to solidify before overnight incubation at 37 °C/5% CO2. Plaque count is performed the next day, and calculations were performed to determine the viable PFU of the formulated VSV.

UV-inactivation of rVSV

To inactivate VSV for XRD analysis, a CL-1000 Ultraviolet Crosslinker (mercury lamp 254 nm wavelength) was employed. 125 μL of stock rVSV (1e10 PFU/mL) was added to a 24 well plate well and completely inactivated at 2000 mJ/cm2. Inactivation was confirmed by plaque assay.

XRD analysis of dried formulations for crystallinity

Dried formulations were prepared with the 4/25 °C dry cycle in 2 mL vials. For XRD analysis, the film was crushed into a powder and extracted from the vials. The powder samples were mounted on top of a zero-background Si wafer for the XRD measurement. Data were collected using a Bruker D8 DISCOVER with DAVINCI.DESIGN diffractometer, equipped with a Co sealed-tube source and Eiger2R 500K area detector. A 2D continuous scan from 12 to 80 degrees 2Th was collected with 0.02deg steps, at 1.2 s per step.

Statistical analysis

Statistical analysis and graphing of data were performed with PRISM software (Graphpad Prism v10.1.2). Comparison of two groups was performed with two-tailed paired t-tests. For analysis of more than two unique groups, ANOVA was utilized with Tukey multiple analyses parameter.