Anhydrous Nucleic Acid Nanoparticles (NANPs) for Storage and Handling at Broad Range of Temperatures

Abstract

Recent advances in nanotechnology now allow for the methodical implementation of therapeutic nucleic acids (TNAs) into modular nucleic acid nanoparticles (NANPs) with tunable physicochemical properties which can match the desired biological effects, provide uniformity, and regulate the delivery of multiple TNAs for combinatorial therapy. Despite the potential of novel NANPs, the maintenance of their structural integrity during storage and shipping remains a vital issue that impedes their broader applications. Cold chain storage is required to maintain the potency of NANPs in the liquid phase, which greatly increases transportation costs. To promote long-term storage and retention of biological activities at higher temperatures (e.g., +50°C), a panel of representative NANPs was first exposed to three different drying mechanisms - vacuum concentration (SpeedVac), lyophilization, and light assisted drying (LAD) - and then rehydrated and analyzed. While SpeedVac primarily operates using heat, lyophilization avoids temperature increases by taking advantage of pressure reduction and LAD involves a near-IR laser for uniform drying in the presence of trehalose. This work compares and defines refinements crucial in formulating an optimal strategy for producing stable, fully functional NANPs and presents a forward advancement in their development for clinical applications.

Short Summary:

Rationally designed nucleic acids have experienced success in various applications. To study the best practices for maintaining their structural integrity over a longer time, we present an analysis of representative nucleic acid nanoparticles (NANPs) dehydrated via vacuum concentration, lyophilization, and light-assisted drying. Our work is a forward advancement in formulating optimal strategies for handling NANPs at broad range of temperatures.

Graphical Abstract

1. Introduction

The advent of personalized medicine has sparked a pressing demand for novel nanomaterials to control molecular behaviors in various biological systems. Among the variety of biomolecular candidates, nucleic acids have been recognized as promising components of formulations intended at addressing critical health problems since these intriguing biopolymers naturally regulate the flow of genetic information in all forms of life. Aside from already approved therapeutic nucleic acids (TNAs)^[1–3], the keener consideration of RNA and DNA for biomedical applications has become even more apparent in the current pandemic, with two mRNA COVID-19 vaccines authorized for global use^[4–8].

Furthermore, the improved understanding of RNA’s folding along with increasingly potent computational algorithms^[9–15] has expanded the repertoire of nucleic acid motifs ^[12,16–18] that when combined^[19,20] can form nucleic acid nanoparticles (NANPs). Rationally designed NANPs, further optimized for biomedical applications^[21–23], established innovative technology now reaching the stage of preclinical development^[21]. The success of NANPs stems from their structural modularity that allows for the precise control over biological and physicochemical properties, ranging from thermodynamic and chemical stabilities to immunorecognition and therapeutic efficacy^[24–26]. In addition to its value in biomedicine, NANPs have been utilized effectively across a broad range of chemical and biotechnological applications. They have found success in the form of nanoscaffolds for TNA coordination^[27], responsive biosensors^[28,29], and logic circuits^[30]. Fine-tuning the unique set of NANPs’ architectural parameters allows for the construction of an expansive library of diverse multifunctional tools with strong potential for therapeutic^[24,27–37], biotechnological^[39], and biochemical applications ^{[24,27–32,38–40]}.

Despite the major advantages offered by NANP technology, one drawback is in the structural diversity of these novel materials which leads to formidable barriers in optimizing protocols for their scaled up production and maintenance^[41]. Currently, the standard approach for the long-term handling and transportation of NANPs relies on a cold chain storage that involves their refrigeration or freezing in the assembly buffer. This route, however, becomes challenging and expensive for long-distance transportation since any breach in the cold chain (e.g., caused by processing through customs regulations) can result in the loss of NANPs’ activity. In addition, cold chain storage may not be feasible in low resource settings due to a lack of available infrastructure. Thus, in preparation for physical transportation or long-term storage at ambient temperatures, NANPs need to be dehydrated to limit all the aforementioned complications^[42]. However, there are no reliable protocols suitable for the variety of NANP structures currently available. An effective strategy to address this deficiency would be to adjust the methods used in NANPs’ preparation for storage.

In this project, we compare several dehydration protocols applied to a set of NANPs and analyze the retention of NANPs’ structural integrities and biological functions after their storage at a broad range of temperatures followed by rehydration (Figure 1). We selected four representative NANPs to determine which of their architectural parameters and physichochemical properties (all summarized in supporting Table S1) are the most vulnerable to anhydrous storage. The effect of NANPs’ dimensionality was assessed by comparing globular cubes^[43,44] to planar rings^[45] and to linear fibers^[45]. The influence of chemical composition, that largely influences NANPs’ integrity at different pH and their resistance to nuclease degradation, was tested for RNA cubes and their DNA analogs, made of identical sequences. The effect of thermodynamic stability was assessed by also comparing RNA (T_m ~56 °C) to DNA (T_m ~36 °C) cubes^[26,46]. The size of NANPs was assessed by comparing RNA cubes (d 12 nm) to RNA rings (d~18 nm)^[26,46] and the contribution from NANPs’ design principles and their connectivity were evaluated by comparing cubes with rings (or fibers). In cubic NANPs, all individual strands assemble only via intermolecular hydrogen bonds and this type of assembly is characteristic for DNA nanotechnology and DNA origami^[47]. Contrarily, RNA ring and fiber NANPs require folding of all monomers via intramolecular hydrogen bonds to expose embedded long-range interaction motifs for their magnesium-dependent intermolecular assembly. This approach is more characteristic of RNA nanotechnology^[48]. We hypothesized that depending on the structural characteristics, some NANPs may express high sensitivity to changes in environmental conditions during the dehydration processing that in turn may cause changes in the physicochemical parameters of NANPs which are linked to their intended biological functions and immunorecognition^[49,50].

Figure 1.

Schematic outline of experimental work. (A) Representative NANPs used in this project were synthesized and verified. (B) NANPs were processed via three different methods: SpeedVac, Lyophilization, and LAD. (C) Along with a control sample stored in-solution, all dehydrated samples were stored for 24 hours at various temperatures then rehydrated (D) and analyzed (E) by native-PAGE and AFM. Biological functionality (RNAi activation) and immunological properties of the rehydrated NANPs were confirmed by cell studies.

For the dehydration step, we selected three different moisture removal techniques that are relatively quick and feasible for applications of larger volumes of NANP solutions. The most straightforward protocol for NANP dehydration utilized a vacuum concentrator (SpeedVac) which centrifuges samples while exposing them to a combination of heat and infrared radiation (IR). Another tested approach was lyophilization (Lyo), a process that involves the freezing and subsequent sublimation of a sample. It has been considered the “gold standard” of protein dehydration^[51] and has also been utilized in the preservation of biopharmaceutical products^[52] and chitosan nanoparticles^[53]. As opposed to SpeedVac, Lyo allows to observe the impact of using dehydration methods that do not involve heating the sample to high temperatures. We also included recently developed light-assisted drying (LAD) which utilizes heat from a near-IR laser to evaporate the water in an aqueous solution^[51]. An important component of this process is trehalose, a disaccharide composed of two glucose molecules. In contrast to crystalline solids, the lack of a lattice pattern in trehalose’s amorphous nature allows it to be more viscous at room temperature^[55]. Because the rigidity of crystalline structures can damage embedded biologics^[52], the property of increased viscosity is highly beneficial in protective applications. Trehalose is commonly utilized as a protectant in conjunction with dehydration techniques because of its ability to restrict molecular motion^[55] via vitrification^[56]. Substances lose hydrogen bonds during the dehydration process due to the removal of water and trehalose is able to temporarily replace these hydrogen bonds^[55]. This mechanism successfully preserves an essential interaction needed to maintain the secondary structures of proteins.

In this work, we show that Lyo and LAD are both efficient alternatives to using SpeedVac to dehydrate NANPs. However, they each have their own respective benefits and drawbacks. Densitometry analysis is used to compare the methods to non-dried NANPs, which are shown to degrade at higher storage temperatures. LAD-processed NANP MD simulations were conducted in order to test the effects of trehalose, vacuum exposure, and rehydration on the molecular dynamics of the NANPs. The result suggest that trehalose introduces additional stability to the NANPs compared to being in vacuum and in water alone. Importantly, trehalose-stabilized NANPs retain their immunostimulatory properties as proven in both reporter cell lines and human peripheral blood mononuclear cells (PBMCs). The comprehensive experimental analysis indicates that optimizing the dehydration step of NANP preparation provides a great benefit to maintaining their overall structural integrity and functionality. The results show that LAD is more suitable for preservation of the DNA-based NANPs when compared to Lyo. When biological activity of is concerned, LAD processed RNA NANPs were more effective in accomplishing their purpose of gene silencing while Lyo ones are more predictably temperature dependent. Overall, our research presents information that would be valuable in deciding the most optimal dehydration method for producing fully functional NANPs and other TNAs which are able to overcome the difficulties of storage and transportation to preserve their original properties.

2. Results and Discussion

We report data on four representative NANPs (DNA cubes, RNA cubes, RNA rings, RNA fibers) dehydrated via SpeedVac, Lyo, and LAD. We stored the dehydrated NANPs and NANPs in solution for 24 hours at five different temperatures ranging from −20 °C to +50 °C and subsequently compared the retention of their structural integrities (by native-PAGE, AFM, and MD simulations) and biological activities (immunology assays, and RNA interference mediated gene silencing) after rehydration (Figure 1).

Assessment of NANPs’ structural integrity upon rehydration.

We first examined the densitometry results acquired from native-PAGE experiments carried out for NANPs dried either via SpeedVac (Figure 2A) of Lyo (Figure 2B). For DNA cubes (Figure 2C), the decrease in band intensity at +37°C and +50°C for samples stored in solution is consistent with the DNA cube’s T_m of ~36.6°C (supporting Table S1). The control stored at +50°C was significantly diminished (p<0.0001) relative to the cold or RT storage conditions. All DNA cubes which were dried via SpeedVac showed a marked decrease in band intensity; the −20 °C, +4 °C, and RT controls had significantly higher band densities than all SpeedVac-dried DNA cubes at any storage temperature (p<0.01, p<0.001, and p<0.001, respectively). However, within the groups of DNA cubes dried via Lyo, there was no significant influence of the storage temperature. The AFM images (Figure 3) indicated the formation of larger structures for both the SpeedVac and Lyo DNA cubes.

Figure 2.

Comparison of SpeedVac and lyo. (A-B) Experimental setup of the SpeedVac and lyo sample preparations. (C) Native-PAGE analysis of NANPs’ integrity after their storage at −20°C, +4°C, RT, +37°C, and +50°C for 24 hours. The intensities of each band from three independent biological repeats were calculated and plotted. Data was normalized to the intensity of the respective control sample (NANP in buffer solution) stored at +4°C. Error bars shown as mean ± SD for n=3 repeats. The significance is denoted in supporting Table S2.

Figure 3.

AFM images of DNA Cubes, RNA Cubes, RNA rings, and RNA fibers. Samples were prepared in solution and processed via SpeedVac, lyo, or LAD, and then stored at room temperature for 24 hours, rehydrated, and analyzed by AFM.

The aggregates are more prominent in lyo than SpeedVac samples. This is also indicated by some aggregation of samples in the wells of the native-PAGE gels (Figure 2C). This aggregation exists to some degree in the majority of the wells, but is most prominent in cubes that have been dried, whether it was via SpeedVac or Lyo. These higher order assemblies were likely formed because the NANPs were subject to prolonged heat exposure. The constant heat at +55 °C, a temperature well above the T_m of DNA cubes, likely caused the denaturation of the strands and partial or full deconstruction of the three-dimensional cube structure. After heat is reduced when the sample cools to ambient temperatures, the broken structures are likely to reform bonds with other nearby structures to form a larger aggregation consisting of parts of multiple cubes. This correlates well with previous research showing that the increase in size in direct correlation with freezing rate^[75]. This factor is important to consider as these methods are utilized in scale-up manufacturing of NANPs.

In the case of the RNA cubes (Figure 2C), Lyo improves structural retention compared to SpeedVac usage for samples stored at all temperatures. Although the densitometry results for the RNA cubes show high variability in Lyo band intensity, both the Lyo and control samples have some significance (p<0.05, Table S2) relative to the SpeedVac samples. Additionally, the distinct patterns on the gels indicate a notable difference in structural retention. The gels show little to no aggregation of RNA cubes in the wells of both the control samples and the Lyo samples. On the contrary, aggregation exists in the wells of all SpeedVac samples. This data is in good agreement with the AFM results (Figure 3) which show how the SpeedVac RNA cubes appear to have formed higher order assemblies while Lyo RNA cubes are comparable to the control sample. This aggregation has appeared for both DNA and RNA cubes, but was non-existent in the both the gels and AFMs for RNA rings and fibers. These entanglements may have been more likely due to the increased dimensionality in the cubes (3D) compared to the rings (2D) and fibers (1D) (supporting Table S1).

For RNA rings, the structural integrity was largely diminished when the control was stored at +50°C, with a decrease in band intensity which was largely significant compared to the −20°C (p<0.001), +4°C, RT, or +37°C (p<0.0001) in-solution storage conditions (Figure 2C). At +50°C, the control was also significantly reduced compared to the Lyo RNA ring kept at any storage condition (p<0.0001). Remarkably, the Lyo RNA rings stored at +50°C had comparable band intensity to the controls stored within the −20 to +37°C range. SpeedVac drying greatly reduced the structural integrity of the ring, with all storage temperatures not significantly differing from the in-solution 50 °C-stored ring. The AFM images (Figure 3) of the SpeedVac-dehydrated rings contains a large number of broken rings presented as open structures. Most of the Lyo rings are comparable to the ones stored in solution and exhibit the round donut-like structure expected of a stable ring. The differences in the dehydration methods are more significant for the samples stored at +50°C. Although samples stored in solution usually provide better yield than SpeedVac-dehydrated samples, the one stored at +50°C was an exception to this trend. The percent intensity of bands for the same dehydration method does not vary significantly across all temperatures while the control samples experience a significant downward trend from RT to +50°C. This indicates that dehydration, whether it be via Lyo or SpeedVac, provides an additional measure of protection against temperature changes. Overall, the structural retention of Lyo RNA rings at every storage temperature was significantly higher than all SpeedVac-processed samples at same conditions.

For all RNA fiber samples, there are no significant differences between the methods of dehydration or storage temperature. The exception is the set of samples stored at +50°C where there is visibly less RNA fibers stored in solution. The band appears to have traveled further down the lane than the rest of the bands on the gel. The AFM images of RNA fibers (Figure 3) do not depict any significant differences between the dehydration methods, correlating well with the native-PAGE data for the samples stored at room temperatures. This is due to the tendency of fibers to assemble in various lengths as made apparent in the AFM images; disassembly and any subsequent aggregation is not physically discernable for this type of NANP structure.

As a third option for sample storage, LAD^[57–60] was used to process the NANPs (Figure 4A), followed by 24 hours of temperature storage. Each NANP was dried at a time as their thermal profiles were monitored (Figure 4B and supporting Figure S3) and as before, all samples were assessed by native-PAGE and AFM. Prior to LAD, NANPs were prepared with trehalose drying solution or were stored with the trehalose drying solution as the controls. For all NANPs, there were no statistically significant differences between the in-solution controls versus LAD-processed samples at any storage temperature, nor were there any differences between storage temperatures within the LAD or control sets. The trend of the +50°C temperature storage showing lower band intensity was maintained for the RNA cube and ring, although there was no significant effect of the storage temperature when the NANPs were stored in the trehalose drying condition. While there may be some effect of the trehalose alone on structural preservation, there was no significant difference found at any temperature between the control with trehalose drying condition and the control without.

Figure 4.

(A) Experimental setup of light-assisted drying (LAD). Components are contained within a controlled low relative humidity chamber. 10 μL of sample is illuminated with a near-IR laser. The temperature of the sample is monitored during processing using the thermal FLIR Camera. (B) Thermal profile of LAD processed NANPs. Images were retrieved from the thermal FLIR camera and presented here at 5 and 30 minutes. Temperatures represented are in degrees Celsius, reaching a maximum of 36.6 ± 0.7 °C. Additional images for the rest of the thirty-minute duration and for the other NANPs are shown in supporting Figure S3. (C) Gel band intensity of NANPs. Samples were stored at −20°C, +4°C, RT, +37°C, and +50°C for 24 hours in solution and after LAD processing. Data was normalized to the intensity of the respective control sample (NANP in buffer solution) stored at +4°C. Error bars shown as mean ± SD for n=3.

In silico studies of NANPs’ rehydration.

We performed 100 ns MD simulations for the DNA cube (Figure 5A), RNA cube (Figure 5B), and RNA ring (Figure 5C) structures in water (Figure 5D), vacuum (Figure 5E), and trehalose (Figure 5F), respectively. It was found that all three NANPs are stable in water since their RMSD only varies in the range of 0–2 Å. A period of oscillation of the RMSD of the DNA cube from 60 ns to 70 ns (Figure 5A) was also observed, implying that RNA rings and RNA cubes are more stable than the DNA cubes in water. When the dynamics of the three structures in trehalose were simulated, it was found that they are even more stable than in water. The enhanced stability is attributable to the highly rigid structure of the trehalose matrix. However, if the three NANPs were simulated in vacuum, their 3D structures are highly unstable. It was observed that the RMSDs of all three NANPs increase from 0 to 15–20 Å in 10 ns. We found that many base pairs have been broken (supporting Figure S4), and this may be the reason why the functions of these NANPs cannot be restored because the original base pairing interaction cannot be easily formed due to the complicated base pair formation mechanisms in RNA^[61]. To further confirm it, we have added water molecules to the structures after the vacuum simulation and performed another 100 ns simulations. We have found that RMSD has increased even higher (supporting Figure S5), and the broken base pairs cannot be restored (supporting Figure S4), indicating that the NANP structures have undergone irreversible deformation in the vacuum simulations.

Figure 5.

Molecular dynamics simulations of RNA ring, RNA cube, and DNA cube in water, vacuum, and trehalose matrix. RMSD of the DNA cube (A), RNA cube (B), and RNA ring (C) are calculated from their 100 ns MD simulation in water (D), vacuum (E), and trehalose matrix (F), respectively. The three models inside (A-C) are the final structures of RNA ring, RNA cube, and DNA cube in vacuum.

Assessment of NANPs’ immunorecognition after rehydration.

To assess the biological activity of NANPs after drying and storage, the panel was introduced to human immune reporter cell lines (Figure 6).

Figure 6.

Immune stimulation of model cell lines by rehydrated DNA cubes, RNA cubes, RNA rings, and RNA fibers delivered using Lipofectamine 2000 (L2K). NANPs were dried via SpeedVac, lyo, or LAD and stored along with NANPs in solution and NANPs with trehalose in solution at −20, 4, RT, 37, or 50 °C for 24 hours. Dried NANPs were then rehydrated and transfected into immune reporter cell lines at 10 nM for analysis via SEAP or Luciferase assays after 24 hours. Schematics on the left panel explain the abridged activation of each PRR. Data set was obtained using the average of three biological repeats for each condition. Within each cell line, all samples are normalized to the cells-only treatment and scaled to each cell line’s positive control (PC) as a maximum value. As additional controls, trehalose and/or L2K and 1X assembly buffer (1XAB) are shown. The same data presented in bar graph format are available in Supporting Figure S6.

The immunorecognition of NANPs as driven by interactions with pattern recognition receptors (PRRs) has been shown to be heavily correlated with their structures which, if maintained, should also yield a comparable level of induction^[26,62,63]. Key parameters have been shown to include dimensionality, with increasing dimensionality from fiber (1D) to ring (2D) to cube (3D) showing higher immune stimulation, and composition, with increased RNA content such as between RNA cube and DNA cube resulting in higher immune stimulation. In order to cross the cell membrane, a transfection agent, Lipofectamine 2000 (L2K), was employed for NANPs of all conditions. Immediately after rehydration, the NANPs were mixed with L2K, transfected into immune reporter cell lines, and their resulting immunorecognition was assessed after 24 hours (Figure 6 and supporting Figure S6). Interestingly, an increase in immunostimulation of NANPs in the trehalose-containing formulations was observed for both the LAD-processed and trehalose control samples at all storage temperatures. The addition of trehalose enhanced the activation of both the IRF and NF-κB pathways (as seen in the THP1-Dual™ cell line), especially as demonstrated by RIG-I and TLR8 reporter cell lines. In tested reporter cell lines, trehalose alone or transfected using L2K shows very little immunostimulation. Between other non-trehalose drying methods, most trends in immunostimulation are maintained between the various storage temperatures and drying methods. RNA cubes remain the most potent activators of immune stimulation in general and of TLR7 in particular. Higher storage temperatures of 50 °C show lower levels of activation for most NANPs in agreement with native-PAGE analysis, although this is not always the case when trehalose is introduced. In addition, despite the structural retention demonstrated by lyo NANPs in the native-PAGE, the lyo conditions of each NANP show overall lower immunostimulation compared to the control or even SpeedVac-dried samples.

To further assess a potential influence of sample preservation procedure on immunological properties of NANPs, and to interrogate the role of trehalose in modulating cytokine responses to NANPs, we conducted an additional set of experiments using freshly isolated human PMBCs. Consistent with our earlier studies, RNA cubes were more potent immunostimulants than RNA ring, and L2K alone induced a chemokine response^[26,27,64]. Since RNA fibers and DNA cubes were previously shown to be non-immunostimulatory^[26], these NANPs were excluded from current experiments. The presence of trehalose in RNA rings and RNA cubes did not result in physiologically significant (i.e., 2-fold) changes in the cytokine responses (Figure 7 and supporting Figure S7). The results were consistent between cultures obtained from three donors except for Donor S8B3, whose cells produced twice more IFNγ and IL-1β in response to both RNA rings and RNA cubes containing trehalose as compared to the same NANPs without trehalose. Cultures of this same donor also demonstrated twice lower levels of chemokines (MCP-1, MCP-2, MIP-1α and MIP-2β) in response to the trehalose-containing NANPs as compared to trehalose-free samples. Although an increase in chemokines (MCP-2, MIP1α, and MIP1β) and interferons (IFNβ, IFNω and IFNλ) was noted in the RNA cubes plus trehalose-treated cultures of Donor S8B3, these changes were not physiologically relevant (i.e., less than 2-fold) and are within the assay variability. Since NANP-mediated cytokines showing differential expression in the presence of trehalose are either produced or influenced by lymphocytes (the cells that proliferate), the observed changes could be attributed to the improved viability of these cells in the presence of trehalose and, therefore, could be limited to in vitro settings.

Figure 7.

Immune stimulation of PBMCs treated with RNA cubes and RNA rings with and without trehalose. (A) Schematic explaining the experimental procedure: blood from three healthy volunteer donors was centrifuged down for the isolation of PBMCs, which were then plated and treated with RNA cubes and RNA rings with and without trehalose. A multiplex ELISA was used to determine the activation of various cytokines. (B) Results from a 15-plex and 4-plex ELISA are shown for each of the three donors. The same data presented in bar graph format are available in supporting Figures S7 and S7.

Lymphocytes present in PBMC cultures from individual donors have different basal level of proliferation which is influenced by a variety of factors such as diet, stress level, vaccination and sub-symptomatic infections^[65], all of which are out of control of our research donor program. The fact that we observed changes in cultures from only one donor suggests that the basal level of lymphocyte proliferation in these cultures were higher than that in PBMCs from the other two donors. Earlier studies demonstrated the use of trehalose for improving survival of transfected cells^[66–67] and reducing cytotoxic effects of some treatments and cell handling procedures^[68–69]. It is not unexpected that being a sugar, trehalose provides an extra energy for proliferating cells, thereby contributing to their biological responses. The changes observed with reporter cell lines, that are also immortalized, proliferating cells, further corroborate this explanation. Therefore, in order to verify if trehalose has immunomodulatory effects on NANPs’ mediated cytokine response or the observed in our study changes are limited to in vitro settings, in vivo studies, and, most importantly, clinical data are necessary. There are other examples in the literature demonstrating that formulation excipients are often non-immunologically inert ^[70–71]. Therefore, our findings warrant further investigation.

Assessment of NANPs’ functional activity after rehydration.

As a last test of retained biological activity, RNA cubes were assembled with six functional Dicer Substrate (DS) RNAs per structure for the silencing of green fluorescent protein (GFP) by RNA interference, as previously demonstrated^[65,71–73]. Again, the DS RNA-functionalized RNA cubes were dried using the three chosen methods and stored for 24 hours at the five storage temperatures for comparison. Upon rehydration and conjugation to L2K, NANPs were transfected into human breast cancer cells (MDA-MB-231) expressing GFP to assess their silencing activity after 72 hours (Figure 8). As seen in the microscopy images, silencing was observed for all drying methods at the 4 °C storage temperature relative to the untreated cells. However, the degree of silencing varied, with LAD-processed samples achieving more silencing than the control cubes or the trehalose control and SpeedVac cubes silencing comparably. The lyophilized cube showed the least silencing activity, although showing the most trend in temperature dependence. Though RNA cubes demonstrated variability in their structural integrity via densitometry, the retained silencing activity may be due to the retention of the functional DS RNAs themselves, which are able to withstand all storage temperatures in solution (supporting Figure S8).

Figure 8.

Retained silencing activity of dried and rehydrated DS RNA-functionalized RNA cubes. RNA cubes functionalized with DS RNAs against GFP were dried via SpeedVac, lyophilized, or processed with LAD and stored along with controls at −20, 4, RT, 37, or 50 °C for 24 hours. Dried NANPs were then rehydrated and all NANPs were conjugated to L2K and transfected into MDA-MB-231 cells expressing GFP at 1 nM. After 72 hours, GFP silencing was assessed by fluorescence microscopy and flow cytometry. Schematic shows the NANP’s processing via RNA interference. Data set was obtained using the average of three biological repeats for each condition. All samples are normalized to the cells-only treatment and are quantified by the normalized mean fluorescence intensity, with GFP-expressing cells equal to 1.0. Representative images of each drying condition at 4 °C are shown as 200 × 200 μm.

Conclusion

We provide additional insight on a traditional alternative (Lyo) and a novel dehydration method (LAD) and their efficiency in processing NANPs. The goal of this study was to develop and test user-friendly and affordable technology for preserving NANPs in a dry state to replace current refrigeration techniques. Successfully optimizing the dehydration process of NANPs will allow for improved stability in storage and transportation. Although we tested various protocols developed for NANPs, our findings may also be beneficial when storing other nucleic acid samples or other nanoparticles of similar characteristics. LAD has the advantage of allowing for selective heating via the fine-tuning of temperature settings, leading to distribution uniformity and more predictable results. On the other hand, Lyo allows for successful drying using room temperature sublimation instead of heating. Lyo also benefits from higher throughput capabilities, since LAD is only currently optimized to process a single sample at a time, a factor that would present a major hurdle for commercial production. While lyo is shown to be the best method for preserving overall structural stability, LAD is able to maintain sufficient biological functionality.

Future research could aim to improve the outcome of the Lyo process to minimize degradation. Alternatives to freezing the product in the laboratory freezer include utilizing a shell bath or direct immersion in liquid nitrogen^[74–76]. Trehalose or other cryoprotectants could also be explored as accessories to improve the lyophilization process and reduce structural damage ^{[56,68,69, 77,78]}. The potential immunomodulatory effect of trehalose on the elicited immune response by NANPs detected in our study is of interest to the field because in addition to the previously described means for controlling NANPs’-mediated immunostimulation^[79–80], it may serve as an additional tool for tailoring immunostimulatory NANPs for favorable applications. However, clinical data that are not yet available are needed to distinguish between the true immunomodulatory effect and in vitro effects arising from the use of trehalose.

Overall, we have shown that both lyo and LAD are more beneficial alternatives to the traditional SpeedVac. With the information obtained in this study, researchers utilizing NANPs would be able to make more informed decisions when it comes to choosing an efficient dehydration method to achieve their intended purposes. Further refinement of the dehydration methods tested will improve the stability of NANPs for a prolonged period and increase their viability for use in biochemical applications and therapeutics.

Materials and Methods

Nucleic Acid Nanoparticle Structure Synthesis and Verification.

All sequences utilized are provided in the Supporting Information. The DNA templates and RNA strands used in RNA NANPs were ordered from Integrated DNA Technologies. Once received, the templates were amplified via PCR with MyTaq™ Mix (Bioline). PCR products were then purified with the DNA Clean and Concentrator™ kit (Zymo Research). RNAs were produced by in vitro run-off transcription using T7 RNA Polymerase in 80 mM HEPES-KOH (pH 7.5), 2.5 mM spermidine, 50 mM DTT, 25 mM MgCl₂, and 5 mM of each rNTP. To stop the transcription, samples were incubated with RQ1 RNase-free DNase (Promega) and then purified using denaturing 8 M urea polyacrylamide gel electrophoresis (PAGE, 15%). After visualizing the bands under UV (short wavelength), they were excised and eluted overnight in a crush and soak buffer (300 mM NaCl, 89 mM tris-borate (pH~8.1–8.2), 2 mM EDTA). For precipitation of the RNAs, the elution was first mixed with 2.5 volumes of 100% EtOH and stored at −20°C. Then, it was rinsed with 90% ethanol, vacuum dried, and dissolved in double-deionized water (17.8 MΩ·cm). To assemble NANPs, the concentrations of all strands were measured using a NanoDrop 2000 (Thermo Scientific). Each NANP was assembled via the combination of each strand in an equimolar ratio along with double-deionized water and assembly buffer a in one-pot thermal anneal. Both the DNA and RNA cubes were heated to 95°C for 2 minutes, mixed with assembly buffer (89 mM tris-borate (pH 8.2), 2 mM MgCl₂, 50 mM KCl), and incubated at 45°C for 30 minutes. The RNA rings and RNA fibers were heated to 95°C for 2 minutes, snap-cooled at 4 °C for 2 minutes, mixed with the previously described assembly buffer, and incubated at 30 °C for 30 minutes before transferring back to 4 °C. All assemblies were confirmed through visualization on 8% (37.5:1) non-denaturing (native-PAGE) polyacrylamide gel with running buffer containing 89 mM tris-borate (pH 8.2) and 2 mM MgCl₂. The gels were run in a Mini-PROTEAN Tetra system (Bio-Rad) for approximately 25 minutes at 300 V in a 4 °C cold room. Afterwards, the gels were removed and stained with ethidium bromide (0.5 μg/mL) for 5 minutes. Gel imaging was completed using a ChemiDoc MP system (Bio-Rad). This procedure was utilized to visualize the NANPs both immediately after the initial assembly and after dehydration and subsequent rehydration. For samples which showed multi-banding or impurities, a second round of purification was applied. For purification, this process was repeated on larger native-PAGE settings of the same composition which ran for approximately one hour at 300 V in the cold room. Samples were then visualized under UV, cut, and eluted in assembly buffer overnight.

Band intensities after dehydration and subsequent rehydration were quantified using ImageLab software (Bio-Rad). For each SpeedVac or lyo processed NANP, the band was normalized to the 4 °C non-dried control on the same gel. For LAD-processed NANPs, the bands were normalized to the 4 °C non-dried control with trehalose drying solution on the same gel. The mean of n=3 bands per NANP condition were plotted with their standard deviation and analyzed with a two-way ANOVA followed by Tukey’s multiple comparisons test using GraphPad Prism version 9.0.0 for Windows. A p-value of less than 0.05 was considered statistically significant.

SpeedVac Drying.

NANP samples (0.5 μM) stored in the previously mentioned standard assembly buffer were aliquoted (10 μL per tube) and centrifuged in a balanced CentriVap micro–IR Vacuum Concentrator (Labconco) with both heat (55 °C) and infrared radiation (IR) for 20 minutes (Figure 2A). Experiments testing the addition of a trehalose drying solution (0.2 M trehalose, 0.33X PBS) to SpeedVac samples are included in supporting Figure S2.

Lyophilization (lyo).

NANP samples (10 μL, 0.5 μM) stored in standard assembly buffer were frozen at −20°C in a laboratory freezer overnight. The samples were then transferred to a high-density polyethylene (HDPE) rack and lyophilized in a Dura-Dry MP freeze dryer (FTS Systems) (Figure 2B). Samples were held at a shelf temperature of 20°C, a −85°C condenser and an 80 mtorr vacuum for two hours. The samples were then equilibrated to ambient conditions with atmospheric air and sealed. Experiments testing the addition of the drying solution to lyo samples are shown in Figure S2.

Light-Assisted Drying (LAD).

The LAD process utilizes a near-infrared laser light to project heat onto a droplet of the sample^{[57–59] (}Figure 3A). All experiments were conducted in a humidity-controlled chamber that was kept at approximately 11% relative humidity. This was achieved by pumping dry air into a chamber containing the experimental setup and monitoring the RH with a temperature and RH logger (ONSET UX100–011). The sample was prepared with the addition of a drying solution, which includes a disaccharide trehalose (0.2 M) in phosphate-buffered saline (PBS) (0.33X) to reach a final concentration of 0.25 μM. For each test, each droplet (10 μL) was deposited onto an 18 mm diameter borosilicate glass coverslip (Fisherbrand 12–546) substrate. The sample was then moved into the humidity chamber for laser irradiation. The glass coverslip allows for easy recovery and rehydration of the NANPs after LAD processing. Inside the chamber, the droplet was heated with a continuous wave (CW) ytterbium fiber laser (IPG Photonics) at 1064 nm (YLR-5–1064). The laser has a factory collimated Gaussian beam with a FWHM spot size of ~4.5 mm which was measured using a BeamTrack 10A-PPS thermal sensor (Ophir Photonics). An FLIR SC655 mid-IR camera recorded the temperature of the samples during the drying process. The temperatures reached during processing for the DNA cubes, RNA cubes, RNA rings, and RNA fibers were 34.1 ± 0.6°C, 36.6 ± 0.7°C, 37.2 ± 1.1°C, and 36.6 ± 1.0°C, respectively. RNA cubes (n=3), RNA rings (n=3), and RNA fibers (n=3) were processed for 30 minutes at 5 W (26.9 W/cm²) and DNA cubes (n=3) were processed for 30 minutes at 4 W (21.5 W/cm²). The maximum wattage of 5 W was appropriate for RNA cubes, RNA rings, and RNA fibers as they have high T_ms (supporting Table S1) which would lead to the sample only being heated to ~35 °C. To accommodate for the lower T_m of DNA cubes (~36.6 °C), the wattage setting was reduced to 4 W to reach a max heating temperature of 33.7 °C.

Storage.

Once processed, the SpeedVac and lyo samples were stored in the same eppendorf tubes (lids closed and sealed with parafilm) along with non-dehydrated control samples. All samples were stored at five different temperatures including in a laboratory refrigerator (−20°C), cold room (+4°C), cabinet drawer (room temperature, RT), and two heat blocks (+37°C and +50°C) (VWR) for ~ 24 hours. The LAD-processed samples were processed on glass coverslips which were placed into sealed plastic containers for storage. The samples were subject to the same temperature conditions as previously mentioned. The coverslips stored at +37°C and +50°C were placed directly on top of sterilized (70% EtOH) heat blocks using sterilized tweezers.

Rehydration.

After 24 hours, the samples stored in Eppendorf tubes were rehydrated with water (10 μL), vortexed, centrifuged, and placed on ice (4°C) until needed for analysis. The samples stored on a glass coverslip were rehydrated by first placing the water (10 μL) onto the dried sample, mixing it with the pipette tip, and waiting (2–3 minutes) before transferring the rehydrated solution into an Eppendorf tube. The tubes were left on ice (4 °C) until needed for analysis.

Atomic Force Microscopy (AFM) Imaging.

Atomic force microscopy was carried out on a freshly cleaved mica surface submerged in an aqueous solution of APS (10(3-aminopropyl)-silatrane) as previously described^[81–83]. AFM imaging was performed with a MultiMode AFM Nanoscope IV system (Bruker Instruments) in tapping mode. Images were processed by FemtoScan Online software package (Advanced Technologies Center, Moscow, Russia)^[84–85].

Nuclease Digestion Assays.

To determine the relative chemical stability of NANPs and their resistance to nuclease degradation, 1μL of enzyme (either RQ1 RNase-free DNase (Promega) or RNase I (NEB)) was added to 20μL of NANP solution and incubated at 37°C. After 1, 15, and 30 mins of incubation, aliquots were taken and the reaction was stopped by snap cooling samples in liquid nitrogen. The samples were loaded in reverse order and analyzed on native-PAGE, as described above.

Stability of NANPs at Different pH.

To determine the relative stability of NANPs at different pH, the pH values of NANP solutions were adjusted to 1.3 and 12.9 by adding 5M HCl and 5M NaOH, respectively. The pH values of 5.1 and 10.9 were established by adding a 1M acetic acid buffer and 1M ammonium acetate buffer to NANP solutions, respectively. To confirm the pH values, all solutions were scaled up in volume and tested with a pH meter at 37°C. NANPs were incubated at 37°C for 30 minutes at different pH and then analyzed by native-PAGE.

Immune reporter cell experiments.

THP1-Dual™, HEK-Blue™ hTLR3, HEK-Blue™ hTLR7, HEK-Blue™ hTLR8, and HEK-Lucia™ RIG-I cells were all purchased from InvivoGen and maintained according to the supplier’s guidelines in an incubator at 37 °C and 5% CO₂. For all experiments, the cells were seeded at ~40,000 cells per well in a flat-bottomed 96-well plate 24 hours before transfection. For a final concentration of 10 nM NANPs per well, the NANPs were rehydrated after 24 hours of storage and immediately incubated with Lipofectamine 2000 (0.375 μL per well) at room temperature for 30 minutes before being added directly to the seeded cells. Positive controls for each cell line were used as follows: 10 ng/mL 3p-hpRNA with 0.375 μL L2K per well (following 30 minutes of incubation) for THP1-Dual™ (IRF) and HEK-Lucia™ RIG-I; 2 μg/mL R848 for THP1-Dual™ (NF-κB), HEK-Blue™ hTLR7, and HEK-Blue™ hTLR8; 2 μg/mL Poly I:C for HEK-Blue™ hTLR3. L2K (0.375 μL per well), trehalose drying solution, trehalose drying solution with L2K, and assembly buffer were used as additional controls. 24 hours post-transfection, the reporter cells were assessed using QUANTI-Blue™ (for THP1-Dual™ and all HEK-Blue™ cell lines) or QUANTI-Luc™ (for THP1-Dual™ and HEK-Lucia™ cell lines) from InvivoGen. For QUANTI-Blue™ assays, 180 μL of the reagent was added with 20 μL of cell supernatant in a flat-bottomed 96-well plate. After three hours of incubation at 37 °C, the plate was read on a Tecan Spark plate reader at an absorbance of 638 nm. All well values were the averages of sixteen-point reads. For QUANTI-Luc™ assays, 50 μL of the reagent was added with 20 μL of cell supernatant in a flat-bottomed black-walled 96-well plate. The plate was read immediately on a Tecan Spark plate reader for luminescence with a 100 ms reading time. Within each plate, all samples were assessed in technical duplicate, averaged, and normalized to the cells-only treatment for assessment of normalized fold induction. Three biological replicates were completed and averaged to evaluate the mean and standard deviation (Figure S6). Heatmaps were prepared using GraphPad Prism version 9.0.0 for Windows. Within each cell line, the scale was adjusted for the positive control to serve as the maximum value.

Analysis of cytokine responses in human PBMC model.

Whole blood was donated by healthy human donor volunteers under NCI-at-Frederick Protocol OH9-C-N046. The blood was collected into vacutainers containing Li-heparin as anticoagulant. PBMC were isolated within two hours of blood collection using Ficoll-Paque density gradient centrifugation and resuspended in complete RPMI medium (RPMI 1640 with 10% FBS, 2 mM L-glutamine, and penicillin/streptomycin); next 1.25 × 10⁶ of live cells/mL were seeded in 96-well U-bottomed plates in a final volume of 160 μL per well. RNA cubes and RNA rings with and without trehalose at particle concentration of 1 μM were complexed with Lipofectamine 2000 (L2K) at a 5:1 v/v ratio. After 30 min incubation at room temperature, complexed NANPs were diluted to 50 nM in OptiMEM; 40 μL of these solutions were added per well on culture plates containing PBMCs. The final concentration of NANPs in PBMC cultures was 10 nM. PBS was used as assay negative control. A mixture of ODN2216, PHA-M and LPS were used as a positive control. Additional controls included L2K and L2K-trehalose. After an overnight incubation at 37 °C, supernatants were collected and analyzed for the presence of cytokines using multiplex chemiluminescent ELISA (Quansys BioSciences, Logan, UT). Two multiplex assays were utilized: 1) a 15-plex panel assessing type II interferon (IFNγ), cytokines (IL-1α, IL-1β, IL-6, IL-10, IL-12, IL-21, IP-10, TNFα), and chemokines (IL-8, MCP-1, MCP-2, MIP1α, MIP-1β, RANTES), and 2) a 4-plex panel assessing type I (IFNα, IFNβ, IFNω) and type III (IFNλ) interferons. This process is depicted in Figure 7A. More details about these experimental procedures are available in earlier reports^{[25,26,79,80,86]}.

Cell silencing experiments.

Cells of the human breast cancer cell line MDA-MB-231 expressing green fluorescence protein (eGFP) were maintained in Dulbecco’s Modified Eagle Medium (MEM) with 2 mM L-glutamine (Gibco), 10% heat-inactivated fetal bovine serum (Atlanta Biologicals), and PenStrep (100 U/mL, 100 μg/mL) (Gibco) in an incubator at 37 °C and 5% CO₂. Cells were seeded at 40,000 cells per well in a 12-well plate 24 hours before transfection. For a final concentration of 1 nM NANPs per well, the NANPs were rehydrated after 24 hours of storage and immediately incubated with L2K (0.375 μL per well) at room temperature for 30 minutes before being brought up in Opti-MEM® (Gibco). Cell media was aspirated and replaced with the treatment in Opti-MEM® for 4 hours in the incubator. Afterwards, the treatment was aspirated and replaced with fresh media for assessment 72 hours post-transfection. An EVOS® FL Cell Imaging System (Thermo Fisher Scientific) was used to image cells with a 20× objective. Brightfield images were collected at 45% while GFP images were collected using an EVOS™ Light Cube, GFP, at 50%. Cells were washed with PBS and detached using trypsin-EDTA (0.25%) (Thermo Fisher Scientific), then resuspended in media and centrifuged at 500 g for 5 minutes. Afterwards, the cell pellet was dispersed in 10% neutral-buffered formalin (VWR) for 15 minutes at room temperature before a second centrifugation at 500 g for 5 minutes. All formalin was aspirated and the resulting cell pellet was resuspended in PBS for analysis via flow cytometry. A BD Accuri C6 flow cytometer equipped with a blue (488 nm) laser and 533/30 optical filters was used to quantitatively assess fluorescence. The cells-only treatment was used to gate the collection of 10,000 events per sample. Three biological replicates were completed and averaged to evaluate the mean. Heatmaps were prepared using GraphPad Prism version 9.0.0 for Windows.

Molecular Dynamics Simulations.

We performed three groups of molecular dynamics simulations. In group 1, the RNA ring, RNA cube, and DNA cube were all solvated with water; in group 2, the three NANP structures were simulated in vacuum; in group 3, the three structures were put inside a trehalose matrix for simulation. The trehalose monomer structure was downloaded from the Cambridge Crystallographic Data Centre (CCDC Number: 668079). Packmol^[87] was used to build up the amorphous trehalose matrix structure. RNA and DNA structures in group 3 were put inside the trehalose matrix, and then trehalose monomers that were sterically overlapped with the RNA/DNA structures were detected and deleted by using PyMOL ^[88]. Initially, structures in all three groups were minimized using steepest descent and conjugate gradient algorithms. Next, structures in group 1 were solvated with explicit water (TIP3P model) in triclinic periodic box, followed by adding neutral ions. Bare neutral ions are also added for structures in group 2. Subsequently, structures of all groups were then subjected to extensive energy minimization. Group 1 and group 3 structures were then subject to NVT and NPT equilibration to adjust the solvent density (“solvent” of group 3 is the trehalose matrix). A cut-off of 1.0 nm for both van der Waals (vdW) and electrostatic interactions and particle mesh Ewald sum with real space cut-off at 1.0 nm was used. LINCS algorithm is used to constrain all bonds involving hydrogen atoms. Group 1 and group 3 structures were equilibrated for 1 ns with a time step of 2 fs. Finally, 100 ns NPT production runs were performed for group 1 and group 3, while NVT production runs are performed for group 2. NPT production runs cannot be performed for group 2 because there is no solvent for temperature and pressure coupling control. All the simulations are performed using Gromacs 2021 simulation software^[89] with Amber ff19SB force field at 37 °C^[90]. Gromacs trajectory analysis tools and in-house scripts were utilized to extract the information from simulation data. The base pairs in the structures were analyzed by MC-Annotate^[91]

Statistical analysis and graphic preparation.

For all experimental analyses, NANPs were assembled, dehydrated, and rehydrated for at least three independent biological replicates. All experimental results were graphed and analyzed using GraphPad Prism version 9.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com. All figures were prepared using Adobe Illustrator CC 2020 v24.0.2.